CANCER RISK ASSESSMENT
CANCER RISK ASSESSMENT Chemical Carcinogenesis, Hazard Evaluation, and Risk Quantification
Edited by CHING-HUNG HSU TODD STEDEFORD
A JOHN WILEY & SONS, INC., PUBLICATION
About the cover: The cover structures are chemicals classified as known human carcinogens in the U.S. National Toxicology Program’s Annual Report on Carcinogens (http://www.ntp.niehs.nih.gov/). The center structure is cyclosporin A (CASRN 59865-13-3). The outer structures going clockwise are benzidine (92-87-5), vinyl chloride (CASRN 75-01-4), tamoxifen (CASRN 10540-29-1), cyclophosphamide (CASRN 50-18-0), benzene (CASRN 71-43-2), and azathioprine (CASRN 446-86-6). These structures were prepared using ACD/ChemSketch (ACD/Labs Release: 11; Product Version: 11.01; http://www.acdlabs.com). Copyright © 2010 John Wiley & Sons, Inc. All rights reserved. Published by John Wiley & Sons, Inc., Hoboken, New Jersey Published simultaneously in Canada No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978) 750-4470, or on the web at www.copyright.com. Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, (201) 748-6011, fax (201) 748-6008, or online at http://www.wiley.com/go/permission. Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives or written sales materials. The advice and strategies contained herein may not be suitable for your situation. You should consult with a professional where appropriate. Neither the publisher nor author shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages. For general information on our other products and services or for technical support, please contact our Customer Care Department within the United States at (800) 762-2974, outside the United States at (317) 572-3993 or fax (317) 572-4002. Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic formats. For more information about Wiley products, visit our web site at www.wiley.com. Library of Congress Cataloging-in-Publication Data: Cancer risk assessment : chemical carcinogenesis, hazard evaluation, and risk quantification / [edited by] Ching-Hung Hsu, Todd Stedeford. p. ; cm. Includes bibliographical references and index. Summary : “With a weight-of-the-evidence approach, cancer risk assessment indentifies hazards, determines dose-response relationships, and assesses exposure to characterize the true risk. This book focuses on the quantitative methods for conducting chemical cancer risk assessments for solvents, metals, mixtures, and nanoparticles. It links these to the basic toxicology and biology of cancer, along with the impacts on regulatory guidelines and standards. By providing insightful perspective, Cancer Risk Assessment helps researchers develop a discriminate eye when it comes to interpreting data accurately and separating relevant information from erroneous”—Provided by publisher. ISBN 978-0-470-23822-6 (cloth) 1. Carcinogens. 2. Health risk assessment. I. Hsu, Ching-Hung. II. Stedeford, Todd. [DNLM: 1. Neoplasms–chemically induced. 2. Risk Assessment–methods. 3. Carcinogens– toxicity. 4. Environmental Exposure. 5. Mutagenicity Tests. QZ 202 C21556 2010] RC268.6.C357 2010 616.99′4071—dc22 2009049268 Printed in the United States of America 10
9
8
7
6
5
4
3
2
1
CONTENTS
PREFACE
xvii
CONTRIBUTORS
xix
ABBREVIATIONS AND ACRONYMS
xxiii
PART I
CANCER RISK ASSESSMENT, SCIENCE POLICY, AND REGULATORY FRAMEWORKS CHAPTER 1
1
CANCER RISK ASSESSMENT
3
Elizabeth L. Anderson, Kimberly Lowe, and Paul Turnham 1.1.
1.2.
1.3.
1.4.
Cancer Risk Assessment 3 1.1.1. Cancer in the United States 3 1.1.2. Historical Perspectives of Cancer Risk Assessment 4 1.1.3. The Defining Steps in Cancer Risk Assessment 9 1.1.4. The Mode of Action (MOA) 11 1.1.5. Accounting for Scientific Uncertainty 11 The Weight of Evidence (WOE) for Determining Carcinogenicity 12 1.2.1. Epidemiologic Studies 12 1.2.2. Animal Models 14 1.2.3. Weight of the Evidence Descriptors 15 Risk Assessment in the 21st Century 16 1.3.1. Using the Advances in Molecular and Computational Biology 1.3.2. Genetic Susceptibility 17 Applications in Risk Management 17 1.4.1. Translating Risk Assessment into Risk Management in the United States 17 1.4.2. International Risk Management 18 1.4.3. Risk–Benefit Analysis 19 1.4.4. Risk Acceptance and Risk Communication 20 References 21
CHAPTER 2
SCIENCE POLICY AND CANCER RISK ASSESSMENT
16
23
Gary E. Marchant 2.1. 2.2.
Introduction 23 Use of Risk Assessment in Regulatory Decision-Making
24
v
vi 2.3. 2.4. 2.5. 2.6. 2.7.
CONTENTS
Role Of Risk Assessment Guidelines 25 Data Quality Requirements 28 Types of Data Used in Risk Assessment 30 Application of “Conservative” Assumptions and Precaution Conclusion 34 References 34
33
HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY WITHIN A REGULATORY CONTEXT
CHAPTER 3
37
Henk Tennekes, Virginia A. Gretton, and Todd Stedeford 3.1. 3.2. 3.3.
3.4.
3.5.
Overview 37 Risk Assessment 37 3.2.1. Principles of Risk Assessment and Management 38 Regulatory Schemes for Industrial Chemicals and Biocides 42 3.3.1. The U.S. Toxic Substances Control Act (TSCA) 42 3.3.2. The EU Registration, Evaluation, Authorization, and Restriction of Chemicals (REACH) 44 3.3.3. Voluntary Initiatives for Evaluating Industrial Chemicals 45 3.3.4. The U.S. Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) 48 3.3.5. The EU Biocidal Products Directive (BPD) 49 Scientific Aspects of Carcinogenic Risk Assessment 50 3.4.1. Dose–Response Relationships in Carcinogenesis and Mechanisms of Carcinogenic Action 50 3.4.2. Mathematical Model for Carcinogenic Risk Assessment 60 Conclusions 61 References 62
USE OF CANCER RISK ASSESSMENTS IN DETERMINATION OF REGULATORY STANDARDS
CHAPTER 4
Robert A. Howd and Anna M. Fan 4.1. 4.2.
4.3.
4.4.
4.5.
4.6.
4.7.
Introduction 66 Air Standards 70 4.2.1. Scientific Issues 70 4.2.2. Regulatory Considerations 72 Water Standards 73 4.3.1. Scientific Issues 73 4.3.2. Regulatory Considerations 75 Food Standards, Pesticide Tolerances, Additives, and Impurities 4.4.1. Scientific Issues 76 4.4.2. Regulatory Considerations 77 Soil Standards 81 4.5.1. Scientific Issues 81 4.5.2. Regulatory Considerations 81 Consumer Product Standards 82 4.6.1. Scientific Issues 82 4.6.2. Regulatory Considerations 83 Recent Developments and Future Directions 84 References 87
76
66
CONTENTS
vii
PART II
CANCER BIOLOGY AND TOXICOLOGY CHAPTER 5
97
THE INTERPLAY OF CANCER AND BIOLOGY
99
James W. Holder 5.1.
5.2. 5.3.
5.4.
Historical Account of Some Important Events in Understanding Cancer 99 5.1.1. Early Cancer Biology History 99 5.1.2. Near-Recent Cancer Biology History 101 Recent Foundations of Biological Mechanisms of Cancer 103 Cell Biology of Cancer 105 5.3.1. In Vitro Systems 105 5.3.2. Programmed Cell Removal 107 5.3.3. Facilitation of Supporting Cells and Cell-to-Cell Communication 5.3.4. Clonal Aspects of Carcinogenesis 116 5.3.5. Biology of Inflammation and Cancer 124 5.3.6. Stem Cell Biology and Cancer 130 5.3.7. Specific Biological Growth and Growth Control Gene Sets and Their Pathways 139 5.3.8. Epigenetic Biology and Nuclear Traffic 142 5.3.9. Biological Initiation of Chemical Carcinogenesis 147 Some Final Thoughts on Biology and Cancer 152 References 155
CHEMICAL CARCINOGENESIS: A BRIEF HISTORY OF ITS CONCEPTS WITH A FOCUS ON POLYCYCLIC AROMATIC HYDROCARBONS
113
CHAPTER 6
168
Stephen Nesnow 6.1. 6.2. 6.3.
A Brief History of Chemical Carcinogenesis 168 James A. and Elizabeth C. Miller and Their Theory of Metabolic Activation 169 6.2.1. Metabolic Activation of PAH and Tumorigenesis 173 The Concepts of Initiation, Promotion, and Progression: The Origin of Multistage Carcinogenesis 182 References 185
CHAPTER 7
HORMESIS AND CANCER RISKS: ISSUES AND RESOLUTION
191
Paolo F. Ricci and Edward J. Calabrese 7.1. 7.2. 7.3. 7.4.
Introduction 191 Evidence for Regulatory Cancer Risk Assessment 194 Hormesis and Cancer Risk Assessment: Models 198 7.3.1. Answers to Our Question 201 Conclusions 203 References 204
THRESHOLDS FOR GENOTOXIC CARCINOGENS: EVIDENCE FROM MECHANISM-BASED CARCINOGENICITY STUDIES
CHAPTER 8
Shoji Fukushima, Min Wei, Anna Kakehashi, and Hideki Wanibuchi
207
viii 8.1. 8.2. 8.3. 8.4. 8.5. 8.6. 8.7.
CONTENTS
Overview 207 Introduction 208 Low-Dose Carcinogenicity of 2-Amino-3,8-Dimethylimidazo[4,5-f ]-Quinoxaline (MEIQX) in the Rat Liver 209 Low-Dose Hepatocarcinogenicity of N-Nitroso Compounds 215 Low-Dose Carcinogenicity of 2-Amino-1-methyl-6-phenylimidazo[5,6-b]pyridine (PHIP) in the Rat Colon 215 Low-Dose Carcinogenicity of Potassium Bromate, KBrO3 in the Rat Kidney 216 Conclusion 219 References 220
PART III
GENETIC TOXICOLOGY, TESTING GUIDELINES AND REGULATIONS, AND NOVEL ASSAYS
223
CHAPTER 9 DEVELOPMENT OF GENETIC TOXICOLOGY TESTING AND ITS INCORPORATION INTO REGULATORY HEALTH EFFECTS TEST REQUIREMENTS
225
Errol Zeiger 9.1. 9.2. 9.3. 9.4. 9.5. 9.6. 9.7.
Introduction 225 Definitions and Usage 226 The Historical Development of Genetic Toxicity Testing Types of Available Tests 228 Testing Approaches 229 Where Are We Now? 232 Summary 235 References 235
227
GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
CHAPTER 10
238
Lutz Müller and Hans-Jörg Martus 10.1. 10.2. 10.3.
10.4. 10.5. 10.6.
10.7.
Historical Overview of Genotoxicity Testing Guidelines 238 Organization for Economic cooperation and Development (OECD) Guidelines for Genotoxicity 243 International Conference of Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) Guidelines for Pharmaceuticals 243 International Workshop on Genotoxicity Tests (IWGT) 248 The International Program on Chemical Safety (IPCS) Under the Auspices of the World Health Organization (WHO) 249 In Vitro Testing 250 10.6.1. In Vitro Tests 250 10.6.2. Evaluation of In Vitro Testing Results 250 10.6.3. Follow-Up to In Vitro Testing 250 In Vivo Testing 251 10.7.1. Follow-Up to In Vivo Testing 251 10.7.2. Strategy for Germ Cell Testing 251
CONTENTS
ix
10.8.
European Union Guideline for Testing of Chemicals Under the Registration, Evaluation, Authorization and Restriction of Chemical (REACH) 252 10.9. Specialty Guidelines for Genotoxicity: Genotoxic Impurities in Pharmaceuticals 256 10.10. The Quintessence for Regulatory Assessment: In Vivo Testing for Risk Assessment 258 10.10.1. Choice of In Vivo Test 261 10.10.2. Evaluation of In Vivo Results 263 10.11. Summary and Outlook 264 References 265
CHAPTER 11
IN VITRO GENOTOX ASSAYS
272
David Kirkland and David Gatehouse 11.1. 11.2. 11.3. 11.4. 11.5. 11.6. 11.7. 11.8. 11.9.
Introduction 272 In Vitro Metabolic Activation 273 In Vitro Tests for Gene Mutation in Bacteria 273 In Vitro Tests for Gene Mutation in Mammalian Cells 276 In Vitro Tests for Chromosome Damage in Mammalian Cells 279 The In Vitro Micronucleus Test 280 In Vitro Test for Unscheduled DNA Synthesis in Rat Hepatocytes In Vitro Comet Assay 284 Strengths and Limitations 285 References 286
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
283
289
Véronique Thybaud 12.1.
12.2. 12.3.
12.4.
12.5.
Introduction 289 12.1.1. Endpoints Used for In Vivo Genetic Toxicology Assays 289 12.1.2. Contribution of In Vivo Genetic Toxicology Assays to Risk Assessment 291 Parameters and Criteria for Valid In Vivo Genotoxicity Assays and Implications for Experimental Design 292 In Vivo Genotoxicity Assays Required in the Standard Battery of Tests 303 12.3.1. Mammalian Erythrocyte Micronucleus Test 304 12.3.2. Bone Marrow Chromosome Aberration Test 308 In Vivo Genotoxicity Assays Used Mainly as Complementary or Follow-Up Tests 310 12.4.1. The Comet Assay 311 12.4.2. DNA Adducts 314 12.4.3. Unscheduled DNA Synthesis Test in Liver Cells 324 12.4.4. Sister-Chromatid Exchange Assay 326 12.4.5. Gene Mutation Assays 328 Conclusion and Perspectives 344 References 345
x
CONTENTS
PART IV
ASSESSING THE HUMAN RELEVANCE OF CHEMICAL-INDUCED TUMORS
361
FRAMEWORK ANALYSIS FOR DETERMINING MODE OF ACTION AND HUMAN RELEVANCE
CHAPTER 13
363
R. Julian Preston 13.1. 13.2.
13.3. 13.4.
Introduction 363 Framework Analysis: Mode of Action and Key Events 364 13.2.1. Definitions 364 13.2.2. An Overview of the Framework for Analyzing Mode of Action 365 13.2.3. Framework for Assessing Human Relevance of Animal MOA 365 13.2.4. Establishing and Applying Key Events in Support of MOA 367 Framework Analysis: Human Relevance 372 Future Directions 375 References 376
CHAPTER 14
EXPERIMENTAL ANIMAL STUDIES AND CARCINOGENICITY
378
Mary Elizabeth (Bette) Meek 14.1. 14.2.
14.3.
14.4. 14.5.
Introduction 378 Current Status of Hazard Testing for Cancer for Regulatory Risk Assessment 14.2.1. The Combined Chronic/Cancer Bioassay in Rats and Mice 379 14.2.2. Perinatal Carcinogenicity Studies 382 14.2.3. Limited In Vivo Studies 382 Application in Risk Assessment 383 14.3.1. Hazard Identification 383 14.3.2. Hazard Characterization 386 14.3.3. Dose–Response Analyses; Selection of Points of Departure 388 Evolution of Testing Strategies 390 Discussion: Closing the GAP Between Hazard Testing and Risk Assessment References 393
CHAPTER 15
CANCER EPIDEMIOLOGY
15.3.
15.4.
15.5. 15.6.
Introduction 397 Considerations for the Epidemiologic Study of Cancer 15.2.1. Demographics 398 15.2.2. Other Variables 400 Epidemiologic Study Methods 403 15.3.1. Types of Epidemiologic Studies 403 15.3.2. Meta-analysis and Case Reports 407 Evaluation of Studies and Their Results 407 15.4.1. Quality of Studies 407 15.4.2. Determining Causal Association 408 Substances Causally Associated with Cancer 411 Future for Cancer Epidemiology 414
391
397
Herman J. Gibb and Jessie P. Buckley 15.1. 15.2.
379
398
CONTENTS
15.6.1. The Effect of Exposure at Different Ages 15.6.2. Molecular Epidemiology 415 15.6.3. Infectious Agents 415 References 416 CHAPTER 16
xi
414
RODENT HEPATOCARCINOGENESIS
419
James E. Klaunig 16.1.
16.2.
16.3. 16.4.
Introduction 419 16.1.1. Initiation 420 16.1.2. Promotion 422 16.1.3. Progression 422 Mechanisms of Action of Hepatic Carcinogens 16.2.1. Genotoxic Agents 424 16.2.2. Nongenotoxic Mechanisms of Action Human Relevance Framework 434 Summary 435 References 435
423 425
MODE OF ACTION ANALYSIS AND HUMAN RELEVANCE OF LIVER TUMORS INDUCED BY PPARα ACTIVATION
CHAPTER 17
439
J. Christopher Corton 17.1. 17.2. 17.3.
17.4.
Overview 439 Introduction 440 Mode of Action Analysis in the EPA Risk Assessment Framework 441 17.3.1. Summary of the Mode of Action and Human Relevance of Liver Tumors Induced by PPARα Activation 441 17.3.2. Detailed Evaluation of the Rodent Mode of Action 443 Relevance of PPARα Activator-Induced Rodent Liver Tumor Response to Humans 467 References 467
ALPHA2U-GLOBULIN NEPHROPATHY AND CHRONIC PROGRESSIVE NEPHROPATHY AS MODES OF ACTION FOR RENAL TUBULE TUMOR INDUCTION IN RATS, AND THEIR POSSIBLE INTERACTION
CHAPTER 18
482
Edward A. Lock and Gordon C. Hard 18.1. 18.2. 18.3. 18.4. 18.5. 18.6.
Introduction 482 Chemicals that Increase the Incidence of Renal Tubule Tumors in Male Rats by an α2u-Globulin Mode of Action 483 Chemicals Increasing the Incidence of Renal Tumors Through Exacerbation of Spontaneous Chronic Progressive Nephropathy (CPN) 489 Chemicals Increasing RTT Incidence Through a Mode of Action Involving Exacerbation of CPN 491 Examples Where the α2u-Globulin and Exacerbated CPN Modes of Action May Be Acting in Concert 493 Relevance of Rat α2u-Globulin Nephropathy and CPN to Humans 495 References 496
xii
CONTENTS
CHAPTER 19
URINARY TRACT CALCULI AND BLADDER TUMORS
501
Samuel M. Cohen, Lora L. Arnold, and Shugo Suzuki 19.1. 19.2. 19.3. 19.4. 19.5. 19.6. 19.7. 19.8.
Introduction 501 Direct and Indirect Formation of Urinary Solids 502 Urinary Factors Influencing the Formation of Urinary Solids Collection of Urine for Detection of Urinary Solids 507 Interspecies Comparison of Urine Composition 508 Urinary Solid Carcinogenesis in Rodents 508 Epidemiology 510 Risk Assessment 511 References 512
505
PART V
METHODS FOR INFORMING CANCER RISK QUANTIFICATION CHAPTER 20 (Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS: A STATE-OF-THE-ART OVERVIEW
515
517
Yin-tak Woo and David Y. Lai 20.1. 20.2.
20.3.
20.4.
20.5.
Introduction 517 Overview of (Q)SAR Analysis and Modeling 518 20.2.1. Types of (Q)SAR 518 20.2.2. Criteria for Assessing Validity and Scientific Soundness of (Q)SAR 519 20.2.3. Difficulties of (Q)SAR Modeling/Prediction of Chemical Carcinogens 520 20.2.4. Importance of Mechanistic Understanding 520 Mechanism-Based SAR Analysis of Chemical Carcinogens, Fibers, and Particles/Nanoparticles 521 20.3.1. Basic Principles 521 20.3.2. SAR of Genotoxic Carcinogens 522 20.3.3. SAR of Nongenotoxic Carcinogens 528 20.3.4. SAR of Fibers, Particles, and Nanomaterials 534 Uses of (Q)SAR in Cancer Hazard/Risk Assessment and Brief Overview of Predictive Systems/Softwares 544 20.4.1. Evolving Uses of (Q)SAR in Cancer Hazard Identification and Risk Assessment 544 20.4.2. Brief Overview of (Q)SAR Systems/Softwares for Predicting Carcinogenic Potential of Chemicals 546 Future Perspectives 548 References 550
PHYSIOLOGICALLY BASED PHARMACOKINETIC (PBPK) MODELS IN CANCER RISK ASSESSMENT
CHAPTER 21
Mathieu Valcke and Kannan Krishnan 21.1. 21.2. 21.3.
Introduction 557 PBPK Modeling: Characteristics and Approaches 558 PBPK Models in Cancer Risk Assessment 563 21.3.1. High-Dose to Low-Dose and Interspecies Extrapolation
565
557
CONTENTS
21.4.
21.5.
21.3.2. Intraspecies Extrapolation 568 21.3.3. Route-to-Route Extrapolation 571 21.3.4. Extrapolation from Individual Carcinogens to Mixtures PBPK Models in Cancer Risk Assessment: Case Studies 574 21.4.1. Dichloromethane (Methylene Chloride) 574 21.4.2. Vinyl Chloride 575 21.4.3. Chloroform 576 21.4.4. Dioxane 577 21.4.5. Trichloroethylene 578 21.4.6. Volatile Organic Chemical Mixtures 578 Concluding Remarks 579 References 580
CHAPTER 22
xiii
571
GENOMICS AND ITS ROLE IN CANCER RISK ASSESSMENT
586
Banalata Sen, Douglas C. Wolf, and Vicki Dellarco 22.1. 22.2. 22.3. 22.4.
22.5. 22.6.
Introduction 586 “-Omics” Technologies 588 Genomics and the New Risk Assessment Paradigm Case Studies 590 22.4.1. Perfluorooctanoic acid (PFOA) 590 22.4.2. Formaldehyde and Glutaraldehyde 591 22.4.3. Conazoles 592 Use of Genomics in Predictive Toxicology 593 Conclusions 594 References 595
CHAPTER 23
589
COMPUTATIONAL TOXICOLOGY IN CANCER RISK ASSESSMENT
597
Jerry N. Blancato 23.1. 23.2. 23.3.
23.4. 23.5.
Introduction 597 Risk Assessment: Historical Perspective 598 Enhancements in Quantitative Risk Assessment 599 23.3.1. Physiologically Based Pharmacokinetic (PBPK) Modeling 23.3.2. Pharmacokinetic Variability and Uncertainty 601 23.3.3. Pharmacodynamic and Dose–Response Modeling 602 Computational Toxicology and Future Risk Assessments 602 23.4.1. 21st-Century Toxicology 603 Conclusion 609 References 610
599
PART VI
GENERAL APPROACHES FOR QUANTIFYING CANCER RISKS CHAPTER 24
LINEAR LOW-DOSE EXTRAPOLATIONS
Michael Dourson and Lynne Haber 24.1. 24.2.
Introduction 615 Historical 616
613 615
xiv 24.3. 24.4.
CONTENTS
Issues Related to Extrapolation from Experimental Data Conclusion 631 References 633
625
QUANTITATIVE CANCER RISK ASSESSMENT OF NONGENOTOXIC CARCINOGENS
CHAPTER 25
636
Rafael Meza, Jihyoun Jeon, and Suresh H. Moolgavkar 25.1.
25.2.
25.3.
Introduction 636 25.1.1. The Hazard or Incidence Function 637 25.1.2. Two-Stage Clonal Expansion (TSCE) Model 637 25.1.3. Multistage Clonal Expansion (MSCE) Model 640 25.1.4. Modeling Dose–Response in the TSCE and MSCE Models 25.1.5. Analysis of Epidemiological Data 643 25.1.6. Analysis of Premalignant Lesions Using the TSCE Model Some Examples and Applications 649 25.2.1. Smoking, Radon, and Arsenic Exposures and Lung Cancer 25.2.2. Folate and Colorectal Cancer 653 25.2.3. Enzyme-Altered Foci in the Rat Liver 653 Concluding Remarks 655 References 655
CHAPTER 26
642 644 649
NONLINEAR LOW-DOSE EXTRAPOLATIONS
659
Ari S. Lewis and Barbara D. Beck 26.1. 26.2.
26.3. 26.4. 26.5.
26.6.
26.7.
Introduction 659 Mechanistic Aspects of Nonlinear Carcinogenesis 661 26.2.1. Pre-DNA Damage Mechanisms 661 26.2.2. Post-DNA Damage Mechanisms 662 26.2.3. Hormesis 663 DNA-Reactive Carcinogens and Nonlinearity 664 Nonmutagenic Carcinogens and Nonlinearity 666 Cancer Risk Assessment 668 26.5.1. Basis for the Linearity Assumption 668 26.5.2. EPA Cancer Risk Assessment and Low-Dose Extrapolation 669 26.5.3. Low-Dose Extrapolation Outside the United States 670 Nonlinearity Principles into Practice 670 26.6.1. Using an RfD or MOE Approach 671 26.6.2. Other Nonlinear Cancer Evaluations: Captan and Chloroform 673 26.6.3. BBDR Modeling 674 26.6.4. Harmonization of Cancer and Noncancer Risks 675 Summary and Conclusion 676 References 677
CANCER RISK ASSESSMENT: MORE UNCERTAIN THAN WE THOUGHT
CHAPTER 27
Edmund A. C. Crouch 27.1. 27.2. 27.3.
Introduction 681 Summary of Previous Analyses 681 Selection of Carcinogenicity Measure—The CD10
684
681
CONTENTS
27.4. 27.5. 27.6. 27.7. 27.8.
The Variation of CD10 Within a Species 684 Extrapolation of the Median CD10 Between Species Extrapolation of the IntraSpecies Variation in CD10 Conclusions 693 Appendix 695 27.8.1. Obtaining the CD10 Estimates 695 27.8.2. Median and Geometric Standard Deviation 27.8.3. Testing Hypotheses about ln(GSD) 696 References 697
xv
686 692
696
COMBINING NEOPLASMS FOR EVALUATION OF RODENT CARCINOGENESIS STUDIES
CHAPTER 28
699
Amy E. Brix, Jerry F. Hardisty, and Ernest E. McConnell 28.1. 28.2. 28.3. 28.4.
28.5.
Introduction 699 Rationale for Combining Neoplasms 701 Usefulness of Differentiating Benign from Malignant Neoplasms and of Subclassifying Neoplasms 702 Criteria for Combining Neoplasms 704 28.4.1. Combinations According to Organ and Tissue 704 28.4.2. Combinations by Site 704 28.4.3. Combining Neoplasms of a Common Cell Type in Different Tissues Summary 711 References 711
CANCER RISK BASED ON AN INDIVIDUAL TUMOR TYPE OR SUMMING OF TUMORS
710
CHAPTER 29
716
Andrew G. Salmon and Lindsey A. Roth 29.1. 29.2. 29.3.
29.4.
29.5.
Introduction 716 Summing of Tumors of Related Types 717 Summing of Unrelated Tumor Types 718 29.3.1. Affected-Animal Count 718 29.3.2. Addition of Independent Potency Values 29.3.3. Distribution-Based Methods 719 Example: 1,3-Butadiene 721 29.4.1. Source Data 721 29.4.2. Affected-Animal Count 722 29.4.3. Distribution-Based Methods 724 Conclusions 732 References 734
718
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
CHAPTER 30
Shannon Gaffney, Jennifer Sahmel, Kathryn D. Devlin, and Dennis J. Paustenbach 30.1. 30.2.
Introduction 736 Exposure Reconstruction Methodology 737 30.2.1. Addressing the Goals of the Exposure Reconstruction 738 30.2.2. Organizing and Ranking Available Exposure Information 738
736
xvi
CONTENTS
30.2.3. 30.2.4.
30.3.
30.4. INDEX
Identifying Key Data Gaps in the Available Exposure Information 740 Selecting the Appropriate Methodology to Reconstruct Exposure Values 740 30.2.5. Conducting an Uncertainty Analysis of the Reconstructed Exposure Values 764 Application of Estimated Historical Exposure Values to Cancer Risk Estimates 766 30.3.1. Estimating Dose 767 30.3.2. Estimating Risk 768 30.3.3. Use of Probabilistic Analysis to Refine Dose Estimates 770 Summary 770 References 772 785
PREFACE
Cancer risk assessment is an ever-changing discipline with standard regulatory practices and defaults giving way to ever-increasing breakthroughs in scientific discovery. The scientific literature is, however, replete with reports of toxicantinduced changes, but discriminating between those reports that are irrelevant or relevant to humans and those that are compensatory versus truly adverse can be an arduous task. This book aims to inform and to provide interpretive guidance on evaluating toxicological data and understanding the relevance of such data to hazard evaluation and cancer risk estimation. The topics presented herein begin with Part I, which provides an overview of cancer risk assessment, followed by a discussion on science policy. The regulatory frameworks for industrial chemicals and biocides are presented along with the general approaches for developing standards for chemicals in air, water, food, soil, and consumer products. In Part II, basic concepts in cancer biology, chemical carcinogenesis, hormesis, and experimental evidence of thresholds for genotoxic carcinogens are provided. Thereafter, Part III describes the testing guidelines and regulations for in vitro and in vivo genotoxicity testing, and Part IV offers interpretive guidance on assessing the human relevance of chemical-induced tumors from rodent studies, along with the necessary criteria for evaluating data from epidemiological studies. Commonly observed modes of action from experimental animal studies, including PPAR-α, α2u-globulin, and so on, are then discussed. In Part V, methods for informing cancer risk quantification, including quantitative structure–activity relationships (QSAR), physiologically based pharmacokinetic (PBPK) modeling, “-omics”, and computational toxicology are discussed. Finally, Part VI addresses general approaches for quantifying cancer risks including linear and nonlinear low-dose extrapolations, summing tumors, and exposure reconstruction for cancer risk estimation. The foregoing topics are critical for keeping abreast of changes that are taking place in cancer risk assessment, as well as in the fields of toxicology and risk assessment in general. For example, with the increased emphasis on describing a chemical’s mode of action for both cancer and noncancer endpoints, an understanding of the human relevance framework is essential, as is the role of rapidly developing technologies (e.g., “-omics”) for informing mode(s) of action. Therefore, readers of this text will take away knowledge that is applicable to cancer risk assessment and more broadly to toxicology and risk assessment. The resources that formed the bases for this text include: peer-reviewed scientific articles, regulatory guidance documents, validated test guidelines, and the many years of experience conveyed throughout by the contributing authors. xvii
xviii
PREFACE
The editors are truly grateful to the contributing authors of this text, who provided their expertise on a gratis basis. If it were not for their dedication and commitment to advancing the knowledge and understanding of cancer risk assessment, the extensive coverage provided herein would not have been possible. Taipei, Taiwan Baton Rouge, Louisiana April 2010
Ching-Hung Hsu Todd Stedeford
CONTRIBUTORS
Elizabeth L. Anderson, Ph.D., FATS Group Vice President and Principal Scientist, Exponent, Inc., Alexandria, Virginia Lora L. Arnold, M.S. Assistant Professor, University of Nebraska Medical Center, Omaha, Nebraska Barbara D. Beck, Ph.D., DABT, FATS Principal, Gradient, Cambridge, Massachusetts Jerry N. Blancato, M.S., Ph.D. Acting Director, Office of Administrative and Research Support, Office of Research and Development (ORD), United States Environmental Protection Agency, Research Triangle Park, North Carolina Amy Brix, D.V.M., Ph.D., DACVP Veterinary Pathologist and Contractor for NTP QA, Experimental Pathology Laboratories, Inc., Research Triangle Park, North Carolina Jessie P. Buckley, M.P.H. Ph.D. Candidate, Gillings School of Global Public Health, University of North Carolina, Chapel Hill, North Carolina Edward J. Calabrese, Ph.D., FATS Professor of Toxicology, Department of Public Health, University of Massachusetts, Amherst, Massachusetts Samuel M. Cohen, M.D., Ph.D. Havlik–Wall Professor of Oncology, Department of Pathology and Microbiology, University of Nebraska Medical Center, Omaha, Nebraska J. Christopher Corton, Ph.D. Senior Research Biologist, Integrated Systems Toxicology Division, National Health and Environmental Effects Research Laboratory (NHEERL), Office of Research and Development (ORD), United States Environmental Protection Agency, Research Triangle Park, North Carolina Edmund A. C. Crouch, Ph.D. Senior Scientist, Cambridge Environmental, Inc., Cambridge, Massachusetts Vicki Dellarco, Ph.D. Science Advisor, Office of Pesticide Programs, United States Environmental Protection Agency, Washington, D.C. Kathryn D. Devlin, M.S. Health Scientist, ChemRisk, LLC, Boulder, Colorado Michael Dourson, Ph.D., DABT, FATS President, Toxicology Excellence for Risk Assessment (TERA), Cincinnati, Ohio xix
xx
CONTRIBUTORS
Anna M. Fan, Ph.D., DABT Chief, Pesticide and Environmental Toxicology Branch, Office of Environmental Health Hazard Assessment (OEHHA), California Environmental Protection Agency, Oakland, California Shoji Fukushima, M.D., Ph.D. Director, Japan Bioassay Research Center, Japan Industrial Safety & Health Association, Hadano, Kanagawa, Japan Shannon H. Gaffney, Ph.D., M.H.S., CIH Managing Health Scientist, ChemRisk, LLC, San Francisco, California David Gatehouse, Ph.D., FRCPath Consultant, Buntingford, Hertfordshire, United Kingdom Herman J. Gibb, Ph.D., M.P.H. President, Tetra Tech Sciences, Arlington, Virginia Virginia A. Gretton Regulatory Advisor, SafePharm Laboratories Ltd., Derbyshire, United Kingdom Lynne Haber, Ph.D., DABT Associate Director, Toxicology Excellence for Risk Assessment (TERA), Cincinnati, Ohio Gordon C. Hard, BVSc, Ph.D., DSc, DACVP, FRCPath, FRCVS, FAToxSci Independent Consultant, Tairua, New Zealand Jerry F. Hardisty, D.V.M., DACVP, IATP President and Veterinary Pathologist, Experimental Pathology Laboratories, Inc., Research Triangle Park, North Carolina James W. Holder, Ph.D. Toxicologist/Cancer, National Center for Environmental Assessment, Office of Research and Development, United States Environmental Protection Agency (Retired), Washington, D.C. Robert A. Howd, Ph.D. Chief, Water Toxicology Section, Office of Environmental Health Hazard Assessment (OEHHA), California Environmental Protection Agency, Oakland, California Jihyoun Jeon, M.S., Ph.D. Staff Scientist, Program in Biostatistics and Biomathematics, Fred Hutchinson Cancer Research Center, Seatte, Washington Anna Kakehashi, Ph.D. Lecturer, Department of Pathology, Osaka City University Medical School, Osaka, Japan David J. Kirkland, Ph.D. Consultant and Professor (University of Wales, Swansea, United Kingdom), Tadcaster, North Yorkshire, United Kingdom James E. Klaunig, Ph.D. Professor and Chair, Environmental Health, Indiana University, Bloomington, Indiana Kannan Krishnan, Ph.D., DABT, FATS Professor, Department of Environmental Health and Health at Work, University of Montréal, Montreal, Quebéc, Canada
CONTRIBUTORS
xxi
David Y. Lai, Ph.D., DABT Senior Toxicologist, Risk Assessment Division, Office of Pollution Prevention and Toxics (OPPTS), United States Environmental Protection Agency, Washington, D.C. Ari S. Lewis, M.S. Senior Scientist, Gradient, Cambridge, Massachusetts Edward A. Lock, MIBiol, Ph.D., FRCPath, FBTS, FATS Professor, School of Pharmacy and Biomolecular Sciences, Liverpool John Moores University, Liverpool, United Kingdom Kimberly Lowe, Ph.D., M.H.S. Senior Scientist, Exponent, Inc., Seattle, Washington Gary E. Marchant, Ph.D., J.D. Lincoln Professor, College of Law, Arizona State University, Tempe, Arizona Hans-Jörg Martus, Ph.D. Head, Genetic Toxicology, Preclinical Safety, Novartis Institutes for BioMedical Research, Basel, Switzerland Ernest E. McConnell, D.V.M., M.S., DACVP, DABT President, Tox Path, Inc., Raleigh, North Carolina Mary Elizabeth (Bette) Meek, M.Sc., Ph.D. Associate Director, Chemical Risk Assessment, McLaughlin Centre for Population Health Risk Assessment, University of Ottawa, Ottawa, Ontario, Canada Rafael Meza, Ph.D. Research Scientist, Division of Mathematical Modeling, University of British Columbia Centre for Disease Control, Vancouver, British Columia, Canada Suresh H. Moolgavkar, M.D., Ph.D. Corporate Vice President and Director, Center for Epidemiology, Biostatistics, and Computational Biology, Exponent, Inc., Bellevue, Washington Lutz Müller, Ph.D. Head Full Development Projects, Non-Clinical Drug Safety, F. Hoffmann-La Roche Ltd., Basel, Switzerland Stephen Nesnow, Ph.D. Senior Research Scientist, Integrated Systems Toxicology Division, National Health and Environmental Effects Research Laboratory (NHEERL), Office of Research and Development (ORD), United States Environmental Protection Agency, Research Triangle Park, North Carolina Dennis J. Paustenbach, Ph.D., CIH, DABT President and Founder, ChemRisk, LLC, San Francisco, California R. Julian Preston, Ph.D. Associate Director for Health, National Health and Environmental Effects Research Laboratory (NHEERL), Office of Research and Development (ORD), United States Environmental Protection Agency, Research Triangle Park, North Carolina Paolo F. Ricci, Ph.D., LL.M., M.P.A. Professor, Holy Names University, Oakland, California, and University of Massachusetts, Amherst, Massachusetts
xxii
CONTRIBUTORS
Lindsey A. Roth, M.A. Research Scientist II, Safer Alternatives Assessment and Biomonitoring Section, Office of Environmental Health Hazard Assessment (OEHHA), California Environmental Protection Agency, Oakland, California Jennifer Sahmel, M.P.H., CIH, CSP Supervising Health Scientist, ChemRisk, LLC, Boulder, Colorado Banalata Sen, Ph.D. Science Education and Outreach Program Manager, Environmental Health Perspectives, DHHS, NIH, NIEHS, Durham, North Carolina Andrew G. Salmon, M.A., D.Phil. Chief, Toxicology and Risk Assessment Section, Office of Environmental Health Hazard Assessment (OEHHA), California Environmental Protection Agency, Oakland, California Todd Stedeford, Ph.D., J.D., DABT Toxicology Advisor & In-House Counsel, Health, Safety & Environment, Albemarle Corporation, Baton Rouge, Louisiana Shugo Suzuki, M.D., Ph.D. Postdoctoral Research Associate, University of Nebraska Medical Center, Omaha, Nebraska Henk Tennekes, M.Sc., Ph.D., RT Consultant in Toxicology, Experimental Toxicology Services, Zutphen, The Netherlands Véronique Thybaud, Ph.D. Scientific Advisor, Disposition-Safety and Animal Research, Preclinical Safety, Sanofi Aventis, Vitry sur Seine, France Paul Turnham, B. Eng., M.S., P.E. Senior Managing Scientist, Exponent, Inc., Alexandria, Virginia Mathieu Valcke, M.Sc. Scientific Advisor, National Institute of Public Health of Québec, Montreal, Quebec, Canada Hideki Wanibuchi, M.D., Ph.D. Professor, Department of Pathology, Osaka City University Medical School, Osaka, Japan Min Wei, M.D., Ph.D. Assistant Professor, Department of Pathology, Osaka City University Medical School, Osaka, Japan Douglas C. Wolf, D.V.M., Ph.D., FIATP, ATS Assistant Laboratory Director, National Health and Environmental Effects Research Laboratory (NHEERL), Office of Research and Development (ORD), United States Environmental Protection Agency, Research Triangle Park, North Carolina Yin-tak Woo, Ph.D., DABT Senior Toxicologist, Risk Assessment Division, Office of Pollution Prevention and Toxics (OPPTS), United States Environmental Protection Agency, Washington, D.C. Errol Zeiger, Ph.D., J.D. Principal, Errol Zeiger Consulting, Chapel Hill, North Carolina
ABBREVIATIONS AND ACRONYMS
AAF 4-ABP ACF ACO ACToR ADAF Ade ADI ADME AFC AHF AhR AI AMS ANOVA AOM apo ARB ARNT ATSDR AUC B[a]A BBDR BDA BE BEEL BEIs BMD BMDL BMR B[a]P BPD BPDE BrDU
2-Acetylaminofluorene 4-Aminobiphenyl Aberrant crypts foci Acyl-CoA oxidase Aggregated chemical toxicity resource Age-dependent adjustment factor Adenine Allowable daily intake Absorption, distribution, metabolism, and excretion Altered foci cells Altered hepatic foci Aryl hydrocarbon receptor Artificial intelligence Accelerator mass spectrometry Analysis of variance Azoxymethane Apolipoprotein Air Resources Board, California EPA Ah receptor nuclear translocator U.S. Agency for Toxic Substances and Disease Registry Area under the curve Benz[a]anthracene Biologically based dose–response Bayesian data analysis Biomonitoring equivalents Biological environmental exposure limit Biological exposure indices Benchmark dose Benchmark dose lower bound Benchmark response Benzo[a]pyrene Biocidal products directive Benzo[a]pyrene diol epoxides 5-Bromo-2-deoxyuridine xxiii
xxiv
ABBREVIATIONS AND ACRONYMS
b.w. CAA CAF CAG CAM CAR CCA CCl4 CD10 CDC CDC CDK CEBS CEO CEO CEPA CERCLA cGys ChAMP CHMP CIIT CMRs CNDR CoA COPC CPDB CPN CPSC CPT-I CPUM CSA CSF CSR CTM Cx CYP 2-D 3-D 4-DAB DAG DAPI 4-DAST DB[a,l]P DC DCB
Body weight U.S. Clean Air Act Cancer-associated fibroblast Carcinogens Assessment Group Cellular adhesion molecule Constitutive androstane receptor Chromated copper arsenate Carbon tetrachloride 10% of Cancer dose Center for Disease Control U.S. Centers for Disease Control and Prevention Cyclin-dependent kinase Chemical effects in biological systems Chloroethylene oxide Cyanoethylene oxide Canadian Environmental Protection Act Comprehensive Environmental Response, Compensation and Liability Act Centigrays Chemical Assessment and Management Program Committee of Human Medicinal Products Chemical Industries Institute of Toxicology Carcinogens, mutagens, or reproductive toxicants Canadian National Dose Registry Acyl coenzyme A Contaminants of Potential Concern Carcinogenic potency database Chronic progressive nephropathy Consumer Product Safety Commission Carnitine palmitoyl transferase-I Colorado Plateau Uranium Miners Chemical Safety Assessment Cancer slope factor Chemical Safety Report Chinese tin miners Connexon Cytochrome P450 Two-dimensional Three-dimensional 4-Dimethylaminoazobenzene Directed acyclic graph 4′,6-Diamidino-2-phenylindole 4-Dimethylaminostilbene Dibenzo[a,l]pyrene Dendritic cells 1,4-Dichlorobenzene
ABBREVIATIONS AND ACRONYMS
DCC DCM 1,3-DCP DDT DEEM DEHA DEHP DEN DEN, DENA DEPM dGua DHEW DINP DINP DMA DMBA DMN DMN DQA DSS DSSTox Dt EAF ECHA ECM ECVAM ED EFSA 2-EH EHEN ELISA EMSA ENNG ENU ENU EPA EPI EPIC ER ERK ESR ESTR EU FDA FDCA
xxv
Deleted in colorectal cancer Dichloromethane or methylene chloride 1,3-Dichloropropene Dichlorodiphenyltrichloroethane Dietary Exposure Evaluation Model Di-(2-ethylhexyl)adipate Di-(2-ethylhexyl)phthalate N-Nitrosodiethylamine N,N-Diethylnitrosamine Dietary Exposure Potential Model Deoxyguanosine U.S. Department of Health Education and Welfare Di-(2-isononyl) phthalate Diisononyl phthalate Dimethylarsenic acid 7,12-Dimethylbenz[a]anthracene or 9,10-Dimethyl-1,2-benz[a] anthracene Dimethylnitrosamine N-Nitrosodimethylamine Data Quality Act Dextran sulfate sodium Distributed structure-searchable toxicity Dose metrics Enzyme-altered foci European Chemicals Agency Extracellular matrix European Centre for the Validation of Alternative Methods Effective dose European Food Safety Authority 2-Ethylhexanol Ethyl hydroxyethylnitrosamine Enzyme-linked immunosorbant assays Electrophoretic mobility shift assay N-Ethyl-N′-nitro-N-nitrosoguanidine Ethylnitrosourea N-Nitroso-N-ethylurea U.S. Environmental Protection Agency Exposure potency index European Prospective Investigation into Cancer and Nutrition Estrogen receptor Extracellular signal-regulated kinases Electron spin resonance Expanded Simple Tandem Repeat European Union U.S. Food and Drug Administration Food, Drug and Cosmetic Act
xxvi
ABBREVIATIONS AND ACRONYMS
FFDCA FGF FGFR3 FIFRA FISH FPG FQPA GAC γ-GGT GI GJIC GJs GLP G6PD GSSG GSH GST GST-P Gua HaSDR HCA HCA HC HCC HCV HEAA HGP HIV HMG-CoA Hmgcr hPPARα HPLC hprt HPV HPV HPVIS HRF HSC HTLV HTS IAEMS IARC ICEM ICCVAM ICH
Federal Food, Drug and Cosmetic Act Fibroblast growth factor Fibroblast growth factor receptor 3 Federal Insecticide, Fungicide and Rodenticide Act Fluorescent in situ hybridization Formamido pyrimidine glycosylase Food Quality Protection Act Genetic alterations in cancer Gamma-glutamyltransferase Gastrointestinal Gap junction intercellular communication Gap junction connections Good laboratory practice Glucose-6-phosphate dehydrogenase Glutathione disulfide Glutathione Glutathione S-transferases Glutathione S-transferase placental form Guanine Health and Safety Data Reporting Hydrocyanic acid High content analysis Health Canada Hepatocellular carcinoma Hepatitis C virus β-Hydroxyacetic acid Human Genome Project Human immunodeficiency virus 3-Hydroxy-3-methylglutaryl-CoA Hydroxymethylglutaryl-CoA reductase Human PPARα High-performance liquid chromatography Hypoxanthine-guanine phosphoribosyl transferase Human papilloma viruses High production volume High Production Volume Information System Human relevance framework Hemocytoblasts Human T-cell lymphotropic virus High-throughput screening International Association of Environmental Mutagen Societies International Agency for Research on Cancer International Conferences on Environmental Mutagens Interagency Coordinating Committee on the Validation of Alternative Methods International Conference on Harmonisation
ABBREVIATIONS AND ACRONYMS
IDS IKK IL1α IL1β ILSI ILSI RSI IND IPCS IR IRIS IRIS ITER ITC IUR IUR IWGT IWR JaCVAM JECFA JEM JNK Kdis LBD LED01 LED10 LET LFC LMS LMW ln(GSD) LNT LOAEL LSC LSS LTA MAC MACT MAP MC MCL MCMC MDA MEHP MeIQx MIBK
xxvii
Immunodefense system IκB kinase Interleukin-1alpha Interleukin-1beta International Life Science Institute International Life Sciences Risk Sciences Institute Exploratory investigational new drug applications International Programme on Chemical Safety Ionizing radiation Integrated Risk Information System U.S. EPA Integrated Risk Information System International Toxicity Estimates for Risk TSCA Interagency Testing Committee Inhalation unit risk Inventory update reporting International Workshop(s) on Genotoxicity Tests Interaction weighting ratio Japanese Center for the Validation of Alternative Methods Joint FAO/WHO Expert Committee on Food Additives Job exposure matrix c-Jun N-terminal kinases Dissolution rate constants Ligand binding domains Lower limit on effective dose01 Lower 95% confidence limit for the dose giving the animals an increased tumor incidence of 10% Linear-energy-transfer Lowest feasible concentration Linearized multistage Low-molecular-weight protein Logarithm of the geometric standard deviation Linear no-threshold Lowest observed adverse effect level Lymphoblast Life-stage study Local tissue array Apoptosis-induced channel Maximum achievable control technology Mitogen-activated protein Mast cell Maximum contaminant level Markov chain Monte Carlo Malondialdehyde Mono-2-ethylhexyl phthalate 2-Amino-3,8-Dimethylimidazo[4,5-f] quinoxaline Methyl isobutyl ketone
xxviii
ABBREVIATIONS AND ACRONYMS
miRNA MLA MLE MMP MMS MN MNU MOA MOE MPV MS MSCE MTBE MTD MUP MVK NAS NAS NBR NCEA NCEs NCEH NCoR NDI NF-kB NHANES NIOSH NIOSH-IREP NNG NNM NOAEL NOEL NPCs NRC NRC NSRLs NTP NTP Cmax OECD OEHHA 8-OH-dG 2-OH-TMP OMB OPP
MicroRNAs Mouse lymphoma tk+/− assay Maximum likelihood estimate Matrix metalloprotease Methyl methanesulfonate Micronuclei Methylnitrosourea Mode of action Margin of exposure Medium-production volume Mass spectrometric Multistage clonal expansion Methyl-tert-butyl ether Maximum tolerable dose Mouse urinary protein Moolgavkar–Venzon–Knudson National Academy of Sciences U.S. National Academies of Science NCI Black–Reiter U.S. EPA National Center for Environmental Assessment Normochromatic erythrocytes National Center for Environmental Health Nuclear receptor corepressor National death index Nuclear factor kappa B National Health and Nutrition Examination Survey U.S. National Institute for Occupational Safety and Health Interactive RadioEpidemiological Program Net nuclear grain N-Nitrosomorpholine No observed adverse effect level No observed effect level Nonparenchymal cells National Research Council U.S. National Research Council No significant risk levels National Toxicology Program U.S. National Toxicology Program Maximum or peak concentration Organisation for Economic Co-operation and Development Office of Environmental Health Hazard Assessment, California EPA 8-Hydroxy-2′-deoxyguanosine 2,2,4-Trimethyl 2-pentanol U.S. Office of Management and Budget U.S. EPA Office of Pesticide Programs
ABBREVIATIONS AND ACRONYMS
OPPTS ORD OSHA OSH Act OSOR OSTP OSWER PAHs PAIR PAPS PBBs PBPK PBTs PCBs PCDD PCE pCi PCNA PD PDF PDGF PEI PELs PFAA PFOA PFOS PGMBE Pgp PHGs PhIP PIR PMR POD PPAR PPAR-α PPL PPREs pRb PRGs PSP PTEN PTL PXR q1* qPCR (Q)SAR
Office of Prevention, Pesticides and Toxic Substances U.S. EPA Office of Research and Development U.S. Occupational Safety and Health Administration U.S. Occupational Safety and Health Act of 1970 One substance, one registration U.S. Office of Science and Technology Policy U.S. EPA Office of Solid Waste and Emergency Response Polycyclic aromatic hydrocarbons Preliminary assessment and information reporting 3′-Phosphoadenosine 5′-phosphosulfate Polybrominated biphenyls Physiologically based pharmacokinetic Persistent, bioaccumulative, and toxic substances Polychlorinated biphenyls Polychlorinated dibenzo dioxin Polychromatic erythrocyte Picocuries Proliferating cell nuclear antigen Cell population growth over time Probability density function Platelet-derived growth factor Polyethyleneimine Permissible exposure limits Perfluoroalkyl acid Perfluorooctanoic acid Perfluorooctanesulfonic acid Propylene glycol monobutyl ether P-glycoprotein Public health goals 2-Amino-1-methyl- 6-phenylimidazo[4,5-b] pyridine Proportionate incidence ratio Proportionate mortality ratio Point of departure Peroxisome proliferator-activated receptor Peroxisome proliferation activating receptor-alpha 32 P-Postlabeling PPARα responsive elements Inactivated retinoblastoma gene product Preliminary remediation goals Poorly soluble particles Phosphatase and tension Priority testing list Pregnane X receptor Upper 95% confidence limit on the cancer potency slope Quantitative polymerase chain reaction Quantitative structure–activity relationships
xxix
xxx
ABBREVIATIONS AND ACRONYMS
RAGS RBC RBP RCF R&D REACH REDs RELs RfD RFLP RIVM RMM RNS R.O.C. RoC ROS RSD RTG RT-PCR RTT RXRα SA SAB SAR SARA SCE SDWA S9 fraction SDS SEER SHEDS SIEF SIR SMR SOT SPP SPS SS IIC STN SV SWCNT SWP T-90 t/a TAA
Risk assessment guidance for Superfund Red blood cell Risk-based prioritizations Refractory ceramic fibers Research and development Registration, evaluation, authorization, and restriction of chemicals Reregistration eligibility documents Recommended exposure limits Reference dose Restriction fragment length polymorphism The Netherlands National Institute for Public Health and Risk management measures Reactive nitrogen species Receiver operating characteristic Report on carcinogens Reactive oxygen species Risk–specific dose Relative total growth Reverse transcription polymerase chain reaction Renal tubule tumors Retinoid X receptor-alpha Structural alert U.S. EPA Science Advisory Board Structure–activity relationship Superfund Amendments and Reauthorization Act Sister chromatid exchange U.S. Safe Drinking Water Act 9000 g Supernatant Safety data sheet Surveillance epidemiology and end results Stochastic human exposure and dose simulation Substance information exchange forum Standardized incidence ratio Standardized mortality ratio U.S. Society of Toxicology Security and prosperity partnership Sanitary and phytosanitary Stoddard solvent IIC Stochastic transition network Simian virus Single-walled carbon nanotubes Safety working party 90% Clearance time Tonnes per annum Thioacetamide
ABBREVIATIONS AND ACRONYMS
TBA TBARS TCA TCDD, dioxin TCE TD TD50 TDI TERA TF TGD TGF TGFβ1 tk TLC TMP TNF TNFα ToxRefDB TPA TRAIL TRI TSCE TTC TWA UCL UDS UF USDA UVR VC VLDL VOC WOE vPvBs VSD WHO Wnt WT1/2 WTO
Tert-butyl alcohol Thiobarbituric reactive substances Trichloroacetate 2,3,7,8-Tetrachlorodibenzo-p-dioxin Trichloroethylene Tolerable dose The dose inducing a tumor incidence of 50% in rodents Tolerable daily intake Toxicology Excellence for Risk Assessment Transcription factor Technical guidance document Transforming growth factor Transforming growth factor beta 1 Thymidine kinase Thin-layer chromatography 2,2,4-Trimethylpentane Tumor necrosis factor Tumor necrosis factor alpha Toxicology reference database Tetradecanoyl phorbol acetate the Environment TNF-related apoptosis-inducing ligand Toxics release inventory Two-stage clonal expansion Threshold of toxicological concern Time-weighted average Upper confidence limit Unscheduled DNA synthesis Uncertainty factor U.S. Department of Agriculture Ultraviolet radiation Vinyl chloride Very low density lipoproteins Volatile organic compound Weight of evidence Very persistent and very bioaccumulative substances Virtually safe dose World Health Organization Wingless type Weighted clearance half-time World Trade Organization
xxxi
PART
I
CANCER RISK ASSESSMENT, SCIENCE POLICY, AND REGULATORY FRAMEWORKS
CH A P TE R
1
CANCER RISK ASSESSMENT Elizabeth L. Anderson Kimberly Lowe Paul Turnham
1.1. 1.1.1.
CANCER RISK ASSESSMENT Cancer in the United States
Cancer is a group of diseases that result from abnormal and prolific cellular division. Based on current U.S. National Cancer Institute’s Surveillance Epidemiology and End Results (SEER) of cancer prevalence, it is estimated that more than 10 million people were living with cancer in the United States in 2005 (NCI 2008). The American Cancer Society predicts that 1 in 2 males and 1 in 3 females will develop some type of cancer in their lifetime, and that 1 in 4 males and 1 in 5 females is at risk of dying from this disease (NCI 2007a,b). Cancer is undoubtedly a substantial threat to public health. Understanding the etiology of cancer, identifying methods of prevention or treatment, and determining the carcinogenicity of the chemicals we use in our everyday lives are the objectives of many of our government divisions, academic institutions, and health-care industries. However, for public health agencies charged with quantifying safe levels of exposure to protect public health, these tasks are not simple matters of using biology to inform the standard-setting process; instead, gaps in science must be filled using a number of assumptions that are based both on scientific inferences and policy judgments. Under Congressional delegation, the broad mission of public health agencies is disease prevention. This includes a wide range of activities from providing education about healthy living to regulating the use and dispersion of agents that are known, or suspected, to cause cancer or other diseases. The basic principle of cancer risk assessment is to characterize both the weight of evidence (WOE) that the agent might be capable of causing cancer and the magnitude of risk, given past, current, or future exposure levels. The fundamental objective is to determine the threshold at which exposure to the agent poses no appreciable risk to humans or, in the absence of mechanistic knowledge, to define an acceptable risk for suspect carcinogens.
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
3
4
CHAPTER 1 CANCER RISK ASSESSMENT
1.1.2.
Historical Perspectives of Cancer Risk Assessment
Imagine a time when there was no exposure assessment, no evaluation of dose– response relationships (potency), and no particular attention paid to mechanisms of action to define the relevance of responses in animals to diseases in humans, as well as a time when the science of risk assessment to address environmental carcinogens was not developed. This time existed when the U.S. Environmental Protection Agency (EPA) was created in 1970, and it existed until the first Federal policy to adopt the use of risk assessment and risk management was announced by the Agency in 1976 (Albert et al. 1977; USEPA 1976). This policy was accompanied by the first guidelines for carcinogen risk assessment (USEPA 1976) and the establishment of an Agency group to carry out these assessments (named the Carcinogens Assessment Group, or CAG). The approach was novel at the time; however, it borrowed from the experience of radiation risk assessment, where a common mechanism of action was known and dose–response relationships in humans had been reasonably well characterized. Of course, large knowledge gaps existed. For most agents suspected of causing cancer, evidence was from high-dose studies in animals that relied on two dose levels to define cancer potential for humans who experienced much lower environmental exposures. Although controversial at the time, the science of risk assessment has developed into the internationally accepted approach to evaluate carcinogen risk associated with of exposure to environmental agents, food contaminants, and occupational contaminants. These approaches also have dictated close scrutiny of the scientific principles that lead to improved methods of addressing potency, mechanisms of action, test methods, exposure, and internal dose relationships. This section describes the landmarks and key events in the evolution of this science. Not long after the EPA was established, it began evaluating carcinogenesis data and translating its findings into public policy. These early decisions spawned the necessity to depart from simple qualitative characterization of tumors in humans or animals to incorporate the reality of exposures at low doses, far below those in the studies, and the potential for harm associated with these low-dose exposures. Because the Agency was newly developed, there was no precedent for regulating carcinogens in the environment. The early years of the EPA were a time of enormous zeal to cleanse the environment, especially of carcinogens that were thought to be the principal cause of a “cancer epidemic.” The Food, Drug, and Cosmetic Act (FDCA) had a provision for regulating intentional food additives to a zero-tolerance level, meaning that evidence of cancer by tumor formation in animals or humans was sufficient cause for banning the agent. The same zero-tolerance policy was attempted for a wide range of environmental agents thought to be potential carcinogens, including three major pesticides: dichlorodiphenyltrichloroethane (DDT), aldrin/dieldrin, and chlordane/ heptachlor, although the cancellation of DDT was probably more compelled by ecologic harm (USEPA 1972, 1975). Between 1970 and 1975, the EPA moved to suspend their use. The cancellation of these three pesticides set the zero-tolerance policy in motion and became what was judged to be the Agency’s cancer policy. However, it quickly became evident that a zero-tolerance policy was impractical.
1.1. CANCER RISK ASSESSMENT
5
For many economically important products, it was impossible to remove all exposure to agents suspected of having the ability to cause cancer (e.g., low-level exposure to benzene, a known human carcinogen, in gasoline). The policy was also highly controversial. Using the qualitative evidence of tumors in animals or humans, attorneys at the EPA had summarized the scientific information needed to characterize an agent as carcinogenic in legal briefs at the conclusions of the hearings to cancel the pesticides listed above. These summary statements were referred to in legal motions as “Cancer Principles.” The intent of these statements was to establish the foundation for the EPA’s authority to protect public health from exposure to environmental carcinogens. This approach received substantial criticism from the scientific community, parts of the private sector, and the Congress (Anonymous 1976). The criticism was largely based on the fact that the complex field of carcinogenesis could not be reduced to simple summary statements (USEPA 1976). In addition, there was concern that the Agency would take a broad approach to cancer regulation by labeling agents as carcinogenic in humans if they were carcinogenic in animals, treating all agents as if they had equal potency, or regulating without information about exposure and the specific threat of a particular agent. Given the large number of chemicals to which people are exposed in their everyday lives, there was a substantial need to establish a basis for setting priorities and balancing the risks associated with their use in terms of social and economic factors, as called for by the specific statutes under which public health agencies operated, including the EPA, which had inherited very broad authorities (Anderson 1983). Ultimately, the failure of the zero-tolerance policy led to the development of the risk assessment framework at the EPA. It was not until 1979 that other federal agencies joined the EPA in an effort to establish interagency guidance for conducting carcinogen risk assessments (Albert et al. 1977; IRLG 1979c; USEPA 1976). This initial risk assessment approach was developed to answer two questions (Anderson 1983): 1. How likely is the agent to be a human carcinogen? This step involves evaluating all of the relevant biomedical data to determine the total weight of evidence (WOE). At that time, the WOE was ranked from strongest to weakest in a scientific context. The strongest evidence was obtained from human data that were supported by animal bioassay results. Substantial evidence of carcinogenicity could be obtained from laboratory animal bioassay results showing replication of effects across species related to dose levels, and suggestive evidence could be obtained from weaker associations in animal studies. Other evidence from in vivo or in vitro studies was also considered. 2. On the assumption that an agent is a human carcinogen, what is the magnitude of its public health impact given current and projected exposures? This step is quantitative in nature and involves establishing a dose–response relationship to extrapolate to low levels of exposure, where environmental exposures generally occur, and evaluating the magnitude of the exposures of interest. Its purpose was to provide regulators a sense of the cancer potency of the agent, and some information about the public health impacts associated with exposures. In this step, risks were bracketed between an upper and lower
6
CHAPTER 1 CANCER RISK ASSESSMENT
bound approaching zero. The upper bounds were expressed both in terms of the individual increased cancer risks in the exposed population and the nationwide impact in terms of the annual increase in cases. Of particular note: (1) These first guidelines called for revising each risk assessment as better information became available, a goal that has been rarely realized. (2) Gaps in scientific knowledge were to be filled with public health protective assumptions to err on the side of safety, an early application of the precautionary principle. Over the last several decades, the Agency has sought to extend guidelines for carcinogens to incorporate improvements in our understanding of the cancer process. Because risk assessment necessarily relies on both science and policy judgments, these guidelines are essential to ensure that a consistent approach to risk assessment is taken. The effort to bring consistency to risk assessment is evolving and has produced revisions of guidelines and standard practices (examples of which are shown in Table 1.1). The most fundamental endorsement of the risk assessments
TABLE 1.1.
Historical Perspectives of the Development of the Risk Assessment Process
Year
Document
Details
1975
Quantitative Risk Assessment for Community Exposure to Vinyl Chloride (Kuzmack and McGaughy 1975) Interim Procedures and Guidelines for Health Risks and Economic Impact Assessments of Suspected Carcinogens (USEPA 1976) Hazardous substances summary and full development plan. United States. Interagency Regulatory Liaison Group (IRLG 1979a) Publications on toxic substances. United States. Interagency Regulatory Liaison Group (IRLG 1979b) Integrated Risk Information System (IRIS)
This was the first risk assessment document to be completed by the EPA.
1976
1978
1979
1980
1983
Risk Assessment in the Federal Government: Managing the Process (NRC 1983)
This document communicated the EPA’s intent to include “rigorous assessments of health risk and economic impacts” in the regulatory process. This document describes laws and legislation regarding hazardous substances and chemicals. This document reports basic facts about toxic substances and describes the publications that are available from many federal agencies. This database reports human health effects that may be related to chemicals found in the environment. Commonly referred to as the “Red Book,” this document was published by the National Academy of Sciences and described methods for risk assessment in the federal government. The EPA adopted and implemented the risk assessment methods that were outlined in this book.
(Continued)
1.1. CANCER RISK ASSESSMENT
TABLE 1.1.
Year
(Continued) Document
1984
Risk Assessment and Management: Framework for Decisionmaking (USEPA 1984)
1985
Chemical Carcinogens: A Review of the Science and Its Associated Principles (OSTP 1985)
1986
The Risk Assessment Guidelines of 1986a (USEPA 1986b)
1986
Guidelines for Carcinogen Risk Assessment (USEPA 1986a)
1989
Risk Assessment Guidance for Superfund, Vol. I: Human Health Evaluation Manual (Part A) (USEPA 1989) Proposed Guidelines for Carcinogen Risk Assessment (USEPA 1996)
1996
1997
7
Exposure Factors Handbook. U.S. EPA (USEPA 1997)
Details Published by the EPA, this document illustrated the strengths and weaknesses of the risk assessment process and emphasized the need to make the process as transparent as possible. Published by the U.S. Office of Science and Technology Policy (OSTP), this document provides a complete review of the application of epidemiology in carcinogen risk assessment. This EPA document provided guidelines for evaluating the human and animal evidence of carcinogenicity, as well as a classification scheme for categorizing the level of risk associated with a particular agent (i.e., limited, inadequate, no data, or no evidence). The purpose of these guidelines was to outline a procedure that EPA scientists could use to assess the cancer risk associated with exposure to chemicals in the environment. This document was also used to inform the public about the process of cancer risk assessment. Published by the EPA Office of Solid Waste and Emergency Response (OSWER), this is the first of a series of guidance documents on risk assessment for the Superfund. Because limitations were identified in the 1986 carcinogen risk assessment guidelines, new cancer risk assessment guidelines were set forth that allowed scientists the flexibility to incorporate relevant biological information into the assessment process. The new guidelines were reviewed by the EPA Science Advisory Board (SAB) in 1997. The guidelines were made available for public comment in 2001 and then were reviewed again by the SAB in 2003. Published by the EPA National Center for Environmental Assessment (NCEA) within the EPA’s Office of Research and Development (ORD), this document provides data on exposure activities and other parameters for assessing exposure to contaminants in the environment. The 1997 handbook updates the 1989 original.
(Continued)
8
CHAPTER 1 CANCER RISK ASSESSMENT
TABLE 1.1. (Continued)
Year
Document
2002
OSWER Draft Guidance for Evaluating the Vapor Intrusion to Indoor Air Pathway from Groundwater and Soils (Subsurface Vapor Intrusion Guidance) (USEPA 2002) World Trade Center Indoor Environment Assessment: Selecting Contaminants of Potential Concern and Setting Health-Based Benchmarks (USEPA 2003)
2003
2005
Guidelines for Cancer Risk Assessment (USEPA 2005)
2008
Child-Specific Exposure Factors Handbook (USEPA 2008a)
2009
The U.S. Environmental Protection Agency’s Strategic Plan for Evaluating the Toxicity of Chemicals (USEPA 2009)
Details Published by the EPA Office of Solid Waste and Emergency Response (OSWER), this document provides guidance for the evaluation of the vapor intrusion exposure pathway. This document, published by the Contaminants of Potential Concern (COPC) Committee of the World Trade Center Indoor Air Task Force Working Group, provides guidelines and methodologies for setting health based standards for chemicals in settled indoor dust. The formal guidelines for cancer risk assessment were initially developed in 1986 and were finalized in 2005. After almost two decades of scientific input and progress, the final guidelines were designed to be flexible, with the ability to evolve as scientific advancement occurs. Published by the National Center for Environmental Assessment (NCEA) within the EPA’s Office of Research and Development (ORD), this document supplements the 1997 Exposure Factors Handbook with child-specific data on exposure activities and other parameters for assessing exposure to contaminants in the environment. In response to modern advances in computational and molecular biology, the EPA developed a strategic plan in 2009 to outline an approach for transforming and improving toxicity testing and risk assessment over the next 10 years. The premise of the proposed new plan is that risk assessors should consider how genes, proteins, and small molecules interact in the molecular pathways to maintain cellular function and how exposure to agents in the environment could disrupt these pathways. The strategic plan is built upon three components: (1) toxicity pathway identification and chemical screening prioritization, (2) toxicity pathway-based risk assessment, (3) institutional transition.
1.1. CANCER RISK ASSESSMENT
9
that had been practiced at EPA since 1976, where approximately 150 carcinogen risk assessments had been completed in the first eight years, occurred in 1983 when the National Research Council (NRC) of the U.S. National Academies of Science (NAS) endorsed risk assessment as a proper process and defined specific steps for hazard identification, dose–response assessment, exposure assessment, and risk characterization as the risk assessment paradigm (NRC 1983). This endorsement created wider applications of risk assessment, which rapidly expanded across all federal regulatory agencies and beyond to state agencies and international communities. The specifics of this process are described in the following section. Present-day risk assessment methodologies have an increasing emphasis on physiologically based pharmacokinetics (PBPK) or toxicokinetic models and mode of action (MOA). Such models have been developed to predict exposure levels in target tissues for a large number of agents. PBPK models are especially useful in the risk assessment context because they allow data to be extrapolated across species, dose levels, and routes of exposure.
1.1.3.
The Defining Steps in Cancer Risk Assessment
The NAS has developed risk assessment strategies and guidelines that are used by many agencies in cancer risk assessment to answer four fundamental questions: (1) Is the agent a carcinogenic hazard? (2) At what dose does the agent become a carcinogenetic hazard? (3) What is the current and expected extent of human exposure to the agent? (4) What is the estimated disease burden expected from exposure to the agent? The strategies used to answer these questions are divided into four actions (NRC 1983): • Hazard Identification. The total weight of the evidence from epidemiologic, animal, and toxicological studies is evaluated to determine the toxicity and carcinogenicity of an agent. In addition, as scientists begin to understand the process by which healthy cells transform into malignant cells, the use of mechanistic information is becoming more common in risk assessment. This may involve identifying the precursor events that may lead to increased cancer risk, as well as the specific genetic or cellular processes that occur during carcinogenesis. • Dose–Response Assessment. The toxic effect of an agent is dependent upon many factors, including the amount of agent that is ingested, the route of exposure, and the specific endpoint under evaluation. Dose–response assessments are primarily focused on determining the safe dose for human exposure for noncarcinogens or acceptable risk levels for carcinogens. Because thresholds for carcinogen activity could not be defined as had traditionally been the case for noncarcinogens, the first risk assessment guidelines at the EPA relied on a linear, nonthreshold, dose extrapolation model for placing plausible upper bounds on risk; the real risks at low doses were thought to be lower, even approaching zero. Dose–response assessments are generally conducted in animals and use empirical, physiologically based toxicokinetic,
10
CHAPTER 1 CANCER RISK ASSESSMENT
or mechanism-based dose–response modeling techniques. In contrast, safety assessments for noncarcinogens historically relied on (a) establishing a no observed effect level (NOEL) or a lowest observed adverse effect level (LOAEL) in animals and (b) reducing this level by application of various safety or uncertainty factors to arrive at a safe dose for humans. Today, there is a convergence of methods for carcinogens and noncarcinogens, at least academically, to utilize understandings of toxicokinetics and toxicodynamics to arrive at safe exposure levels. • Exposure Assessment. The fate of an agent in the environment and the extent to which humans will be exposed to the agent is determined through exposure assessment. The primary interests in exposure assessments are to determine the magnitude, frequency, and duration of the exposure. This assessment involves determining the environmental fate and transport of the agent, as well as evaluating the routes of potential exposure (i.e., inhalation in the air, ingestion in food or water, and through dermal contact). The most detailed guidance for exposure assessment is found in the EPA’s Risk Assessment Guidance for Superfund, Volume I (USEPA 1989) and the EPA’s Exposure Factors Handbook (USEPA 1997). • Risk Characterization. Using both (a) the results of the qualitative hazard identification to express the WOE that an agent poses a cancer risk and (b) the quantitative information obtained from the dose–response modeling together with the results of the exposure assessment, the risk characterization step fundamentally describes the risk associated with exposure to an agent at various levels of exposure for the circumstances of concern. The fact that there are scientific uncertainties in these steps has long been recognized. While there are no formal methods to fully characterize the uncertainties in the hazard assessment and dose–response stages (USEPA 2005), methodology and mathematical techniques exist for accounting for uncertainty (and variability) in the exposure assessment stages. Monte Carlo risk analysis modeling, for example, is a mathematical tool that can be used to describe the impact of uncertainty in a specific exposure scenario. It provides a probability distribution for each uncertainty parameter in the model and then can calculate thousands of probability scenarios. This tool allows risk assessors to model the unavoidable uncertainties that are inherent in the risk assessment process, including the occasion when conflicting expert opinions needs to be combined (Vose 1997). The NAS also defined a separate step, Risk Management, where the level of acceptable risk is established. For suspected carcinogens, an acceptable risk range of one in a million to one in ten thousand has been chosen by the EPA and most other public health agencies as the acceptable risk range for regulatory purposes, with risk becoming less acceptable as it rises above the presumptively safe level of one in ten thousand (40_CFR_Part_61 1989). In addition, the results of any necessary risk–benefit analyses and scientific uncertainty analyses, as well as other social and economic issues as defined by the enabling statute, may be considered at this stage in the process.
1.1. CANCER RISK ASSESSMENT
1.1.4.
11
The Mode of Action (MOA)
As described in the Guidelines for Cancer Risk Assessment (USEPA 2005, pp. 1–10), the MOA is defined as “a sequence of key events and processes, starting with interaction of an agent with a cell, proceeding through operational and anatomical changes, and resulting in cancer formation.” In fact, the severity of effect associated with exposure to an agent largely depends on the interaction between the biology of the organism and the chemical properties of the specific agent (USEPA 2005). In terms of cancer risk assessment, theoretically the potential carcinogen effect of an agent can be identified through modes of action that influence mutagenicity, mitogenesis, inhibition of cell death, cytoxicology, and immune function (USEPA 2005). Conclusions about the MOA for a particular agent are based on the following questions (USEPA 2005): (1) Do animal tests sufficiently support the hypothesized MOA? (2) If the MOA is supported by animal models, is the same action relevant to humans? (3) Are there specific populations or life stages in which humans are more vulnerable to the MOA? This information is included in the final risk assessment narrative that summarizes the total weight of the evidence regarding the potential carcinogenicity of an agent. Because the MOA is based on physical, chemical, and biological processes, it is possible for an agent to have more than one MOA at different sites within the body. This makes it impossible to generalize the results obtained for one endpoint to other sites within the body. Information on the MOA often includes tumor data in humans, tumor data in animals and observations from in vitro test systems, and the structural analogue of the agent (USEPA 2005). As with all components of risk assessment, establishing the MOA of an agent can only be defined with confidence where complete data packages, rather than generic assessments or general knowledge of the agent, provide the foundations. When determining if the MOA observed in animal models is relevant to humans, risk assessors must rely on many sources of information including consideration of the tumor type, the number of studies conducted at each site, and the subgroups evaluated (gender, species, etc.), the metabolic activation and detoxification process observed in the animal model and in humans, the route of exposure, the dose, and the effect of dose and time on the progression of the tumor (see Chapter 13) (USEPA 2005). Only rarely are complete data sets available for defining the MOA. Most often the available information can provide only partial certainty about the MOA and its contribution to the WOE evaluation.
1.1.5.
Accounting for Scientific Uncertainty
One of the greatest challenges of risk assessment is to account for and manage the scientific uncertainty associated with each step in the assessment process. Uncertainty is an unavoidable consequence of evaluating the fate of an agent in our dynamic environment and complex human systems. Sources of uncertainty in assessing the carcinogenicity of an agent include: (1) the parameter values resulting from data that are limited or inadequate, (2) the parameter modeling caused by inherent limitation
12
CHAPTER 1 CANCER RISK ASSESSMENT
in the models that are used to evaluate exposures and outcomes, and (3) the completeness of the assessment because of the often infeasible task of exhaustively evaluating all possible components of risk (USEPA 1997). In addition, there is uncertainty associated with applying the results of laboratory animal studies to humans (i.e., interspecies extrapolation), estimating the risk of low-dose ambient exposures from high-dose animal studies (i.e., dose extrapolation), and accounting for the needs of susceptible populations (i.e., intraspecies extrapolation). Given these intrinsic challenges, it may be impossible to guarantee that the best outcome identified in the risk assessment process will actually occur; however, it is imperative that public health decisions are made despite these uncertainties. The consequence of not doing so would be paralysis of the public health and regulatory systems (Bean 1988).
1.2. THE WEIGHT OF EVIDENCE (WOE) FOR DETERMINING CARCINOGENICITY 1.2.1.
Epidemiologic Studies
Results from well-conducted epidemiologic studies provide the strongest weight of evidence (WOE) in cancer risk assessment. Epidemiology is the science of understanding the distribution of disease among humans and the factors that increase or decrease the risk of disease incidence (see Chapter 15). Because epidemiologic studies always measure an exposure (i.e., to a toxic agent) and an outcome (i.e., a specific cancer type), they are of great value to the cancer risk assessment process. Nevertheless, most observations in human populations have occurred when populations have been inadvertently exposed at high levels, above those commonly experienced in the environment. Epidemiologic studies are conducted in humans; therefore there are no issues related to species-to-species variation; however, other factors must be considered when estimating how the carcinogen potential of an agent may change when exposures are far lower or when population circumstances are at issue—for example, when lifestyle factors of the individual or population are concurrently assessed. The best evidence comes from well-conducted epidemiologic studies that are sufficiently powered to test a specific hypothesis and are backed up by confirmatory animal studies. However, well-conducted epidemiology studies are available for only a limited number of substances and often have limited uses because of difficulties involved in interpretation. Unlike animal studies that are conducted in a controlled setting within the laboratory, epidemiologic studies seek to evaluate humans in their natural environments. This is both advantageous and challenging for the risk assessment process. Well-conducted epidemiologic studies will often have many of the following attributes (USEPA 2005): The objectives and the hypothesis are clearly stated, the people included in the study have been properly selected, the exposure has been characterized, the length of the study is long enough to ensure adequate time for the disease to occur, design flaws that may bias the results have been identified and minimized, factors that may confound the relationship between the exposure and the outcome have been properly accounted for, enough people have been enrolled in the study
1.2. THE WEIGHT OF EVIDENCE (WOE) FOR DETERMINING CARCINOGENICITY
13
to detect the desired measure of effect, the data have been collected and analyzed using appropriate methods, and the results have been clearly documented. Because it is possible for one or more of these factors to be inadequate, epidemiologic studies that show no association between exposure to an agent and a cancer outcome do not prove that an agent has no carcinogenic potential. Therefore, the limitations of epidemiologic studies that are used in the risk assessment process must be identified and considered. The types of epidemiologic studies used by risk assessors include case–control studies, cohort studies, descriptive epidemiologic studies, and case reports: • Case–control studies enroll people who have the disease (i.e., cases) and people who do not have the disease (i.e., controls) and then look retrospectively to assess the differences in exposure between the two groups. It is possible to determine causality from a well-conducted case–control study; overall evidence of causality is judged as a WOE that takes account of all qualified epidemiologic studies. • Cohort studies enroll people who have been exposed to the agent of interest and people who have not been exposed to the agent, and then they follow the two groups through time to see which group (if either) has a higher incidence of disease. It is possible to determine causality from a well-conducted cohort study; overall evidence of causality is judged as a WOE that takes account of all qualified epidemiologic studies. • Descriptive epidemiologic studies do not have a temporal component like case–control or cohort studies. Rather, this type of study evaluates factors that may influence the incidence of a disease, such as demographic or socioeconomic characteristics. It is not possible to determine causality from a descriptive epidemiologic study. Rather, this type of study is often used to generate a hypothesis that can be tested in case–control or cohort studies. • Case reports are used to describe specific events or outcomes that occurred in a small number of people. It is not possible to determine causality from case reports, but they are useful for identifying unique events, such as the effects of a unique exposure or the incidence of an unusual tumor and for generating hypotheses that may be tested in follow-up, appropriately designed studies. The premise of epidemiology is to determine if there is an association between an exposure and an outcome. However, the goal of risk assessment is to determine if the WOE from all human studies establishes that the agent is known to cause the outcome. In 1965, Sir Bradford Hill developed a list of criteria that is used to help scientists and epidemiologists assess whether the relationship between an exposure and an outcome in epidemiological studies is causal (Hill 1965). Meeting each criterion does not provide a definitive determination of causation, but it does provide substantial information that can be used when the weight of the evidence is evaluated. In addition, the Hill criteria are intended for use in the evaluation of human data, not the combination of human and animal data. As listed below, the EPA has slightly modified the original list that was developed by Hill so it can be used in modern-day risk assessments (USEPA 2005):
14
CHAPTER 1 CANCER RISK ASSESSMENT
1. The association is observed across many different independent studies. 2. The magnitude of the association is large. 3. There is specificity in the observed association such that one exposure leads to one outcome. (Note: This is currently believed to be the weakest of all of Hill’s criteria.) 4. The exposure precedes the outcome, which leads to a temporal relationship between the two factors. 5. There is a biological gradient that is the result of a strong correlation between the exposure and the outcome. 6. The relationship between the exposure and the outcome is biologically plausible. 7. The relationship between the exposure and the outcome is observed in animal studies or other types of studies. 8. There is experimental evidence of causation from human populations. (Note: Given the ethical boundaries associated with using humans in experiments, data from these types of studies are rarely generated.) 9. Information of the structural analogues of an agent can provide information about causality. In addition, given the complexity of the risk assessment process and the growing amount of scientific literature on this topic, the use of meta-analyses is becoming a necessary skill of risk assessors. Meta-analysis is a valuable statistical technique, in which the potential health effects of an exposure are quantitatively evaluated across the entire body of relevant epidemiologic literature. Meta-analysis differs from a qualitative review of the literature because it is data-driven rather than narrative-based. Conducting a meta-analysis can be a very time-consuming and tedious process, especially when there is a large body of literature available on a specific topic. However, there are many benefits to applying this tool to cancer risk assessment. First, because the results of epidemiologic studies are sometimes conflicting, meta-analysis allows the scientific experts to formally identify sources of heterogeneity across studies. Second, meta-analysis provides researchers with an opportunity to examine selected subgroups of studies and to determine how specific studies influence the overall trend observed in the literature at large. This is especially valuable in cancer risk assessment because factors beyond exposure to the agent may be influencing the risk of cancer. Additional uses of epidemiology information in cancer risk assessments are described in the later part of this book (Chapter 15).
1.2.2.
Animal Models
Whole-animal test models are commonly used to determine the potential carcinogenicity of an agent (see Chapter 14). Animal models provide a platform to evaluate cancer outcomes after long-term exposure to the agent at various doses, as well as to identify possible modes of action. Although epidemiologic studies are favored
1.2. THE WEIGHT OF EVIDENCE (WOE) FOR DETERMINING CARCINOGENICITY
15
because they are conducted in human population, data from animal studies are often the primary data available and do provide valuable information to the risk assessment process because they allow the relationship between the agent and the cancer to be evaluated in a highly controlled environment. In addition, because ethical considerations are different for animals from humans, it is possible to learn a great deal about the factors that influence the carcinogenicity of an agent (i.e., detrimental doses and lengths of exposure that increase the risk of tumor initiation and promotion in the chosen laboratory model). If the outcome of an animal study is the presence of an uncommon tumor type, tumors at multiple anatomical locations within the same animal, development of tumors by more than one route of entry, tumors in multiple species, tumors in both genders, progression of a preneoplastic lesion to a malignant tumor, metastatic disease, unusual tumor response, a high proportion of malignant tumors, or clear evidence of dose-related increases in tumor incidence in replicated studies, then substantial credence is given to the carcinogenic potential of an agent (USEPA 2005). On the contrary, an agent is reasonably deemed as having no carcinogenic potential if no malignancies develop from well-conducted, long-term animal studies in more than two species.
1.2.3.
Weight of the Evidence Descriptors
As part of the risk assessment process, the total weight of the evidence from the aforementioned studies is used to determine the agent’s carcinogenic potential. In an effort to maintain consistency in the assessment and reporting process, agents are typically categorized in some way. The EPA has defined categories that are very similar to categorical schemes used by the U.S. National Toxicology Program (NTP), the International Agency for Research on Cancer (IARC), and the European Union (EU) (USEPA 2005). The example from EPA is as follows. It is possible for an agent to be classified into more than one group if its association with cancer varies by dose or route of exposure. • Carcinogenic to Humans. There is strong evidence of human carcinogenicity. To meet this classification, there must be evidence of causality from epidemiologic studies. If there is not, an agent can still meet this classification if all of the following conditions are met: (1) There is strong evidence of an association but not enough evidence to show exposure to the agent causes cancer, (2) there is extensive evidence that the agent is carcinogenic to animals, (3) the MOA and precursor have been identified in animals, and (4) there is strong evidence that the key precursor events that initiate the MOA in animals also occur in humans. • Likely to Be Carcinogenic to Humans. There is strong evidence of human carcinogenicity, but the weight of the evidence is not sufficient to meet the conditions of the “Carcinogenic to Humans” category. For example, there is strong evidence to support an association between exposure to the agent and cancer, but epidemiologic causality cannot be confirmed. In this category, the agent has generally been carcinogenic to more than one species of animal.
16
CHAPTER 1 CANCER RISK ASSESSMENT
• Suggestive Evidence of Carcinogenic Potential. There is evidence to suggest that an agent is carcinogenic, but the data cannot support strong conclusions about its effect. In this category, there are weak associations (that may or may not be statistically significant) between the agent and the cancer outcome in animal or human studies. • Inadequate Information to Assess Carcinogenic Potential. Agents are categorized into this group if there are inadequate or conflicting data of cancer outcomes associated with exposure to a particular agent. • Not Likely to Be Carcinogenic to Humans. Agents are categorized into this group if there is evidence to suggest that there is no association between exposure to an agent and cancer. In some cases, an agent may be carcinogenic in animals, but the MOA is not similar in humans.
1.3.
RISK ASSESSMENT IN THE 21ST CENTURY
1.3.1. Using the Advances in Molecular and Computational Biology In 2009, the EPA released a strategic plan to use new molecular and computational biology technologies in toxicity testing and risk assessment (USEPA 2009). The goal of the strategic plan is to use knowledge about the toxicity pathway to improve how risk assessments are conducted over the next 10 years. Although the complexity of the human body is well appreciated, specific information about toxicity pathways has been lacking. As a result of scientific and technological advances, valuable information about how genes, proteins, and small molecules interact to form pathways that maintain cellular function is quickly emerging (see Part IV). Understanding the manner in which exposure to agents in the environment disrupt these pathways is of high value to the sustained public health. The goal of the strategic plan is to replace whole-animal studies with in vitro tests in human cell lines. This approach would allow the rapid evaluation of new chemicals, chemical mixtures, different exposure scenarios, and the influence of chemicals on sensitive populations. If successful, this approach will be ideal for areas where data from animal and epidemiologic studies are nearly impossible to obtain and the existing knowledge base is lacking for many substances, such as in the fields of developmental toxicology, neurotoxicology, immunotoxicity, and reproductive toxicity. In the new plan, animal models will be used for evaluating mechanisms and the MOA. The plan is built upon three components (USEPA 2009): • Chemical Screening and Prioritization. There is urgent need for the rapid and cost-efficient screening of chemicals so they can be prioritized for risk assessment. This includes chemicals that are produced in high volumes, toxicants in the air, the drinking water Contaminant Candidate list, and chemicals found at Superfund sites. • Toxicity Pathway-Based Risk Assessment. Current risk assessment strategies are challenged by issues related to species extrapolation, dose extra-
1.4. APPLICATIONS IN RISK MANAGEMENT
17
polation, and quantifying cancer risk in susceptible populations. In the new plan, disruptions in the baseline biological processes that are likely associated with toxicity pathways will be identified, and their association with adverse health effects will be measured. • Institutional Transition. Adopting a new paradigm for toxicity testing and risk assessment will require changes to the EPA’s operations, organization, and outreach. The EPA is expecting that this transition will likely require more than a decade for full implementation.
1.3.2.
Genetic Susceptibility
Carcinogenesis is a complex and multistep process that often cannot be simplified into the basic exposure–outcome matrix. The effect an agent has on cancer risk is dependent upon several factors, including, but not limited to, the nature of the individual who was exposed, the dose the individual received, and the length of the exposure. During the formal risk assessment process, it is relatively straightforward to quantify or model the dose levels and the length of exposure an individual may experience under circumstances with defined parameters. In fact, the exposure assessment process has been well informed by guidelines as well as the availability of exposure factors to be used in determining the average concentration an individual might experience over the applicable duration and frequency of exposure. However, determining the genetic factors that may influence cancer risk and then accounting for these findings during the regulation process is challenging. Furthermore, the role of background genetic factors in cancer causation may be far more important than the role of the agent in question. With the completion of the Human Genome Project (HGP) in 2003, a substantial amount of evidence came to light that illustrated the importance of genetic factors in cancer susceptibility and risk. In fact, a person’s genetic background is now considered to be a major factor in determining their risk of developing cancer. Genetic variants in key DNA repair genes and carcinogen metabolism genes have been associated with an increase in risk for some types of cancer.
1.4.
APPLICATIONS IN RISK MANAGEMENT
1.4.1. Translating Risk Assessment into Risk Management in the United States Risk management and public policy decisions related to the regulation of carcinogenic agents are largely based on quantitative risk assessments and qualitative assessments of the biomedical evidence (Anderson 1983). Risk assessment is now commonly used to set priorities, determine if there is residual risk present after the best available technologies have been implemented, balance the risks and benefits of using a carcinogenic agent, set standards and target levels of risk to protect public health, and provide information regarding the urgency of situations where populations have been inadvertently exposed to toxic agents (Anderson 1983).
18
CHAPTER 1 CANCER RISK ASSESSMENT
The determination that an agent has the potential to be labeled a suspect or known human carcinogen does not alone provide the quantitative basis for determining a safe level of exposure. As noted from the early work of the Carcinogen Assessment Group (CAG) at the EPA, there are hundreds of agents that show some evidence of carcinogenic potential; however, the relative potency of these agents has been found to vary enormously (Anderson 1983). In fact, some of the chemicals that have the strongest qualitative evidence of carcinogenicity have a relatively low potency. Consequently, risk managers must be cautious and must consider relative potency in setting quantitative standards. In the absence of the MOA, the quantification of risk has defaulted to a linear nonthreshold dose–response model to establish a public health protective level of risk. The best-defined approaches for evaluating the risk and setting a level of protective risk have been defined under the EPA programs for cleanup of hazardous waste sites. Given the protective nature of the inference judgments, the outcomes of the risk assessment process are intended to be biased toward public health protection, and consequently they are best used as plausible upper bounds on risk (USEPA 2005). The EPA has commonly used an acceptable range of risk of one in a million to one in ten thousand, becoming presumptively less acceptable as risk rises above this level. However, public health agencies across national and international boundaries may arrive at different levels of acceptable risk as a generic matter or for particular agents, depending upon the application of the precautionary principal. For risk management purposes, low risk defined by the linear nonthreshold model in association with conservatively evaluated exposure can define, with a reasonable degree of confidence, when a risk to public health is acceptable and not of concern as a causal agent of disease. However, because these approaches rely partly on science and partly on inference-based public health protective assumptions, they cannot be used to determine causality. Therefore, it is inappropriate to imply that associated levels of exposure are causally related to disease occurrence when the acceptable risk ranges used by public health agencies to quantify standards for exposure and remediation are marginally exceeded (USEPA 2008b).
1.4.2.
International Risk Management
In the United States, the science of risk assessment has evolved out of the necessity to make public health decisions in the face of scientific uncertainty. Risk assessment methodologies have been established over the past three decades, and their applications have impacted virtually every aspect of public health and environmental protection in many countries. An example of the far-reaching applications of risk assessment can be found in the World Trade Organization (WTO) Agreement on Sanitary and Phytosanitary (SPS) (Anderson and St. Hilaire 2004; Measures 1994). This agreement requires counties to either (1) adopt the harmonized international standards or (2) use standards based on risk assessment, scientific principles, and scientific evidence if they choose to adopt stricter regulations than the international standards (GATT 1947; Howse 2000; Measures 1994). The WTO provides a platform for resolving discrepancies that arise over the appropriateness of national
1.4. APPLICATIONS IN RISK MANAGEMENT
19
standards that are more restrictive than other national or international standards. As of July 2008, the WTO had 153 members (www.wto.org). In 2007 the regulation on Registration, Evaluation, Authorization, and Restriction of Chemicals (REACH) was enacted in an effort to improve the framework in which chemicals are regulated in the EU. REACH requires industry to be responsible for the assessment and management of risks that may be posed by chemicals, as well as to provide the necessary safety information to their users. The overall goal of REACH is to enhance the manner in which public health and the environment are protected from the risks that are associated with the use of synthetic chemicals. It requires that companies work together to complete the registration requirements for all substances that are made in or imported into the EU. REACH requires participation in the Substance Information Exchange Forum (SIEF), which obligates companies to share information from vertebrate studies. In addition, REACH promotes the framework of “One Substance, One Registration” (OSOR), which minimizes the administrative issues that can be associated with this type of regulation. REACH has also established parameters for submitting chemical safety reports that encourage the collection, evaluation, and dissemination of all data based on the elements of risk assessment and public health protection (Environment_ Directorate-General_of_the_European_Commission 2009). Most developed countries have developed their own guidelines and practices for risk assessment. The Society for Risk Analysis and its flagship journal, Risk Analysis: An International Journal, serve as an academic forum to share the rapidly advancing sciences in the field. Also, the importance of these sciences and their applications and development is found in the curricula of most major universities.
1.4.3.
Risk–Benefit Analysis
Determining the level of risk associated with an agent may not be the only factor that is evaluated when determining when, how, and where the agent will be used. Risk–benefit analyses may play various roles in risk management, to determine if the risk of an agent outweighs its benefits. The enabling statutory language and a variety of other social and economic factors play roles in risk–benefit analysis. Generally speaking, the risk associated with an agent will be tolerated at a higher level if the agent poses substantial benefit (and vice versa). The U.S. Food and Drug Administration (FDA), the U.S. Occupational Safety and Health Administration (OSHA), and the EPA use risk–benefit analyses as permitted by the applicable statute to determine the standard of regulation for a given agent. For example, if the contraindication for a specific type of heart medication is liver cancer in 1 per 10,000 individuals, the risk associated with its use will likely be deemed as more acceptable if the drug reduces the mortality associated with heart attack by 80% than if it reduces mortality associated with heart attack by only 10%. The EPA’s regulation of pesticides is governed by the Federal Insecticide, Fungicide, and Rodenticide Act (FIRFA). Because there are public health benefits associated with controlling pests as well as risk associated with the chemicals used for this purpose, FIRFA requires that the EPA balance the risk and benefits of an agent when determining how it will be regulated. Resulting decisions include
20
CHAPTER 1 CANCER RISK ASSESSMENT
quantifying the risk of disease in the general population that is associated with exposure to the agent after normal use, the risk of disease experienced by the applicators of the agent, and comparative risk for a substitute agent, if available. The challenge of risk–benefit analyses is to ensure that all costs are accounted for at the social and environmental levels. In addition, one must consider risks and benefits at both the individual and population levels. Certainly, the level of risk that a person is willing to accept is a private and personal decision.
1.4.4.
Risk Acceptance and Risk Communication
Information obtained from risk assessments is used to aid public health officials in developing management decisions. However, the public will often view the risks associated with an agent differently than will the scientific experts, even after costly and time-consuming risk assessment efforts have been implemented. These discrepancies may be attributable to difference in how the public and scientific communities define risk, or they may stem from the fundamental lack of trust the public has toward the risk assessment process (Slovic 1991). Regardless, risk perception is an important topic that invariably must be considered before the implementation of regulations or public health management decisions. The manner in which an individual or different cultures perceive risk is often influenced by demographic, psychological, social, or political factors (Slovic 1991). The perception of risk can vary between and within individuals, such that two people may perceive the risk of the same agent differently, and a single person may view the risk of an agent differently depending on the current events in their life. Research in this area has consistently revealed many issues that are known to affect how risk is perceived, including (Asante-Duah 2002b): Are exposures to the risk factor voluntary or involuntary? Are the potential or known effects of exposure to the risk factor immediate or delayed? Is the risk factor natural or manmade? Can the risk factor be controlled? If it is controllable, how does the individual perceive their control over the risk factor? Is the type of risk factor new to the individual or are they familiar with it? Are there benefits associated with the risk factor? Are the consequences of exposure to the risk factor manageable or catastrophic? Is the individual exposed to similar risk factors? Are the effects of the risk factor reversible? Are there alternatives to the risk factor? Does the individual view the distribution of the risk factor as equitable within the population? Is exposure to the risk factor continuous or intermittent? Are the consequences associated with exposure to the risk factor tangible? Understanding and considering these issues is a challenging but essential component of risk management. However, effective risk communication is central to the successful implementation and acceptance of management actions. Risk communication often takes shape in the form of written communication (i.e., newsletters, public notices, warning labels) or verbal communication (i.e., focus groups, public meetings, workshops) (Asante-Duah 2002a). In terms of cancer risk assessment, effective risk communication strategies include, but are not limited to: involving all stakeholders and the public early in the decision-making process; taking the necessary steps to ensure that there is a two-way dialogue between the scientific experts
REFERENCES
21
and the interested parties; anticipating and preparing for the mitigation of controversy; delivering clear, honest, and factual information about the risk factors; and implementing a system to evaluate how all parties perceived the risk communication efforts (Asante-Duah 2002a). The precautionary nature of risk management decisions made by public health authorities can approach a zero risk tolerance that is not based on the outcome of the risk assessment process or the certainty of the data that underlie the assessment process but rather on social and political influences. The original purpose of risk assessment was to separate important from less important risks and provide a basis for making decisions to protect the public health. With the adoption of risk assessment and risk management as a process for making public health decisions, the concept of achieving zero risk for suspect carcinogens was abandoned as a workable, achievable policy. The important role of risk assessment is to inform the public health decision process so that responsible decisions in the interest of public health can be made. Extreme application of the precautionary principle, whether motivated by public expectations or regulatory desire to achieve ever lower risk, can lead to a virtual zero tolerance policy; it is the role of risk assessment founded on scientific principles to advise the reasonableness of these policy decisions.
REFERENCES Anonymous (1976). Editorial: Seventeen principles about cancer, or something. Lancet 13, 571–573. 40_CFR_Part_61 (1989). National Emissions Standards for Hazardous Air Pollutants; Benzene Emissions from Maleic Anhydride Plants, Ethylbenzene/Styrene Plants, Benzene Storage Vessels, Benzene Equipment Leaks, and Coke By-Product Recovery Plants. Albert, R. E., Train, R. E., and Anderson, E. L. (1977). Rationale developed by the environmental protection agency for the assessment of carcinogenic risk. J Natl Cancer Inst 58, 1537–1541. Anderson, E. L. (1983). Quantitative approaches in use to assess cancer risk. Risk Anal 3, 277–295. Anderson, E. L., and St. Hilaire, C. (2004). The contrast between risk assessment and rules of evidence in the context of international trade disputes: Can the U.S. experience inform the process? Risk Anal 24, 449–459. Asante-Duah, K. (2002a). Design of public health risk managment programs. In Public Health Risk Assessment for Human Exposures to Chemicals, Kluwer Academic Publishers, London, pp. 237–256. Asante-Duah, K. (2002b). Principles and concepts in risk assessment. In Public Health Risk Assessment for Human Exposures to Chemicals, Kluwer Academic Publishers, London, pp. 43–70. Bean, M. (1988). Speaking of risk. ASCE Civil Eng 589, 59–61. Environment_Directorate-General_of_the_European_Commission (2009). What is REACH? Vol. 2009. European Commission. GATT (1947). General Agreement of Tariffs and Trade. Art X Oct 30. Hill, A. (1965). The environment of disease: Association or causation? Proc R Soc Med 58, 295–300. Howse, R. (2000). Democracy, science and free trade risk regulation on trial at the World Trade Organization. Michigan Law Rev 98, 23–29. IRLG (1979a). Hazardous Substances Summary and Full Development Plan. United States. Interagency Regulatory Liaison Group, Washington, D.C. IRLG (1979b). Publications on Toxic Substances: United States. Interagency Regulatory Liaison Group, Washington, D.C. IRLG (1979c). Scientific basis for the identification of potential carcinogens and estimation of risk. J Natl Cancer Inst 63, 243–268.
22
CHAPTER 1 CANCER RISK ASSESSMENT
Kuzmack, A. M., and McGaughy, R. E. (1975). Quantitative Risk Assessment for Community Exposure to Vinyl Chloride. U.S. Environmental Protection Agency, Washington, D.C. Measures, S. A. P. (1994). Marrakesh Agreement Establishing the World Trade Organization. Art 7 and Annex B. Reprinted from H.R. Doc. No. 103–316 at 69–81. NCI (2007a). SEER cancer statistics review 1975–2004. In Lifetime Risk (Percent) of Being Diagnosed with Cancer by Site and Race/Ethnicity: Males, 17 SEER Areas, 2002–2004 (Table I-15) and Females, 17 SEER Areas, 2002–2004 (Table I-16), National Cancer Institute, ed. NCI (2007b). SEER cancer statistics review 1975–2004. In Lifetime Risk (Percent) of Dying from Cancer by Site and Race/Ethnicity: Males, Total U.S., 2002–2004 (Table I-18) and Females, Total U.S., 2002–2004 (Table I-19), National Cancer Institute, ed. NCI (2008). SEER Cancer Statistics Review 1975–2005, Ries, L. A. G., Melbert, D., Krapcho, M., Stinchcomb, D. G., Howlader, N., Horner, M. J., Mariotto, A., Miller, B. A., Feuer, E. J., Altekruse, S. F., Lewis, D. R., Clegg, L., Eisner, M. P., Reichman, M., and E, B. K., eds., National Cancer Institute, Bethesda, MD. NRC (1983). Risk Assessment in the Federal Government: Managing the Process, National Academy Press, Washington, DC. OSTP (1985). Chemical carcinogens: A review of the science and its associated principles. Fed Reg 50, 10371–10442. Slovic, P. (1991). Risk perception and trust. In Fundamentals of Risk Analysis and Risk Management, Molak, V., ed., Lewis Publishers New York, pp. 233–245. USEPA (1972). Respondents brief in support of proposed findings, conclusions, and order at 63–64. In Re: Stevens Industries Inc. et al. Consolidated DDT hearings (5 April 1972). USEPA (1975). Respondents motion to determine whether or not the registration of mirex should be cancelled or amended. Attachment A (9 September 1975). USEPA (1976). Interim procedures and guidelines for health risk and economic impact assessments of suspected carcinogens. Fed Reg 41, 21402–21405. USEPA (1984). Risk Assessment and Management: Framework for Decisionmaking, EPA/600/9–85/002, Washington D.C. USEPA (1986a). Guidelines for Carcinogen Risk Assessment, US Environmental Protection Agency, Risk Assessment Forum Washington, D.C. USEPA (1986b). The risk assessment guidelines of 1986a. Fed Reg 51, 33992–34005. USEPA (1989). Risk Assessment Guidance for Superfund, Vol. I: Human Health Evaluation Manual (Part A), US EPA Office of Emergency and Remedial Response. USEPA (1996). Proposed guidelines for carcinogen risk assessment. Fed Reg 61, 17960–18011. USEPA (1997). Exposure Factors Handbook, US EPA Office of Research and Development/National Center for Environmental Assessment. USEPA (2002). OSWER Draft Guidance for Evaluating the Vapor Intrusion to Indoor Air Pathway from Groundwater and Soils (Subsurface Vapor Intrusion Guidance), US EPA Office of Solid Waste and Emergency Response. USEPA (2003). World Trade Center Indoor Environment Assessment: Selecting Contaminants of Potential Concern and Setting Health-Based Benchmarks. Contaminants of Potential Concern (COPC), Committee of the World Trade Center Indoor Air Task Force Working Group. USEPA (2005). Guidelines for Cancer Risk Assessment, US Environmental Protection Agency, Risk Assessment Forum, Washington, D.C. USEPA (2008a). Child-Specific Exposure Factors Handbook, US EPA Office of Research and Development/National Center for Environmental Assessment. USEPA (2008b). IRIS Limitations, US Environmental Protection Agency. USEPA (2009). The U.S. Environmental Protection Agency’s Strategic Plan for Evaluating the Toxicity of Chemicals, EPA100/K-09/001, Washington, D.C. Vose, D. (1997). Monte Carlo risk analysis modeling. In Fundamentals of Risk Analysis and Risk Management, Molak, V., ed., Lewis Publishers, New York.
CH A P TE R
2
SCIENCE POLICY AND CANCER RISK ASSESSMENT Gary E. Marchant
2.1.
INTRODUCTION
Cancer risk assessment is primarily a scientific undertaking, but as recognized in 1983 with the publication of the U.S. National Research Council’s (NRC’s) “Red Book” on risk assessment in the federal government (NRC 1983), policy inputs are necessary to bridge the uncertainties and assumptions that are inherent in risk assessment science. As cancer risk assessments are increasingly used to support regulatory decisions with substantial real-world health and economic consequences, the policy inputs into risk assessment become more critical, scrutinized, and contested. Of course, risk management decisions that often utilize risk assessments, such as determining an acceptable level of risk, are also laden with policy issues, but those are beyond the scope of this chapter, which focuses on policy issues relating to how risk assessments are conducted, not on the related issue of how they are used. Policy inputs into risk assessment generally seek to achieve one or more of the following goals: (i) to ensure that risk assessments are scientifically credible and robust, given the inherent uncertainties in risk assessment; (ii) to support a particular policy goal or outcome, such as ensuring greater protection of human health or avoiding inefficient or unwarranted regulatory burdens or liabilities; or (iii) to make risk assessments more efficient, timely, and legally defensible. Many science policy controversies related to risk assessment focus on specific, narrow questions, such as whether to use animal or human data, the shape of the dose-response curve, how data on mechanism or mode of action should affect the risk assessment, and how to treat susceptible subpopulations. Many of these specific questions are discussed elsewhere in this volume. The analysis here, while touching on many of these specific questions as examples, will instead emphasize the broader structural and institutional aspects of how science policy issues affect risk assessment. These broader issues include: • Use of risk assessment in regulatory decision-making • Role of risk assessment guidelines
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
23
24
CHAPTER 2 SCIENCE POLICY AND CANCER RISK ASSESSMENT
• Data quality requirements • Types of data that can be used in risk assessment • Application of “conservative” assumptions and precaution In discussing these issues, this chapter will draw primarily on U.S. examples and experience. Although all industrial nations have confronted similar issues in their risk assessment and regulatory programs, the United States has generally had the most advanced and express policies for carcinogen risk assessment, although other nations are now more openly addressing similar issues (Millstone et al. 2008).
2.2. USE OF RISK ASSESSMENT IN REGULATORY DECISION-MAKING Regulatory statutes are generally silent on whether regulatory agencies can, or must, use risk assessment in making regulatory decisions. Rather, the role of risk assessment must be inferred from the statutory regime and interpretation by the courts. Three general approaches are specified by statutes for setting regulatory standards: (i) “acceptable” risk; (ii) cost–benefit analysis; and (iii) feasibility (or best available technology). The first two of these approaches are premised on risk assessment: The first (acceptability) involves identifying risks and then determining what types and levels of risk are acceptable, while the second (cost–benefit analysis) weighs the benefits of reducing risks against the costs of those reductions. Both of these approaches require identification, if not quantification, of risks; thus both approaches presumably permit and arguably require risk assessment. The third approach (feasibility) requires the regulatory agency to reduce risks as low as technologically (or perhaps economically) feasible, and it appears to be oblivious to what the actual risks are. Statutory programs utilizing this approach would therefore presumably not require risk assessment and may even prohibit such consideration. In recent years, there has been a trend in the United States away from riskbased regulatory approaches and toward feasibility or “best available technology” approaches that are not based on risk assessment (Wagner 2000). This trend is largely due to the inherent uncertainties and controversies over risk assessment. The sentiment underlying this trend was expressed by U.S. Senator David Durenberger, who, during the 1990 reauthorization of the U.S. Clean Air Act (CAA) in which Congress abandoned the previous risk-based approach for regulating hazardous air pollutants in favor of a technology-based Maximum Available Control Technology (MACT) requirement, stated (Durenberger 1990): “I’d be glad to declare risk assessment dead.” Courts have enforced the distinction between regulatory programs that permit (or require) risk assessment from those which prohibit reliance on risk assessment. In the seminal 1980 case reviewing the benzene standard promulgated by the U.S. Occupational Safety and Health Administration (OSHA), the U.S. Supreme Court held that the U.S. OSHA must use risk assessment to demonstrate that workers were exposed to a “significant risk” before taking regulatory action (IUD 1980). The U.S. OSHA had proposed to reduce exposures to the lowest levels feasible after
2.3. ROLE OF RISK ASSESSMENT GUIDELINES
25
determining that risk assessment was too uncertain to provide a reliable basis for regulatory action, but the Supreme Court held that the extraordinary power that a regulatory agency wielded could only be exercised after a threshold finding of significant risk. Specifically, the Court held that “the risk from a toxic substance must be quantified sufficiently to enable the Secretary to characterize it as significant in an understandable way” (IUD 1980, p. 646). Conversely, the courts have also overturned agencies for relying on risk assessment in statutory programs that do not authorize such assessments, in particular under statutes mandating a best available technology approach. For example, the D.C. Circuit recently overturned a regulation promulgated by the U.S. Environmental Protection Agency (EPA) that provided a “low-risk” exemption from hazardous air pollutant MACT standards for sources that could demonstrate with a risk assessment that their emissions would impose a maximum individual risk of less than 1 in one million (NRDC 2007). The Court held that such a risk-based approach was impermissible because Congress mandated a technology-based approach that left no room for standards based on risk assessment.
2.3.
ROLE OF RISK ASSESSMENT GUIDELINES
Every risk assessment involves a complex mix of data sets, toxicological methods, models, data gaps, uncertainties, and assumptions. Given this complexity, every risk assessment is, on the whole, unique. At the same time, there are common or at least similar issues that arise over and over again in different risk assessments. Risk assessments conducted on a truly individualized basis, in which the appropriate assumptions and methods to apply are determined de novo based on scientific judgment in light of all available data for that particular risk assessment, would be very difficult, if not impossible, for regulatory agencies. Evaluating all the data and then selecting the appropriate assumptions, methods, and data to apply in a risk assessment is a very resource-intensive undertaking, and it is one that will consume much time and resources and inevitably invite scientific disagreement and controversy (Flamm 1989). To address this tension, some regulatory agencies have developed risk assessment guidelines to provide efficiency, consistency, and predictability in cancer risk assessment. The adoption of these guidelines was heavily influenced by the U.S. NRC’s “Red Book,” which recommended that regulatory agencies adopt risk assessment guidelines containing “inference options” to bridge data or theoretical gaps in the risk assessment process (NRC 1983). The inference options would apply as defaults in the absence of adequate data or theoretical information needed in risk assessment. In response to this report, the U.S. EPA issued its initial carcinogen risk assessment guidelines in 1986 (EPA 1986), which consisted largely of a series of intentionally vague “generalities” about the cancer process and the methods to be used for calculating cancer risk (Albert 1994). These generalities, referred to as “default options,” established a presumptive set of assumptions that were to be applied to address the inherent uncertainties in risk assessment. The default options in the U.S. EPA guidelines were not intended to be inflexible, binding rules, but
26
CHAPTER 2 SCIENCE POLICY AND CANCER RISK ASSESSMENT
rather presumptive principles that would apply in the absence of sufficient data to establish an alternative assumption. “Convincing proof” was required to depart from a default option in the guidelines. In contrast, OSHA adopted cancer guidelines as binding rules, and other agencies (including regulatory agencies in most other countries) have not adopted risk assessment guidelines (NRC 2007b). Some of the “most important” default options included in the U.S. EPA’s 1986 carcinogen guidelines (NRC 1994) included the following: • Laboratory animals are assumed to be a valid surrogate for humans in assessing risk; thus, positive cancer results in animal bioassays are taken as evidence of human carcinogenicity. • Humans are assumed to be as sensitive as the most sensitive animal species, and strain or sex is evaluated in an appropriately designed animal bioassay. • Benign tumors are assumed to be as significant as malignant tumors. • The dose–response curve of humans to potential carcinogens is assumed to be linear all the way to the zero exposure levels with no threshold. • A given intake of a substance is assumed to have the same effect regardless of the rate or route of intake. • Individual substances are assumed to exert their effect independently of other substances to which the body is exposed. Most of these default options selected by the U.S. EPA were deliberately chosen to be “conservative,” in that they were intended to estimate the plausible upper bound of actual risk. The adoption of standardized default options in the form of explicit guidelines provided many benefits to the Agency. The highly publicized issuance of the guidelines temporarily quelled much of the brewing controversy about the credibility of the U.S. EPA’s risk assessment practices (Albert 1994). The guidelines helped to sanitize risk assessment by removing the suspicion that the U.S. EPA would manipulate risk assessment principles on a case-by-case basis to support predetermined regulatory outcomes (Goldstein 1989). Risk assessment guidelines also encouraged consistency in risk assessment approach and procedure across the broad array of the U.S. EPA regulatory programs that use risk assessment. Risk assessment guidelines also furthered the objective of efficiency, by sparing the Agency the need to revisit the same controversial issues in each successive rulemaking proceeding. Risk assessment guidelines were also justified on the basis that they would provide regulated businesses greater predictability and certainty about regulatory requirements. The administrative convenience and consistency provided by risk assessment guidelines have, however, come at the expense of flexibility and change in response to emerging science. A 1994 study of the U.S. EPA’s risk assessment practices by the NRC, required by the 1990 CAA Amendments, endorsed the U.S. EPA’s use of default options in its carcinogen risk assessment guidelines, but criticized the agency for applying the guidelines too rigidly (NRC 1994). The U.S. NRC report found that the U.S. EPA rarely, if ever, departed from the defaults in the guidelines, had failed to explain the basis for each default in its guidelines, and had developed no clear
2.3. ROLE OF RISK ASSESSMENT GUIDELINES
27
criteria for departing from default options. The report recommended that the U.S. EPA adopt a more “structured approach” by articulating an explicit set of guidelines or principles for deciding when and how to depart from the default options (NRC 1994, p. 91). Such criteria are needed, according to the U.S. NRC report, “to lessen the possibility of ad hoc, undocumented departures from default options that would undercut the scientific credibility of the agency’s risk assessments” (NRC 1994, p. 105). In 1996, two years after the U.S. NRC report, the U.S. EPA undertook a comprehensive rewrite of its carcinogenic risk assessment guidelines that were finalized almost a decade later in 2005 (EPA 2005a). The revised guidelines incorporated a much more case-by-case approach that considers all relevant evidence. As explained by the EPA, the revised guidelines “are intended to be both explicit and more flexible than in the past concerning the basis for making departures from defaults, recognizing that expert judgment and peer review are essential elements of the process” (EPA 1996). The revised guidelines incorporate “a weight-of-the-evidence approach that considers all relevant data in reaching conclusions about the potential human carcinogenicity of an agent” (Wiltse and Dellarco 1996). Controversial risk assessment policy issues often come to the forefront in the development of risk assessment guidelines. For example, a major controversy in the development of the U.S. EPA’s 2005 revised carcinogen risk assessment guidelines was how to deal with susceptible subpopulations. The U.S. EPA initially took the position that because the guidelines generally apply conservative defaults, they will provide a margin of safety that will protect susceptible subgroups. Environmental organizations and the U.S. EPA’s own Science Advisory Board were critical of this approach, and the controversy was responsible for much of the delay between the 1996 proposal for the revised guidelines and the finalization of those guidelines almost 10 years later in 2005. The U.S. EPA addressed this issue of susceptible subpopulations in part in the final guidelines by publishing an accompanying Supplemental Guidance for Assessing Susceptibility from Early-Life Exposure to Carcinogens that proposed to apply an additional age-specific safety factor to account for the increased susceptibility of children, but only for carcinogens that exhibit a mutagenic mode of action (EPA 2005b). The U.S. EPA makes explicitly clear that its risk assessment guidelines are not binding rules, and it is free to depart from the guidelines as necessary (EPA 2005a). However, by adhering to its guidelines, the U.S. EPA not only provides some consistency and predictability to its risk assessments, but also provides some immunity in legal challenges to its risk assessments. For example, adoption of risk assessment guidelines provides a standard by which reviewing courts can review the reasonableness of an agency’s risk assessments. As one federal appeals court noted, “EPA’s specific enunciation of its underlying analytical principles, derived from its experience in the area, yields meaningful notice and dialogue, enhances the administrative process and furthers reasoned agency decision making” (EDF 1976). If the U.S. EPA complies with its own guidelines, the Agency’s decision is likely to be upheld (Ausimont 1988; NRDC 1987). Conversely, if the U.S. EPA violates its own guidelines without a reasoned explanation, its action is susceptible to judicial invalidation (CCC 2000).
28
2.4.
CHAPTER 2 SCIENCE POLICY AND CANCER RISK ASSESSMENT
DATA QUALITY REQUIREMENTS
The quality, adequacy, and consistency of risk assessments are promoted by both internal and external oversight mechanisms. The risk assessment guidelines discussed in the previous section are an important internal mechanism. Congress, the Executive branch, and courts all impose some external oversight to ensure that agency risk assessments meet minimum requirements for data quality. Although, as discussed above, Congress has not provided specific requirements relating to risk assessment in most regulatory statutes, section 300g-1(b)(3)(A) of the U.S. Safe Drinking Water Act specifies that the U.S. EPA is to use “the best available, peer-reviewed science and supporting studies conducted in accordance with sound and objective scientific practices” (SDWA 1996). The D.C. Circuit relied on this statutory language to reject a U.S. EPA regulation for chlorinated byproducts in drinking water, which applied a linear dose–response model, even though the “best available” scientific evidence suggested a nonlinear relationship (CCC 2000). In 2000, the U.S. Congress enacted the Data Quality Act (DQA; sometimes also known as the Information Quality Act) requiring the U.S. Office of Management and Budget (OMB) to issue guidance for “ensuring and maximizing the quality, objectivity, utility, and integrity of information … disseminated by Federal agencies” (DQA 2000). The U.S. OMB subsequently issued a directive to federal agencies to “adopt a basic standard of quality (including objectivity, utility, and integrity) as a performance goal,” and to help agencies in this endeavor also provided a model guideline describing substantive standards for information quality (OMB 2001). Each federal agency adopted its own standards for ensuring data quality based on the U.S. OMB guidance. Federal agency risk assessments are therefore subject to these data quality guidelines, and they can be challenged if they fail to meet the applicable standards. The Act requires agencies to provide a mechanism for interested parties to challenge agency actions that purportedly fail to meet the data quality standards, and so each agency provides for members of the public to petition the agency under the Act. Several risk assessment documents prepared by the U.S. EPA and other federal agencies have been challenged under these provisions to date. In some cases the agency has revised the document in response to the DQA petition, whereas in other cases the agency has rejected the petition and upheld the risk assessment document in its original form. A key factor in the application of the DQA is whether agency decisions on petitions are judicially reviewable. Although the statute is silent on judicial reviewability, the initial court cases have held that there is no right of judicial review (SI 2006), which significantly limits the force of the DQA statute. The U.S. OMB has promulgated additional measures that seek to further influence agency risk assessments, ostensibly for the purpose of enhancing the scientific credibility and validity of agency actions. In 2005, the OMB issued a bulletin mandating peer review of scientific information disseminated by the federal government (OMB 2005). This bulletin imposed stringent peer review requirements for “influential scientific information” that included “scientific assessments” such as health risk assessments, which are required to be externally peer reviewed prior to dissemination pursuant to stated criteria for “scientific integrity” and “process
2.4. DATA QUALITY REQUIREMENTS
29
integrity.” Scientific integrity was defined as issues such as “expertise and balance of the panel members; the identification of the scientific issues and clarity of the charge to the panel; the quality, focus and depth of the discussion of the issues by the panel; the rationale and supportability of the panel’s findings; and the accuracy and clarity of the panel report.” “Process integrity” was defined as issues relating to “transparency and openness, avoidance of real or perceived conflicts of interest, a workable process for public comment and involvement, and adherence to defined procedures.” The U.S. OMB also issued a highly controversial Proposed Risk Assessment Bulletin in January 2006 specifying uniform government-wide requirements for agency risk assessments with the stated objective of enhancing “the technical quality and objectivity of risk assessments prepared by federal agencies” (OMB 2006). The U.S. NRC published a highly critical review of the proposed bulletin in 2007 and concluded that it is “fundamentally flawed and should be withdrawn.” (NRC 2007b). One key criticism was that the “one size fits all” approach of the draft guidance fails to accommodate the significant differences between agencies in the types and goals of risk assessments. In response, the U.S. OMB decided to not issue its risk assessment bulletin in final form and instead issued an “Updated Principles for Risk Analysis,” which revised an earlier 1995 document issued to federal agencies on general risk analysis principles (OMB 2007). In addition to the legislative and executive branches, courts also exercise some oversight over the quality of risk assessments. Courts are generally at their most deferential in reviewing risk assessments and other science-based decisions. The U.S. Supreme Court has instructed that “when examining this kind of scientific determination, as opposed to simple findings of fact, a reviewing court must generally be at its most deferential” (BGEC 1983). As another court opinion acknowledged, “substantive review of mathematical and scientific evidence by technically illiterate judges is dangerously unreliable” (Ethyl 1976). Notwithstanding this general deferential approach to scientific risk assessments, reviewing courts will occasionally overturn agency decisions on the ground that they are based on risk assessments that are outdated or otherwise flawed. For example, the U.S. EPA’s proposal to list methylene diphenyl diisocyanate (MDI) as a “high-risk” pollutant under the U.S. CAA was rejected by a reviewing court because the agency applied a generic air dispersion model to calculate human exposure that had “no rational relationship to the known properties of MDI” (CMA 1994). The Agency’s model assumed that MDI will behave as a gas under the relevant conditions, whereas in fact the undisputed evidence before the Agency showed that MDI would be a solid at the relevant temperature. Courts view their role as ensuring that regulatory agency practices decisions “must remain attuned to our rapidly expanding knowledge and technology” (EDF 1978) and must “accurately reflect the latest scientific knowledge useful in indicating the kind and extent of all identifiable effects on public health” (LIA 1980). The controversy over the shape of the dose–response curve in risk assessment is an example that shows the influential, yet somewhat sporadic and unpredictable, role of the courts in risk assessment policy issues. In the regulatory context, agencies such as the U.S. EPA have traditionally applied a linear, no-threshold dose–response
30
CHAPTER 2 SCIENCE POLICY AND CANCER RISK ASSESSMENT
model [specifically, the linearized multistage (LMS) model] as a conservative default assumption in carcinogen risk assessment. While the U.S. EPA’s new 2005 carcinogen risk assessment guidelines signal a more flexible, data-based approach to the selection of dose–response model, most of the Agency’s risk assessments reviewed by courts to date have employed the linear, no-threshold model (EPA 2005a). Most judicial decisions reviewing the use of linear dose–response models in the regulatory context have upheld the Agency’s reliance on such models (IFI 1992; PCHRG 1986). A few court decisions, however, have been more skeptical of the linear model. For example, the U.S. EPA’s use of the linear, no-threshold model in its risk assessment for drinking water chlorinated byproducts was rejected by the court because it was contrary to evidence suggesting a nonlinear model that had been accepted by both the U.S. EPA and its Science Advisory Board (CCC 2000). On the other hand, the U.S. OSHA’s departure from the linear, no-threshold model in its formaldehyde risk assessment was likewise rejected by the court (IU 1989). The court held that the U.S. OSHA had improperly used the maximum likelihood estimate (MLE) rather than the upper confidence limit (UCL) to calculate risk, and the UCL but not the MLE model was consistent with a linear dose–response assumption. The court held that the U.S. OSHA had failed to justify its departure from its traditional linear, no-threshold dose–response assumption. Judicial decisions in nonregulatory contexts such as toxic tort and product liability suits are likewise inconsistent in their consideration of the linear, no threshold model. As in the regulatory context, most cases find no problem with an expert’s reliance on a risk assessment using the linear model. In a handful of cases, however, the court rejects reliance on a linear dose–response assumption. For example, one court in addressing the cancer risks from a low concentration of benzene in Perrier® held that “there is no scientific evidence that the linear no-safe threshold analysis is an acceptable scientific technique used by experts in determining causation in an individual instance” (Sutera 1997). Another court decision concluded that “[t]he linear non-threshold model cannot be falsified, nor can it be validated. To the extent that it has been subjected to peer review and publication, it has been rejected by the overwhelming majority of the scientific community. It has no known or potential rate of error. It is merely an hypothesis” (Whiting 1995). The inconsistency and unpredictability of judicial review of risk assessments adds an additional element of uncertainty into the risk assessment process.
2.5.
TYPES OF DATA USED IN RISK ASSESSMENT
Another important policy issue for risk assessment is the type of data that can be used in risk assessment. The context in which the risk assessment is used will often dictate what types of data may be used. In the regulatory context, agencies such as the U.S. EPA tend to make determinations based on the “weight of evidence” that considers all available evidence, including human epidemiological and clinical data (when available), animal studies, and cellular and molecular assays (EPA 2005a). While the U.S. EPA states a preference for human data, it recognizes that
2.5. TYPES OF DATA USED IN RISK ASSESSMENT
31
human data are often not available, and thus the Agency most commonly makes regulatory decisions based on risk assessments using animal data. Courts have generally upheld this reliance on animal data, citing the preventive and prophylactic function of regulatory agencies in preventing toxic exposures and risks (IFI 1992; PCHRG 1986). Cancer risk assessments are also sometimes used in toxic tort and product liability litigation. In this context, courts express a much stronger preference for risk assessments based on human data and are more skeptical of animal studies. For example, the U.S. Supreme Court rejected the reliance of plaintiffs’ experts on animal studies showing that polychlorinated biphenyls (PCBs) can cause cancer, holding that the studies were “so dissimilar” to the human exposure and toxicity at issue in that case as to be without any value (GE 1997). This difference in the evidentiary approach of courts and agencies flows from the different institutional objectives (Allen 1996): Regulatory [agencies] … make prophylactic rules governing human exposure. This methodology results from the preventive perspective that the agencies adopt in order to reduce public exposure to harmful substances. The agencies’ threshold of proof is reasonably lower than that appropriate in tort law, which “traditionally make[s] more particularized inquiries into cause and effect” and requires a plaintiff to prove “that it is more likely than not that another individual has caused him or her harm.”
Another policy issue relating to the type of data used in risk assessment concerns the incorporation of new types of data and methods. Regulatory agencies generally require new test methods and types of data to be validated before they can be used in regulatory risk assessments. While this validation requirement has traditionally tended to be ad hoc and informal, there has been a trend in recent years toward more formal validation requirements (Balls and Fentem 1999). The U.S. Congress enacted legislation in 2000 that required a formalized and harmonized validation system for new toxicological methods relied on by federal regulatory agencies, implemented through the Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM) (Congress 2000). The statute requires that a federal agency “that requires or recommends acute or chronic toxicological testing …. shall ensure that any new or revised acute or chronic toxicity test method, including animal test methods and alternatives, is determined to be valid for its proposed use prior to requiring, recommending, or encouraging the application of such test method” (§4(c)). The European Union has likewise created the European Centre for the Validation of Alternative Methods (ECVAM) to validate new toxicological test methods, and the Organization for Economic Cooperation and Development (OECD) has also adopted formal guidelines for the validation of test methods for use in regulatory decision-making (OECD 2005). The need for formal validation of new test methods can help ensure the validity and consistency of risk assessment methods, but it also carries the risk of further slowing the adoption of new methods. As science has rapidly progressed over the past few decades in its understanding of how toxic agents cause cancer and other adverse effects, risk assessment has struggled to keep up. Despite the rapid development of risk assessment methodologies and their underlying science, regulatory
32
CHAPTER 2 SCIENCE POLICY AND CANCER RISK ASSESSMENT
agencies have been slow to incorporate the most recent scientific data and methods. This lag results from a variety of factors, including the inflexibility built into many governmental programs, legal risks created by changes in agency practices, and the tendency for most new developments in risk assessment “science” to be exculpatory by usually downgrading the magnitude or even existence of risk from particular agents. An example of the delay in accepting new data and models was the U.S. EPA’s long and torturous process for accepting data suggesting that male rat kidney cancers caused by a variety of compounds including unleaded gasoline via a mechanism involving the male rat-specific protein α2u-globulin may not be relevant to humans (EPA 1991). The acceptance of this alternative assumption involved a process lasting almost ten years, which included a peer review workshop convened by the U.S. EPA and review by different committees of the Agency’s Science Advisory Board (McClellan 1996). In response to criticisms from the U.S. NRC (NRC 1994) and others that agencies take too long and are too conservative in adopting new data and methods, there is increasingly awareness of the importance of creating incentives for risk assessment scientists to develop and use better toxicological methods. An example of this tension between adhering to established approaches and creating incentives for new types of data can be seen with the U.S. EPA’s response to new toxicogenomic data. The U.S. EPA issued an “Interim Genomics Policy” in 2002 which recognizes “that genomics will have an enormous impact on our ability to assess the risk from exposure to stressors and ultimately to improve our risk assessments” (EPA 2002). The Agency stated it would consider genomic data in risk assessment “on a case-by-case basis,” but “these data alone are insufficient as a basis for decisions” at this time. Moreover, the guidance adds that “[b]efore such information can be accepted and used, agency review will be needed to determine adequacy regarding the quality, representativeness, and reproducibility of the data.” Another recent report issued by the U.S. NRC urged EPA and other regulatory agencies to be more aggressive in supporting and utilizing genomic data, suggesting that toxicogenomic data should not be held to more rigorous standards than other types of toxicological data (NRC 2007a). The report (p. 199) recommended: “Although caution, scrutiny, and validation are required to protect against premature, inappropriate, and unethical use of toxicogenomic data in regulatory and litigation contexts, care should also be taken to ensure that a higher standard of proof is not imposed for toxicogenomic data relative to other types of toxicological data used in regulation and litigation.” In toxic tort and product liability litigation, the admissibility of new risk assessment methods and data must be approved by the trial judge before being presented to the jury. The U.S. Supreme Court announced a new standard for the admission of scientific data in 1993 in its Daubert decision (Daubert 1993). Under this new standard, federal judges must serve as a “gatekeeper” to ensure that scientific evidence is reliable and relevant, which includes an assessment of whether the evidence (i) has been empirically tested, (ii) has a known rate of error, (iii) has been peerreviewed and published, and (iv) is generally accepted within the relevant scientific field. Many state courts have adopted a similar standard, although some still apply the earlier standard on admissibility (Frye 1923), which is whether the evidence is
2.6. APPLICATION OF “CONSERVATIVE” ASSUMPTIONS AND PRECAUTION
33
“generally accepted” in the relevant field of expertise. These admissibility standards for scientific evidence are likely to present a barrier to the introduction of new risk assessment methods or data that have not yet been widely accepted.
2.6. APPLICATION OF “CONSERVATIVE” ASSUMPTIONS AND PRECAUTION Another policy issue in cancer risk assessment is the role of “conservative” (i.e., upper-bound or worst-case) assumptions in risk assessment, a long-standing controversy that has been rekindled by the recent adoption and proliferation of the precautionary principle (Marchant 2003). Risk assessment inevitably involves uncertainties, and agencies such as the U.S. EPA have traditionally sought to bridge such uncertainties with conservative assumptions that represent a plausible upperbound of risk. As the U.S. EPA explains its approach, “[o]ur risk estimates are designed to ensure that risks are not underestimated which means that a risk estimate is the upper bound on the estimated risk” (EPA 2004). The use of conservative assumptions has been criticized by some for inserting risk management policies (i.e., err on the side of safety) into the risk assessment process and also because the compounding of multiple worst-case assumptions may produce risk estimates that are implausibly high (Nichols and Zeckhauser 1988). Other experts have expressed concern that while the use of conservative assumptions may be appropriate initially when uncertainty is large, it is problematic if the initial use of such assumptions prevents revision of risk estimates when new data become available for political reasons: “One implication of the inherent conservatism in risk assessment is that the inevitable consequence of most scientific advances related to the assessment of risk for individual chemicals is to lower the calculated risk …. The worst thing that we can do is to set up a situation so that we cannot use this increased scientific information because of the political aspects of changing numbers” (Goldstein 1989). Still others argue that risk assessments are not conservative enough and that they underestimate risks because, for example, they fail to fully account for susceptible subpopulations and the synergistic effects of some toxic exposures (Finkel 1996; Latin 1988). The courts have generally been sympathetic to the use of conservative assumptions in agency risk assessments, although with some limitations. The U.S. Supreme Court, in its 1980 decision on the U.S. OSHA’s occupational health standard for benzene which ushered in the era of regulatory risk assessment by requiring a threshold showing of “significant risk,” wrote that “so long as they are supported by a body of reputable scientific thought, the Agency is free to use conservative assumptions” in calculating cancer risk (IUD 1980, p. 656). A subsequent court decision interpreted that decision to say that “OSHA may use assumptions, but only to the extent that those assumptions have some basis in reputable scientific evidence” (AFL-CIO 1992). Courts have in some cases rejected the use of conservative assumptions when those assumptions are contradicted by available data. In one case, the court held that the U.S. EPA could not ignore accurate information “at hand” on the relevant risk “in favor of blanket, highly conservative assumptions” (LIA 1994).
34
CHAPTER 2 SCIENCE POLICY AND CANCER RISK ASSESSMENT
The court added that “[w]hile the EPA ‘may err on the side of overprotection,’ it ‘may not engage in sheer guesswork.’ ” The debate about the use of conservative assumptions in risk assessment has now melded into the global debate on the precautionary principle. The precautionary principle is an ill-defined concept, adopted into the laws of the European Union and several other countries, which calls for greater precaution in controlling unknown risks, in some versions shifting the burden of proof to the proponent of a product or technology to demonstrate its safety (Marchant 2003). The United States has taken the position that the use of conservative assumptions in risk assessment and other existing regulatory protections provide a sufficient exercise of precaution, and the precautionary principle is unnecessary and ill-advised (Graham 2002). The European Union, the global leader in promoting the precautionary principle, has taken the position that the precautionary principle is a risk management tool that does not even apply to risk assessment; rather, the precautionary principle is considered only after a full scientific risk assessment (CEC 2000). Yet a third position is that the precautionary principle requires that risk assessment be revised to further incorporate precaution, such as by, for example, expanding the scope of potential harms and subjects, giving greater weight to early indications of potential harms that have not yet been demonstrated, and paying more attention to synergistic and cumulative effects of toxic exposures (Goldman 2003; Tickner 2002). Finally, a fourth position supports using the precautionary principle to replace risk assessment altogether (O’Brien 2000). Like many of the issues discussed in this chapter, this controversy is likely to rage for some time given the divergent opinions and interests, important stakes, and strong emotions at issue.
2.7.
CONCLUSION
Science policy issues and controversies underlie almost every aspect of cancer risk assessment. These policy issues are primarily a function of the scientific uncertainties inherent in risk assessment. As new scientific methods and data begin to fill in some of the data gaps and uncertainties in risk assessment, the role of policy will gradually recede, although there is no prospect of policy issues being mooted entirely in the foreseeable future. Moreover, the extent to which we substitute novel scientific data and models for preexisting policy inferences is itself an ongoing policy debate, as is the appropriate role of precaution and conservatism in risk assessment.
REFERENCES AFL-CIO (1992). American Federation of Labor and Congress of Industrial Organizations v. Occupational Safety and Health Administration, U.S. Department of Labor, 965 F.2d 962 (C.A. 11). Albert, R. E. (1994). Carcinogen risk assessment in the U.S. Environmental Protection Agency. Crit Rev Toxicol 24, 75–85. Allen (1996). Allen v. Pennsylvania Engineering Corp., 102 F.3d 194 (C.A.5 (La.)). Ausimont (1988). Ausimont USA Inc. v. E.P.A., 838 F.2d 93 (C.A.3).
REFERENCES
35
Balls, M., and Fentem, J. H. (1999). The validation and acceptance of alternatives to animal testing. Toxicol in Vitro 13, 837–846. BGEC (1983). Baltimore Gas and Elec. Co. v. Natural Resources Defense Council, Inc., 462 U.S. 87. CCC (2000). Chlorine Chemistry Council v. E.P.A., 206 F.3d 1286 (C.A.D.C.). CEC (2000). Commission of the European Communities, Communication from the Commission on the precautionary principle, Brussels, 02.02.2000 COM(2000) 1, pp. 1–29. CMA (1994). Chemical Mfrs. Ass’n v. E.P.A., 28 F.3d 1259 (C.A.D.C.). Congress (2000). ICCVAM Authorization Act (Public Law 106-545), 106th Congress. Daubert (1993). Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579. DQA (2000). Data Quality Act, Pub. L. No. 106–554, § 515(a), 114 Stat. 2763A, pp. 153–154. Durenberger, D. (1990). 1990 Clean Air Act Amendments—Conference Report, Cong. Rec. S16895, S16932 (Oct. 27, 1990). EDF (1976). Environmental Defense Fund, Inc. v. Environmental Protection Agency, 548 F.2d 998. EDF (1978). Environmental Defense Fund, Inc. v. Costle, 578 F.2d 337. EPA (1986). Guidelines for carcinogen risk assessment, EPA/630/R-00/004, 1–38. EPA (1991). Alpha2u-globulin: Association with chemically induced renal toxicity and neoplasia in the male rat, EPA/625/3-91/019F, 1–118. EPA (1996). Proposed guidelines for carcinogen risk assessment, EPA/600/P-92/003C, 1–142. EPA (2002). Science Policy Council—Interim policy on genomics. http://www.epa.gov/osainter/spc/ pdfs/genomics.pdf, 1–4. EPA (2004). An examination of EPA risk assessment principles and practices. EPA/100/B-04/001, 1–182. EPA (2005a). Guidelines for carcinogen risk assessment, EPA/630/P-03/001F, 1–166. EPA (2005b). Supplemental guidance for assessing susceptibility from early-life exposure to carcinogens, EPA/630/R-03/003F, 1–126. Ethyl (1976). Ethyl Corp. v. Environmental Protection Agency, 541 F.2d 1 (C.A.D.C.). Finkel, A. M. (1996). Who’s exaggerating? Discover May 1, 48–54. Flamm, W. G. (1989). Critical assessment of carcinogenic risk policy, Regul Toxicol Pharmacol 9, 216–224. Frye (1923). Frye v. U.S., 293 F. 1013 (C.A.D.C.). GE (1997). General Electric. Co. v. Joiner, 522 U.S. 136. Goldman, L. R. (2003). The red book: A reassessment of risk assessment. Hum Ecolo Risk Assess 9, 1273–1281. Goldstein, B. D. (1989). Risk assessment and the interface between science and law. Columbia J. Environ Law 14, 343–355. Graham, J. D. (2002). The role of precaution in risk assessment and management: An American’s view. Remarks prepared for “The US, Europe, Precaution and Risk Management: A Comparative Case Study Analysis of the Management of Risk in a Complex World” Conference Organizers: The European Commission (Group of Policy Advisers), the U.S. Mission to the EU, the German Marshall Fund with the European Policy Centre and the Center for Environmental Solutions, Duke University (January 11–12, 2002). IFI (1992). International Fabricare Institute v. U.S. E.P.A., 972 F.2d 384. IU (1989). International Union, United Auto, Aerospace and Agricultural. Implement Workers of America, UAW v. Pendergrass, 878 F.2d 389. IUD (1980). Industrial Union Department, AFL-CIO v. American Petroleum Institute. 448 U.S. 607. Latin, H. (1988). Good science, bad regulation, and toxic risk assessment. Yale J. Regul 5, 89–148. LIA (1980). Lead Industries Association, Inc. v. Environmental Protection Agency, 647 F.2d 1130 (C.A.D.C.), certification denied, 449 U.S. 1042. LIA (1994). Leather Industries of America, Inc. v. E.P.A., 40 F.3d 392 (C.A.D.C.). Marchant, G. E. (2003). From general policy to legal rule: Aspirations and limitations of the precautionary principle. Environ Health Perspect 111, 1799–1803. McClellan, R. O. (1996). Reducing uncertainty in risk assessment by using specific knowledge to replace default options. Drug Metab Rev 28, 149–179. Millstone, E., van Zwanenberg, P., Levidow, L., Spok, A., Hirakawa, H., and Matsuo, M. (2008). Riskassessment policies: Differences across jurisdictions. EUR 23259 EN, Joint Research Centre, Institute for Prospective Technological Studies, 1–84.
36
CHAPTER 2 SCIENCE POLICY AND CANCER RISK ASSESSMENT
Nichols, A. L., and Zeckhauser, R. J. (1988). The perils of prudence: How conservative risk assessments distort regulation. Regul Toxicol Pharmacol 8, 61–75. NRC (1983). Risk Assessment in the Federal Government: Managing the Process Working Papers, The National Academy Press, http://www.nap.edu/openbook.php?isbn=POD115&page=R1, Washington, D.C. NRC (1994). Science and Judgment in Risk Assessment, The National Academies Press, http://books.nap. edu/openbook.php?record_id=2125&page=R1, Washington, D.C. NRC (2007a). Applications of Toxicogenomic Technologies to Predictive Toxicology and Risk Assessment, The National Academies Press, http://books.nap.edu/openbook.php?record_id=12037&page=R1, Washington, D.C. NRC (2007b). Scientific Review of the Proposed Risk Assessment Bulletin from the Office of Management and Budget, The National Academies Press, http://books.nap.edu/openbook.php?record_id=11811& page=R1, Washington, D.C. NRDC (1987). Natural Resources Defense Council, Inc. v. E.P.A., 824 F.2d 1211 (C.A.D.C.). NRDC (2007). Natural Resources Defense Council v. E.P.A., 489 F.3d 1364 (C.A.D.C.). O’Brien, M. (2000). Making Better Environmental Decisions—An Alternative to Risk Assessment, The MIT Press, Cambridge, MA. OECD (2005). Guidance document on the validation and international acceptance of new or updated test methods for hazard assessment, ENV/JM/MONO(2005)14. OECD Series on Testing and Assessment, No. 34, pp. 1–96. OMB (2001). Guidelines for ensuring and maximizing the quality, objectivity, utility, and integrity of information disseminated by federal agencies (October 1, 2001), http://www.whitehouse.gov/omb/ fedreg/final_information_quality_guidelines.html. OMB (2005). Final information quality bulletin for peer review. Federal Register 70, 2664–2677. OMB (2006). Proposed risk assessment bulletin, http://www.whitehouse.gov/omb/inforeg/proposed_ risk_assessment_bulletin_010906.pdf, pp. 1–26. OMB (2007). Updated principles for risk analysis; M-07-24 Memorandum for the heads of executive departments and agenices (September 19, 2007), http://www.whitehouse.gov/omb/memoranda/fy2007/ m07-24.pdf, pp. 1–13. PCHRG (1986). Public Citizen Health Research Group v. Tyson, 796 F.2d 1479 (C.A.D.C.). SDWA (1996). Safe Drinking Water Act, 42 U.S.C. § 300F to 300J-26. SI (2006). Salt Institute v. Leavitt, 440 F.3d 156 C.A.4 (Va.). Sutera (1997). Sutera v. Perrier Group of America Inc., 986 F.Supp. 655 D.Mass. Tickner, J. A. (2002). Precaution, Environmental Science, and Preventive Public Policy, Island Press, Washington, D.C. Wagner, W. E. (2000). The triumph of technology-based standards, U. Ill. L. Rev., pp. 83–113. Whiting (1995). Whiting v. Boston Edison Co., 891 F.Supp. 12 D.Mass. Wiltse, J., and Dellarco, V. L. (1996). U.S. Environmental Protection Agency guidelines for carcinogen risk assessment: Past and future. Mutat Res 365, 3–15.
CH A P TE R
3
HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY WITHIN A REGULATORY CONTEXT Henk Tennekes Virginia A. Gretton Todd Stedeford
3.1.
OVERVIEW
The first section of this chapter provides a discussion of hazard assessment, classification of potentially dangerous substances, and the process of risk assessment. A summary of the mandatory and voluntary initiatives for regulating chemicals and biocides in the United States and Europe is also included together with information on the regulatory aspects of hazard communication. The second section deals with the scientific aspects of hazard identification and risk assessment of carcinogenic chemicals within the regulatory context.
3.2.
RISK ASSESSMENT
To enable materials to be stored and used safely, the risks to human health and the environment must be assessed. Risk assessment of both new and existing substances comprises the following steps (NRC 1983): 1. Hazard Identification. Identification of intrinsic hazardous properties. 2. Dose–response Assessment. Determination of the dose/concentration–response characteristics. 3. Exposure Assessment. Exposure assessment for humans (i.e., workers, consumers, and those exposed indirectly via the environment) and for the different environmental compartments (air, soil, water) likely to be exposed to the substance. Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
37
38
CHAPTER 3 HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY
4. Risk Characterization. Comparison of information on hazardous properties and effective dose levels/concentrations with exposure levels in order to characterize the degree of risk posed by the substance to human health or the environment.
3.2.1.
Principles of Risk Assessment and Management
A particular substance may have several hazards, which are categorized as physicochemical, toxicological, or environmental. Physicochemical hazards arise from intrinsic physical or chemical properties of the substance. Toxicological hazards result from a chemical causing harmful effects when ingested, inhaled, or absorbed through the skin. Toxic effects may be acute or chronic, local or systemic, reversible or irreversible. Environmental hazards relate to persistence, bioaccumulation, and toxicity to terrestrial and aquatic organisms. Test guidelines are available for conducting studies aimed at evaluating the physical/chemical properties, environmental fate, and potential human health and ecological hazards of chemical substances [reviewed by Knight and Thomas (2003)]. The most frequently used guidelines are the harmonized test guidelines published by the U.S. Environmental Protection Agency’s (EPA’s) Office of Prevention, Pesticides, and Toxic Substances (OPPTS) (
), the Organization for Economic Co-operation and Development’s (OECD’s) guidelines for the testing of chemicals ( < http://www.oecd.org/document/40/0 ,3343,en_2649_34377_37051368_ 1_1_1_1,00.html>), and the European Commission’s testing methods in Annex V to Directive 79/831/EEC (). Risk management measures (RMMs) are implemented after a risk–benefit evaluation and are in the form of instructions for safe use, labeling, or occupational exposure limits. Hazards of substances and preparations must be communicated to users, both workers and the general public. This is achieved by standardized classification and labeling (e.g., EC 2008) of potentially dangerous chemicals and by providing a Safety Data Sheet (SDS). Most developed countries also have legal provisions for banning or restricting the use of chemicals to safe conditions. 3.2.1.1. Classification of Carcinogens. Carcinogenic chemicals are classified on the basis of the weight of evidence. The quality and nature of the evidence determines the category of the classification, not potency. Various bodies in Europe and the United States have subtly different definitions for their categorizations (Persad et al. 2007; Stedeford and Persad 2007); however, international efforts are under way to harmonize classification schemes. For example, under the Globally Harmonised System of Classification and Labelling of Chemicals Regulation No. 1272/2008 (EC 2008), substances are categorized as known or presumed human carcinogens (category 1) if there is sufficient epidemiological and/or animal data to establish a causal association between human exposure to a substance and development of cancer. Substances are categorized as suspected human carcinogens (category 2) based on evidence obtained from epidemiological and/or animal data that are not sufficiently convincing to warrant category 1. Numerous factors need to be taken into account when assessing the weight of evidence for classification as
3.2. RISK ASSESSMENT
39
category 1 or 2 or nonclassification. These factors include but are not limited to the following: • Carcinogenic effects were noted only at very high dose levels exceeding the maximal tolerated dose. • There was appearance of tumors, especially at high dose levels, only in particular organs of certain species known to be susceptible to a high spontaneous tumor formation. • There was appearance of tumors only at the site of application in very sensitive test systems (e.g., intraperitoneal or subcutaneous application of certain locally active compounds), and the particular target is not relevant to man. • There was lack of genotoxicity in short-term tests in vivo and in vitro. • There is a secondary mechanism of action with the implication of a practical threshold above a certain dose level (e.g., hormonal effects on target organs or on mechanisms of physiological regulation, chronic stimulation of cell proliferation). • There is a species-specific mechanism of tumor formation (e.g., by specific metabolic pathways) irrelevant for humans. Carcinogenicity evaluations by the International Agency for Research on Cancer (IARC) serve as the international benchmark for classifying chemicals as carcinogens. IARC assesses and classifies chemicals according to the following scheme (Illing 2001; IARC 2006): • Group 1. The agent (mixture) is carcinogenic to humans. The exposure circumstance entails exposures that are carcinogenic to humans. • Group 2A. The agent (mixture) is probably carcinogenic to humans. The exposure circumstance entails exposures that are probably carcinogenic to humans. • Group 2B. The agent (mixture) is possibly carcinogenic to humans. The exposure circumstance entails exposures that are possibly carcinogenic to humans. • Group 3. The agent (mixture or exposure circumstance) is not classifiable as to its carcinogenicity to humans. • Group 4. The agent (mixture) is probably not carcinogenic to humans. A list of chemicals assessed by IARC is available at the following URL: . In the United States, the two most prominent authorities for classifying the carcinogenicity of chemicals are the U.S. EPA and the U.S. National Toxicology Program (NTP). The weight-of-evidence descriptors used in the EPA’s Guidelines for Carcinogen Risk Assessment are based on the following (EPA 2005): • Carcinogenic to Humans. Strong evidence of human carcinogenicity when, for example, there is convincing epidemiologic evidence of a causal association between exposure and cancer.
40
CHAPTER 3 HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY
• Likely to Be Carcinogenic to Humans. The weight of evidence is adequate to demonstrate carcinogenic potential to humans but does not reach the weight of the above descriptor. The use of the term “likely” as a weight of evidence descriptor does not correspond to a quantifiable probability. • Suggestive Evidence of Carcinogenic Potential. A concern for potential carcinogenic effects in humans is raised, but the data are judged not sufficient for a stronger conclusion. This descriptor covers a spectrum of evidence associated with varying levels of concern for carcinogenicity, ranging from a positive cancer result in the only study on an agent to a single positive cancer result in an extensive database that includes negative studies in other species. • Inadequate Information to Assess Carcinogenic Potential. This descriptor is appropriate when available data are judged inadequate, lack pertinent information, or provide conflicting evidence, for applying one of the above descriptors. • Not Likely to Be Carcinogenic to Humans. The available data are considered for deciding there is no basis for human hazard concern. In some instances, there can be positive results in experimental animals when there is strong, consistent evidence that each mode of action in experimental animals does not operate in humans. In other cases, there can be convincing evidence in both humans and animals that the agent is not carcinogenic. On a biennial basis, the U.S. NTP issues the congressionally mandated Report on Carcinogens (RoC). This document provides a list of known human carcinogens and substances that are reasonably anticipated to be human carcinogens, along with a brief profile for each substance. A summary of the classification criteria used in the 11th RoC is provided below (NTP 2008): • Known to Be Carcinogenic to Humans. Sufficient evidence of carcinogenicity from studies in humans which indicates a causal relationship between exposure to the agent, substance, or mixture, and human cancer. • Reasonably Anticipated to Be a Human Carcinogen. This designation may be made based on either (1) limited evidence of carcinogenicity from studies in humans, (2) sufficient evidence of carcinogenicity from studies in experimental animals, or (3) less than sufficient evidence of carcinogenicity in humans or laboratory animals. 3.2.1.2. Current Principles of Carcinogenic Risk Assessment. Weightof-evidence-based systems which classify carcinogenic hazards are part of, but do not substitute for, the risk assessment process (Di Marco et al. 1998). Carcinogen risk assessment is based on an evaluation of appropriate toxicological and exposure data sets, which should meet certain criteria for data quality and relevance (ECHA 2008a; EPA 2003; Klimisch et al. 1997). However, national policy frameworks can differ to the extent that risk assessment outcomes may be quite different for the same chemical(s). As discussed in Chapter 2, differences in science policy have been greater for cancer risk assessment compared to other toxic endpoints, with a
3.2. RISK ASSESSMENT
41
tendency to differentiate cancer risk assessment on the basis of presumed mechanism (i.e., genotoxic or nongenotoxic) and relevance to humans (some carcinogenic responses in animals may be considered irrelevant to human risk assessment) (Di Marco et al. 1998; EPA 1991, 1998; IARC 1999). Historically, risk assessment for noncancer endpoints has been based on the identification of a “no observed adverse effect level” (NOAEL) from a toxicity study with an animal model. The NOAEL is then divided by appropriate uncertainty factors to take potential inter- and intraspecies differences in response into account. However, this approach does not take into account the size of the toxicity study or the shape of the dose–response curve. The benchmark dose (BMD) approach has been suggested as an alternative to a NOAEL (Crump 1984). A BMD is a dose or concentration that produces a predetermined change (e.g., 10% or 1 standard deviation) in response rate of an adverse effect (called the benchmark response or BMR). A BMDL is the statistical lower confidence limit on the dose or concentration at the BMD. The BMD and BMDL are calculated using mathematical dose–response models, which make appropriate use of sample size and the shape of the dose– response curve (EPA 2009b, 2000a). The BMDL is like a NOAEL (i.e., as a point of departure) and is divided by an appropriate composite uncertainty factor to derive a reference value. The European Union (Commission Directive 93/67/EEC, Article 3, paragraph 1; repealed) and WHO (1994) have used the NOAEL/uncertainty factor approach for nongenotoxic carcinogens that are believed to have an effect threshold (WHO 1994). For genotoxic carcinogens, however, the regulatory default is applied that is based on the assumption that if “one hit” could cause a mutation and eventually result in cancer, then any exposure level could be associated with a finite cancer probability. Under such circumstances, a mathematical model (that quantitatively describes the relation between dose [exposure] and cancer [probability]) would be required to determine a “virtually” safe dose (VSD), a dose associated with an insignificantly small cancer risk. The choice of the model has an impact on risk predictions, because it usually involves extrapolation to low doses for which no data may be available, and has remained controversial. The U.S. EPA applies an alternative dose–response evaluation of carcinogens using a low-dose, linear model (EPA 2005). The linear extrapolation is applied under two circumstances: (1) when there are data to indicate that the dose-response curve has a linear component below the point of departure or (2) as a default for a tumor site where the mode of action is not established. For a linear extrapolation, a straight line is drawn from the point of departure to the origin. The slope of the line, known as the slope factor, is an upper-bound estimate of risk per increment of dose that can be used to estimate risk probabilities for different exposure levels. The slope factor is equal to 0.01/LED01, for example, if the LED01 is used as the point of departure. The lower limit on effective dose01 (LED01) is the 95% lower confidence limit of the dose of a chemical needed to produce an adverse effect in 1% of those exposed to the chemical, relative to control. If, however, there are sufficient data to ascertain that a chemical’s mode of action supports modeling at low doses, a reference dose or concentration may be developed in lieu of a cancer slope factor.
42
CHAPTER 3 HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY
3.3. REGULATORY SCHEMES FOR INDUSTRIAL CHEMICALS AND BIOCIDES Many countries have schemes requiring new chemical substances to be notified to the national regulatory authority with a standard set of hazardous properties data and an assessment of the hazardous properties. This enables the risks to humans and the environment to be assessed with a view to deciding which RMMs are necessary. There are national inventories of existing substances that can be supplied without notification, by definition, all others are new. There are nevertheless schemes, both national and international, to evaluate high production volume (HPV) existing substances [reviewed in Knight and Thomas (2003)]. For example, the U.S. EPA and the OECD evaluate HPV chemicals produced in amounts greater than or equal to one million pounds per year or 2.2 million pounds per year, respectively. More recently, voluntary and mandatory chemical initiatives have been proposed or implemented that evaluate lower volume chemicals—that is, those produced in amounts greater than 25,000 pounds per year (EPA 2008a) or ∼2200 pounds per year (EC 2006). The discussion that follows provides an overview of laws in the United States and Europe for regulating industrial chemicals, followed by a discussion of voluntary initiatives for evaluating medium- and high-production-volume industrial chemicals. Thereafter, the laws for regulating biocides in the United States and Europe are discussed.
3.3.1.
The U.S. Toxic Substances Control Act (TSCA)
The U.S. Toxic Substances Control Act (TSCA) of 1976 (15 U.S.C. 2601 et seq.) provides the U.S. EPA with the authority to regulate industrial chemicals and mixtures (TSCA 1976). Section 2(b)(1) of TSCA states that it is the policy of the United States that “adequate data should be developed with respect to the effect of chemical substances and mixtures on health and the environment and that the development of such data should be the responsibility of those who manufacture [which is defined by statute to include import] and those who process such chemical substances and mixtures [.]” The core sections of TSCA that provide authority for implementing the above policy are discussed below. Prior to regulating chemicals under TSCA, it was foreseen that the chemicals in commerce would have to be known. Section 8(b) of TSCA addresses this need and grants the U.S. EPA authority to: “…compile, keep current, and publish a list of each chemical substance which is manufactured or processed in the United States.” TSCA Section 3(2)(A) defines a “chemical substance” as “… any organic or inorganic substance of a particular molecular entity, including (i) any combination of such substances occurring in whole or in part as a result of a chemical reaction or occurring in nature and (ii) any element or uncombined radical.” Foods, drugs, cosmetics, tobacco and tobacco products, radioactive materials, and pesticides are generally excluded, under Section 3(2)(B)(i)–(vi) of TSCA. Chemical substances not included on the TSCA inventory are considered new chemical substances, with some exemptions, and require compliance with premanufacturing notification requirements set forth under Section 5(a)(1)(A) of
3.3. REGULATORY SCHEMES FOR INDUSTRIAL CHEMICALS AND BIOCIDES
43
TSCA. The U.S. EPA may also promulgate rules after determining that a “significant new use” is proposed for a chemical on the TSCA inventory, under Section 5(a)(1) (B) and Section 5(a)(2). The U.S. EPA requires the submission of EPA Form 771025 (Rev. 5-95) for premanufacturing notices, significant new use notices, and any applicable exemptions at least 90 days before manufacturing or processing the new chemical substance. Along with this form, the submitter is “… required to submit all test data in [the submitter ’s] possession or control and to provide a description of all other data known to or reasonably ascertainable by [the submitter], if these data are related to the health and environmental effects on the manufacture, processing, distribution in commerce, use, or disposal of the new chemical substance.” Several possible outcomes may occur during the U.S. EPA’s 90-day review period and include the following: (1) No additional information is requested, (2) additional information may be requested, or (3) an administrative order may be issued that prohibits or limits the manufacture, processing, distribution, or disposal of a substance, pending the development of information. If a notice has been completed and the submitter has commenced commercial manufacture, the submitter is required to submit a Notice of Commencement of Manufacture or Import (EPA Form 7710-56 (8-95)) within 30 calendar days of the date the substance is first manufactured or imported for commercial purposes. Continued reporting requirements are placed on persons that manufacture, process, or distribute in commerce any chemical substance or mixture and include: (1) maintaining records of significant adverse reactions to health or the environment, alleged to have been caused by the substance or mixture (Section 8(c) of TSCA) and (2) immediately informing the U.S. EPA of “… information which reasonably supports the conclusion that such chemical substance or mixture presents a substantial risk of injury to health or the environment,” unless the person has actual knowledge that the U.S. EPA has been adequately informed (Section 8(e) of TSCA). Section 4(a)(1) of TSCA mandates that the U.S. EPA require by rule that manufacturers and/or processors of new or existing chemicals substances and mixtures conduct testing if the Administrator of the U.S. EPA finds that “[t]he manufacture, distribution in commerce, processing, use, or disposal of a chemical substance or mixture, or any combination of such activities, may present an unreasonable risk of injury to health or the environment [emphasis added].” A TSCA Section 4 test rule may require manufacturers and processors to conduct testing on environmental fate, ecotoxicity, acute toxicity, genetic toxicity, repeated dose toxicity, or developmental and reproductive toxicity. When a statutory finding under TSCA Section 4(a)(1) (i.e., “may present an unreasonable risk of injury to health or the environment”) cannot readily be made, the U.S. EPA may request that the TSCA Interagency Testing Committee (ITC), an independent advisory committee to the Administrator of the U.S. EPA, add the chemical to the TSCA Section 4(e) Priority Testing List (PTL). Once a chemical is added to the PTL, the U.S. EPA must promulgate a TSCA Section 8(a) Preliminary Assessment and Information Reporting (PAIR) rule and a TSCA Section 8(d) Health and Safety Data Reporting (HaSDR) rule. Section 8(a) PAIR rules require producers and importers to submit to the U.S. EPA one-time reports on production/importation volumes, end uses, and exposure-related data for the listed chemicals. Section 8(d)
44
CHAPTER 3 HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY
HaSDR rules require producers and importers to submit to the U.S. EPA copies and lists of certain types of unpublished health and safety studies for the listed chemicals. Submitters under the HaSDR rule are also requested to provide robust summaries of health and environmental effects studies. In addition, when the ITC designates chemicals for testing, the U.S. EPA is required to initiate proceeding under a TSCA Section 4(a) test rule, if the PAIR and HaSDR data trigger a finding of unreasonable risk of injury to health or the environment.
3.3.2. The EU Registration, Evaluation, Authorisation, and Restriction of Chemicals (REACH) EU chemical control legislation has recently (2007–2008) been revised under the new scheme for Registration, Evaluation, Authorisation, and Restriction of Chemicals (REACH) (EC 2006). REACH places a duty on companies that produce, import, and use chemicals to assess the risks arising from their use (with new studies in justified cases) and take the necessary RMMs. The burden of proof for putting safe chemicals on the market has been transferred from the regulators to industry. Animal testing data must be shared to avoid duplication. Registration of information on the properties, uses, and safe handling of chemical substances will be an integral part of the system. A phase-in system lasting up to 11 years is planned for existing chemicals, known as “phase-in substances.” Higher tonnage substances (≥1000 tonnes per annum [t/a]), as well as lower tonnage substances that are very toxic to the aquatic environment (≥100 t/a) or classified as CMRs (i.e., carcinogens, mutagens, or reproductive toxicants; ≥1 t/a), will require more data and have to be registered by November 30, 2010. All other phase-in substances manufactured or imported in quantities ≥100 t/a or ≥1 t/a must be registered by May 31, 2013 or May 31, 2018, respectively. New chemicals or “nonphase-in substances” will be evaluated on an ongoing basis and must be registered before being manufactured or imported into the EU. Under REACH, all substances manufactured in or imported into the EU at ≥ 1 t/a must be registered with the European Chemicals Agency (ECHA). A Chemical Safety Report (CSR) is required for substances registered at 10 t/a unless the substance is present only in a preparation at below 0.1% or below the concentration limit(s) triggering classification of the preparation as dangerous. A CSR is a risk assessment that must follow (a) the general provisions of Annex I of REACH and (b) the ECHA’s guidance for writing a CSR (ECHA 2008c). Substances of very high concern classified as category 1 or 2 CMRs are amongst the substances subject to tighter controls, including an authorization regime, along with persistent, bioaccumulative, and toxic substances (PBTs) and very persistent and very bioaccumulative substances (vPvBs). PBTs and vPvBs are classified as such based on the criteria set forth in Annex XIII of REACH. When substances are classified as CMR, PBT, or vPvB, they shall appear on the list of substances subject to authorization under Annex XIV of REACH. Other substances of concern, such as endocrine disrupters, will also be added to this list (Annex XIV) on an ad hoc basis. Substances subject to authorization will have to be approved for a specific use, with decisions based on a risk assessment and consideration of socioeconomic factors. For existing substances in Annex XIV of REACH, a “Sunset Date” will be
3.3. REGULATORY SCHEMES FOR INDUSTRIAL CHEMICALS AND BIOCIDES
45
set after which the substance may no longer be used or sold, unless an application is submitted to and subsequently approved by ECHA. The classification of a substance as a CMR, PBT, or vPvB will be a factor in deciding what further higher-tier testing is required (e.g., Sections 8 and 9 of Annexes VIII, IX, and X), as will be the mandatory performance of an exposure assessment and subsequent risk characterization (ECHA 2008b). Most of the benefits expected from REACH are based on the expected significant increase in cancer prevention. Postle et al. (2003) estimated that the economic benefits of preventing between 2167 and 4333 deaths per annum due to cancer over a 30-year period will be between œ18 billion and œ54 billion; by comparison, the benefits of preventing deaths from all other diseases combined were between œ23 million and œ225 million (Postle et al. 2003). Clearly, carcinogenicity is of paramount importance within regulatory frameworks. Due to the importance of the EU as a trade bloc, it is expected that other jurisdictions will adapt their legislation to be compatible.
3.3.3. Voluntary Initiatives for Evaluating Industrial Chemicals 3.3.3.1. The U.S. EPA’s Former Chemical Assessment and Management Program (ChAMP). In August 2007, Canada, Mexico, and the United States committed, under the Security and Prosperity Partnership (SPP), to accelerate and strengthen national and regional risk-based assessment and management of chemicals. ChAMP was the name given by the U.S. EPA to identify its efforts to meet the SPP commitments. By 2012, the U.S. EPA, under ChAMP, planned to assess and prepare screening-level characterizations of hazard, exposure, and risk, and to use this information to develop initial risk-based prioritizations (RBPs). The U.S. EPA planned to evaluate over 4000 organic Medium-Production Volume (MPV) chemicals produced at volumes greater than 25,000 pounds per year, but less than one million pounds per year. Health and environmental hazard and environmental fate characterizations would be informed by existing data, Canadian categorization results, U.S. EPA Structure Activity analysis input, and knowledge gained under the U.S. EPA’s HPV Chemical Challenge. The U.S. EPA planned to also evaluate 2750 organic HPV chemicals produced at or above one million pounds per year. HPV challenge submissions were to provide the base hazard data for evaluations under ChAMP, whereas the 2006 Inventory Update Reporting (IUR), under TSCA, was to provide the use and exposure information. Beyond organic MPVs and select HPVs, the U.S. EPA planned to assess ∼750 inorganic HPV chemicals, which were first reported under the 2006 IUR cycle. The general voluntary approach used in the HPV Chemical Challenge was expected to be used (see Section 3.3.2, e.g., sponsorship commitments, development of test plans, public review step, completion of data package, and submission to EPA). The OECD’s inorganic HPV guidance would serve as a benchmark for preparing submissions. During the preparations for implementing ChAMP, the U.S. EPA was in the process of “resetting” the TSCA Inventory (EPA 2008b). The original inventory was compiled in 1979 and consisted of 62,000 chemicals. Since then, ∼21,000 new chemicals have been added to the TSCA inventory. Though the U.S. EPA plans to
46
CHAPTER 3 HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY
reset the TSCA Inventory in order to better understand the universe of chemicals actually in commerce, ChAMP was superseded on September 29, 2009, with a regulatory management approach due to concerns about the sufficiency of data from which the U.S. EPA could evaluate these chemicals. This determination may eventually lead to changes in TSCA similar to the approach for chemical management under REACH (i.e., no data, no market, tiered testing, etc). 3.3.3.2. The U.S. EPA’s High Production Volume (HPV) Chemicals Challenge. In 1998, the U.S. EPA launched the voluntary High Production Volume (HPV) Chemicals Challenge Program. This initiative was created to ensure the public availability of a baseline set of data on over 2800 HPV chemicals, which are freely accessible at the following URL: http://iaspub.epa.gov/oppthpv/ public_search.html_page. The U.S. EPA defined HPV chemicals as those being manufactured or imported in amounts greater than or equal to one million pounds per year, based on volumes reported under the TSCA, 1990 IUR. The initiative was called a “challenge” because the U.S. EPA challenged U.S. manufacturers and importers of HPV chemicals to voluntarily sponsor chemicals under the program. The data sought under this program were based on internationally agreed-upon test data known as the Screening Information Data Set (SIDS), as developed by the OECD. SIDS data sets enable regulators to assess human and environmental hazards and are intended to provide enough information to assign a priority for further work, if necessary. These data include the following: acute toxicity; repeated dose toxicity; developmental and reproductive toxicity; mutagenicity (gene mutation and chromosomal aberration/damage assays); ecotoxicity (studies in fish, invertebrates, and algae); and environmental fate [including physical/chemical properties (melting point, boiling point, vapor pressure, n-octanol/water partition coefficient, and water solubility), photolysis, hydrolysis, transport/distribution, and biodegradation]. The U.S. EPA has issued several guidance documents, which aid sponsors with preparing submissions (EPA 2009a). As part of the commitment under the U.S. EPA’s program, sponsors submit data summaries of existing data, along with a test plan that proposes a testing strategy to fill data gaps. Once submitted to the U.S. EPA, the documents are posted on a public database and a 120-day comment period begins whereby all stakeholders (e.g., the U.S. EPA, industry, environmental protection groups, animal welfare groups, private citizens, etc.) have an opportunity to provide input. Comments are intended to provide feedback, which may be used to revise test plans and data summaries. In the event that an HPV chemical lacks necessary testing data, the U.S. EPA may, at their discretion, issue a test rule to obtain the required data. 3.3.3.3. The OECD’s Work on Investigation of HPV Chemicals. The OECD defines HPV chemicals as those manufactured or imported in quantities greater than or equal to 2.2 million pounds per year in at least one member country or an EU region. In an effort to undertake the investigation of HPV chemicals in a cooperative manner, the OECD modified its work on HPV chemicals through an OECD Council Decision in 1991 (OECD 1991). The OECD’s cooperative approach includes involvement from member countries in four basic components: (1) selection of chemicals to be evaluated, (2) collection of data from governments, public
3.3. REGULATORY SCHEMES FOR INDUSTRIAL CHEMICALS AND BIOCIDES
47
sources, and industry, (3) completion of the SIDS dossier, and (4) assessment of the potential hazards for the chemical investigated. The current OECD HPV chemicals list includes 4853 substances (OECD 2004). The status of each chemical in the process may be monitored via the OECD’s publicly available HPV database . The OECD’s program enables a member country to sponsor the HPV chemicals produced by industries within its borders and, in turn, to benefit through data sharing from the sponsorship of other countries. This process eliminates duplicative testing. The OECD has developed extensive guidance, which aids sponsors by (1) following procedures to improve the efficiency of investigating HPV chemicals, (2) data gathering and testing for the SIDS, (3) evaluating the quality of data in the SIDS dossier, (4) assessing the initial hazards of chemicals, (5) preparing the SIDS initial assessment report and SIDS profile, and (6) overseeing any additional postSIDS activities (OECD 2007). 3.3.3.4. The U.S. EPA’s Voluntary Children’s Chemical Evaluation Program (VCCEP). On December 26, 2000, the U.S. EPA announced the Voluntary Children’s Chemical Evaluation Program (VCCEP) and requested sponsorship commitments from manufacturers or importers for 23 chemicals. The VCCEP was designed to provide data to enable the public to better understand the potential health risks to children associated with certain chemical exposures. In support of this pilot program, the U.S. EPA designed a tiered-testing approach, which consisted of the following: Tier 1: acute toxicity, in vitro gene mutation, combined repeated dose toxicity with reproductive and developmental toxicity screens or repeated dose oral toxicity and reproductive toxicity (one generation); Tier 2: 90-day subchronic toxicity in rodents; reproduction and fertility effects; prenatal developmental toxicity (two species); in vivo mammalian bone marrow chromosomal aberrations or in vivo mammalian erythrocyte micronucleus (triggered off results from in vitro mammalian chromosomal aberration test if conducted in tier 1); immunotoxicity; metabolism and pharmacokinetics; Tier 3: carcinogenicity or chronic toxicity/carcinogenicity; neurotoxicity screening battery; developmental neurotoxicity (EPA 2000b). The U.S. EPA chose these studies based on their appropriateness for evaluating the toxicity of chemicals to which children have significant potential for exposure. The VCCEP process consists of following basic steps, outlined below (EPA 2000b): Step 1: Chemical Selection. After receiving comments from various stakeholders, the U.S. EPA selected chemicals judged by the U.S. EPA to present the relatively greatest potential for exposures that may impact children. Step 2: Tier 1 Commitment. A manufacturer or importer of a VCCEP chemical submits a letter to the U.S. EPA indicating their commitment to sponsoring a chemical. Step 3: Submission of Tier 1 Data. A VCCEP chemical sponsor submits to the U.S. EPA a Tier 1 Hazard Assessment, a Tier 1 Exposure Assessment, a Tier 1 Risk Assessment, and a Data Needs Assessment. Step 4: Peer Consultation Regarding Tier 2 Data Needs. At the U.S. EPA’s request, a third-party contractor convenes a Peer Consultation to evaluate
48
CHAPTER 3 HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY
whether Tier 1 data needs were met, whether the Tier 1 submission fully characterized the chemical’s potential risk to children, and whether there are remaining Tier 2 data needs. The results and comments from the Peer Consultation are compiled by the third-party contractor and submitted to the U.S. EPA. A possible conclusion is that no more work is needed. Step 5: U.S. EPA Review of Peer Consultation. The U.S. EPA reviews the sponsor ’s submission and the third-party contract report and announces a the Tier 2 Data Needs Decision. If the U.S. EPA disagrees with the conclusions from the third-party peer consultation report, sponsors and other stakeholders are given 60 days to comment on the U.S. EPA’s Tier 2 Data Needs Decision. Following a review of comments, the U.S. EPA mails its final decision to the sponsor and posts the decision on the VCCEP website. If the U.S. EPA requires further testing under Tier 2 or Tier 3, steps 2 through 5, above, are repeated with consideration of the appropriate tier ’s testing requirements. If a chemical is recommended for Tier 2 or Tier 3 testing, but is not sponsored by a manufacturer or importer of the chemical, the U.S. EPA may require the data by issuing a TSCA Section 4 test rule. A summary of the submissions on VCCEP chemicals is available at the following URL: http://www.epa.gov/oppt/vccep/pubs/ chemmain.html.
3.3.4. The U.S. Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) The U.S. Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) (7 U.S.C. 136 et seq.) grants the U.S. EPA the authority to regulate the registration, distribution, sale, and use of pesticides (FIFRA 1972). Under Section 2(u), FIFRA defines a pesticide as “… any substance or mixture of substances intended for preventing, destroying, repelling, or mitigating any pest ….” Biocides are included under a separate definition for antimicrobial pesticides, which are defined under Section 2(mm) as pesticides intended to “(i) disinfect, sanitize, reduce, or mitigate growth or development of microbiological organisms; or (ii) protect inanimate objects, industrial processes or systems, surfaces, water, or other chemical substances from contamination, fouling, or deterioration caused by bacteria, viruses, fungi, protozoa, algae, or slime ….” Under FIFRA Section 3, every pesticide product must be registered with the U.S. EPA or specifically exempted under FIFRA Section 25(b) before being sold or distributed in the United States. An applicant for a new registration or an existing registrant must demonstrate to the U.S. EPA’s satisfaction that, among other things, the pesticide product, when used in accordance with widespread and commonly recognized practice, will not cause “unreasonable adverse effects” to humans or the environment. This safety determination requires the U.S. EPA to weigh the risks of the use of the pesticide against any benefits. A general overview of the core provisions of FIFRA that aid with the U.S. EPA’s weighing of risks versus benefits is provided below, along with provisions specific to antimicrobial pesticides. Under Section 3(c)(2) of FIFRA, the U.S. EPA is granted broad authority to require scientific testing and submission of the resulting data to the U.S. EPA by
3.3. REGULATORY SCHEMES FOR INDUSTRIAL CHEMICALS AND BIOCIDES
49
applicants for registration of pesticide products. An applicant must furnish the U.S. EPA with substantial amounts of data on the pesticide, its composition, toxicity, potential for human exposure, environmental properties, and ecological effects, as well as information on its product performance in certain cases. Section 3(c)(2)(B) authorizes the U.S. EPA to require a registrant to develop and submit additional data to maintain a registration. However, Section 3(c)(2) does not require the U.S. EPA to develop data requirements for an “antimicrobial pesticide” as defined under Section 2(mm). Though Section 3(h) describes the registration requirements for antimicrobial pesticides, the scope is limited to requirements for process improvements, timeframes for review purposes, and other regulatory matters, but does not contain provisions for data requirements. Under Title 40 of the Code of Federal Regulations (CFR) Part 158 et seq., the U.S. EPA’s final rules on the data requirements for conventional pesticides (EPA 2007b), biochemical pesticides (Subpart U) (EPA 2007a), and microbial pesticides (Subpart V) (EPA 2007a) are listed. As part of those rules, the U.S. EPA preserved the original Data Requirements for Pesticides to provide continued regulatory coverage for antimicrobial pesticides and transferred the original 1984 data requirements of Part 158 into a new Part 161 titled “Data Requirements for Antimicrobial Pesticides” (EPA 2007c). Part 161 contains the current data requirements for antimicrobial pesticide chemicals, although it is intended to be replaced with Subpart W of Part 158 once the U.S. EPA issues a final rule. This specific action for antimicrobial pesticides is in process because the U.S. EPA determined that the original data requirements of 1984 failed to adequately address the unique applications, use patterns, and other factors germane to antimicrobial pesticides. Once a product is approved for registration, the registrant is required, under Section 6(a)(2), to inform the U.S. EPA if the registrant obtains “… additional factual information regarding unreasonable adverse effects on the environment of the pesticide ….” Section 2(bb) of FIFRA defines “unreasonable adverse effects on the environment” to include unreasonable risk to humans. Section 4 of FIFRA requires that the U.S. EPA reregister each pesticide that the U.S. EPA first registered before November 1984. This date was chosen because pesticides registered after 1984 were subject to the Part 158 testing requirements. Section 3(g) of FIFRA also requires the U.S. EPA to periodically review the registrations of all pesticides due to changes in science, public policy, and pesticide use practices, which occur over time. The U.S. EPA promulgated a new registration review program in 2006, as detailed in 40 CFR Part 155, Subpart C, which began to replace the U.S. EPA’s reregistration program as the mechanism for systematic review of existing pesticides.
3.3.5.
The EU Biocidal Products Directive (BPD)
Biocide active substances or products are generally exempt from REACH evaluation and authorization procedures because they are regulated in the EU under the Biocidal Products Directive (BPD) (EC 1998). This Directive has an established process for evaluation of active substances (for listing in Annex I of the BPD), followed by national approvals of the formulated biocidal products containing them. However,
50
CHAPTER 3 HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY
REACH substance evaluation can apply to the active substance if it has been prioritized and placed on the Community Rolling Action Plan, and active substances used in biocidal products may be included in Annex XIV of REACH if they are classified as CMR, PBT, vPvB, or endocrine disrupters. The dossier for an active substance includes information on the applicant, the identity of the active substance and biocidal product, their physical and chemical properties, methods of detection and identification, effectiveness against target organisms, intended uses, their toxicological profiles, their ecotoxicological and environmental fate and behavior properties, measures necessary to protect humans, animals and the environment, EU classification and labeling, and an overall summary and evaluation. The study reports are rewritten as Robust Summaries, and any deviations from the standard methods have to be explained and justified; there are data waiver forms to justify omitting the standard studies. The common core data set for active substances is specified in Annex IIA of the BPD. Additional data selected from Annex IIIA of the BPD are needed for particular product types, as specified in the data requirements Technical Guidance Document (TGD) (EC 1993) to reflect particular exposures, in order to conduct an adequate risk assessment. Furthermore, additional studies, not necessarily restricted to those listed in Annex IIIA of the BPD, may be needed to investigate further ambiguous findings from the standard data set or as an outcome of the risk assessment. Similarly, the common core data set for biocidal products is specified in Annex IIB of the BPD with additional data selected from Annex IIIB for particular product types according to the TGD. Risk assessment is a key part of the EU approval process for Annex I of the BPD listing of active substances and national authorization of biocidal products. The EU risk assessment procedures for biocides are the same as those used for chemical substances. Under the BPD criteria for Annex I inclusion, an active substance may not be authorized for use by the general public if it is classified as a category 1 or 2 carcinogen. In addition, professional use may only be authorized if exposure to humans is unlikely or exposure is below the threshold for the effect.
3.4. SCIENTIFIC ASPECTS OF CARCINOGENIC RISK ASSESSMENT 3.4.1. Dose—Response Relationships in Carcinogenesis and Mechanisms of Carcinogenic Action Risk assessment frequently involves estimating safe exposure concentrations for exposure durations that were not tested experimentally. Generally applicable biologically based models have to be applied. Before developing such a model, extensive data are needed to build its form as well as to estimate how well it conforms to the observed data to support confidence in results. The first benchmark study of dose–response relationships in chemical carcinogenesis was reported by Druckrey (1943) with 4-dimethylaminoazobenzene (4-DAB), also known as “butter yellow,” in BDIII rats. Within the range of daily dosages from 3 to 30 mg per rat, the time up to the appearance of liver cancer (t) was found to be inversely proportional to the daily dose (d). The product of the
51
3.4. SCIENTIFIC ASPECTS OF CARCINOGENIC RISK ASSESSMENT
TABLE 3.1.
Daily Dose, d (mg/rat) 30 20 10 5 3
Induction of Liver Cancer in BDIII Rats by 4-DAB
Median Tumor Induction Time, t (days)
Total Dose, D (mg/rat)
34 52 95 190 350
1020 1040 950 950 1050
daily dosage and the median tumor induction time, which corresponds to the sum of all daily doses—that is, to the total dose, D—was found to be practically constant (Table 3.1): dt = D ~ 1000 mg = constant
(3.1)
Assuming a linear relationship between the daily dosage (d) and the 4-DAB concentration (c) at the site of carcinogenic action, Eq. (3.1) would read as ct = constant
(3.2)
Equation (3.2), that the product of exposure concentration and duration produces a constant toxic effect, is known as Haber ’s rule (Haber 1924), named after the German chemist Fritz Haber, who in the early 1900s characterized the acute toxicity of gases used in chemical warfare. The smaller the effects (Haber) product, c × t, the greater the toxicity of the gas. However, not all gases have a constant effects product. Flury (1921) pointed out that the effects product for hydrocyanic acid (HCA) does not remain a constant, but instead increases with decreasing concentrations of the toxicant in the inspired air. Apparently, agents such as HCA, which are toxic only after their resorption, are better tolerated as the concentration at which they are inhaled becomes smaller, suggesting that detoxification processes are more efficient at low concentrations than they are at higher concentrations. Flury introduced a constant detoxification factor (e) in Haber ’s rule, which appeared sufficient to describe the observations made:
(c − e) t = constant
(3.3)
Clark (1937) further expanded Haber ’s rule for the action of a number of drugs:
(c − cm ) (t − tm ) = constant
(3.4)
where cm is a threshold concentration and tm is a minimum time of response. Clark commented at the time (Clark 1937): The formula ct = constant is indeed an impossible one in the case of drugs acting on biological material because it implies that an infinitely small concentration of a drug will produce the selected action in infinite time, and conversely that a sufficiently high concentration will produce an instantaneous effect. In some cases ct = constant gives an approximate fit, but this merely implies that cm and tm are so small as not to produce a measurable error.
52
CHAPTER 3 HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY
So, the perfect fit of Haber ’s rule to the carcinogenic action of 4-DAB suggested that threshold concentration (cm) and minimum time of response (tm) were (in Clark’s words) so small as not to produce a measurable error. Druckrey and Küpfmüller (1948) provided a theoretical explanation for ct = constant. By denoting the initial concentration of specific receptors that 4-DAB reacts with as R, the concentration of receptors that 4-DAB has reacted with as CR, and the mean 4-DAB concentration at the site of action as C, the reaction kinetics in the case of a bimolecular reaction are dC R dt = K ( R − C R ) C − C R TR
(3.5)
where K is the reaction constant for association and TR is the time constant for dissociation. Druckrey and Küpfmüller then inferred that their experiment had shown that the carcinogenic action of 4-DAB was irreversible, and that as TR → ∞ we obtain dC R dt = K ( R − C R ) C
(3.6)
Now, assuming that up to the time of action we have CR << R, which appears reasonable, then R remains practically constant; therefore, dC R dt = KRC
(3.7)
Since the dose level was kept constant throughout the study, C probably remained constant as well. Integration yields C R = KRCt
(3.8)
which is Haber ’s rule. Druckrey and Küpfmüller also pointed out that if the carcinogenic dose remained constant despite very long latency periods (during which time many cell divisions would have occurred in the liver), the specific receptors (i.e., the targets for chemical carcinogens) would have to be “substances capable of self-replication” (zur Selbstreproduktion befähigte Substanzen) (Druckrey and Küpfmüller 1948). Miller and Miller (1947) and Mueller and Miller (1948) reported 4-DAB metabolism by rat liver microsomes and covalent binding of aminoazo dyes to rat liver proteins (Miller and Miller 1947; Mueller and Miller 1948). It was not until the 1960s that DNA was recognized by Brookes and Lawley (1964) (Brookes and Lawley 1964), as the target for chemical carcinogens, as recently inferred by Wunderlich (2007). Though Eq. (3.8) provided a theoretical explanation for Haber ’s rule, it assumed proportionality between the concentration of occupied receptors and the effect. This may not always be the case. The reversibility of receptor binding can have the same significance for dose–response characteristics as the reversibility of the effect. Denoting the time constant for the reversibility of the effect as Tr, three types of dose–response characteristics were identified by Druckrey and Küpfmüller (1949) when the time constants approach either zero or infinity (Table 3.2) (Druckrey and Küpfmüller 1949). Thus, an effect may be dose-dependent when the time constants for receptor binding and the effect are quite small. Haber ’s rule (ct = constant) may be obtained when either receptor binding or the effect is irreversible. If,
3.4. SCIENTIFIC ASPECTS OF CARCINOGENIC RISK ASSESSMENT
TABLE 3.2.
53
Dose–Response Characteristics According to Druckrey and Kupfmuller (1949)
Reversibility of Receptor Binding
Receptor Binding in Relation to Compound Concentration
TR → 0
CR ∼ C
TR → ∞
CR ∼ ∫C dt
TABLE 3.3.
Tr → Tr → Tr → Tr →
0 ∞ 0 ∞
E E E E
∼ ∼ ∼ ∼
CR ∫CR dt CR ∫CR dt
Effect in Relation to Compound Concentration E E E E
∼ ∼ ∼ ∼
C ∫C dt ∫C dt ∫∫C dt
Dose–Response Characteristics Dose-dependent Ct = constant Ct = constant Reinforced by time
Induction of Ear-Duct Carcinomas in BDII Rats by 4-DAST
Daily Dose, d (mg/kg) 3.4 2.0 1.0 0.5 0.28 0.2 0.1
Reversibility of the Effect
Effect in Relation to Receptor Binding
Median Tumor Induction Time, t (days)
Total Dose, D (mg/kg)
250 340 407 550 607 700 900
850 680 407 275 170 140 90
Source: After Druckrey and Dischler (1963).
however, both receptor binding and the effect are irreversible, the effect would be proportional to the double integral of compound concentration over time. The implication would be that time would reinforce the effect even when the compound is no longer present at the site of action. The first indication that time may reinforce the action of a chemical carcinogen was a significantly smaller total carcinogenic dose (700 mg instead of 1000 mg) at a 4-DAB dose of 1 mg/day with a median liver tumor induction time of 700 days, but Druckrey and co-workers reported more pertinent evidence in 1963 with benchmark studies of the production of ear duct and liver carcinomas by 4-dimethylaminostilbene (4-DAST) (Druckrey and Dischler 1963) and diethylnitrosamine (DENA) (Druckrey et al. 1963), respectively, in BDII rats (Tables 3.3 and 3.4). In contrast to the 4-DAB carcinogenicity study, where the total carcinogenic dose remained practically constant at daily dosages from 3 to 30 mg, the total carcinogenic dose decreased with decreasing daily 4-DAST or DENA dose levels, even though the median tumor induction times increased with decreasing daily dose levels. In a logarithmic system of coordinates, there was a linear relationship between the median tumor induction time (t) and the daily dosage (d):
54
CHAPTER 3 HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY
TABLE 3.4.
Induction of Liver Carcinomas in BDII Rats by DENA
Daily Dose, d (mg/kg)
Median Tumor Induction Time, t (Days)
Total Dose, D (mg/kg)
101 137 192 238 355 457 609 840
963 660 460 285 213 137 91 64
9.6 4.8 2.4 1.2 0.6 0.3 0.15 0.075
Source: After Druckrey et al. (1963).
Median tumor induction time (t), (days)
1000
100 0.1
1
10
Daily dose of diethylnitrosamine (mg/kg-bw) Figure 3.1. A linear relationship in a logarithmic system of coordinates between the median liver tumor induction time (t) in BD II rats and the daily dosage of diethylnitrosamine (d). Linearity leads to Eq. (3.10).
ln d = ln k − n ln t
(3.9)
dt n = constant
(3.10)
or
where the time exponent n was 3.0 and 2.3 for 4-DAST and DENA, respectively (Figure 3.1). There was no evidence for a subthreshold dose for these carcinogens, since no deviation from linearity was recognizable even at the lowest carcinogenic dose level despite latency periods approaching the average life expectancy of the test species (2.5 years). Dose and time dependencies in the form of Druckrey’s equation [Eq. (3.10)] were confirmed in the ED01 study, the largest toxicology experiment ever conducted, examining the carcinogenicity of 2-acetylaminofluorene in about 25,000 mice
3.4. SCIENTIFIC ASPECTS OF CARCINOGENIC RISK ASSESSMENT
55
(Littlefield et al. 1980) and in the BIBRA study with 4000 rats to investigate the carcinogenicity of nitrosamines (Peto et al. 1991). Equation (3.10) was also found to apply to nonmelanoma skin cancer induced by solar ultraviolet radiation (UVR). UVR is usually subdivided into ultraviolet A (UVA) wavebands (UVA2: 315–340 nm and UVA1: 340–400 nm) and the ultraviolet B (UVB) waveband (280–315 nm). The relationship between the daily dose (d) and the median nonmelanoma skin tumor induction time (t) in hairless mice for both UVA1 (de Laat et al. 1997) and UVB (Sterenborg et al. 1988) has been demonstrated to be d r t = constant
(3.11)
dt 1 r = constant
(3.12)
or
where r = 0.62 for UVB and 0.35 for UVA1 and 1/r = 1.6 for UVB and 2.9 for UVA1. In the early 1960s, the interaction of dialkylnitrosamines with nucleic acids had been established. Dialkylnitrosamines are hydroxylated and dealkylated by microsomal drug-metabolizing enzymes, and the resulting diazoalkanes are powerful alkylating agents. Magee and Farber (1962) demonstrated alkylation of nucleic acids at the N7 position of guanine in rats orally exposed to dimethylnitrosamine (Magee and Farber 1962). Warwick and Roberts (1967) observed covalent binding of a 4-DAB metabolite to DNA (Warwick and Roberts 1967). Thus, by the late 1960s, there was strong experimental evidence to indicate that DNA was the target for carcinogens (Wunderlich 2007). UVR has been shown to cause mutations of the gatekeeper tumor suppressor gene that encodes the p53 protein, an outcome that compromises its critical role in the orchestration of the cellular responses to genotoxicity and cytotoxicity (Brash et al. 1991). If time reinforces a carcinogenic effect even when the compound is no longer present at the site of action, it should be possible, in principle, to induce cancer with a single dose of a chemical carcinogen. This was achieved by Druckrey et al. (1970) in single-dose experiments with N-nitroso-N-ethylurea (ENU) in BD IX rats. ENU is a direct-acting ethylating nitrosamide, which is rapidly lost from the blood after intravenous injection; it has an in vivo half-life of 5–6 minutes (Figure 3.2). Upon single-dose treatment in early life (1, 10, or 30 days after birth), the overwhelming majority of animals died from malignancies of the nervous system (Druckrey et al. 1970). However, extra-neural cancer was frequently observed when the animals received single-dose treatment at 3 months after birth (Tennekes and Ivankovic 1984). At a constant age at treatment, the median induction time of neurogenic malignancies decreased with increasing dose levels, as expected (Table 3.5). At a constant dose, the median induction time of neurogenic malignancies increased with increasing age at treatment (Table 3.5). Tennekes and Ivankovic (1984) established a quantitative relationship (Figure 3.3) between (post-conception) age at treatment (a), dose (d), and the median induction period of neurogenic malignancies (t), which can be described as follows:
56
CHAPTER 3 HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY O CH3
N N H2N
Ethylnitrosourea (ENU) O
Heterolysis N O
N
HO NH
CH3
N
CH3
H2O
HO
Diazoethane N
+
-
-
N
CH
CH3 +
H
+
N
Ethyldiazonium
N
H2C
+
CH3
Carbonium ion
CH3
N
N
OH N
N
Guanine H2N
N
N R
Figure 3.2. Alkylation of guanine by ethylnitrosourea.
a t = Kd r , with K = constant, where r = 0.426
(3.13)
d ( t a ) = constant, where n = 1 r = 2.35
(3.14)
or n
Equation (3.14) indicates that the velocity of carcinogenesis is determined by the initiating dose and the state of the relevant targets at treatment, reminiscent of Wilder ’s law of initial value (the direction of response of a body function to any agent depends to a large degree on the initial level of that function). Target cells for
3.4. SCIENTIFIC ASPECTS OF CARCINOGENIC RISK ASSESSMENT
TABLE 3.5.
57
Induction of Neurogenic Malignancies in BD IX Rats by ENU
Single Dose, d (mg/kg)
Age at Treatment, a (days post conception)
Median Tumor Induction Time, t (days)
Age-Dependent Velocity of Carcinogenesis, a/t
25 (1)a 25 (1) 34 (10) 25 (1) 34 (10) 54 (30) 25 (1) 34 (10) 54 (30) 25 (1) 34 (10) 54 (30) 115 (91) 115 (91)
500 310 510 240 360 600 205 300 410 183 215 350 447 420
0.050 0.080 0.067 0.104 0.094 0.090 0.122 0.113 0.131 0.137 0.158 0.154 0.257 0.273
5 10 10 20 20 20 40 40 40 80 80 80 210 250–300 a
Numbers in parentheses denote days after birth.
Source: Druckrey et al. (1970) and Tennekes and Ivankovic (1984).
Median age-dependent velocity of malignant neurogenic tumor formation (a/t)
1
0.1
0.01 1
10
100
1000
Single dose of ethylnitrosourea (d) (mg/kg-bw)
Figure 3.3. Median age-dependent velocity of malignant neurogenic tumor formation (a/t) in BD IX rats versus single dose levels of ethylnitrosourea (d), on logarithmic coordinates. Linearity leads to Eq. (3.14).
ENU in early life are subependymal cells in the brain and subpial cells in the spinal cord, which are destined to differentiate into glial cells—that is, astrocytes or oligodendrocytes. The evidence suggests that critical genetic changes induced by the carcinogen in these target cells lead to inhibition of normal cell differentiation processes, as has been observed in leukemogenesis (Graf et al. 1978; Choe et al. 2003). Yuspa and Morgan (1981) reported that cells resistant to terminal differentiation can be readily isolated from skin of BALB/c mice exposed to an initiating dose
58
CHAPTER 3 HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY
of carcinogen in vivo but not from control mouse skin (Yuspa and Morgan 1981). More recent experimental evidence indicates that resistance to terminal differentiation may be related to activation of a proto-oncogene by the chemical carcinogen. Topical treatment of mouse epidermis with dimethylbenzanthracene (DMBA) results in the activation of the Ha-ras gene (Nelson et al. 1992), and the expression of an activated v-Ha-ras in primary mouse keratinocytes can alter the differentiation program induced by Ca2+ (Yuspa et al. 1985). Though the cells undergo morphological differentiation and cell cycle withdrawal, they do not express certain differentiation markers and they can also revert to the basal cell phenotype if the Ca2+ concentration is reduced, suggesting that they are blocked at an early and reversible stage of differentiation. The v-Ha-ras block of keratinocyte differentiation correlates with altered regulation of both cyclin D1 (an important regulator of G1 to S-phase transition) and cdk2 (a kinase that initiates the principal transitions of the cell cycle) (Martinez et al. 1999). The experiments with a single ENU dose demonstrated that the “initiating” interaction is almost “timeless” and that all events following “initiation” proceed entirely autonomously. This evidence lends supports to the mutator phenotype hypothesis proposed by Loeb (1991) that mutations in caretaker tumor-suppressor genes (involved in DNA repair, mismatch repair, DNA replication or chromosomal segregation) are an early event in carcinogenesis and initiate a cascade of further mutations, resulting in yet greater genetic instability (Loeb 1991). Genetic variation in tumor-suppressor genes could also affect the incidences of common (spontaneous) neoplasms in rats, which show time-dependent trends far more frequently in outbred strains than in inbred strains (Tennekes et al. 2004a,b). For example, small variations in DNA repair are highly heritable. DNA repair efficiency is correlated with cancer risk, and there are widespread amino acid polymorphisms in the known repair genes (Frank 2004). This could explain why numerous carcinogenicity studies with nongenotoxic substances have yielded positive results. A nongenotoxic carcinogen does not form DNA adducts, does not induce DNA repair, and is negative in in vitro or in vivo tests for mutagenicity, but may induce cell proliferation (Melnick 1992) or nuclear polyploidization (an increase in DNA content by whole number multiples of the entire set of chromosomes) in its target organ (Lalwani et al. 1997). Cell proliferation could fix DNA damage into heritable mutations, particularly when DNA repair mechanisms are insufficient, and also lead to clonal expansion of initiated cells. Genetic instability in polyploid cells—in particular, in those with critical mutations in gatekeeper tumor-suppressor genes that encode the p53 and Rb proteins (proteins that regulate cellular responses that prevent the survival or proliferation of potential cancer cells, i.e., apoptosis and cellular senescence)—might provide a route to aneuploidy (any chromosome number that is not an exact multiple of the haploid number) and thereby contribute to the development of cancer (Storchova and Pellman 2004). A crucial question for risk assessment is whether or not the dose–response characteristics for nongenotoxic carcinogens are different from those observed for genotoxic carcinogens, as has been inferred frequently (Ashby and Purchase 1992). Using tumor data from chronic feeding studies with more than 1500 mice at six
59
3.4. SCIENTIFIC ASPECTS OF CARCINOGENIC RISK ASSESSMENT
TABLE 3.6. The Dose–Response Characteristics of Liver Tumor Enhancement in CF-1 Mice by the Nongenotoxic Pesticide Dieldrin
Background Dose Equivalent (d0), in ppm Dieldrin in Dieta 10.2 10.2 10.2 10.2 10.2 10.2 10.2
(935.34)b (871.08) (835.38) (771.12) (664.02) (471.24) (314.16)
Sum of Background Dose (d0) and Actual Dieldrin Dose (δx) (in ppm in diet)
Actual Dieldrin Dose (δx), in ppm in Diet 0 0.1 1 2.5 5 10 20
(8.54) (81.9) (189) (325.5) (462) (616)
b
10.2 10.3 11.2 12.7 15.2 20.2 30.2
(935.34)b (879.62) (917.28) (960.12) (989.52) (933.24) (930.16)
Median Liver Tumor Induction Time, tx (weeks) 131 122 117 108 93 66 44
a The background dose equivalent (d0) for the liver tumor induction period (t0) in control mice, estimated to be equipotent to a level of 10.2 ppm dieldrin in the diet, was based on an observed linear relationship between the reciprocal median liver tumor induction period (= the velocity of tumor development) and dieldrin dose. b
Total tumorigenic dose (in mg/kg body weight) in parentheses, calculated on the basis of an average daily food intake of 100 g/kg body weight. Source: Tennekes et al. (1985).
levels of continuous exposure, Tennekes et al. (1985) showed that the dose–response characteristics of liver tumor enhancement in mice by the nongenotoxic pesticide dieldrin (Table 3.6) were consistent with Haber ’s rule if a background dose equivalent (d0) for the liver tumor induction period (t0) in control mice was determined, (Tennekes et al. 1985). There was a linear relationship on logarithmic coordinates (Figure 3.4) between the sum of background dose (d0) and actual dieldrin dose (δx), denoted as dx, and the median liver tumor induction period (tx): d x t x = d0 t0 ~ 935 mg kg = constant
(3.15)
The perfect fit of Haber ’s rule to the carcinogenic action of dieldrin in mouse liver provided no evidence of a subthreshold dose. Therefore, nongenotoxic carcinogens should not be assumed to exhibit a threshold per se; rather they must be evaluated with regard to mode of action and human relevance, as discussed in Chapter 13. There was a strong correlation between the kinetics of proliferation (i.e., nuclear polyploidization) and tumor formation in livers of mice exposed to dieldrin (van Ravenzwaay et al. 1987). More recent experimental evidence underpins the critical role of proliferation in murine hepatocarcinogenesis. Ha-ras gene mutation is detected frequently in spontaneous and carcinogen-induced mouse liver tumors (Bauer-Hofmann et al. 1990; Reynolds et al. 1987; Wiseman et al. 1986), and this mutation could be a very early, perhaps even the first, critical event during murine hepatocarcinogenesis (Buchmann et al. 1989). However, despite the rapid appearance of large dysplastic hepatocytes, no hepatocellular carcinomas developed from activated H-ras transgenic mice at least for 1 year after infection (Harada et al.
60
CHAPTER 3 HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY
Median liver tumor induction period (t x), (weeks)
1000
100
10 10
100
Background dose (d0 ) + dieldrin dose = dx, (ppm in diet)
Figure 3.4. The sum of background dose (d0) and dieldrin dose (0, 0.1, 1, 2.5, 5, 10, and 20 ppm in the diet) versus the median liver tumor induction period (tx) in CF-1 mice, on logarithmic coordinates. d0 was found to be equipotent to a level of 10.2 ppm dieldrin the diet. The tangent of the angle (45°) is 1. Linearity leads to Eq. (3.15).
2004). Activated H-ras alone appeared insufficient for clonal expansion of dysplastic hepatocytes, but an additional mutation in the ß-catenin gene (encoding a downstream activator in the Wnt signaling pathway), which stabilizes β-catenin, leads to hepatocellular carcinoma at a 100% incidence within 2 months after the infection (Harada et al. 2004). Apparently, this latter mutation promotes proliferation of the dysplastic cells generated by activated H-ras. This interpretation is supported further by the fact that cyclin D1, one of the target genes of Wnt signaling, was overexpressed in hepatocellular carcinomas in these mice.
3.4.2. Mathematical Model for Carcinogenic Risk Assessment The available evidence indicates that Druckrey’s equation [Eq. (3.10)] with n ≥ 1 can serve as a regulatory basis for risk assessment of genotoxic and nongenotoxic carcinogens. Carlborg (1981) pointed out that this equation is implied by a Weibull model for dose–response functions in carcinogenesis, as follows. The simple form of the Weibull model is P = 1 − e exp , with the exponent = − ( α + βx m )
(3.16)
where x is the dose, P is the tumor rate, and m, α, and β are parameters to be estimated from the data. The parameter α is determined by the background tumor probability, β is a scale parameter related to the units measuring the dose, and m is
ACKNOWLEDGMENTS
61
the important shape parameter. At very low doses the excess risk over background is (approximately) βxm. The VSD corresponding to a one-in-a-million risk over background is then given by VSD = (10 −6 β )
1m
(3.17)
The extended form of the Weibull model includes the age of the animals at death (t): P = 1 − e exp , with the exponent = − ( α + βx m ) t k
(3.18)
where k is a new parameter. Now suppose that t measures the time to a tumor. Also, suppose that the background tumor rate is zero (α = 0). For a test group at some dose x, consider the median time to tumor—that is, the value of t such that P = 0.50. The extended Weibull model for this dose and time is 0.50 = 1 − e exp , with the exponent = −βx m t k
(3.19)
[ − ln 0.50 β]1 m = xt k m = xt n
(3.20)
This reduces to
where n = k/m and the left side of the equation is a constant. This is Druckrey’s equation [Eq. (3.10)].
3.5.
CONCLUSIONS
A fundamental goal of toxicology is to determine safe levels of human exposure to toxic substances. Carcinogen risk assessment is based on assessment of appropriate toxicological and exposure data sets. International differences in science policy have been greater for cancer risk assessment compared to other toxic endpoints. There is also a tendency to differentiate cancer risk assessment on the basis of mode of action and relevance to humans (some carcinogenic responses in animals may be considered irrelevant to human risk assessment) (see Section IV) (Di Marco et al. 1998). From a regulatory standpoint, low-dose linear extrapolations are an appropriate default for carcinogens in the absence of mode of action data and a complete human relevance evaluation. Based on benchmark studies of dose– response relationships in chemical carcinogenesis, it is proposed to use the Weibull model for dose–response functions in carcinogenesis for estimating carcinogenic risks.
ACKNOWLEDGMENTS The authors gratefully acknowledge the critical review of the manuscript by Professor Volker Wunderlich of the Max Delbrück Center for Molecular Medicine (MDC), Berlin. Editorial help from Dr. Derek J. Knight and Duncan Harris of the Department of Registration Services at SafePharm Laboratories Ltd. is acknowledged.
62
CHAPTER 3 HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY
REFERENCES Ashby, J., and Purchase, I. F. (1992). Nongenotoxic carcinogens: An extension of the perspective provided by Perera. Environ Health Perspect 98, 223–226. Bauer-Hofmann, R., Buchmann, A., Wright, A. S., and Schwarz, M. (1990). Mutations in the Ha-ras proto-oncogene in spontaneous and chemically induced liver tumours of the CF1 mouse. Carcinogenesis 11, 1875–1877. Brash, D. E., Rudolph, J. A., Simon, J. A., Lin, A., McKenna, G. J., Baden, H. P., Halperin, A. J., and Ponten, J. (1991). A role for sunlight in skin cancer: UV-induced p53 mutations in squamous cell carcinoma. Proc Natl Acad Sci USA 88, 10124–10128. Brookes, P., and Lawley, P. D. (1964). Evidence for the binding of polynuclear aromatic hydrocarbons to the nucleic acids of mouse skin: relation between carcinogenic power of hydrocarbons and their binding to deoxyribonucleic acid. Nature 202, 781–784. Buchmann, A., Mahr, J., Bauer-Hofmann, R., and Schwarz, M. (1989). Mutations at codon 61 of the Ha-ras proto-oncogene in precancerous liver lesions of the B6C3F1 mouse. Mol Carcinog 2, 121–125. Carlborg, F. W. (1981). Dose–response functions in carcinogenesis and the Weibull model. Food Cosmet Toxicol 19, 255–263. Choe, K. S., Radparvar, F., Matushansky, I., Rekhtman, N., Han, X., and Skoultchi, A. I. (2003). Reversal of tumorigenicity and the block to differentiation in erythroleukemia cells by GATA-1. Cancer Res 63, 6363–6369. Clark, A. J. (1937). General pharmacology. In Handbuch der Experimentellen Pharmakologie, Vol. 4, Heubner, W., and Schuller, J., eds., Springer Verlag, Berlin/New York, pp. 123–142. Crump, K. S. (1984). A new method for determining allowable daily intakes. Fundam Appl Toxicol 4, 854–871. de Laat, A., van der Leun, J. C., and de Gruijl, F. R. (1997). Carcinogenesis induced by UVA (365-nm) radiation: The dose–time dependence of tumor formation in hairless mice. Carcinogenesis 18, 1013–1020. Di Marco, P. N., Priestly, B. G., and Buckett, K. J. (1998). Carcinogen risk assessment. Can we harmonise? Toxicol Lett 102–103, 241–246. Druckrey, H. (1943). Quantitative Grundlagen der Krebserzeugung. Klin Wochenschr 22, 532. Druckrey, H., and Dischler, W. (1963). Dosis-wirkungsbeziehungen bei der Krebserzeugung durch 4-dimethylaminostilben bei Ratten. Z Krebsforsch 65, 272. Druckrey, H., and Küpfmüller, K. (1948). Quantitative Analyse der Krebsentstehung. Z Naturforschung 3b, 254–266. Druckrey, H., and Küpfmüller, K. (1949). Dosis und Wirkung. Beitrage zur theoretischen Pharmakologie, Editio Cantor GmbH, Freiburg im Breisgau, Germany. Druckrey, H., Schagen, B., and Ivankovic, S. (1970). Induction of neurogenic malignancies by one single dose of ethyl-nitrosourea (ENU) given to newborn and juvenile BD IX-strain rats. Z Krebsforsch 74, 141–161. Druckrey, H., Schildbach, A., Schmaehl, D., Preussmann, R., and Ivankovic, S. (1963). Quantitative analysis of the carcinogenic effect of diethylnitrosamine. Arzneimittelforschung 13, 841–851. EC (1993). Technical guidance documents in support of the risk assessment directive (93/67/EEC) for new substances notified in accordance with the requirements of Council Directive 67/548/EEC. Brussels, Belgium, Commission of the European Communities. EC (1998). Directive 98/8/EC of the European Parliament and of the Council of 16 February 1998 concerning the placing of biocidal products on the market. Official Journal of the European Communities L123. EC (2006). Regulation (EC) No 1907/2006 of the European Parliament and of the Council of 18 December 2006 concerning the registration, Evaluation, Authorisation and Restriction of Chemicals (REACH), establishing a European Chemicals Agency, amending Directive 1999/45/EC and repealing Council Regulation (EEC) No 793/93 and Commission Regulation (EC) No 1488/94 as well as Council Directive 76/769/EEC and Commission Directives 91/155/EEC, 93/67/EEC, 93/105/EC and 2000/21/ EC. Official Journal of the European Communities L136, 3.
REFERENCES
63
EC (2008). Regulation (EC) No 1271/2008 of the European Parliament and of the Council of 16 December 2008 on classification, labeling and packaging of substances and mixtures, amending and repealing Directives 67/548/EEC and 1999/45/EC, and amending regulation (EC) no 1907/2006, Official Journal of the European Union L353, pp. 1–353. ECHA (2008a). Guidance on information requirements and chemical safety assessment, Chapter R.4: Evaluation of available information. Guidance for the implementation of REACH, pp. 1–23. ECHA (2008b). Guidance on information requirements and chemical safety assessment. Part A: Introduction to the Guidance Document. Guidance for the implementation of REACH, http://reach.jrc. it/docs/guidance_document/information_requirements_part_a_en.pdf?vers=20_08_08, pp. 1–44. ECHA (2008c). Guidance on information requirements and chemical safety assessment. Part F: Chemical Safety Report. Guidance for the Implementation of REACH, http://reach.jrc.it/docs/guidance_document/information_requirements_part_f_en.pdf?vers=20_08_08, 1–10. EPA (1991). Alpha2u-globulin: Association with chemically induced renal toxicity and neoplasia in the male rat, EPA/625/3-91/091F, pp. 1–118. EPA (1998). Assessment of thyroid follicular cell tumors, EPA/630/R-97/002, pp. 1–52. EPA (2000a). Benchmark dose technical guidance document, EPA/630/R-00/001, pp. 1–96, http://www. epa.gov/nceawww1/pdfs/bmds/BMD-External_10_13_2000.pdf. EPA (2000b). Voluntary Children’s Chemical Evaluation Program. Fed Reg 65, 81700–81718. EPA (2003). A summary of general assessment factors for evaluating the quality of scientific and technical information, EPA100/B-03/001, pp. 1–18. EPA (2005). Guidelines for carcinogen risk assessment, EPA/630/P-03/001F, pp. 1–166. EPA (2007a). Data requirements for biochemical and microbial pesticides. Fed Reg 72, 60988–61025. EPA (2007b). Data requirements for conventional pesticides. Fed Reg 72, 60933–60988. EPA (2007c). Redesignation of part 158; technical amendments. Fed Reg 72, 60251–60255. EPA (2008a). Chemical Assessment and Management Program (ChAMP). http://www.epa.gov/champ/. EPA (2008b). TSCA inventory reset and inorganic high production volume challenge programs; notice of public meeting. Fed Reg 73, 70640–70643. EPA (2009a). High production volume (hpv) challenge—guidance documents. http://www.epa.gov/hpv/ pubs/general/guidocs.htm. EPA (2009b). Registration for Download of Benchmark Dose Software | National Center for Environmental Assessment, U.S. Environmental Protection Agency, Washington, D.C. http://www.epa.gov/ncea/ bmds/progreg.html. FIFRA (1972). Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA), 7 U.S.C. §136 et seq. Flury, F. (1921). Ueber Kampfgasvergiftungen: I. Ueber Reizgase. Z gesamte experimentelle Medizin 13, 1–15. Frank, S. A. (2004). Genetic predisposition to cancer—Insights from population genetics. Nat Rev Genet 5, 764–772. Graf, T., Ade, N., and Beug, H. (1978). Temperature-sensitive mutant of avian erythroblastosis virus suggests a block of differentiation as mechanism of leukaemogenesis. Nature 275, 496–501. Haber, F. (1924). Zur Geschichte des Gaskrieges. In Fünf Vorträge aus den Jahren 1920–1923, Julius Springer, Berlin, pp. 76–92. Harada, N., Oshima, H., Katoh, M., Tamai, Y., Oshima, M., and Taketo, M. M. (2004). Hepatocarcinogenesis in mice with beta-catenin and Ha-ras gene mutations. Cancer Res 64, 48–54. IARC (1999). Species differences in thyroid, kidney and urinary bladder carcinogenesis. IARC Sci Publ 147, 1–225. IARC (2006). IARC Monographs on the Evaluation of Carcinogenic Risks to Humans—Preamble. http:// monographs.iarc.fr/ENG/Preamble/CurrentPreamble.pdf, Lyon, France. Illing, P. (2001). Toxicity and Risk: Context, Principles and Practice, Taylor and Francis, London. Klimisch, H. J., Andreae, M., and Tillmann, U. (1997). A systematic approach for evaluating the quality of experimental toxicological and ecotoxicological data. Regul Toxicol Pharmacol 25, 1–5. Knight, D. J., and Thomas, M. B. (2003). Practical Guide to Chemical Safety Assessment. Rapra Technology, Shrewsbury, U.K. Lalwani, N. D., Dethloff, L. A., Haskins, J. R., Robertson, D. G., and de la Iglesia, F. A. (1997). Increased nuclear ploidy, not cell proliferation, is sustained in the peroxisome proliferator-treated rat liver. Toxicol Pathol 25, 165–176.
64
CHAPTER 3 HAZARD AND RISK ASSESSMENT OF CHEMICAL CARCINOGENICITY
Littlefield, N. A., Farmer, J. H., Gaylor, D. W., and Sheldon, W. G. (1980). Effects of dose and time in a long-term, low-dose carcinogenic study. J Environ Pathol Toxicol 3, 17–34. Loeb, L. A. (1991). Mutator phenotype may be required for multistage carcinogenesis. Cancer Res 51, 3075–3079. Magee, P. N., and Farber, E. (1962). Toxic liver injury and carcinogenesis. Methylation of rat-liver nucleic acids by dimethylnitrosamine in vivo. Biochem J 83, 114–124. Martinez, L. A., Chen, Y., Fischer, S. M., and Conti, C. J. (1999). Coordinated changes in cell cycle machinery occur during keratinocyte terminal differentiation. Oncogene 18, 397–406. Melnick, R. L. (1992). Does chemically induced hepatocyte proliferation predict liver carcinogenesis? FASEB J 6, 2698–2706. Miller, E. C., and Miller, J. A. (1947). The presence and significance of bound aminoazo dyes in the livers of rats fed p-dimethylaminoazobenzene. Cancer Res 7, 468–480. Mueller, G. C., and Miller, J. A. (1948). The metabolism of 4-dimethylaminoazobenzene by rat liver homogenates. J Biol Chem 176, 535–544. Nelson, M. A., Futscher, B. W., Kinsella, T., Wymer, J., and Bowden, G. T. (1992). Detection of mutant Ha-ras genes in chemically initiated mouse skin epidermis before the development of benign tumors. Proc Natl Acad Sci USA 89, 6398–6402. NRC (1983). Risk Assessment in the Federal Government: Managing the Process Working Papers, National Academy Press Washington, D.C., http://www.nap.edu/openbook.php?isbn=POD115& page=R1. NTP (2008). Report on Carcinogens (RoC)—National Toxicology Program. http://ntp.niehs.nih.gov/ntp/ roc/eleventh/intro.pdf. OECD (1991). Decision-recommendation of the Council on the co-operative investigation and risk reduction of existing chemicals. C(90)163/Final, http://webdomino1.oecd.org/horizontal/oecdacts.nsf/ linkto/C(90)163. OECD (2004). The 2004 OECD list of high production volume chemicals. Environment Directorate, Organisation for Economic Co-operation and Development, pp. 1–143, http://www.oecd.org/dataoecd/55/38/33883530.pdf. OECD (2007). Manual for investigation of HPV chemicals—OECD Secretariat, July 2007. http://www. oecd.org/document/7/0,3343,en_2649_34379_1947463_1_1_1_1,00.html. Persad, A. S., Stedeford, T., and Dourson, M. (2007). Classifying chemicals as carcinogens: An analysis of the weight-of-evidence descriptors used by IARC, IRIS, and NTP. Nat Resour Law J 1, 156–223. Peto, R., Gray, R., Brantom, P., and Grasso, P. (1991). Effects on 4080 rats of chronic ingestion of Nnitrosodiethylamine or N-nitrosodimethylamine: A detailed dose-response study. Cancer Res 51, 6415–6451. Postle, M., Vernon, J., Zarogiannis, P., and Salado, R. (2003). Assessment of the impact of the new chemicals policy on occupational health [final report]: Prepared for the European Commission— Environment Directorate—General, Risk and Policy Analysts Limited, Norfolk, U.K., pp. 1–96. Reynolds, S. H., Stowers, S. J., Patterson, R. M., Maronpot, R. R., Aaronson, S. A., and Anderson, M. W. (1987). Activated oncogenes in B6C3F1 mouse liver tumors: Implications for risk assessment. Science 237, 1309–1316. Stedeford, T., and Persad, A. S. (2007). The influence of carcinogenicity classification and mode of action characterization on distinguishing “like products” under Article III:4 of the GATT and Article 2.1 of the TBT agreement. N.Y.U. Environ Law J 15, 377–419. Sterenborg, H. J., van Weelden, H., and van der Leun, J. C. (1988). The dose–response relationship for tumourigenesis by UV radiation in the region 311–312 nm. J Photochem Photobiol B 2, 179–194. Storchova, Z., and Pellman, D. (2004). From polyploidy to aneuploidy, genome instability and cancer. Nat Rev Mol Cell Biol 5, 45–54. Tennekes, H., Gembardt, C., Dammann, M., and van Ravenzwaay, B. (2004a). The stability of historical control data for common neoplasms in laboratory rats: Adrenal gland (medulla), mammary gland, liver, endocrine pancreas, and pituitary gland. Regul Toxicol Pharmacol 40, 18–27. Tennekes, H., Kaufmann, W., Dammann, M., and van Ravenzwaay, B. (2004b). The stability of historical control data for common neoplasms in laboratory rats and the implications for carcinogenic risk assessment. Regul Toxicol Pharmacol 40, 293–304.
REFERENCES
65
Tennekes, H., van Ravenzwaay, B., and Kunz, H. W. (1985). Quantitative aspects of enhanced liver tumour formation in CF-1 mice by dieldrin. Carcinogenesis 6, 1457–1462. Tennekes, H. A., and Ivankovic, S. (1984). A quantitative relationship between dose, age at treatment, and the median induction period of neurogenic malignancies in single dose experiments with N-nitrosoN-ethylurea in BD IX rats in post-natal life, and its implications [unpublished report]. The German Cancer Research Centre, Heidelberg, Germany. TSCA (1976). Toxic Substances Control Act, 15 U.S.C. §2601 et seq. van Ravenzwaay, B., Tennekes, H., Stohr, M., and Kunz, W. (1987). The kinetics of nuclear polyploidization and tumour formation in livers of CF-1 mice exposed to dieldrin. Carcinogenesis 8, 265–269. Warwick, G. P., and Roberts, J. J. (1967). Persistent binding of Butter Yellow metabolites to rat liver DNA. Nature 213, 1206–1207. WHO (1994). Assessing human health risks of chemicals: Derivation of guidance values for health-based exposure limits. Environmental Health Criteria 170, http://www.inchem.org/documents/ehc/ehc/ ehc170.htm. Wiseman, R. W., Stowers, S. J., Miller, E. C., Anderson, M. W., and Miller, J. A. (1986). Activating mutations of the c-Ha-ras protooncogene in chemically induced hepatomas of the male B6C3 F1 mouse. Proc Natl Acad Sci USA 83, 5825–5829. Wunderlich, V. (2007). “Substances capable of self-reproduction” as cellular targets of chemical carcinogens. Ntm 15, 271–283. Yuspa, S. H., Kilkenny, A. E., Stanley, J., and Lichti, U. (1985). Keratinocytes blocked in phorbol esterresponsive early stage of terminal differentiation by sarcoma viruses. Nature 314, 459–462. Yuspa, S. H., and Morgan, D. L. (1981). Mouse skin cells resistant to terminal differentiation associated with initiation of carcinogenesis. Nature 293, 72–74.
CH A P TE R
4
USE OF CANCER RISK ASSESSMENTS IN DETERMINATION OF REGULATORY STANDARDS Robert A. Howd Anna M. Fan
4.1.
INTRODUCTION
Risk assessment has been identified as an appropriate tool for assigning health risks their rightful order and priority, and it should be a principal component of public programs (Rodricks 1994). The application of such a tool can be more successful when a broader view of its nature and content is adopted. It can help to guide research and the allocation of research funds, as well as to inform decisions about public health. However, many of the laws guiding federal programs require the agencies to answer questions for which direct, relevant empirical evidence is not available. For example, data are lacking on human exposure to low chemical levels, for which risk cannot be directly measured and for which there is only animal data in most cases. Regulators are faced with ever-increasing detections of chemicals in the environment because of enhanced analytical capabilities, which has also resulted in more laws relating to the presence of these toxic chemicals. Basing regulatory decisions merely on the presence of chemicals—related ultimately to detection limit—is not useful. Quantitative cancer risk assessment is widely employed to provide a scientific basis for regulatory decisions, for specific chemicals and exposure situations, and for establishing broader policies related to regulations of chemicals in the environment. Determination of cancer risk for setting regulatory standards is a complex process, in addition to the underlying biological complexities of tumor induction and growth. In the face of limited data, assumptions have to be used to assess risk. To achieve consistency, specific sets of assumptions and default values are widely applied to carcinogens, especially to extrapolate from animal data to human risks.
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
66
4.1. INTRODUCTION
67
The no-threshold concept for carcinogens is one of these assumptions, which has influenced public policy and resulted in carcinogens being singled out for special treatment. Use of defaults can be considered as a practical discipline to enable action in the face of residual uncertainties. Health-protective assumptions are considered necessary in order to avoid underestimating risks. The default assumption of a linear slope for cancer dose–response through zero dose is often criticized, but general alternatives have not yet been justified. Following adoption of defaults, including extrapolation to low doses, a range of risk levels can be developed and considered for risk management purposes. The risks considered sufficiently low to protect public health (e.g., insignificant or “de minimis” risk) are defined. The current default methods for cancer risk assessment are little changed from the 1980s despite the great advances that have been made since then in understanding of tumor biology. This is because the correspondence between induction of tumors in a particular animal strain and in the varied human population is still not adequately understood. Some defaults have been seen as too conservative, or not adequately conservative, by various interest groups. Different assumptions can be used for specific risk assessments, but the hurdle that must be overcome to utilize different assumptions has, in practice, been very high. Risk reduction measures to be developed in any particular case might consider the strength of evidence for the default assumptions in that case. The requirements of the applicable laws must also be considered. Testing for carcinogenicity of chemicals is mostly conducted in chronic studies in rodents (Milman and Weisburger 1994; Doull et al. 2007). Rats and mice were chosen for these studies because of their common use in toxicology studies, a good understanding of their husbandry and biological parameters, the fecundity and health of the chosen strains, and their relatively short lifespan. However, there are many differences between rodents and humans—as well as between mice and rats—which lend great uncertainty to extrapolation of the results of rodent bioassays to humans. Risk assessment practice assumes concordance between species in carcinogenic effects, while acknowledging that different tissues and organs may be affected in different species (Rodricks and Turnbull 1994; Gray et al. 1995; EPA 2005a). One underlying assumption is that chemicals that can interact with DNA, leading to tumors, should be able to do this in many different cell types. However, differences in cellular function and metabolic processes among tissues and among species clearly alter the relative chemical reaction rates. In addition, not all cancer results from interactions with DNA. Tumors may be generated by direct alkylation of DNA by a reactive chemical or its metabolite, by increasing cell turnover secondary to cytotoxicity or hormonal effects, or by many other alterations of cellular homeostasis, which is inherently tissue-specific. Therefore, risk assessors need to consider the differences among tissues and metabolic processes in order to apply the results of cancer bioassays to risk assessment for humans. Consideration of the differences in tumor response to a particular chemical between rats and mice and between males and females of each species helps in this process. The animal tests are designed to provide a certain level of sensitivity in tumor detection, and the tumor incidence data are used to develop cancer potency factors
68
CHAPTER 4 USE OF CANCER RISK ASSESSMENTS
(Hamm 1994; Fan 2005). With 50 animals per group, a 10% increase in tumor incidence is about the minimal detectable rate, with a low background of tumors in the responsive tissue. The extrapolation of this rate to a level of concern for humans— usually one cancer case per 10,000 to one case per million—obviously introduces very large uncertainties. (Risk assessment from human studies is discussed separately below.) Yet an increase in the number of animals to increase the sensitivity for detection of tumors has not seemed desirable, because the vastly increased effort and expense would then limit the number of animal tests that can be performed. Instead, risk assessment attempts to understand the process of tumor induction in this limited sample in order to inform the extrapolation process. One way this is done is through the choice of doses. Several doses ranging from the maximum tolerated dose (MTD) to a minimally toxic or nontoxic level are planned. Evaluation of the noncarcinogenic toxic effects versus the tumor responses at each dose is used to probe the biological responses to a chemical. For example, if cytotoxicity and hyperplasia occur in response to increasing doses, and these effects correlate with tumor incidence, it may mean that the tumors are occurring as a side effect of the cytotoxicity, and the chemical is not a direct-acting carcinogen. This could mean that noncytotoxic doses will not cause tumors and that a linear extrapolation of cancer risk to zero dose is not appropriate, as many researchers have concluded for chloroform (Schmidt 1999; Pontius 2000; EPA 2001a; Beddowes et al. 2003). Supporting data from studies in isolated rodent tissues and the standard genotoxicity assays in Salmonella and other species can help inform the mechanistic interpretations. However, a risk assessor must evaluate these data with caution. Are the results from liver (where regenerative hyperplasia is most often observed) relevant to tumors that may be observed in other organs? If a chemical caused tumors in liver, bladder, and lung, for example, but correlative cytotoxicity was only documented in liver, what should be concluded from the liver cytotoxicity data? The data may seem solid for a cytotoxic threshold for liver tumors in rats, while kidney tumors are observed without notable renal cytotoxicity in mice, and a genotoxic metabolite is detected. In this case, a cautious risk assessor would not likely conclude that a nonlinear extrapolation is appropriate for determination of a regulatory standard, based on the mouse data. These interpretations of the biological responses in animals must be extended, for risk assessment, to humans. To carry on the above example of a chemical for which liver tumors are concluded to be secondary to cytotoxicity in rat liver, but doubts are introduced by the mouse data, it would be relevant to question whether humans are more similar to the rat or to the mouse. Comparative studies with liver slices of various species can be useful in evaluation of the metabolism of chemicals (Steensma et al. 1994). Cytochrome P450s, which are responsible for much of the Phase 1 metabolism of chemicals in liver, are often the source of genotoxic metabolites (Alvares 1981; Schut and Castonguay 1984; Green et al. 2001). Variations in cytochrome P450s occur because of differences in genetic background, lifestyle (e.g., smoking), life stage (infants, adolescents, adults, the elderly), sex, and disease (Vesell 1978; Dollery et al. 1979; Kato and Kamataki 1982; Shimada et al. 1994). One might discover that a genotoxic metabolite is detectable in liver slices at low
4.1. INTRODUCTION
69
levels only in smokers and in a small percent of adult male nonsmokers from a Mediterranean background. How should these data be applied to the risk assessment? Should a linear assumption be made for cancer calculation in males, and a threshold assumption in females? According to standard practice, the linear assumption would be applied to the entire population, and it would be assumed that there is a cancer risk from exposure to the chemical over the entire lifetime, although the metabolite was only detected in adult males. Estimation of population risk from studies in intact humans is treated much the same as animal data. Occupational exposures to carcinogens are generally at higher doses than occur from environmental exposures to chemicals, and a linear extrapolation to zero dose is the standard method to apply the available cancer incidence data to the general population. Ecological human studies (which involve observations of increased tumor rates across a population, without individual exposure measurements) may provide good evidence of carcinogenicity of a chemical, but usually provide inadequate data to establish a dose–response (Greenland and Robins 1994; Lubin 2002). In such a case, the cancer risk assessment could be derived from animal tumor data, despite the fact that the most compelling data on human tumor risks are from the human study. A judgment is made as to whether the extrapolation from animals is more or less uncertain than the extrapolation from the ecological study, although this may or may not be explicit in the risk assessment (Milman and Weisburger 1994; Wartenberg et al. 2000; EPA 2005a). In other cases, such as for arsenic, asbestos, and tobacco smoke, the human data are inherently superior. The potential for a nonlinear extrapolation from human data should also be considered. Some risk assessors have argued that benzene-induced leukemia and lymphoma is a threshold phenomenon, based on mechanistic considerations, and that the human data demonstrate such a threshold (Cox and Ricci 1992; Cox 1996; Yokley et al. 2006). However, regulatory risk assessors have not yet accepted these arguments for benzene (Bailer and Hoel 1989; EPA 1998; OEHHA 2001). Considerations include the multiple genotoxic metabolites of benzene, a nonlinear production of protein and DNA adducts in humans (saturated at higher benzene doses), and the difficulty of establishing dose–response relationships at lower doses with the available animal and human data (Bailer and Hoel 1989; Henderson et al. 1992; Turteltaub and Mani 2003; Rappaport et al. 2005; Lin et al. 2007). The cancer risk assessment principles and practices have been documented over the last few years in the U.S. Environmental Protection Agency (EPA) draft guidelines. After several iterations, these were recently finalized (EPA 2005a). The guidelines could still be considered a work in progress, to the extent that the Supplemental Guidance for children’s exposures to carcinogens (EPA 2005c) is a separate document, and the recommended correction factors for children have not yet been incorporated into approved risk assessments. Exposure assessments for cancer still utilize the traditional default of 2 L/day for drinking water consumption, despite recent more specific guidance on drinking water consumption rates (EPA 2004a, 2008a,b). The following discussion covers how these cancer risk assessment methods have been applied in various programs, for many different purposes.
70
CHAPTER 4 USE OF CANCER RISK ASSESSMENTS
4.2. 4.2.1.
AIR STANDARDS Scientific Issues
Data on increased cancer in humans after inhalation exposure to carcinogens have contributed greatly to identification of human carcinogens, but use of this information in estimation of carcinogenic potency has been more limited (Morello-Frosch et al. 2000). This is due to the difficulty of estimating long-term exposure to human populations. Documentation of exposures in occupational situations through regular air sampling has, however, made it possible to directly calculate carcinogenic potency of some chemicals in humans. Examples of this are benzene (Rinsky et al. 1987; Hayes et al. 1997; Lan et al. 2004), asbestos (Gilson 1976; de Klerk et al. 1989; Hodgson and Darnton 2000; Hein et al. 2007), and hexavalent chromium (Crump et al. 2003; Park et al. 2004; Park and Stayner 2006). Potency estimates for other common air pollutants, such as diesel exhaust, have been difficult for a number of reasons, including the complexity of quantitating exposures to the variable mixture of compounds (Dawson and Alexeeff 2001; Ris 2007). For the great majority of chemicals, therefore, animal data have been used to estimate human potency. This has led to intensive research on anatomical differences in airways between test species and humans (Lippmann and Schlesinger 1984; Reznik 1990; Harkema 1990; Fang et al. 1993; Jarabek et al. 2005). The main issues in extrapolation of animal toxicity data to humans include (a) interspecies scaling related to body size, breathing rate, and metabolic rate, (b) anatomical differences leading to differences in deposition in both upper and lower airways, including terminal bronchiole, (c) particle overload of lungs in experimental studies, (d) crossroute extrapolations, and (e) multiroute exposures in inhalation studies. Interspecies scaling of inhalation exposures has been studied for decades (Kliment 1973; De Sesso 1993; Jarabek et al. 2005), but has recently been improved by development of models which incorporate differences among species in both anatomy and physiology (Frederick et al. 1998, 2002; Andersen et al. 2002; Conolly et al. 2004; Tsujino et al. 2005). Differences in lung deposition among species are related to airflow rates, size of the conducting vessels, and breathing patterns (e.g., breathing through the mouth versus the nose) (Lippmann and Schlesinger 1984; Jarabek et al. 2005). Net uptake of solvent vapors appears to be similar among species, at about 50–70% of the respiratory volume (Raabe 1986; Dallas et al. 1994; Fisher et al. 2000; Bouchard et al. 2001); uptake of water-soluble compounds can be greater, probably due to solubilization in the upper airways (Johanson 1991; Perkins et al. 1995; Overton et al. 2001). The intention in toxicity studies to evaluate a dose range including the MTD leads to the phenomenon of lung particle overload for low-toxicity, poorly soluble particles (Vincent et al. 1985; ILSI 2000). In this situation, the mucociliary transport mechanism, which normally clears the lung of insoluble particles, is overwhelmed by high inhalation exposures to the particles. This leads to deposition and accumulation of particles in any accessible lung areas (dependent on particle diameters), pulmonary inflammation, and possibly tumors (Oberdörster 1995a,b). It has been controversial whether the high-particle-load studies, particularly in rat, are predictors
4.2. AIR STANDARDS
71
of human tumorigenicity (Valberg and Watson 1996a,b; Oberdörster 1995a, 2002; Brockmann et al. 1998; Borm et al. 2004; Valberg et al. 2006). Consideration of the anatomical and physiological differences between rats and humans in particle handling can help in interpretation of these results (Brown et al. 2005). Cross-route extrapolation is very important in cancer risk assessment because some chemicals have been found to be carcinogenic only by a single route, especially inhalation, whereas exposures may be multiroute. Estimates of systemic dose from inhalation versus oral exposure for cross-route potency calculations are straightforward for most chemicals (EPA 1994, 2002). However, route-specific carcinogenicity may occur through inhalation because (1) deposition of insoluble particles in lungs causes cancer, which is not applicable to any other exposure route, and (2) local irritant or cytotoxic effects can provide unique effects (i.e., tumors of nasal turbinates) that may or may not be applicable for other exposure routes. Formaldehyde is acknowledged as a carcinogen by inhalation, but is not considered to be carcinogenic by oral exposure (Conolly et al. 2004; EPA 2008a). For other chemicals, such as hexavalent chromium and nickel, the results have been more controversial and difficult to interpret (Sunderman 2001; Costa and Klein 2006; Sedman et al. 2006, Beaumont et al. 2008). Regulatory decisions on whether a chemical found to produce tumors at a route of entry should be considered to be a multiroute carcinogen are based on several considerations. Among these are genotoxicity of the chemical, whether the tumors are preceded by tissue lesions and regenerative hyperplasia, and prevalence of precursor lesions in other tissues. Adequacy of studies by other exposure routes (both positive and negative) is also relevant. Simultaneous exposure by multiple sources and routes may also be relevant for a contaminant found in air, with different doses and potency values for each exposure. An example of this is the radioactive element radon. As a gas, it seeps into homes from the soil, adheres to particles in the air, and in this form can be retained in the lung and cause lung cancer (Nero et al. 1986; Field 2001). Radon is also found in tap water, and from this source it can both be drunk and provide an additional radon source to air (Nazaroff et al. 1987); however, increased tumor rates from the oral exposure have not been reported. A comprehensive risk assessment would combine risks from both sources, perhaps accompanied by a cost–benefit analysis for each in order to determine how to best allocate resources (EPA 1999a). Hexavalent chromium is another example with both inhalation and oral exposures. Several human studies demonstrate its carcinogenic potency by inhalation, while its oral carcinogenicity is based on an oral study in animals (NTP 2007). The estimated inhalation potency of hexavalent chromium is much greater than the oral potency. Inhalation exposure to suspended droplets in showering represents a small dose, but it might contribute to the carcinogenicity from drinking the water, because of the huge potency difference. Conversely, because inhaled droplets can be trapped in the respiratory tree, cleared from the lung by the “mucociliary escalator,” and swallowed (Jarabek et al. 2005), there could be a gastric cancer risk from occupational inhalation exposures (Sedman et al. 2006). Evaluation of this contingency should be incorporated into analysis of the risk from airborne contaminants. Consideration of multiple exposure routes is even more important in animal inhalation studies, because chemicals may be accumulated in fur, and also might be
72
CHAPTER 4 USE OF CANCER RISK ASSESSMENTS
significantly absorbed by the dermal route (Phalen et al. 1984; Ballantyne et al. 2006). This effect is of particular concern for aerosols, because they can be entrained in fur and licked off by the animals, so that the major dose may be oral instead of inhalation. Although some moderately lipophilic chemicals may be dermally absorbed at significant rates, this is more relevant for studies in small animals such as mice than for humans because of the greater surface–volume ratio of the mice. Both effects may result in greater doses than can be accounted for by concentration times breathing rate (Pauluhn 2003b). Nonhomogenous distribution of aerosols in large chambers can also result in variability in doses among animals (Phalen 1976; Pauluhn and Mohr 2000; Pauluhn 2003a,b). As a result, individual “nose-only” exposure chambers are used for some studies instead of whole-body exposure chambers, although these may be stressful to the animals, which also can cause increased or more variable uptake (Phalen et al. 1984; Dorato and Wolff 1991; Pauluhn 2005). Each of these issues must be considered in interpretation of cancer potency data, from which standards for exposure to chemicals in air (or other routes) could be derived. However, utilization of these data in development of air standards presents many further challenges, as discussed below.
4.2.2.
Regulatory Considerations
Evaluation of carcinogenicity by inhalation has been facilitated by a wealth of information from occupational epidemiology studies, which in some respects defined the field of cancer risk assessment. Early information on lung cancer caused by cigarette smoking, emissions from coke ovens, and asbestos (Doll 1957; Lloyd et al. 1970; Knox et al. 1968), along with data on liver angiosarcomas from vinyl chloride (Tabershaw and Gaffey 1974), focused attention on cancer by the inhalation route after both occupational and nonoccupational exposures. These data provided support for the development of various regulatory measures, including (a) the federal Clean Air Act (1970) and its amendments and (b) the Toxic Substances Control Act of 1976. A permissible exposure limit (PEL) of 12 fibers per cubic centimeter of asbestos in air (later lowered) was included in the first promulgation of standards by the U.S. Occupational Safety and Health Administration (OSHA) in 1971. However, this was intended to protect employees against asbestosis, not cancer (OSHA 2008). It should be noted that the National Ambient Air Quality Standards authorized by the Clean Air Act are not based on carcinogenicity. Emissions from individual sources can be closely regulated, both within a facility (occupational standards) and outside it (with New Source Performance Standards at the federal level and/or Air Toxic Hot Spots regulations in California). The National Institute of Occupational Safety and Health (NIOSH) began in 1970, with the passage of the Occupational Safety and Health Act, to develop recommended exposure limits (RELs) for chemicals in the workplace. In 1974, NIOSH joined with OSHA to update the OSHA program for PELs for a wide variety of substances, incorporating cancer potency data as it became available over subsequent years. Their evaluations were published in criteria documents, Special Hazard Reviews, and summarized in a “Compendium of Policy Documents and Statements” (NIOSH 1992). Available information is periodically updated in the NIOSH Pocket
4.3. WATER STANDARDS
73
Guide to Chemical Hazards (NIOSH 2005). The RELs are intended to be “based on human and/or animal data, as well as on the consideration of technological feasibility for controlling workplace exposures to the REL … [whereas] [u]nder the old policy, RELs for most carcinogens were nonquantitative values labeled ‘lowest feasible concentration (LFC)’ ” (NIOSH 2005). However, the paucity of recent cancer potency evaluations or updates makes it appear that this is not a NIOSH priority (NIOSH 2008). The State of California, under its Air Toxics Hot Spots Program, has provided a compilation of the most currently available cancer potency values for a long list of chemicals in air. The latest version of the Technical Support Document provides cancer unit risks and potency factors for 122 of the 201 carcinogenic substances or substance groups which are regulated in the California Air Toxics program (OEHHA 2005). These values are used by the California Air Resources Board (ARB) in prioritization of regulatory actions under the Toxic Air Contaminant Program (ARB 2008a). In addition, under the Air Toxics Hot Spots Program, emissions from individual stationary sources are addressed. In this program, air modeling is used to evaluate cancer risks from emissions; nearby residents are notified of excess risks from emissions, and facilities are pressed to reduce those emissions to de minimis risk levels (i.e., less than 10−6 lifetime extra risk) (ARB 2008b).
4.3.
WATER STANDARDS
4.3.1.
Scientific Issues
Risk assessment of chemicals for development of drinking water standards has customarily given precedence to exposures by the oral route. This is important because of the potential for route-specific effects, especially those resulting from site-of-contact toxicity. Local irritation followed by cytotoxicity, tissue lesions, and repair, resulting in increased cell turnover, is a well-known mechanism of tumor promotion (Jones et al. 1983; Cohen and Ellwein 1988; Albert et al. 1991; Butterworth et al. 1995; Scott and Cogliano 2000) and certainly should be presumed to be a viable mechanism of carcinogenesis for chemicals in drinking water (Bogen 1990). Such effects that occur with inhalation exposures (irritation in the nasal passages or lungs, followed by development of tumors) may not be observed after oral exposures (Liteplo and Meek 2003; Conolly et al. 2004). Risk assessment for development of drinking water standards must therefore specifically consider whether the effects observed in the available cancer bioassays are applicable to the drinking water route. Reasons for route-specific toxicity include: • • • • •
Concentration-dependent effects Vehicle-dependent effects Instability of some chemicals in aqueous solution Binding of chemicals to other ingested substances Conversion of ingested substances to less active forms by gastrointestinal enzymes or the acidic environment of the stomach
74
CHAPTER 4 USE OF CANCER RISK ASSESSMENTS
Concentration-dependent effects related to administration of chemicals in drinking water include (a) use of low concentrations because of low solubility in the aqueous phase or poor palatability (bad taste), (b) lack of accumulation of chemicals in the gastrointestinal (GI) tract, compared to accumulation in lung after inhalation administration, or (c) poor gastrointestinal absorption. A classic example of these factors is represented by asbestos fibers. Accumulation of the fibers in lung through inhalation exposures results in carcinogenicity (mesothelioma), but the fibers are very poorly absorbed during passage through the gastrointestinal tract, and few (if any) tumors are observed (OEHHA, 2003). The fine asbestos fibers are suspended in water when administered; but in other cases, chemicals are dissolved to help ensure homogeneity of the dosing medium. Low solubility in water would help limit the potential for irritant effects or cytotoxicity. However, in most cases, a solvent or suspending agent (like Tween 80) is added to ensure that the concentration can be high enough to reach an MTD. This strategy has limitations for administration in drinking water, to the extent that a bad taste or odor at higher concentrations may lead to decreased drinking. Decreased drinking will also depress food consumption, which may result in lower body weight gain. This can result in leaner animals, with increased lifespan and decreased cancer rates (Jorgenson et al. 1985). To overcome palatability problems, oral gavage administration of chemicals may be utilized. However, gavage administration commonly results in higher peak concentrations and potential for site-of-contact effects, because of the bolus doses. In rats and mice, the rodent species most commonly used for cancer bioassays, both water and food are consumed over several hours, which results in lower, sustained absorption of chemicals. The high peak of absorbed concentrations observed with gavage dosing may result in a greater potential for concentration-related tissue damage, especially in the liver, compared to administration in drinking water or the diet. Saturation of metabolism of the chemical may occur, which can result in a greater or lesser delivery of a reactive chemical to sensitive tissues, depending on whether the carcinogenic moiety comes before or after the rate-limiting step. Peak blood concentrations are especially dependent on systemic absorption rate for rapidly metabolized or cleared substances. Chemicals may therefore be observed to cause increases in tumor rates after administration by gavage, but not when administered in drinking water or food. Whether this implies a threshold effect (and use of pharmacokinetic adjustments or nonlinear cancer potency estimation methods) must be decided in the risk assessment process. Vehicle-dependent effects are relevant not only because solvents and suspending agents allow delivery of higher concentrations, but also because the choice of vehicle can alter absorption rate. This may be related to decreased stomach emptying rate (for vegetable oils) or altered partitioning among fluid phases in the GI tract and across the epithelial cell membranes. Lipophilic compounds dissolved in oil will not be absorbed until the oil phase is acted upon by bile salts, converted to chylomicrons, and absorbed. If the compounds are dissolved or suspended in water, they may partition much more rapidly into cells; this may be facilitated by rapid absorption of a water-miscible solvent such as alcohol or DMSO. Instability of chemicals in aqueous solution can result in a finding that a chemical found to be carcinogenic by another route, such as inhalation, is not carcinogenic
4.3. WATER STANDARDS
75
by oral ingestion. Formaldehyde is one example of this, where hydration of formaldehyde in solution greatly alters its properties, compared to formaldehyde gas. It can be considered to be a site-of-contact carcinogen by inhalation (Kimbell et al. 2001; Conolly et al. 2004), but studies by the oral route are negative or inconclusive (EPA 2008a). Binding of chemicals to other substances can greatly affect toxicity. The classic example of this is the administration of charcoal or burnt toast to people who have ingested poisons, to adsorb the chemical and prevent GI absorption. Lipophilic chemicals for which this mechanism would be most likely (e.g., dioxins), tend to bind tightly to sediments, and therefore are filtered out in water treatment plants, so they do not get delivered to the tap. If found in tap water, their GI absorption is also likely to be low. Conversion of ingested substances to less active forms by GI enzymes or the acidic environment of the stomach is a factor for some environmental contaminants. Hexavalent chromium provides a good example, because it can be rapidly reduced to the relatively nontoxic trivalent form in the stomach. The recently completed positive cancer bioassay on hexavalent chromium conducted by the U.S. National Toxicology Program (NTP) appears to indicate, however, that this conversion is not complete at the concentrations administered (NTP 2007). All the aforementioned factors must be considered in determining whether excess tumors observed in both human and animal studies are appropriate for use in setting regulatory limits for drinking water, or just how they will be considered.
4.3.2.
Regulatory Consideration
It has been customary in development of U.S. regulatory levels of chemicals in drinking water to assume that humans consume 2 liters per day of water for a 70year lifetime (EPA 1997a). This has been considered to be a health-protective assumption (compared to use of mean intake) because 2 L/day corresponds to about the 75th percentile of total water consumption in adults, according to the survey of Ershow and Cantor (1989). Current movement is toward use of more specific, databased values in risk assessment, especially for sensitive subgroups such as infants (EPA 2004a), with chemicals likely to have short-term effects. For cancer, the exposures can be averaged over a lifetime. Another exposure issue is the concept of relative source contribution—that is, adjustment of the water regulatory level to account for exposures to a chemical through other media (i.e., air, food, dermal uptake). This procedure is standard for noncancer endpoints, where it is assumed that the total exposures must exceed a toxicological threshold for effects to be observed. Cancer risk calculations usually do not assume a threshold, and the cancer risk is calculated based on “extra risk” from the water route, independent of other exposures. Therefore, no relative source contribution is included in the calculation. This convention may have to be altered if cumulative risk assessment becomes the norm, which evaluates combined risk from exposure to all sources of similarly acting chemicals. Establishment of drinking water standards by the EPA involves determination of a maximum contaminant level goal (MCLG), followed by establishment of the
76
CHAPTER 4 USE OF CANCER RISK ASSESSMENTS
actual regulatory level, called the maximum contaminant level (MCL), as defined by the Safe Drinking Water Act of 1974 and its 1986 amendments (EPA 2001b). The MCL must be set as close as possible to the MCLG, considering both cost and technical feasibility. State standards must be set equal to or lower than the federal MCL. For carcinogens, federal policy is to set the MCLG to zero, based on the premise that the goal should be zero exposure to all carcinogens through tapwater. However, recognizing that this is impractical, especially for disinfection byproducts and widespread environmental contaminants such as arsenic, EPA guidelines call for MCLs to be established not to exceed a lifetime cancer risk of 1 in 10,000 (Donohue and Miller 2008). This is much larger than the de minimis risk of one in a million utilized in some other contexts because of potential risk–benefit considerations as well as practicality. Disinfection of water with chlorine, for example, prevents many more illnesses and deaths from cholera and other water-borne diseases than would be caused by the disinfection byproducts (trihalomethanes and dozens of minor halogenated compounds). Nevertheless, standards are set as low as feasible to limit potential for cancer (and other potential adverse effects). California performs risk assessments to develop public health goals (PHGs) for chemicals in drinking water, similar to the federal process for development of MCLGs, except that the PHG for carcinogens is established at the one-in-a-million lifetime risk level. The PHG is used in development of a California MCL (equal to or less than the federal MCL). The use of a specific (nonzero) value provides a more explicit comparison to the cancer risk represented by the final regulatory standard. This approach also makes it readily apparent whether the MCL actually achieves the maximum target risk of less than 1 in 10,000. Theoretical lifetime cancer risks from drinking water at the MCLs are shown in Table 4.1. States are allowed to set lower regulatory limits under the federal Safe Drinking Water Act in order to provide lower risks from chemicals in drinking water when it is feasible to do so. Thus, a state that has low levels of arsenic in its drinking water supplies may wish to set a limit much lower than the federal level of 10 ppb. This also provides a regulatory mechanism through which pollution from point sources within a state can be controlled, so as to avoid degradation of cleaner ground or surface water supplies.
4.4. FOOD STANDARDS, PESTICIDE TOLERANCES, ADDITIVES, AND IMPURITIES 4.4.1.
Scientific Issues
Risk assessment of low levels of chemical carcinogens in food presents a particular challenge for consistent scientific input into the regulatory process. There is no comprehensive guideline for such assessment in common foods or dietary items, nor a regulatory program or publication that systematically evaluates methods or conducts independent investigations. In the United States, risk assessment for carcinogens in foods has been on a case-by-case basis using the general approach and methodology and quantifying the concentration of the chemical in food (and other
4.4. FOOD STANDARDS, PESTICIDE TOLERANCES, ADDITIVES, AND IMPURITIES
77
TABLE 4.1. Estimated Lifetime Cancer Risks of Chemicals in Drinking Water at the Federal MCL
Cancer Risk per Million People Exposed Chemical
MCL (ppb)
Arsenic Benzene Benzo[a]pyrene Carbon tetrachloride Chlordane Chromium 1,2-Dichloroethane Dichloromethane Di(2-ethylhexyl)phthalate Ethylene dibromide Heptachlor Heptachlor epoxide Hexachlorobenzene Polychlorinated biphenyls (PCBs) Pentachlorophenol Toxaphene Tritium Vinyl chloride
10 5 0.2 5 2 100 5 5 6 0.05 0.4 0.2 1 0.5 1 3 20,000 pCi/L 2
U.S. EPAa 500 8 42 19 20 NA 13 1 2 125 52 52 46 5 3 96 NA 84
OEHHAa 2500 7 50 5 3 1670b 1 1 0.3 5 1 2 33 6 2 100 50 10
a
EPA values from Integrated Risk Information System (IRIS) (2008), OEHHA values (rounded to 1 significant figure) from Public Health Goal documents, available at http://www.oehha.ca.gov/water/phg/allphgs.html.
b
Based on calculation for hexavalent chromium, the predominant form found in finished drinking water supplies.
sources) to assess exposure. Food contains many carcinogenic constituents; some are intentionally added (bromate; also see additives discussed below), some are naturally occurring (aflatoxin), and some occur as unintentional contaminants (polychlorinated biphenyls, dioxin, acrylamide). Relevant considerations include concentration of the parent compound in the raw foods, formation of toxic breakdown products, concentration of residues during processing, and transfer of chemicals from food packaging materials. Each of these considerations may affect how the presence of the carcinogen is regulated in the final food product, although in every case the ultimate intent is to limit dietary exposures to carcinogens.
4.4.2.
Regulatory Considerations
The precautionary approach to food safety in the United States is administered by the U.S. Department of Agriculture (USDA) for most meat products, by the U.S. Food and Drug Administration (FDA) for most prepared foods, and by the EPA for pesticides applied to both plants and animals. Regulations involving a precautionary
78
CHAPTER 4 USE OF CANCER RISK ASSESSMENTS
approach are found in 21 CFR 70.40, 70.50, 170.20, and 170.22. Regulations control use of a variety of chemicals directly on animals and plants, use of food additives (flavorings, stabilizers, pigments and dyes, etc.), levels of chemicals absorbed from packaging, and tolerances for chemicals in the final food product. The controls may be in the form of outright bans on the use of a chemical or treatment process, binding regulatory limits, or advisory action levels to control unavoidable contaminants in food that may be poisonous or deleterious to health (21 U.S.C. 342, 346; 21 CFR 109. 4). These regulations have limited permissible levels of aflatoxin in a way that applies pressure upon producers of peanuts and corn to minimize the formation of aflatoxin during growing and storage (39 FR 42751. 1974). Tolerances were set for polychlorinated biphenyls (PCBs) (21 CFR 109.15, 109.30) and for lead in ceramic ware (109.16) as well as other action levels. In 1999, the U.S. Department of Food and Agriculture and the FDA restricted imports of products that might be dioxin-contaminated. In certain cases, the FDA has applied a negligible risk concept for food additives. This is demonstrated in the case of dimethyl dicarbamate, a yeast inhibitor for use in beverages (FDA 2000). The additive eventually decomposes to methanol and carbon dioxide, but in the presence of ammonium ions (not uncommon in certain beverages) a carcinogenic chemical may also be formed in small amounts. The FDA used formal quantitative risk assessment procedures to estimate the upper-bound limit of carcinogenic risk to humans posed by urethane generated by decomposition of the additive. It was concluded that the potential risk was sufficiently low that the additive would be safe for the requested use, and the FDA’s final rule approved its use (56 FR 40502 1988). A Risk Assessment of Genotoxic Carcinogens in Food Task Force has been set up to improve quantitative assessment of cancer risk from low dietary exposure to genotoxic carcinogens and to contribute to the prevention of diet-related cancer (ILSI 2008). In 2005, a European Food Safety Authority/World Health Organization (EFSA/WHO) international conference was held with support of ILSI Europe on “Risk Assessment of Compounds that are both Genotoxic and Carcinogenic: New Approaches.” The conference concluded that in order to better define the level of health concern associated with a certain MOE (margin of exposure) (or range of MOEs), it would be informative to calculate MOEs for selected examples. The task force set up an Expert Group that worked in close collaboration with WHO/IPCS and EFSA to carry out this work. The expert group prepared case studies on 12 different chemicals, including among others acrylamide, benzene, and leucomalachite green. Between 2005 and 2008, the Expert Group was developing a framework for the evaluation of genotoxic carcinogens and a set of recommendations to integrate mechanistic information in the risk assessment for publication in a scientific journal. In 2008, a workshop was held in Rhodes, Greece on “Application of the Margin of Exposure (MOE) Approach to Genotoxic Carcinogens in Food,” to critically review the work of the Expert Group. The overall objectives of the workshop were to critically appraise the MOE approach in the light of these assessments, provide guidance on the application of the MOE approach, and further characterize the interpretation of the numerical value of the MOE (e.g., banding). A manuscript was discussed at the workshop. The revised manuscript and case studies are intended
4.4. FOOD STANDARDS, PESTICIDE TOLERANCES, ADDITIVES, AND IMPURITIES
79
for publication in the peer-reviewed scientific literature. In addition, a workshop summary report will be published in the ILSI Europe Report Series. O’Brien et al. (2006) and Barlow et al. (2006) described some of the related earlier work in this area. Progress on regulating carcinogens in foods has been most intense for pesticides during the last few decades. The overall impact of these efforts includes great advances in the biological science, improving cancer risk assessment methodology, and using cancer risk assessment to guide legislation. To some extent this was driven by the U.S. Food Quality Protection Act (FQPA) of 1996, which amended the Federal Food, Drug, and Cosmetic Act (FFDCA) of 1938 and its Delaney Clause (added in 1958), which prohibited carcinogenic food additives (Jordan and Timm 2006), as summarized below. The FQPA established a single set of standards and procedures applicable to all tolerance-setting decisions. The earlier FFDCA had two provisions on food additives and pesticides that did not fit reasonably together. The U.S. FDA was required in 1954 to establish tolerances or maximum permissible levels for pesticides in raw agricultural commodities, which allowed for weighing the risk of consumption versus the benefit for growing the commodity. A food was considered adulterated if it contained either a pesticide for which no tolerance had been established or an amount of pesticide that exceeded the tolerance. In 1958, Congress added a provision subjecting additives to a similar regulatory scheme; however, it precluded consideration of benefits of using the additives. The additive provision also included a strict ban on additives that had been found to induce cancer in animals or humans, regardless of the potency, the relevance to humans, and the level of risk it might pose, which became known as the Delaney Clause. A paradox thus existed in that pesticide residues that concentrate in foods during processing would be considered food additives, and thus were subject to the Delaney Clause and considered unsafe, whereas the same level of carcinogenic pesticide residues in fresh food could be considered safe. In 1970, the responsibility for administration of the pesticide provision of the FFDCA was transferred to the EPA as part of the reorganization plan that established the EPA. During the 1970s and 1980s, more advanced risk assessment methods became available, and more toxicological data were generated that identified some pesticides as animal carcinogens. In some cases, EPA did consider potency, human relevance, and negligible exposure in setting pesticide tolerances, rather than a strict application of the Delaney Clause. A need for studying the impact of implementing the Delaney Clause on public health was identified by both the EPA and its critics. The U.S. National Academy of Sciences (NAS) was requested by the EPA to study this issue; NAS (1987) concluded that application of the Delaney Clause had decreased public health protection overall, because it had led the EPA to approve the use of noncarcinogenic pesticides to address agricultural pest problems that were more risky than potentially carcinogenic alternatives. Although the “de minimis” (negligible risk) approach was advanced by the EPA to address carcinogenic pesticides, it was challenged and disallowed until the regulatory restructuring under the FQPA as discussed below [see EPA (2000c)]. A program was established under the USDA in 1991 to sample fresh fruits and pesticides for analysis of pesticide residues. Other food products and processed
80
CHAPTER 4 USE OF CANCER RISK ASSESSMENTS
commodities were added later. These data facilitated the calculation of consumer dietary exposures to pesticides. In 1993, NAS (1993) released a report, commissioned by the U.S. Congress, which concluded that EPA’s approach to assess risk to children and infants had failed to keep pace with scientific developments. The basis for this conclusion included the differences between children and adults and variations in dietary exposures. The report recommended changes to the EPA’s approaches used to assess dietary risk in the following areas: toxicity studies, uncertainty factors, food consumption data, pesticide residue data, and risk assessment methods. Major advancements have since been made in the USDA’s Pesticide Data Program and the national survey of food consumption data. During the same period, starting in the 1970s, there was a growing awareness that certain environmental contaminants may have the potential to interact with estrogen receptors and affect human health and wildlife. The effects include interaction with the endocrine system, endocrine-related diseases, developmental abnormalities, and interference with normal reproduction, development, and growth. The chemicals of potential concern have been discussed and referred to as endocrine disruptors. The public wanted an aggressive program to address the potential endocrine effects of pesticides. The issues relating to the Delaney Clause, the 1993 NAS report, and endocrine disruptors created an environment for the passage of new legislation. A consensus developed for elimination of the Delaney Clause for pesticides, language to compel the EPA to move ahead with the recommendations of the 1993 NAS report, and a program to assess the potential endocrine effects of pesticides. The resulting legislation, the FQPA of 1996, provided a single risk-only pesticide safety standard for all types of food, effective immediately. It applies to decisions on proposed new tolerances and all tolerances in existence when it became a law. The reevaluation involved 9600 tolerances within 10 years. Elements in the FQPA on children’s sensitivity and endocrine effects are not related to cancer risk assessment, a focus of this chapter, and are not further discussed. However, it can be noted that for cancer risk assessment, the EPA has developed a guideline document for addressing exposure early in life (EPA 2005c). This Supplemental Guidance describes approaches that EPA could use in assessing cancer risks from exposures to children from 0 to 16 years of age. This document is consistent with the National Research Council’s 1994 recommendation that “EPA assess risks to infants and children whenever it appears that their risks might be greater than those of adults” (NRC 1994). The NTP plays an important role in influencing the scientific decisions of regulatory programs. Its toxicology and carcinogenicity studies provide critical data used in setting environmental standards for carcinogens. In addition, the FDA uses the NTP studies in its evaluation of food additives. For example, the FDA cited the NTP studies in regard to the following impurities in food additives: p-chloroaniline hydrochloride, an impurity in the food additive CI pigment Red 202 (2,9-dichloro5,12-dihydroquinone[2,3-b]-acridine-7,14-dione), used as a colorant in polymers in contact with food; tetrachloroethylene, an impurity in the food additive 4,5-dichloro3H-1,2-dithiol-3-one, used as a slimicide in the manufacture of food-contact paper and paperboard; benzene, an impurity in ethylene–norborene copolymers which could contact dry food (Wolfe and Portier 2006).
4.5. SOIL STANDARDS
4.5. 4.5.1.
81
SOIL STANDARDS Scientific Issues
Cancer risk assessment for exposure to chemicals in soil involves several unique exposure issues. Volatile chemicals in soil may evaporate off soil particles and thus may result in a potential inhalation hazard, but little or no hazard from dermal or oral exposure. Semi-volatile and nonvolatile chemicals may bind tightly to soil, and thus be less available for absorption. Unlike air, water, and food, soil is not directly consumed (in most cases), but is contacted incidentally while carrying out other activities. Thus exposure scenarios and soil exposure factors are more variable and important for chemicals in soil than in other media. The EPA Exposure Factors Handbook (EPA 1997a) has summarized these principles; especially relevant in the handbook are Chapter 4 on soil ingestion, Chapter 6 on dermal exposures, and Chapter 17 on residential exposures. Recently, this has been supplemented with child-specific exposure guidance (EPA 2008b). Estimation of cancer risk from exposure to chemicals in soil has been especially problematical because of the standard assumption of lifetime (70-year) exposures. With exposure in most cases being dependent on individual activity patterns, which vary greatly over a lifetime, it is difficult to document appropriate health-protective scenarios. Residential exposures to vapors from soil have been somewhat less problematical, because exposures can be assumed to be passive and long-term, but modeling methods are still an issue (Olson and Corsi 2001; Shan and Javandel 2005; Mills et al. 2007). Dermal exposure to soil is greatest for children who play in the dirt, as well as for farmers and other workers with direct, regular, soil contact. Residential scenarios also consider home gardeners. Exposure estimates for ingestion of soil (including dust) are based on incidental ingestion, and they are greatest for infants who have repeated hand-to-mouth contact. Purposeful soil ingestion, called pica, is not accounted for in site risk assessments, both because it is of relatively low incidence and because it would be prohibitively expensive or impossible to clean up soil to the point that people who eat soil would be protected from all contaminants (LaGoy 1987; Calabrese et al. 1997; EPA 2005b, 2008b). Cancer risks from soil exposure must also take into account the availability of the chemicals. Metals in soil are often in insoluble forms and can bind tightly to other soil minerals. Highly lipophilic chemicals also bind tightly and thus have limited bioavailability. These factors are applicable to both the dermal and oral exposure routes (Dean and Ma 2007; Ljung et al. 2007; Moody et al. 2007; Saghir et al. 2007; Reifenrath et al. 2008).
4.5.2.
Regulatory Considerations
Application of cancer risk assessment values in the context of soil contamination is conducted in the United States under the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA) of 1980, also known as the Superfund program. This law was enacted to address abandoned hazardous waste sites through development of a mechanism to provide funding and a process to rate the hazard of
82
CHAPTER 4 USE OF CANCER RISK ASSESSMENTS
sites and follow through on their stabilization or cleanup. Later amendments— namely, the Superfund Amendments and Reauthorization Act (SARA) of 1986 and the Small Business Liability Relief and Brownfields Revitalization Act of 2002— further delineate the legal processes, as described by EPA (2008c). The Resource and Conservation Recovery Act of 1976, as amended, governs the disposal of hazardous waste (EPA 2008d). Violations of this law and its enabling regulations also may involve cancer risk assessments, which are generally carried out under guidelines developed for Superfund cleanups. The risk assessment procedures for Superfund sites were described in the Risk Assessment Guidance for Superfund (RAGS) manuals (EPA 1989), which have been expanded and clarified in many subsequent publications (i.e., EPA 1997b, 2000a, 2001c, 2003; ECOS-DoD 2007). Cancer potency values utilized in the site characterizations are derived from the EPA’s IRIS (IRIS 2008) or a hierarchy of other sources (EPA 2003). Risk assessment is conducted for carcinogens using the same methods as for other programs, with exposure defaults and the assumption of linear extrapolation through zero risk, unless scientific consensus supports a nonlinear extrapolation (EPA 2005a). Multiroute exposure assessments are conducted, which include evaluation of exposure to soil from direct dermal contact, incidental soil ingestion, inhalation of dust derived from the site, and breathing of chemical vapors (when applicable) (EPA 1989). Risk estimates are conducted for different site-specific scenarios, involving possible industrial and/or residential exposure patterns, and with different assumptions involving soil exposure (surface soil replacement, clay caps, paved surfaces, etc.). Site cleanup decisions generally involve cleanup to theoretical lifetime cancer risk levels in the range of 10−4 to 10−6. In the United States, the EPA is the lead agency on major waste sites, but there are thousands of smaller sites across the United States, for which states often take the lead. Because it is not possible to do a comprehensive cancer and noncancer risk assessment for each site (abandoned gas stations with petroleum hydrocarbon in the soil, for instance), the risk assessment process has been simplified with predefined cleanup standards for most common contaminants. These are often called Preliminary Remediation Goals (PRGs) (EPA 2007). These may be modified by individual states to meet their own exposure or risk standards (CleanupLevels 2008).
4.6. 4.6.1.
CONSUMER PRODUCT STANDARDS Scientific Issues
Many national standards for consumer products other than food have been based on limiting exposure to carcinogens. However, product standards have rarely utilized a specific cancer risk level, and thus there appear to be few scientific issues unique to development of product standards based on cancer. The major consideration would be the difficulty in arriving at accurate population-based exposure estimates for cancer risk from products; but since specific cancer risks are not generally utilized, this point may be moot.
4.6. CONSUMER PRODUCT STANDARDS
4.6.2.
83
Regulatory Considerations
Rather than regulating carcinogens in products based on estimated cancer risk levels, the carcinogen or product has more often simply been banned; in other cases a practical level of control has been established, with or without promulgation of regulations. The value of the product in commerce appears to have been the major consideration in many cases; in others, offsetting public health benefits are more relevant. Carcinogenic potency, its applicability to humans, and the amount of exposure may also have been considered, but perhaps in a more qualitative fashion. The use of benzene in gasoline provides a good example of these considerations. Benzene is clearly a human carcinogen, and its leukemic potency is known reasonably well from occupational studies (EPA 1998; OEHHA 2001). However, it is a near-unavoidable component of the aromatic fraction added to gasoline to increase octane rating. The Clean Air Act amendments (1990) recognized benzene as a hazardous air pollutant, and regulated gasoline formulation to limit emissions, based on a practical level of control rather than a specific objective for decreasing cancer risk. The most recent “Mobile Source” regulations continue this approach to reduce benzene levels as far as practical, based on cancer risk, without a specific risk target level (EPA 2008e). Manufacturers voluntarily removed benzene from other consumer products decades ago, without the necessity for a strict ban (CPSC 1981). Regulation of asbestos in products has had an interesting history. Although lung disease from inhalation of asbestos fibers has been noted since Roman times, increased cancer risk was not well documented until the 1970s. The first promulgated air exposure standard for asbestos (in 1971) of 12 fibers/cm3 was based on prevention of occupational asbestosis, not cancer. Through a lengthy period of regulatory proposals, often followed by court decisions overturning them (NIOSH 2005; OSHA 2008), the standard was eventually lowered to a reference exposure level of 0.1 fiber/cm3. Use of asbestos declined sharply in the 1970s and 1980s, and in 1989 the EPA promulgated a rule to ban and phase out all product uses of asbestos. The asbestos phase-out was protested by industry, and after an adverse decision by the Fifth U.S. Court of Appeals in 1991, the rule was repealed. Use of asbestos was then still legal in vinyl asbestos floor tile, pipeline wrap, roof coatings, brake pads, and many other products, but not in “new” product uses (EPA 1999b). However, asbestos is also regulated under a host of other laws (EPA 2008f), and asbestos litigation has resulted in billions of dollars in legal settlements for asbestosis and lung cancer. Recognizing their legal liabilities, industrial manufacturers have by now reformulated all these products to avoid asbestos use (White 2004). Regulation of pesticides in household products is conducted under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) and its amendments (EPA 2008g.) Application to specific products has been somewhat less rigorously controlled than for pesticides used in food products, as discussed above. Over the last couple of decades, registrations for the most troublesome carcinogenic pesticides have been canceled, because less-toxic replacement chemicals became available (EPA 2000b, 2008h). The EPA periodically reviews pesticide uses and hazards for continued product registration. Results are summarized in the Reregistration
84
CHAPTER 4 USE OF CANCER RISK ASSESSMENTS
Eligibility Documents (REDs), which provide discussions of risks versus benefits. Cancer risks greater than 10−6 are generally deemed worthy of mitigation, which might include reformulation, use restrictions, cancellation, or voluntary withdrawal of the product by the manufacturer. Control of cancer risk from pesticides can be illustrated by the example of chromated copper arsenate (CCA), used as a wood preservative. The risk estimate was particularly contentious, but despite arguments over the specific numbers, a cancer hazard to children from the arsenic in this product used in decks and play structures could not be denied. These and other residential-related uses were finally phased out in late 2003 (EPA 2008i). CCA is still available for commercial construction. Arsenic continues to be available for household use in ant baits, although the potential for human exposure in this product is relatively low. 1,4-Dichlorobenzene (DCB) in mothballs provides a different type of pesticide example. A cancer bioassay conducted by NTP (1987) found excess tumors in both rats and mice. The International Agency for Research on Cancer (IARC) concluded that DCB is reasonably anticipated to be a human carcinogen, and it is listed by the State of California as a chemical known to cause cancer (see below). DCB in mothballs results in low vapor levels in areas where it is used, and vapors emanate from clothes worn after storage with DCB. An estimated human inhalation cancer risk value is available (OEHHA, 2005). Despite the widespread exposure and recognized cancer risk, there are no restrictions on the use of DCB for this purpose. Cancer risk estimates have been made for many chemicals under the California Safe Drinking Water and Toxic Enforcement Act of 1986 (Proposition 65), introduced and passed by California citizens. The Proposition requires companies to warn citizens about significant exposures to carcinogens caused by their products or other activities, which has been defined by regulation as exposure exceeding a 1 in 100,000 lifetime cancer risk level. (Note: Warnings about exposure to reproductive and developmental toxicants are also required). A specific procedure for calculating cancer potency is in the regulations implementing the statute. Following this procedure, No Significant Risk Levels (NSRLs) are developed and published regularly by California’s Office of Environmental Health Hazard Assessment (OEHHA). This law has had considerable national impact because many products have been reformulated for national sales to decrease exposure to carcinogens and avoid the mandatory labeling for sales in California. Several hundred chemicals are currently listed as carcinogens, and NSRLs have been published for many of them (OEHHA 2008).
4.7. RECENT DEVELOPMENTS AND FUTURE DIRECTIONS Recent developments in toxicology and risk assessment will have an impact on the future direction of carcinogenicity assessment and chemical regulations. Some of these developments have been discussed by Fan and Howd (2008), including advances in the areas of mode of action (MOA), genomics, nanotoxicology, bioinformatics, and sensitive populations. Additional discussion regarding recent
4.7. RECENT DEVELOPMENTS AND FUTURE DIRECTIONS
85
developments is presented below on genomics, structure-based thresholds of toxicological concern, the precautionary principle, and the European Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) regulation. In the area of carcinogenicity assessment, genomics will have a significant regulatory impact on possible nonlinear extrapolation for nongenotoxic carcinogens (EPA 2004b). Currently, nongenotoxic carcinogens without plausible MOA data are subjected to the same linear low-dose extrapolation applied to genotoxic carcinogens. The basic premise for the linear default is that electrophilic compounds are presumed to produce single DNA modifications in single cells that could potentially lead to cancer. Biomarkers of response that are predictive of specific outcomes for genotoxic materials can be identified through gene expression pattern recognition, correlated with histological changes in tissues and the eventual progression to tumors. Chemicals that do not activate these patterns should then be assumed to act through a nongenotoxic MOA, which may involve threshold mechanisms. An example is the work to identify genes associated with peroxisome proliferators, such as the peroxisome proliferator-activated receptor alpha (PPAR-α) that are linked to alterations in mouse hepatocellular growth following exposure to the proliferators, as well as to a host of other regulatory mechanisms (Michalik et al. 2004; Carlberg and Dunlop 2006). The threshold of toxicological concern (TTC) concept has been developed to provide criteria for risk assessment decision-making in the absence of detailed information on chemicals. The approach involves estimating a tolerable human exposure value for all chemicals below which there is a very low probability of an appreciable risk to human health (Kroes et al. 2004, 2005), based on their chemical structures, compared to an extensive toxicity database. As utilized by U.S. FDA in their Threshold of Regulation procedure, structural alerts for high-potency carcinogenicity are included, to increase the assurance of safety. The TTC concept was adopted by the Joint FAO/WHO Expert Committee on Food Additives (JECFA) to evaluate flavoring agents in food, and it is also now used by the European Food Safety Authority. In the TTC decision-tree approach of Kroes et al. (2004), proteins, heavy metals, and dioxins were excluded because the database used to derive TTC values did not include proteins and heavy metals, and the extreme species-dependence of the dioxins and related compounds made it less useful for this category (compared to the existing toxicity equivalence factor method). This approach could be much more widely used to categorize trace chemicals in the environment as well as help prioritize the thousands of untested chemicals for further evaluation. While scientific advances are being made, political influences are also important. This can be seen in some cases such as the court-forced incorporation of the nonlinear method for EPA’s assessment of chloroform instead of the linear extrapolation method. This brings to mind the importance of evaluation of carcinogenesis mechanisms by scientists rather then being “settled” in the courts. International decisions and deliberations are also affecting regulation of carcinogens (or at least the debate about them) in the United States. An example is the precautionary principle, which is often applied in the context of the impact of human actions on human health and the environment, in which the consequence of actions
86
CHAPTER 4 USE OF CANCER RISK ASSESSMENTS
may be unpredictable. Under this principle, if an action or policy might cause severe or irreversible harm to the public or to the environment, in the absence of a scientific consensus that harm would not ensue, the burden of proof falls on those who advocate taking the action (Raffensberger and Tickner 1999). The government environmental and regulatory programs that conduct risk assessments for carcinogens to support the establishment of drinking water and air quality standards can also be considered as means of applying the precautionary principle. Another aspect of international interest is a new European Community Regulation on chemicals, known as the REACH (Regulation EC 1907/2006). Directive 67/548 concerns the classification, packaging, and labeling of dangerous substances and applies in parallel with REACH. The two most important aims are to improve protection of human health and the environment from the risks of chemicals while enhancing the competitiveness of the EU chemicals industry. REACH is very wide in its scope, covering all substances whether manufactured, imported, used as intermediates, or placed on the market, either on their own, in preparations, or in articles, unless they are radioactive, subject to customs supervision, or are nonisolated intermediates, in quantities of 1 tonne or more per year. Waste is specifically exempted. Food that meets the definition of a substance, on its own or in a preparation, will be subject to REACH. However, such substances are largely exempted from registration, evaluation, and authorization. REACH gives greater responsibility to industry to manage the risks from chemicals and to provide safety information on the substances. Manufacturers and importers will be required to gather information on the properties of their chemical substances, which will allow their safe handling, and to register the information in a central database. The regulation also calls for progressive substitution of the most dangerous chemicals when suitable alternatives have been identified. In the United States, as risk assessment methodologies have been refined, alternatives to the current methods for reaching public policy decisions have been proposed (Silbergeld 1993). The considerations include using a technology-based approach to investigate solutions or simply banning the most toxic chemicals without waiting for quantitative calculations. The technology-based approach envisions simpler rules to estimate risk, which can include applying a safety or uncertainty factor without having to select a mechanism-based approach. One can also use tools for risk reduction that do not require setting point estimates for standards, such as California’s Proposition 65, which triggers disclosure provisions rather than specific control actions. A similar approach is the Toxics Release Inventory (TRI), which requires reporting of environmental releases of chemicals. The TRI is a publicly available EPA database that contains information on toxic chemical releases and waste management activities reported annually by certain industries as well as federal facilities. It contains detailed information on nearly 650 chemicals and chemical categories that 23,000 industrial and other facilities manage through disposal or other releases, recycling, energy recovery, or treatment. TRI data help to (a) identify potential concerns and gain a better understanding of potential risks, (b) identify priorities and opportunities to work with industry and government to reduce toxic chemical disposal or other releases and potential risks associated with them, and (c) establish reduction targets and measure progress toward those targets. These
REFERENCES
87
approaches are seen as having the advantage of timely and efficient resolution of controversial issues in environmental and occupational health. While these approaches have several merits, they do not totally exclude the components of a risk assessment. Risk assessment continues to be an appropriate tool for addressing relevant human health risk issues and remains a major component of public health programs. Meanwhile, new methods are being developed, and guidelines to address various scientific issues continue to be published by the EPA to improve environmental policies and regulations. More precise methods becoming available for calculating cancer risk should help eliminate some of the traditional defaults. Because new federal methods and proposed guidelines are often first implemented by individual states rather than the federal government, this may help justify formal adoption of the new science at the federal level.
REFERENCES Albert, R. E., Miller, M. L., Cody, T., Andringa, A., Shukla, R., and Baxter, C. S. (1991). Benzo[a] pyrene-induced skin damage and tumor promotion in the mouse. Carcinogenesis 12(7), 1273–1280. Alvares, A. P. (1981). Cytochrome P-450s: Research highlights of the last two decades. Drug Metab Rev 12(2), 431–436. Andersen, M. E., Green, T., Frederick, C. B., and Bogdanffy, M. S. (2002). Physiologically based pharmacokinetic (PBPK) models for nasal tissue dosimetry of organic esters: Assessing the state-ofknowledge and risk assessment applications with methyl methacrylate and vinyl acetate. Regul Toxicol Pharmacol 36(3), 234–245. ARB (2008a). Toxic Air Contaminant (TAC) Identification Program. Air Resources Board, California Environmental Protection Agency, Sacramento, CA. Accessed at http://www.arb.ca.gov/toxics/id/ id.htm. ARB (2008b). AB 2588 Air Toxics “Hot Spots” Program. Air Resources Board, California Environmental Protection Agency, Sacramento, CA. Accessed at http://www.arb.ca.gov/ab2588/ab2588.htm. Bailer, A. J., and Hoel, D. G. (1989). Metabolite-based internal doses used in a risk assessment of benzene. Environ Health Perspect 82, 177–184. Ballantyne, B., Snellings, W. M., and Norris, J. C. (2006). Respiratory peripheral chemosensory irritation, acute and repeated exposure toxicity studies with aerosols of triethylene glycol. J Appl Toxicol 26(5), 387–396. Barlow, S., Renwick, A. G., Kleiner, J., Bridges, J. W., Busk, L., Dybing, E., Edler, L., Eisenbrand, G., Fink-Gremmels, J., Knaap, A., Kroes, R., Liem, D., Müller, D. J., Page, S., Rolland, V., Schlatter, J., Tritscher, A., Tueting, W., and Würtzen, G. (2006). Risk assessment of substances that are both genotoxic and carcinogenic. Report of an International Conference organized by EFSA and WHO with support of ILSI Europe. Food Chem Toxicol 44(10), 1636–1650. Beaumont, J. J., Sedman, R. M., Reynolds, S. D., Sherman, C. D., Li, L.-H., Howd, R. A., Sandy, M. S., Zeise, L. A., and Alexeeff, G. V. (2008). Cancer mortality in a population exposed to hexavalent chromium in drinking water. Epidemiology 19(1), 12–23. Beddowes, E. J., Faux, S. P., and Chipman, J. K. (2003). Chloroform, carbon tetrachloride and glutathione depletion induce secondary genotoxicity in liver cells via oxidative stress. Toxicology 187, 101–115. Bogen, K. T. (1990). Risk extrapolation for chlorinated methanes as promoters vs initiators of multistage carcinogenesis. Fundam Appl Toxicol 15(3), 536–557. Borm, P. J., Schins, R. P., and Albrecht, C. (2004). Inhaled particles and lung cancer, part B: Paradigms and risk assessment. Int J Cancer 110(1), 3–14. Bouchard, M., Brunet, R. C., Droz, P. O., and Carrier, G. (2001). A biologically based dynamic model for predicting the disposition of methanol and its metabolites in animals and humans. Toxicol Sci 64(2), 169–184.
88
CHAPTER 4 USE OF CANCER RISK ASSESSMENTS
Brockmann, M., Fischer, M., and Müller, K. M. (1998). Exposure to carbon black: A cancer risk? Int Arch Occup Environ Health 71(2), 85–99. Brown, J. S., Wilson, W. E., and Grant, L. D. (2005). Dosimetric comparisons of particle deposition and retention in rats and humans. Inhal Toxicol 17(7–8), 355–385. Butterworth, B. E., Conolly, R. B., and Morgan, K. T. (1995). A strategy for establishing mode of action of chemical carcinogens as a guide for approaches to risk assessments. Cancer Lett 93(1), 129–146. Calabrese, E. J., Stanek, E. J., James, R. C., and Roberts, S. M. (1997). Soil ingestion: A concern for acute toxicity in children. Environ Health Perspect 105(12), 1354–1358. Carlberg, C., and Dunlop, T. W. (2006). An integrated biological approach to nuclear receptor signaling in physiological control and disease. Crit Rev Eukaryot Gene Expr 16(1), 1–22. Clean Air Act Amendment of 1990. Public Law 101–549. Accessed at http://www.epa.gov/air/caa/. CleanupLevels. (2008). Cleanup Levels for Hazardous Waste Sites. Information provided by CleanupLevels.com. Accessed at http://www.cleanuplevels.com. Cohen, S. M., and Ellwein, L. B. (1988). Cell growth dynamics in long-term bladder carcinogenesis. Toxicol Lett 43(1–3), 151–173. Conolly, R. B., Kimbell, J. S., Janszen, D., Schlosser, P. M., Kalisak, D., Preston, J., and Miller, F. J. (2004). Human respiratory tract cancer risks of inhaled formaldehyde: Dose–response predictions derived from biologically-motivated computational modeling of a combined rodent and human dataset. Toxicol Sci 82(1), 279–296. Comment in Toxicol Sci 82(1), 1–2. Costa, M., and Klein, C. B. (2006). Toxicity and carcinogenicity of chromium compounds in humans. Crit Rev Toxicol 36(2), 155–163. Comment in: Crit Rev Toxicol 36(9), 777–778; discussion 779. Cox, L. A., Jr. (1996). Reassessing benzene risks using internal doses and Monte-Carlo uncertainty analysis. Environ Health Perspect 104(Suppl 6), 1413–1429. Cox, L. A., Jr., and Ricci, P. F. (1992). Reassessing benzene cancer risks using internal doses. Risk Anal 12(3), 401–410. Comment in Anonymous (1993) Risk Anal 13(2), 145–146. CPSC (1981). Proposal to Withdraw Proposed Ban (on benzene in consumer products). Consumer Product Safety Commission, Washington, D.C., 46 FR 3034-01. Crump, C., Crump, K., Hack, E., Luippold, R., Mundt, K., Liebig, E., Panko, J., Paustenbach, D., and Proctor, D. (2003). Dose–response and risk assessment of airborne hexavalent chromium and lung cancer mortality. Risk Anal 23(6), 1147–1163. Dallas, C. E., Muralidhara, S., Chen, X. M., Ramanathan, R., Varkonyi, P., Gallo, J. M., Bruckner, J. V. (1994). Use of a physiologically based model to predict systemic uptake and respiratory elimination of perchloroethylene. Toxicol Appl Pharmacol 128(1), 60–68. Dawson, S. V., and Alexeeff, G. V. (2001). Multi-stage model estimates of lung cancer risk from exposure to diesel exhaust, based on a U.S. railroad worker cohort. Risk Anal 21(1), 1–18. Comment in Risk Anal 21(1), 19–23; Risk Anal 21(2), 213–216. de Klerk, N. H., Armstrong, B. K., Musk, A. W., and Hobbs, M. S. (1989). Cancer mortality in relation to measures of occupational exposure to crocidolite at Wittenoom Gorge in Western Australia. Br J Ind Med 46(8), 529–536. De Sesso, J. M. (1993). The relevance to humans of animal models for inhalation studies fo cancer in the nose and upper airways. Qual Assur 2, 213–231. Dean, J. R., and Ma, R. (2007). Approaches to assess the oral bioaccessibility of persistent organic pollutants: A critical review. Chemosphere 68(8), 1399–1407. Doll, R. (1957). Smoking and lung cancer. Proc R Soc Med 50(7), 503–504. Dollery, C. T., Fraser, H. S., Mucklow, J. C., and Bulpitt, C. J. (1979). Contribution of environmental factors to variability in human drug metabolism. Drug Metab Rev 9(2), 207–220. Donohue, J. M., and Miller, W. M. (2008). Summary of the development of federal drinking water regulations and health-based guidelines for chemical contaminants. In Risk Assessment for Chemicals in Drinking Water, Howd, R. A., and Fan, A. M., eds., John Wiley & Sons, Hoboken, NJ, pp. 17–33. Dorato, M. A., and Wolff, R. K. (1991). Inhalation exposure technology, dosimetry, and regulatory issues. Toxicol Pathol 19(4 Pt 1), 373–383. Doull, J., Borzelleca, J., Becker, R., Daston, G., DeSesso, J., Fan, A., Fenner-Crisp, P., Holsapple, M., Holson, J., Llewellyn, G., MacGregor, J., Seed, J., Walls, I., and Woo, Y. T. (2007). Framework for use of toxicity screening tools in context-based decision-making. J Food Chem Toxicol 45, 759–796.
REFERENCES
89
ECOS-DoD (2007). Identification and selection of toxicity values/criteria for CERCLA and hazardous waste site risk assessments in the absence of IRIS values. Emerging Contaminants Task Group Risk Assessment Provisional Values Subgroup Issue Paper. Environmental Council of States, Department of Defense Sustainability Work Group. Accessed at http://www.ecos.org/files/2733_file_FINAL_ ECOS_PV_Paper_4_23_07.doc. EPA (1989). Risk Assessment Guidance for Superfund, Vol. I: Human Health Evaluation Manual, Part A, Interim Final. Office of Emergency and Remedial Response, U.S. Environmental Protection Agency, Washington, D.C., EPA/540/1-89/002, December 1989. EPA (1994). Methods for Derivation of Inhalation Reference Concentrations and Application of Inhalation Dosimetry, US Environmental Protection Agency, Washington, D.C., EPA/600/8-90/066F. Accessed at http://oaspub.epa.gov/eims/eimscomm.getfile?p_download_id=473051. EPA (1997a). Exposure Factors Handbook, Revised. National Center for Environmental Assessment, US Environmental Protection Agency, Washington, D.C. Accessed at http://cfpub.epa.gov/ncea/cfm/ recordisplay.cfm?deid=12464. EPA (1997b). Clarification of the Role of Applicable or Relevant and Appropriate Requirements in Establishing Preliminary Remediation Goals under CERCLA. OSWER No. 9200.4-23. Office of Solid Waste and Emergency Response, US Environmental Protection Agency, Washington, D.C. Accessed at http://www.epa.gov/superfund/health/contaminants/radiation/pdfs/aras.pdf. EPA (1998). Carcinogenic Effects of Benzene: An Update. National Center for Environmental Assessment, Office of Research and Development, US Environmental Protection Agency, Washington, D.C., EPA/600/P-97/001F. Accessed at http://www.epa.gov/ncea/pdfs/benzenef.pdf. EPA (1999a). Proposed Radon in Drinking Water Rule. Office of Water, US Environmental Protection Agency, Washington, D.C., (4607) EPA 815-F-99-006. Accessed April 2008 at http://www.epa.gov/ safewater/radon/fact.html. EPA (1999b). EPA Asbestos Materials Bans: Clarification. US Environmental Protection Agency, Washington, D.C. Accessed at http://www.epa.gov/asbestos/pubs/asbbans2.pdf. EPA (2000a). Science Policy Council Handbook: Risk Characterization. Office of Science Policy, Office of Research and Development, US Environmental Protection Agency, Washington, D.C., EPA 100B-00-002. Accessed at http://www.epa.gov/osa/spc/pdfs/rchandbk.pdf. EPA (2000b). Status of chemicals in special review. Office of Prevention, Pesticides and Toxic Substances, US Environmental Protection Agency, EPA 738-R-00-001. Accessed at http://www.epa.gov/oppsrrd1/ special_review/sr00status.pdf. EPA (2000c). Agrichemicals. Food Quality Protection Act—Pesticide Reviews. Ag Center Fact Sheet. US Environmental Protection Agency, EPA 305-F-00-006 (October 2000). Accessed at http://www. epa.gov/oecaagct/factsheets/epa-305-f-00-006ag.html. EPA (2001a). Toxicological review of chloroform (CAS No. 67-66-3), In support of summary information on the Integrated Risk Information System (IRIS). US Environmental Protection Agency Washington, D.C., EPA/635/R-01/001. Accessed at http://www.epa.gov/iris/toxreviews/ 0025-tr.pdf. EPA (2001b). Controlling Disinfection By-Products and Microbial Contaminants in Drinking Water, Clark, R. M., and Boutin, B. K., eds., Office of Research and Development, US Environmental Protection Agency, Cincinnati, Ohio, EPA/600/R-01/110, December 2001. Accessed at http://www. epa.gov/nrmrl/pubs/600r01110/600r01110.pdf. EPA (2001c). Risk Assessment Guidance for Superfund (RAGS): Part D. Standardized Planning, Reporting and Review of Superfund Risk Assessments, Final. Office of Solid Waste and Emergency Response, US Environmental Protection Agency, Washington, D.C. Accessed at http://www.epa.gov/ oswer/riskassessment/ragsd/index.htm. EPA (2002). A Review of the Reference Dose and Reference Concentration Processes. US Environmental Protection Agency, Washington, D.C., EPA/630/P-02/002F=. Accessed at http://www.epa.gov/ncea/ iris/RFD_FINAL[1].pdf. EPA (2003). Human Health Toxicity Values in Superfund Risk Assessments. Memo from Michael B. Cook, Director, Office of Superfund Remediation and Technology Innovation, to Superfund National Policy Managers, Regions 1–10. OSWER Directive 9285.7-53. Office of Solid Waste and Emergency Response, US Environmental Protection Agency, Washington, D.C. Accessed at http://www.epa.gov/ oswer/riskassessment/pdf/hhmemo.pdf.
90
CHAPTER 4 USE OF CANCER RISK ASSESSMENTS
EPA (2004a). Estimated Per Capita Water Ingestion and Body Weight in the United States—An Update. US Environmental Protection Agency, Washington, D.C., EPA-822-R-00-001. EPA (2004b). Potential Implications of Genomics for Regulatory and Risk Assessment Applications at EPA. US Environmental Protection Agency, Washington, D.C., EPA 100/B-04-002. EPA (2005a). Guidelines for Carcinogen Risk Assessment. Risk Assessment Forum, US Environmental Protection Agency Washington, D.C., EPA/630/P-03/001F. EPA (2005b). Summary Report of the EPA Colloquium on Soil/dust Ingestion Rates and Mouthing Behavior for Children and Adults. EPA contract No. EP-C-04-027. National Center for Environmental Assessment, US Environmental Protection Agency, Washington, D.C. Accessed at http://cfpub.epa. gov/ncea/cfm/recordisplay.cfm?deid=146585. EPA (2005c). Supplemental Guidance for Assessing Susceptibility from Early-Life Exposure to Carcinogens. Risk Assessment Forum, US Environmental Protection Agency, Washington, D.C., EPA/630/R-03/003F. EPA (2007). Region 9: Superfund. Preliminary Remediation Goals. US Environmental Protection Agency, San Francisco, CA. Accessed at http://www.epa.gov/region09/waste/sfund/prg/. EPA (2008a). Formaldehyde. Integrated Risk Information System. Office of Water, US Environmental Protection Agency, Washington, D.C. Accessed at http://www.epa.gov/NCEA/iris/subst/0419.htm. EPA (2008b). Child-Specific Exposure Factors Handbook, US Environmental Protection Agency, Washington, D.C. Accessed at http://oaspub.epa.gov/eims/eimscomm.getfile?p_download_id=484738. EPA (2008c). Superfund Laws, Policy and Guidance. US Environmental Protection Agency, Washington, D.C. Accessed at http://cfpub.epa.gov/compliance/resources/policies/cleanup/superfund/. EPA (2008d). Wastes: Laws and Regulations. Office of Solid Waste, US Environmental Protection Agency, Washington, D.C. Accessed at http://www.epa.gov/osw/laws-regs/index.htm. EPA (2008e). Control of hazardous air pollutants from mobile sources: Early credit technology requirement revision. Fed Reg 73(49), 13132–13136, March 12, 2008. Accessed at http://www.epa.gov/ fedrgstr/EPA-AIR/2008/March/Day-12/a4917.htm. EPA (2008f). Asbestos Laws and Regulations. US Environmental Protection Agency, Washington, D.C. Accessed at http://www.epa.gov/asbestos/pubs/asbreg.html. EPA (2008g). FIFRA Statute, Regulations, & Enforcement. US Environmental Protection Agency, Washington, D.C. Accessed at http://www.epa.gov/compliance/civil/fifra/fifraenfstatreq.html. EPA (2008h). Regulating Pesticides: Special Review Process. US Environmental Protection Agency, Washington, D.C. Accessed at http://www.epa.gov/oppsrrd1/special_review/. EPA (2008i). Regulating Pesticides: Chromated Copper Arsenate (CCA). US Environmental Protection Agency, Washington, D.C. Accessed at http://www.epa.gov/oppad001/reregistration/cca/index.htm. Ershow, A. G., and Cantor, K. P. (1989). Total Water and Tapwater Intake in the United States: Population-Based Estimates of Quantities and Sources. Life Sciences Research Office, Federation of American Societies for Experimental Biology, Bethesda, MD. Fan, A. M. (2005). Cancer potency factor. In Encyclopedia of Toxicology, 2nd edition, Wexler, P., ed., Elsevier, Oxford, UK, pp. 401–404. Fan, A., and Howd, R. (2008). Risk assessment: Emerging issues, recent advances, and future challenges. In Risk Assessment of Chemicals in Drinking Water, Fan, A., and Howd, R., eds., John Wiley & Sons, Hoboken, NJ, pp. 325–363. Fang, C. P., Wilson, J. E., Spektor, D. M., and Lippmann, M. (1993). Effect of lung airway branching pattern and gas composition on particle deposition in bronchial airways. III. Experimental studies with radioactively tagged aerosol in human and canine lungs. Exp Lung Res 19(3), 377–396. FDA (2000). United States Food Safety System: Precaution in US Food Safety Decisionmaking. Annex II to the United States’ National Food Safety System Paper, US Food and Drug Administration, Washington, D.C. Accessed at http://www.foodsafety.gov/∼fsg/fssyst4.html. Field, R. W. (2001). A review of residential radon case–control epidemiologic studies performed in the United States. Rev Environ Health 16(3), 151–167. Fisher, J. W., Dorman, D. C., Medinsky, M. A., Welsch, F., and Conolly, R. B. (2000). Analysis of respiratory exchange of methanol in the lung of the monkey using a physiological model. Toxicol Sci 53(2), 185–193. Frederick, C. B., Bush, M. L., Lomax, L. G., Black, K. A., Finch, L., Kimbell, J. S., Morgan, K. T., Subramaniam, R. P., Morris, J. B., and Ultman, J. S. (1998). Application of a hybrid computational
REFERENCES
91
fluid dynamics and physiologically based inhalation model for interspecies dosimetry extrapolation of acidic vapors in the upper airways. Toxicol Appl Pharmacol 152(1), 211–231. Frederick, C. B., Lomax, L. G., Black, K. A., Finch, L., Scribner, H. E., Kimbell, J. S., Morgan, K. T., Subramaniam, R. P., and Morris, J. B. (2002). Use of a hybrid computational fluid dynamics and physiologically based inhalation model for interspecies dosimetry comparisons of ester vapors. Toxicol Appl Pharmacol 183(1), 23–40. Gilson, J. C. (1976). Asbestos cancers as an example of the problem of comparative risks. IARC Sci Publ 13, 107–116. Gray, G. M., Li, P., Shlyakhter, I., and Wilson, R. (1995). An empirical examination of factors influencing prediction of carcinogenic hazard across species. Regul Toxicol Pharmacol 22: 283–291. Green, T., Lee, R., Toghill, A., Meadowcroft, S., Lund, V., and Foster, J. (2001). The toxicity of styrene to the nasal epithelium of mice and rats: Studies on the mode of action and relevance to humans. Chem Biol Interact 137, 185–202. Greenland, S., and Robins, J. (1994). Ecologic studies—Biases, misconceptions, and counterexamples. Am J Epidemiol 139, 747–760. Hamm, T. E., Jr. (1994). Design of a long-term animal bioassay for carcinogenicity. In Handbook of Carcinogen Testing, 2nd edition, H. A., Milman, E. K., Weisburger, eds., William Andrew Publishing, Norwich, NY, pp. 270–285. Harkema, J. R. (1990). Comparative pathology of the nasal mucosa in laboratory animals exposed to inhaled irritants. Environ Health Perspect 85, 231–238. Hayes, R. B., Yin, S. N., Dosemeci, M., Li, G. L., Wacholder, S., Travis, L. B., Li, C. Y., Rothman, N., Hoover, R. N., and Linet, M. S. (1997). Benzene and the dose-related incidence of hematologic neoplasms in China. Chinese Academy of Preventive Medicine—National Cancer Institute Benzene Study Group. J Natl Cancer Inst 89(14), 1065–1071. Hein, M. J., Stayner, L. T., Lehman, E., and Dement, J. M. (2007). Follow-up study of chrysotile textile workers: Cohort mortality and exposure–response. Occup Environ Med 64, 616–625. Henderson, R. F., Sabourin, P. J., Medinsky, M. A., Birnbaum, L. S., and Lucier, G. L. (1992). Benzene dosimetry in experimental animals: Relevance for risk assessment. Prog Clin Biol Res 374, 93–105. Hodgson, J. T., and Darnton, A. (2000). The quantitative risks of mesothelioma and lung cancer in relation to asbestos exposure. Ann Occup Hyg 44(8), 565–601. ILSI (2000). The relevance of the rat lung response to particle overload for human risk assessment: A workshop consensus report. Inhal Toxicol 12(1), 1–17. ILSI (2008). Risk Assessment of Genotoxic Carcinogens in Food Task Force. International Life Sciences Institute. Accessed at http://europe.ilsi.org/activities/taskforces/riskassessment/ RiskAssessmentCarcinogens.htm and http://europe.ilsi.org/NR/rdonlyres/D1716BE4-A26F-4F90AA2B-917E4C173DA5/0/Genotox2008.pdf. IRIS (2008). Toxicity Reviews. Integrated Risk Information System. Office of Water, US Environmental Protection Agency, Washington, DC. http://www.epa.gov/ncea/iris/index.html. Jarabek, A. M., Asgharian, B., and Miller, F. J. (2005). Dosimetric adjustments for interspecies extrapolation of inhaled poorly soluble particles (PSP). Inhal Toxicol 17(7–8), 317–334. Johanson, G. (1991). Modelling of respiratory exchange of polar solvents. Ann Occup Hyg 35(3), 323–339. Jones, T. D., Griffin, G. D., and Walsh, P. J. (1983). A unifying concept for carcinogenic risk assessments. J Theor Biol 105(1), 35–61. Jordan, W. L., and Timm, G. (2006). Food Quality Protection Act and the Safe Drinking Water Act. In Toxicology and Regulatory Process, Green, S., ed., Taylor & Francis, New York, pp. 43–75. Jorgenson, T. A., Meierhenry, E. F., Rushbrook, C. J., Bull, R. J., and Robinson, M. (1985). Carcinogenicity of chloroform in drinking water to male Osborne–Mendel rats and female B6C3F1 mice. Fund Appl Toxicol 5, 760–769. Kato, R., and Kamataki, T. (1982). Cytochrome P-450 as a determinant of sex difference of drug metabolism in the rat. Xenobiotica 12, 787–800. Kimbell, J., Subramaniam, R., Gross, E., Schlosser, P., and Morgan, K. (2001). Dosimetry modeling of inhaled formaldehyde: Comparisons of local flux predictions in the rat, monkey, and human nasal passages. Toxicol Sci 64, 100–110.
92
CHAPTER 4 USE OF CANCER RISK ASSESSMENTS
Kliment, V. (1973). Similarity and dimensional analysis, evaluation of aerosol deposition in the lungs of laboratory animals and man. Folia Morphol (Praha) 21(1), 59–64. Knox, J. F., Holmes, S., Doll, R., and Hill, I. D. (1968). Mortality from lung cancer and other causes among workers in an asbestos textile factory. Br J Ind Med 25(4), 293–303. Kroes, R., Kleiner, J., and Renwick, A. (2005). The threshold of toxicological concern concept in risk assessment. Toxicol Sci 86(2), 226–230. Kroes, R., Renwick, A. G., Cheeseman, M., Kleiner, J., Mangelsdorf, I., Piersma, A., Schilter, B., Schlatter, J., van Schothorst, F., Vos, J. G., Würtzen, G., and European Branch of the International Life Sciences Institute. (2004). Structure-based thresholds of toxicological concern (ITTC): guidance to substances present at low levels in the diet. Food Chem Toxicol 42, 65–83. LaGoy, P. K. (1987). Estimated soil ingestion rates for use in risk assessment. Risk Anal 7(3), 355–359. Lan, Q., Zhang, L., Li, G., Vermeulen, R., Weinberg, R. S., Dosemeci, M., Rappaport, S. M., Shen, M., Alter, B. P., Wu, Y., Kopp, W., Waidyanatha, S., Rabkin, C., Guo, W., Chanock, S., Hayes, R. B., Linet, M., Kim, S., Yin, S., Rothman, N., and Smith, M. T. (2004). Hematotoxicity in workers exposed to low levels of benzene. Science 306(5702), 1774–1776. Lin, Y. S., Vermeulen, R., Tsai, C. H., Waidyanatha, S., Lan, Q., Rothman, N., Smith, M. T., Zhang, L., Shen, M., Li, G., Yin, S., Kim, S., and Rappaport, S. M. (2007). Albumin adducts of electrophilic benzene metabolites in benzene-exposed and control workers. Environ Health Perspect 115(1), 28–34. Lippmann, M., and Schlesinger, R. B. (1984). Interspecies comparisons of particle deposition and mucociliary clearance in tracheobronchial airways. J Toxicol Environ Health 13(2–3), 441–469. Liteplo, R. G., and Meek, M. E. (2003). Inhaled formaldehyde: Exposure estimation, hazard characterization, and exposure–response analysis. J Toxicol Environ Health B Crit Rev 6(1), 85–114. Ljung, K., Oomen, A., Duits, M., Selinus, O., and Berglund, M. (2007). Bioaccessibility of metals in urban playground soils. J Environ Sci Health A Tox Hazard Subst Environ Eng 42(9), 1241–1250. Lloyd, J. W., Lundin, F. E., Redmond, C. K., and Geiser, A. B. (1970). Long-term mortality study of steelworkers. IV. Mortality by work area. J Occup Med 12, 151–157. Lubin, J. H. (2002). The potential for bias in Cohen’s ecological analysis of lung cancer and residential radon. J Radiol Prot 22, 141–148. Michalik, L., Desvergne, B., and Wahli, W. (2004). Peroxisome-proliferator-activated receptors and cancers: complex stories. Nat Rev Cancer 4, 61–70. Mills, W. B., Liu, S., Rigby, M. C., and Brenner, D. (2007). Time-variable simulation of soil vapor intrusion into a building with a combined crawl space and basement. Environ Sci Technol 41(14), 4993–5001. Milman, H. A., and Weisburger, E. K., eds. (1994). Handbook of Carcinogen Testing, 2nd edition, William Andrew Publishing, Norwich, NY. Moody, R. P., Joncas, J., Richardson, M., and Chu, I. (2007). Contaminated soils (I): In vitro dermal absorption of benzo[a]pyrene in human skin. J Toxicol Environ Health A 70(21), 1858–1865. Morello-Frosch, R. A., Woodruff, T. J., Axelrad, D. A., and Caldwell, J. C. (2000). Air toxics and health risks in California: The public health implications of outdoor concentrations. Risk Anal 20(2), 273–291. NAS (1987). Regulating Pesticides in Food—The Delaney Paradox, National Academy of Sciences, National Research Council. National Academies Press, Washington, D.C. NAS (1993). Pesticides in the Diets of Infants and Children, National Academy of Sciences, National Research Council. National Academies Press, Washington, D.C. Nazaroff, W. W., Doyle, S. M., Nero, A. V., and Sextro, R. G. (1987). Potable water as a source of airborne 222Rn in U.S. dwellings: A review and assessment. Health Phys 52(3), 281–295. Nero, A. V., Schwehr, M. B., Nazaroff, W. W., and Revzan, K. L. (1986). Distribution of airborne radon222 concentrations in U.S. homes. Science 234(4779), 992–997. NIOSH (1992). NIOSH recommendations for occupational safety and health: Compendium of policy documents and statements. US Department of Health and Human Services, Public Health Service, CDC, Cincinnati, OH. DHHS publication no (NIOSH)92-100. Accessed at http://198.246.98.21/ niosh/92-100.html.
REFERENCES
93
NIOSH (2005). NIOSH Pocket Guide to Chemical Hazards. US Department of Health and Human Services, Public Health Service, CDC, Cincinnati, OH. Accessed at http://www.cdc.gov/niosh/npg/ default.html. NIOSH (2008). NIOSH Publications (list). Accessed at http://www.cdc.gov/niosh/pubs/all_date_desc_ nopubnumbers.html. NRC (1994). Science and Judgment in Risk Assessment. Committee on Risk Assessment of Hazardous Air Pollutants, Commission on Life Sciences, National Research Council. National Academies Press, Washington, D.C. NTP (1987). Toxicology and carcinogenesis studies of 1,4-dichlorobenzene (CAS No. 106-46-7) in F344/N rats and B6C3F1 mice (gavage studies). TR-319. National Toxicology Program Technical Report Series 319, 198. NTP (2007). Draft NTP Technical Report on the Toxicology and Carcinogenesis Studies of Sodium Dichromate Dihydrate in F344/N Rats and B6C3F1 Mice. NTP TR 546. NIH Publication No. 07-5887, National Toxicology Program, Research Triangle Park, NC. Accessed at http://ntp.niehs.nih.gov/ files/546_board_web.pdf. Oberdörster, G. (1995a). Lung particle overload: Implications for occupational exposures to particles. Regul Toxicol Pharmacol 21(1), 123–135. Oberdörster, G. (1995b). The NTP talc inhalation study: A critical appraisal focused on lung particle overload. Regul Toxicol Pharmacol 21(2), 233–241. Oberdörster, G. (2002). Toxicokinetics and effects of fibrous and nonfibrous particles. Inhal Toxicol 14(1), 29–56. O’Brien, J., Renwick, A. G., Constable, A., Dubing, E., Muller, D. J. G., Schlatter, J., Slob, W., Tueting, W., van Benthem, J., Williams, G. M., and Wolfreys, A. (2006). Approaches to the risk assessment of genotoxic carcinogens in food: A critical appraisal. Food Chem Toxicol 44(10), 1613–1635. OEHHA (2001). Public Health Goal for benzene in drinking water. Office of Environmental Health Hazard Assessment, California Environmental Protection Agency, Oakland and Sacramento, CA. Accessed at http://www.oehha.ca.gov/water/phg/pdf/BenzeneFinPHG.pdf. OEHHA (2003). Public Health Goal for asbestos in drinking water. Office of Environmental Health Hazard Assessment, California Environmental Protection Agency, Oakland and Sacramento, CA. Accessed at http://www.oehha.ca.gov/water/phg/pdf/Ph4asbestos92603.pdf. OEHHA (2005). Air Toxics Hot Spots Program Risk Assessment Guidelines, Part II. Technical Support Document for Describing Available Cancer Potency Factors. Office of Environmental Health Hazard Assessment, California Environmental Protection Agency, Oakland, CA. Accessed at http://www. oehha.ca.gov/air/hot_spots/pdf/May2005Hotspots.pdf. OEHHA (2008). Proposition 65 Safe Harbor Levels: No significant risk levels for carcinogens and maximum allowable dose levels for chemicals causing reproductive toxicity. Office of Environmental Health Hazard Assessment, California Environmental Protection Agency, Oakland, CA. Accessed at http://www.oehha.ca.gov/prop65/pdf/Feb2008StatusReport.pdf. Olson, D. A., and Corsi, R. L. (2001). Characterizing exposure to chemicals from soil vapor intrusion using a two-compartment model. Atmos Environ 35, 4201–4209. OSHA (2008). Occupation Exposure to Asbestos, Section 1. Regulatory History. U.S. Occupational Safety and Health Administration, Washington, D.C. Accessed at http://www.osha.gov/pls/oshaweb/ owadisp.show_document?p_table=PREAMBLES&p_id=775. Overton, J. H., Kimbell, J. S., and Miller, F. J. (2001). Dosimetry modeling of inhaled formaldehyde: the human respiratory tract. Toxicol Sci 64(1), 122–134. Park, R. M., Bena, J. F., Stayner, L. T., Smith, R. J., Gibb, H. J., and Lees, P. S. (2004). Hexavalent chromium and lung cancer in the chromate industry: A quantitative risk assessment. Risk Anal 24(5), 1099–1108. Park, R. M., and Stayner, L. T. (2006). A search for thresholds and other nonlinearities in the relationship between hexavalent chromium and lung cancer. Risk Anal 26(1), 79–88. Pauluhn, J., and Mohr, U. (2000). Inhalation studies in laboratory animals—Current concepts and alternatives. Toxicol Pathol 28(5), 734–753. Pauluhn, J. (2003a). Overview of testing methods used in inhalation toxicity: From facts to artifacts. Toxicol Lett 140–141, 183–193. Pauluhn, J. (2003b). Issues of dosimetry in inhalation toxicity. Toxicol Lett 140–141, 229–238.
94
CHAPTER 4 USE OF CANCER RISK ASSESSMENTS
Pauluhn, J. (2005). Overview of inhalation exposure techniques: Strengths and weaknesses. Exp Toxicol Pathol 57(Suppl 1), 111–128. Perkins, R. A., Ward, K. W., and Pollack, G. M. (1995). Comparative toxicokinetics of inhaled methanol in the female CD-1 mouse and Sprague–Dawley rat. Fundam Appl Toxicol 28(2), 245–254. Phalen, R. F., Mannix, R. C., and Drew, R. T. (1984). Inhalation exposure methodology. Environ Health Perspect 56, 23–34. Phalen, R. F. (1976). Inhalation exposure of animals. Environ Health Perspect 16, 17–24. Pontius, F. W. (2000). Chloroform: science, policy, and politics. J Am Water Works Assoc 92(5), 12. Raabe, O. G. (1986). Inhalation uptake of selected chemical vapors at trace levels. Final Report, CARB contract No. A3-132-33. California Air Resources Board, Sacramento, CA. Raffensberger, C., and Tickner, J. (1999). Protecting public health and the environment: Implementing the precautionary principle. Island Press. Washington, D.C. Rappaport, S. M., Waidyanatha, S., Yeowell-O’Connell, K., Rothman, N., Smith, M. T., Zhang, L., Qu, Q., Shore, R., Li, G., and Yin, S. (2005). Protein adducts as biomarkers of human benzene metabolism. Chem Biol Interact 153–154, 103–109. Regulation EC 1907/2006 (2006). Regulation (EC) No 1907/2006 of the European Parliament and of the Council of 18 December 2006 concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH), establishing a European Chemicals Agency, amending Directive 1999/45/EC, and repealing Council Regulation (EEC) No 793/93 and Commission Regulation (EC) No1488/94 as well as Council Directive 76/769/EEC and Commission Directives 91/155/EEC, 93/67/EEC, 93/105/ EC and 2000/21/EC. Reifenrath, W. G., Kammen, H. O., Reddy, G., Major, M. A., and Leach, G. J. (2008). Interaction of hydration, aging, and carbon content of soil on the evaporation and skin bioavailability of munition contaminants. J Toxicol Environ Health A 71(8), 486–494. Reznik, G. K. (1990). Comparative anatomy, physiology, and function of the upper respiratory tract. Environ Health Perspect 85, 171–176. Rinsky, R. A., Smith, A. B., Hornung, R., Filloon, T. G., Young, R. J., Okun, A. H., and Landrigan, P. J. (1987). Benzene and leukemia. An epidemiologic risk assessment. N Engl J Med 316, 1044–1050. Ris, C. (2007). U.S. EPA health assessment for diesel engine exhaust: A review. Inhal Toxicol 19(Suppl 1), 229–239. Rodricks, J. V. (1994). Risk assessment, the environment, and public health. Environ Health Perspect 102(3), 258–264. Rodricks, J. V., and Turnbull, D. (1994). Risk assessment: biological considerations. In Handbook of Carcinogen Testing, 2nd edition, Milman, H. A., and Weisburger, E. K., eds., pp. 630–650, William Andrew Publishing, Norwich, NY. Saghir, S. A., Bartels, M. J., Budinsky, R. A., Harris, E. E., Clark, A. J., Staley, J. L., Chai, Y., and Davis, J. W. (2007). Effect of organic carbon content, clay type, and aging on the oral bioavailability of hexachlorobenzene in rats. Environ Toxicol Chem 26(11), 2420–2429. Schmidt, C. W. (1999). Chloroform: An EPA test case. Environ Health Perspect 107(7), A358–A360. Schut, H. A., and Castonguay, A. (1984). Metabolism of carcinogenic amino derivatives in various species and DNA alkylation by their metabolites. Drug Metab Rev 15(4), 753–839. Scott, C. S., and Cogliano, V. J. (2000). Trichloroethylene and cancer: Epidemiologic evidence. Environ Health Perspect 108(Suppl 2), 159–160. Sedman, R. M., Beaumont, J., McDonald, T. A., Reynolds, S., Krowech, G., and Howd, R. (2006). Review of the evidence regarding the carcinogenicity of hexavalent chromium in drinking water. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev 24(1), 155–182. Shan, C., and Javandel, I. (2005). A multilayered box model for calculating preliminary remediation goals in soil screening. Risk Anal 25(2), 339–349. Shimada, T., Yamazaki, H., Mimura, M., Inui, Y., and Guengerich, F. P. (1994). Interindividual variations in human liver cytochrome P-450 enzymes involved in the oxidation of drugs, carcinogens and toxic chemicals: Studies with liver microsomes of 30 Japanese and 30 Caucasians. J Pharmacol Exp Ther 270(1), 414–423. Silbergeld, E. K. (1993). Risk assessment: The perspectives and experience of U.S. environmentalists. Environ Health Perspect 101(2), 100–104.
REFERENCES
95
Steensma, A., Beamand, J. A., Walters, D. G., Price, R. J., and Lake, B. G. (1994). Metabolism of coumarin and 7-ethoxycoumarin by rat, mouse, guinea pig, cynomolgus monkey and human precision-cut liver slices. Xenobiotica 24(9), 893–907. Sunderman, F. W., Jr. (2001). Nasal toxicity, carcinogenicity, and olfactory uptake of metals. Ann Clin Lab Sci 31(1), 3–24. Tabershaw, I., and Gaffey, W. (1974). Mortality study of workers in the manufacture of vinyl chloride and its polymers. J Occup Med 16, 509–518. Tsujino, I., Kawakami, Y., and Kaneko, A. (2005). Comparative simulation of gas transport in airway models of rat, dog, and human. Inhal Toxicol 17(9), 475–485. Turteltaub, K. W., and Mani, C. (2003). Benzene metabolism in rodents at doses relevant to human exposure from urban air. Res Rep Health Eff Inst 113, 1–26; discussion 27–35. Valberg, P. A., Long, C. M., and Sax, S. N. (2006). Integrating studies on carcinogenic risk of carbon black: Epidemiology, animal exposures, and mechanism of action. J Occup Environ Med 48(12), 1291–1307. Valberg, P. A., and Watson, A. Y. (1996a). Lung cancer rates in carbon-black workers are discordant with predictions from rat bioassay data. Regul Toxicol Pharmacol 24(2 Pt 1), 155–170. Valberg, P. A., and Watson, A. Y. (1996b). Analysis of diesel-exhaust unit-risk estimates derived from animal bioassays. Regul Toxicol Pharmacol 24(1 Pt 1), 30–44. Vesell, E. S. (1978). Disease as one of many variables affecting drug disposition and response: Alterations of drug disposition in liver disease. Drug Metab Rev 8(2), 265–291. Vincent, J. H., Johnston, A. M., Jones, A. D., Bolton, R. E., and Addison, J. (1985). Kinetics of deposition and clearance of inhaled mineral dusts during chronic exposure. Br J Ind Med 42(10), 707–715. Wartenberg, D., Reyner, D., and Scott, C. S. (2000). Trichloroethylene and cancer: Epidemiologic evidence. Environ Health Perspect 108(Suppl 2), 161–176. White, M. J. (2004). Asbestos and the Future of Mass Torts. National Bureau of Economic Research. NBER Working Paper No. W10308. Accessed 4/30/08 through the Social Science Research Network at: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=510023. Wolfe, M. S., and Portier, C. J. (2006). National toxicology program. In Toxicology and Regulatory Process, Green, S., ed., Taylor & Francis, New York, pp. 129–160. Yokley, K., Tran, H. T., Pekari, K., Rappaport, S., Riihimaki, V., Rothman, N., Waidyanatha, S., and Schlosser, P. M. (2006). Physiologically-based pharmacokinetic modeling of benzene in humans: A Bayesian approach. Risk Anal 26(4), 925–943.
PART
II
CANCER BIOLOGY AND TOXICOLOGY
CH A P TE R
5
THE INTERPLAY OF CANCER AND BIOLOGY James W. Holder
Human cancer is a highly diverse and complex disease based on multiple etiologies, multiple cell targets, and distinct developmental stages with various adjunct pathways feeding into each stage, depending on the environmental conditions of the organ and of the organism. This chapter examines essential characteristics that cancer cells commonly share as well as cellular features that override normal controls of proliferation and homeostasis. Emphasis is placed on specific biological mechanisms that can be toxicologically affected, rather than presenting an anecdotal listing of specific toxic actions. Many publications referred to in this overview contain excellent illustrations and diagrams of pathways that are described in this chapter. This discussion of the interplay of cancer and biology uses a historical path of discovery theme. The long progress in Biology, as a logical sometimes but always eclectic field of endeavor, may be measured in part from the knowing, the unknowing, or adventitious contributions of many cancer investigations (Morange 2007). Elemental experimental findings and conclusions are discussed first, culminating in a synthesis of biological mechanisms that form the bases of the current biological cancer model.
5.1. HISTORICAL ACCOUNT OF SOME IMPORTANT EVENTS IN UNDERSTANDING CANCER 5.1.1.
Early Cancer Biology History
Cancer was known to the ancients even before the Greek era. In about 500 b.c. the word for cancer(s) was the Greek word “karchinos” for crab because of the appearance of the central mass and its emanations with vascular fenestrations supplying blood to and from the abnormality. Due to accessibility, the first surgical work of Hippocrates (460–375 b.c.) was with superficial cancers of the skin, lip, and mouth and of the female breast, but he also knew of cancers of the stomach and of the uterus. Galen (131–203 a.d.) followed Hippocrates’ observations and descriptions
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
99
100
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
and postulated that cancers arose because of imbalances in the Four Humors of Empedocles: air, fire, earth, and water. These forces on the body led to heat, cold, wetness, and dryness often seen in patients exhibiting imbalances which were thought to be manifest by the four body humors: blood, phlegm, yellow bile, and black bile. Galen associated cancer with black bile but also referred to it as melanchole. Galen’s humoral concept of holistic medicine lasted until the 15th century. Cancer was perceived as a “swelling,” but unfortunately cancers were included with a variety of other swellings—for example, ulcers, carbuncles, boils, warts, aneurysms, and masses such as buboes. All were designated as “tumors contrary to nature.” Because of the lack of epistemological precision, this led to diagnostic confusion and misguided remediation. By the 17th century, the iatrochemical hypothesis was advanced by Paracelsus (1493–1541), who later was recognized as the Father of Toxicology. As the basic nature iatrochemical concept came to prominence, it substituted balances of humors with balances of body chemicals, which unfortunately included a component of mysticism because chemistry was poorly understood at the time. One mystic approach was that cancer was contagious, an idea that lingered until recent times in the 1900s, bolstered by developments in microbiology begun by Joseph Lister and furthered by Louis Pasteur, among others. Another carcinogenesis concept was based on the cancer involvement of the lymph nodes resulting in stasis of lymph flow producing persistent irritation. Leaving empiricism, the modern era of cancer study began with Claude Deshais Gendron, who in 1700 eschewed past causative hypotheses and described cancer as “a loss in ordered tissue structure.” He described cancers as arising locally that were hard, growing masses that penetrated deep into tissues. Cancers expanded by “filaments,” according to Gendron, and finally ulcerated or spread (Musracchi and Shimkin 1956). Claude Gendron concluded that only localized lesions removed completely, along with their extensions, were curable, which was well known to Hippocrates. At this time, Biology in general was greatly aided by the biological epistemological systemization of species definitions, nomenclatures, and rankings by Linnaeus (1707–1778). Bernardino Ramazzini founded occupational/industrial medicine with “cause-and-effect studies” of occupational diseases. Advocacy of worker protective measures led to eventual passage of Italian factory safety and workmen’s compensation laws. Industrial hygiene began in 1700 when Ramazzini wrote the first important book on occupational diseases. The book De morbis artificum diatriba describes the health hazards of irritating chemicals, dusts, metals, and other abrasive agents encountered by workers in 52 occupations. These included miners, potters, masons, wrestlers, farmers, nurses, soldiers, and many others. He further introduced the concept that lifestyle choices affect the biology of cancer incidence and outcomes. Clinical recognition of overt cancers was firm by 1775 on the basis of gross morphology. Scrotal cancer was the first well-documented instance of an occupational cancer that was referred to as “sootwart” (Shimkin 1980). Percivall Pott, the English surgeon, described chimney sweep’s scrotal cancer in young boys who were small enough to gain entry to clean chimneys. Chimney sweep scrotal cancer was the first case-based occupational cancer described to be caused by a known external agent (soot). Pott’s description also suggested that it is artificial to separate too
5.1. HISTORICAL ACCOUNT OF SOME IMPORTANT EVENTS IN UNDERSTANDING CANCER
101
sharply clinical observations from experimental medicine. Indeed, Pott scientifically recorded and interpreted an adventitious experiment that occurred to many unfortunate youths in pursuit of a trade. The sootwart incidences were modified by social and legal measures against child labor and poor hygiene, but did not eliminate it. Pott also recorded his suspicions that cancer of the breast was associated with nulliparity as compared with child-bearing women and that cancers of the lip and nose were initiated by tobacco smoking. Pott’s complete definitive works, and Ramazzini’s, eventually led to the fields of epidemiology and the study of chemical carcinogenesis (see Chapter 6). The rise of gross pathology dated from Morgagni’s great work of 1761 that included exacting descriptions of cancer of many internal organs (Triolo 1965). The London school of John Hunter produced Baillie’s illustrated atlas of pathology in 1793, while the Paris school introduced Bichat’s concept of tissues, rather than whole organs only, as the origins of the cancer lesions. Meticulous biological descriptions of specific cancer types, made from gross observations on many autopsies, led to several systematic classifications. Thus, many neoplastic disease states became clearly identified, thus separating them from purely inflammatory and other pathologic processes. Unsuccessful attempts were even made to transfer cancer between humans and from man to dog.
5.1.2.
Near-Recent Cancer Biology History
In Germany in 1838, Johannes Müller was the first scientist to view tumor histological slices of humans or animals via an achromatic microscope. Müller recognized the cellular nature of tissues and tumor structures and made the prescient observation that tumor cells resembled normal embryonic cells. Müller revolutionized the understanding of neoplastic growth, as differing from normal growth, and spawned the field of cellular pathology. Müller ’s comprehensive and precise developments of the cellular basis of pathology encouraged his students, who made major contributions to cellular pathology. His students Schleiden and Schwann identified the cell as the individual and basic functional unit of all living things. Two of his other students, Virchow and Cohnheim, studied inflammations and cancers by histogenesis, a study of the formation and development of organs (organogenesis) and various somatic tissues from undifferentiated cells. Cohnheim characterized and defined the three primary germ layers as the endoderm, mesoderm, and ectoderm and advanced an embryological theory of cancer biology. These investigations formed the foundation for the field of histology. As proposed by Mathias Schlieiden (re: plants, 1839) and later by Theodor Schwann (animals, published 1847), the “cell organization theory” became one of the most fundamental theories of modern biology (Schleiden and Schwann 1839). They proposed that all living things possess cells that are “organisms” within themselves which are separate from, but related to, the whole organism in which they reside. The whole organism is constituted from collections of cells and organs that are ordered within the organism according to definite laws. These earlier works and the pathology studies of Rudolf Virchow formed the basis of the hypothesis that the cell, with its central nucleus, is the functional biological unit. Rudolf Virchow supported “the cell theory and its central nucleus” as the origin
102
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
of many diseases, including cancer. He is best known for his 1858 developmental pronouncement Omnis cellula e cellula (“every cell originates from another existing cell like it”) which adumbrates stem cell biology. Virchow proposed an immunological origin of carcinogenesis which he called: “The irritation theory of carcinogenesis” and was the first scientist to recognize leukemia as a cancerous disease. By the 20th century, histopathology had become the final arbiter of diagnosis, specific histogenesis, and prognosis in cancer (Shimkin 1980; Triolo 1965). Gregor Johann Mendel (1822–1884), often called the father of genetics, is known for his systematic study of the inheritance of pea plant traits. From 1856 to 1863, Mendel cultivated and tested some 29,000 pea plants (i.e., Pisum sativum) and showed, by looking at various traits, any of which could occur in one of two observable forms, that segregation of each trait was distributed on average as follows: 1 in 4 pea plants had purebred recessive alleles, 2 in 4 were hybrid alleles, and 1 in 4 was purebred dominant alleles. In 1866, Mendel’s experiments on pea plant traits were generalized into Mendel’s Law of Inheritance, which stated that vertical inheritance [horizontal inheritance was not discovered for another 93 years, (Kiba et al. 1960)] or segregation of phenotypic traits in progeny followed definite quantitative laws. The full significance of Mendel’s work was not recognized until the turn of the 20th century (Punnett 1907). Its rediscovery prompted the contribution of genetics to biological understanding of the basic mechanisms of heredity, of evolution, and of certain mechanisms of chemical carcinogenesis (Punnett 1907). Theodor Heinrich Boveri (1862–1915), a German biologist, demonstrated in sea urchins that it was necessary to have all chromosomes present, and intact, for proper embryonic development to occur (Satzinger 2008). This discovery was an important part of the Boveri–Sutton chromosome theory, which found stainable or colored histologically distinct elements in the nucleus; they called these structures chromosomes, which appeared to function according to Mendel’s principles as carriers or actors of heredity [later (in 1909) called “genes” by Danish botanist Wilhelm Johannsen]. Chromosomes seemed to be the agents of Mendel’s typological traits segregations which later in the collective were referred to as the biological phenotype (Boveri 1929). Boveri reasoned that a cancerous tumor begins with a single cell in which the make-up of its chromosomes can become scrambled (outof-plane polar bodies), thus causing the cells to divide uncontrollably (Boveri 1929). It was only much later in the 20th century that leading researchers came to believe that Boveri was essentially correct (Baltzer 1964). In 1879, human chemical exposures were studied during the Industrial Revolution period. It was found that chronic dermal contact with shale oil, coal distillates, petroleum products, or chimney soot could cause skin cancer. An inordinate prevalence of lung cancer was exhibited among coal miners and was the first internal cancer associated with a known occupational exposure. An iatrogenic cancer of the skin, due to long-term ingestion of potassium arsenite from Fowler ’s solution (used as a tonic in small doses), was recorded by 1887. In 1895, excessive cancer of the urinary bladder was identified in workers from the aniline dye industry. Following Lister ’s phenolic surgical sterilization methods, after 1867, Theodor Billroth in Vienna typified the period in formal and documented surgeries for systematic surgical invasion of all body cavities for the removal of internal cancers.
5.2. RECENT FOUNDATIONS OF BIOLOGICAL MECHANISMS OF CANCER
103
In the biomedical sciences, the latter part of the 19th century was dominated by the bacteriologist Louis Pasteur and was later expanded by Robert Koch. These microbial techniques applied to cancer circa 1880–1910. Claims were made of many microbial isolates that were thought to be the cause of various cancers. These were false starts in the study of the etiology of cancer. Nevertheless, the work stimulated interest in the subject, and it expanded facilities for further disease work. In 1895, the Curies discovered ionizing radiation from purified pitchblende, an ore containing uranium that emitted Roentgen X rays. From these works, two periodic elements radium and polonium were discovered in 1899. X-ray sources were later found to be experimentally carcinogenic in animals and humans. Marie Curie died of radiation-caused aplastic anemia with complications due to her radiation exposure doing ore extractions. In 1915 the experimental reproduction of cancer, on the inner surface of rabbit ears painted with coal tar, was first achieved in Japan (Yamagiwa and Ichikawa 1918). A “tar-period of cancer biology research” followed; for example, skin exposure to the benzene extract of coal tar increased the number of lung tumors in mice (Murphy and Sturm 1925). The responding lung site was distal to the application site. The lung carcinogensis involved transport and later-to-be-discovered metabolic activation of the carcinogen (Watabe 1983). The early history of polycyclic aromatic hydrocarbon isolation and identification from coal tar and cancer testing is best described by Kennaway (Cook et al. 1932) (see Chapter 6). A pioneering work on dimethylnitrosamine (DMNA) showed the efficacy of this potent multiorgan carcinogen (Magee and Barnes 1967). 7,12-Dimethylbenz[a]anthracene or 9,10-dimethyl1,2-benzanthracene (DMBA) cancer work precisely quantified the dose–response and predictability of potent carcinogens (Druckrey 1967). Aflatoxins are among the most active carcinogens of “natural” origin being elaborated by the microorganism, Aspergillus flavus. Its carcinogenic property was demonstrated well before aflatoxin B1 was chemically identified as the carcinogenic agent of Aspergillus flavus (Shimkin 1980; Shimkin and Triolo 1969). This organism is widely distributed and grows on protein foods in humid conditions and cereal products. The high incidence of hepatatocellular carcinomas among human populations in tropical Africa and Asia appears to be related to aflatoxin ingestion (Druckrey 1967; Magee and Barnes 1967; Smela et al. 2001).
5.2. RECENT FOUNDATIONS OF BIOLOGICAL MECHANISMS OF CANCER Early experimentation showed that injured body parts, if persistently irritated, could produce cancer (Deelman 1927). Isaack Berenblum (1903–2000) was the first to experimentally describe the biology of cancer onset. Cancer occurs in sequential steps of cellular changes from normal, to precancerous, cancerous, premalignant, malignant, and metastatic stages with variations of duration and dose of a cocarcinogen, croton oil, a tumor promotor (Berenblum 1941, 1954; Berenblum and Shubik 1949; Foulds 1957). DMBA is a mutagenic, complete carcinogen at least at sufficiently high enough doses. When DMBA was applied once, or repeatedly, at low
104
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
subcarcinogenic doses (200 nmol in 0.2 ml acetone/mouse) onto the shaved dorsal area, no papillomas appeared on mice skin. If DMBA were applied first at a sufficient dose followed by multiple applications (twice/week) of 1% croton oil, a powerful irritant, then papillomas appeared to such an extent to involve all the mice (i.e., 100% incidence) and most had multiple papillomas (multiplicity) in 10–14 weeks (Boutwell 1964). If the carcinogen DMBA dose was too low in the initiating dose, no consequent tumor resulted. Because this experiment has been repeated hundreds of times, this proves that there is a finite threshold dose for the DMBA initiation process (abbreviated as I). A second control group used just 1% croton oil twice/ week, and this noninitiated group produced no tumors in the same time period. The initiation stage of the skin by a sufficient amount of DMBA was permanent (i.e., a fixed I* stage was set); however, the “promotion phase” (P stage) could only be produced with multiple doses of the irritant 1% croton oil. The croton oil could be delayed for 20 weeks after initiation and when commenced, as before, would produce papillomas at the same incidence and multiplicity in 10–14 additional weeks. This result demonstrates that the I* stage once set or “fixed” is not reversible. No cellular changes were ever observed in the I stage in the 1960s, at low subcarcinogenic doses of the DMBA carcinogen; however, hyperplasia was observed in mouse skin during the P stage treated with sufficient amounts of the active tumor promoter croton oil or its active component, tetradecanoyl phorbol acetate (TPA) (Boutwell 1974). In the P stage, increased mitoses first occurred in the basal or germinal layer of skin and then in the more differentiated upper layers with accompanying de novo syntheses of DNA, RNAs, and proteins (Boutwell 1976). These studies and others suggested that the P stage was additive because more TPA developed more tumors in shorter durations (t = time) according to Haber ’s Law: c · t = k, where k is a constant specific for the type of agent applied during the P stage (see Chapter 3). The P stage was reversible because the papillomas would regress after stopping promoter applications at 5–6 weeks or about halfway to 100% incidence (Boutwell 1985). The I* and P stages together constitute obligatory and sequential stages in the carcinogenic process. This cancer stage model was generalized by the demonstration of the I–P protocol in an organ other than skin, the liver (Solt et al. 1983). I-stage treatment was performed in the liver by diethylnitrosoamine (DEN, 200 mg/kg. b.w., once) and completed by feeding F-344 rats 0.02% 2-acetylaminflourene (2-AAF) in order to “fix” I* (make the I* stage permanent). Then a strong mitogenic stimulus was given by either (1) performing partial surgical hepatectomy or (2) applying a dose of carbon tetrachloride (CCl4) at a rate of 2 ml/kg b.w. mixed 1 : 1 with corn oil followed by 1 more week of 0.02% 2-AAF feeding in the diet and then followed by a basal diet. Liver hyperplasia, nodules, and hepatocellular carcinoma (HCC) appeared in succession. Following further exposures, liver nodules got uniformly bigger with time which could be stained for γ-GGT (gamma gutamyl transferase), a neonatal form of the glutathione transferases (Enomoto and Farber 1982). Farber noted that early nodules could either remodel or become persistent, and a number of other initiators could be substituted for DEN. Initiator chemical examples are ethionine, 1,2-dimethylhydrazine, 2-AAF, benzo[a]pyrene (B[a]P), and methylnitrosourea (MNU). Other promoters could be substituted for CCl4 such as orotic acid or phenobarbital (Farber 1984; Slaga 1983, 1984a–c)). The whole I-stage process required
5.3. CELL BIOLOGY OF CANCER
105
fixed liver mutation(s) (I → I*) followed by P-stage mitogenesis (among other gene expression changes) that could be produced in a number of ways. The liver sites reacted hyperplastically, adapted, remodeled, or continued to grow with sustained P-stage influences. Finally, with dose and time, the tumors acquire malignant properties, becoming more autonomous from surrounding liver to the point that liver cancer is irreversibly formed. As many as six distinct stages were exhibited in the rat liver I–P protocol; and because of cancer staging similarities of the rat and human, it was thought the rat might be useful for further study as an experimental model for the biological development of human cancer (Farber 1987). Quantitative cytological relationships were applied to the evaluation of F-344 rat liver foci during I and P stages with measurements done for cell numbers, cellular volumes, and de novo synthesis of growth enzymatic phenotypes (ATPase, γ-GGT, and π-GST [glutathione S-transferase pi]) in altered hepatic foci (AHF). This was achieved by quantitative cytological stereological methods accomplished by tissue mensuration in 2-D and 3-D (Campbell et al. 1986; Pitot et al. 1987). The liver sections exhibited congruent enzyme-altered volumes in the AHF marking tissue growth during the P stage (Pitot et al. 1996). The markers also demonstrated a small number of “foci within foci” and demonstrated irregular multiple lesion types in the chemical disruption regions, or “hits” to use an old radiation term, thus showing the manifest diversity of AHF (Dragan et al. 1997). It has been theorized that promotion susceptibility and resistance factors may be inherited, and foci within foci may indicate a local additional loss of resistance to disturbing the homeostasis, paracrine, or hormonal balance (Angel and DiGiovanni 1999; Furth 1953). Biological definitions of I and P as well as the essential operative properties of the biological processes in cancer are discussed by Pitot et al. (1985). Molecular biology cancer models from established animal models have been reviewed by Yuspa and Poirier (1988). A more recent, balanced carcinogenesis overview of “the classic biology of chemical carcinogensis” is presented by Malarkey and Maronpot (2005), and for bioassaying cancer potential the reader is referred to Maronpot (2007).
5.3.
CELL BIOLOGY OF CANCER
This section on the cell biology of cancer covers many biochemical, microbiological, and medical subfields that have often evolved and proceeded autonomously in the past but now are interrelated in current times. This review attempts to bring these fields more into a concerted focus because they each represent different aspects of one complex cancer model. These include (1) in vitro cell and tissue work, (2) necrosis, apoptosis, and tissue remodeling, (3) supporting or stromal cells and cell-to-cell interactions, (4) persistent inflammation, (5) stem cell biology, and (6) epigenetic biology and nuclear trafficking.
5.3.1.
In Vitro Systems
Over four decades ago, it was first reported that carcinogens in vitro could induce malignant transformation in normal cultured mesenchymal cells (Berwald and Sachs
106
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
1965). These transformation findings were interpreted by many as the first in vitro experimental demonstrations of carcinogenic processes. Sequential steps to transformation processes were observed in vitro, and this experimental work was consistent with the previous in vivo animal work (Barrett and Ts’o 1978a; Boreiko et al. 1980; Casto et al. 1977). After carcinogen exposure, in vitro proliferation was required for transformation (Barrett and Ts’o 1978b). The effectiveness of exposure was cell-cycle specific, and genotoxic effects of exposures, including mutagenesis, could be measured directly and in conjunction with cell transformation. A complex correlation with the mutagenic carcinogens was found in inducing transformation processes (Barrett and Ts’o 1978a; Boreiko et al. 1980; Casto et al. 1977; Rubin 2001; Yuspa and Poirier 1988). Because two or more hit types were necessary for complete transformation, the observed cancer-related in vitro processes appeared to be multiple-staged (Barrett and Ts’o 1978b; Mondal and Heidelberger 1980; Yuspa and Poirier 1988). The complexity, at that time, was attributed to the following: (1) Not all cells are exactly the same even in monoculture, (2) stochastic factors determine which cells are hit, (3) all cells participate in an evolving competition to eliminate the perturbant carcinogen, and (4) cells compete for in vitro nutrients and survival. Together, these studies indicated that when results were separated from cytotoxicity, they supported the hypothesis that somatic mutation mechanisms were involved in the transformation steps and that the whole transformation process was more complex than a single gene mutation (Rubin 1992, 1999). Carcinomas arise from epithelial germ cell types (i.e., stem cells) in vivo. It is known that affected epithelial cells give rise to 80% of all human cancers. Four epithelial cell culture models (e.g., skin, liver, trachea, and breast) have yielded substantial insights into mechanistic cellular changes intrinsic to cancer development in each respective target site. Clonal epidermal cell methods for various species have been developed: mouse, hamster, rat, and human. Of interest in the in vitro models, which can be scrutinized closely, has been the consistent finding that early carcinogen-induced changes in epithelial cells are subtle, usually unrecognizable by most criteria studied, but become obvious when ongoing selective pressures for cell growth or survival, such as exposure to a carcinogen, are applied to the culture system. Selection of different specific colonies for study led to the discovery that carcinogens can alter or transform epithelial cells—long before these cells become tumorigenic and deviate from their normal response to signals for tissue maintenance, differentiation, and growth. These deviations can appear as altered foci cells (AFC), which sometimes have a characteristic structural appearance and/or biochemical alterations, and can often be used as markers for early carcinogenesis. AFC occur in a field of normal cells, which are equivalent to those in control plates—that is, cells not exposed to the carcinogen. This gives rise to the clonal theory of cancer, which began with Theodor Boveri: A single cell can be fundamentally altered in a clone that in time gives rise to the neoplastic phenotype. Metabolically activated carcinogens covalently bind in vitro to mouse epidermal DNA similar to in vivo (Nakayama et al. 1984). In vitro DNA repair processes operate to fix single-strand DNA. There is base excision, nucleotide excision, mismatch repair as well as double-strand repair (for double-strand breaks), and UV and
5.3. CELL BIOLOGY OF CANCER
107
chemical damage repair (oxidation, alkylation, hydrolysis, UV) comparably with in vivo processes. At sufficient exposures, carcinogens can induce net mutations in human and mouse epidermal cell cultures (Yuspa and Poirier 1988) whereupon some epidermal cells, the transformants, exhibit deviations from their normal program of differentiation; that is, the deviants have altered growth requirements. At some dosages and culture conditions, normal cells slough from the culture dish because they terminally differentiate (via senescence), but transformed cells do not terminate and remain on the culture dish (Kulesz-Martin et al. 1980). When transformed cells (derived from carcinogen-exposed cells) are passed through various plating and regrowth cycles and then transplanted to syngeneic or immunosuppressed hosts in vivo, some transformant cells produce squamous cell carcinomas (Yuspa et al. 1980). While retaining many in vitro characteristics of normal skinderived epidermal cells, keratinocyte in vitro transformants exhibit the following: (1) They have different keratin protein content (the main set of epidermal proteins], (2) they are able to grow in low Ca+2 in which normal keratinocytes will not, (3) they become anchorage-independent in a small percentage of cells (showing altered differentiation and cell-to-cell and cell-to-substrate binding interactions), (4) they do not terminate normally but rather show immortal characteristics that vary with different ACF, and (5) the polar keratinocyte membrane protein pemphigoid in hemidesmosomes decreases. Because pemphigoid protein normally participates in binding of the basement membrane and stromal cell elements in the skin basal layer, its decrease indicates cellular independency within the organ environment. Other in vitro model cell systems have been developed such as those derived from the breast and liver. In vitro studies permit focused experimentation that cannot be done in the complex milieu of the in vivo environment, but in vivo studies have the importance of being done in the whole animal where all control and specific environmental factors are operational. Both approaches complement the other. Each organ shows its unique differentiation and growth properties that one might expect of cells that have specialized in development and ontogeny through gene restriction and appropriate gene controls to become functional units within the whole of the local environments that each their specific organs provides. The development of multiple experimental in vitro model systems to study the initiation (I) and promotion (P) stages has expanded our awareness of the variety of pathways which ultimately lead to the development of different types of cancer. In most cell-derived in vitro systems, the standard I-P protocols involving many chemical classes of initiators and promoters produce many types of preneoplastic changes. Sometimes, in vivo there is a low percentage of animals with malignant cancers and even less frequently a lower number exhibiting metastases (Burns et al. 1983; Kaufmann et al. 1985; Peraino et al. 1975). More will be discussed on the decreasing cumulative probability of each successive stage later in the chapter.
5.3.2.
Programmed Cell Removal
It is the design of species with fixed sizes and constant organ/body weight (b.w.) that a proper balance is maintained between mitosis and programmed cell death
108
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
Figure 5.1. Cellular apoptosis. Various critical cellular apoptotic integrated processes are schematically shown. From membrane, to mitochondria, and caspase cascades, and the nucleus all become involved. The cell is blebbing at bottom right in the process of programmed cell death. (Image acquired with permission from from BioOncology of Genentech Corporation (www.biooncology.com.) See insert for color representation of this figure.
(apoptosis). The various steps in apoptosis are graphically represented as a cell system model in Figure 5.1. The net growth occurring over time is zero in mature higher organisms. In the biology of cells, there are many mechanisms participating in both sides of this partial differential process (δN/δt → c) where N is number cells in a compartment, t is time, and c is a variable constant usually approaching zero for net zero growth. It is a partial differential process because many necessary factors are maintained constant
5.3. CELL BIOLOGY OF CANCER
109
in order to achieve homeodynamic cell equilibrium. In a world of many chaotic environments, the steady state is often referred to as homoeostasis; but the preferred term today is homeodynamic equilibrium, which allows that cells are rarely at true equilibrium, where the differential equals zero, but rather are at metastable equilibriums, which are dynamic but versatile and rigorous states throughout the life history cycle of a cell compartment. Many scientists thought the most obvious way for cancer to happen is to presume, and therefore investigate accordingly, that excessive cell production was the sole explanation for tumors. This is not necessarily true ab initio because one may surgically remove a lobe of rat liver, close the incision, allow the rat to recuperate, and, in sufficient time for liver repair, reenter the rat to find that the lobe not only grew back but did so with same size and anatomical configuration as before. Obviously, there is the execution of an exquisite set of guidance balances with shaping control growth mechanisms. Normal regrowth knows where and when to stop and finally homeodynamic equilibrium resumes. A cancer locus, however, loses the ability to stop growth within the design and constaints of the organ. The solution as to where the information for this controlled growth and architectural form of the lobe might be stored is not yet understood, but the biological cellular memory in the liver is a likely candidate and will be reviewed later in this chapter (Holliday 2002, 2006; Jablonka and Lamb 2002). The sum total of apoptosis (programmed cell death) in the balanced healthy human body (with no cancers or illness) is about 1010 cells/day (Renehan et al. 2001). Normal abrasive cell loses account for normal daily cell loses such as when one washes the skin where the top squamous layer sheds (some) as well as the losses of epithelial cells of the alimentary tract due to the flow of food and drink and any necrotic loses as in bruises. The normal cell replacement rate may be on the order of 1011 cells/day for an average adult. Apoptosis is a controlled and timely biological process of cell ablation from an organ that plays a fundamental role in homeodynamic equilibrium of metazoans at constant body weight. Blocking apoptosis (negative effects) or enhancing apoptosis (positive effects) can contribute to a number of human diseases including cancer, autoimmune conditions, neurodegenerative, and wasting disorders (e.g., cachexia). In cancer where there is a net increase in cells over time, if apoptosis is blocked, then cells are not removed according to the homeodynamic set point determined by the cell memory for that specific organ. Thus, even at constant mitotic input, cells can accumulate in that locus. Nonetheless, tissue cells can only increase mitosis rates by two- to fourfold at maximum. This limits the hypercellularity accretion rate in tumors to be accounted for by uncontrolled mitosis alone; thus, inferentially, inhibition of apoptosis may play a role in cancer progression. As a process, apoptosis can remove one or many cells over orders of magnitude, and thus a fine control in apoptosis activity is requisite in order to counterbalance mitosis and maintain homeodynamic equilibrium. If apoptosis were evolved optimally, it would detect all unusual cells and remove them before cancer could become a problem. However, cancer is insidious in that it does not alter membrane epitopes in the early premalignant or focal phases; and later in tumor progression, apoptosis is actively inhibited by specific oncogenic agents until the cancer takes control and is independent (Criollo et al. 2007; Klein et al. 2007; Renehan et al. 2001).
110
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
Tumor Necrosis Factor (TNF-a)
En
do
Trail [APO-2L]
pla
sm
ic
ret ic
ulu
Gap junction intercellular communication of cytotoxins
m
CASPASE 10
Caspase 9 Calpai n
Cytochrome C
ROCK1
Apoto some
BAX
CASPASE 8
BID (BH3 Interacting Domain Death Agonist)
and Ca +2
CASPASE 8
BCL2
Nucle
us
Chromatin/DNA Cleavage into nucleosomes
X
CASPASE 3, 6, 7 is
Inhibits DNA Repair and Synthes
Cell shrinkage with Membrane blebbing
N-killer cells signals, or Contiguous cell(s) signal(s) receiving inadequate Cell-to-cell communication
Figure 5.2. Apoptotic processes. Various critical cellular apoptotic integrated processes are schematically shown. The extrinsic and intrinsic pathways to apoptosis are shown (see text). All organs depend on the homeodynamic balance of cell synthesis and programmed cell death (apoptosis). See insert for color representation of this figure.
The apoptotic program is executed by a family of proteases known as caspases, which are activated in a proteolytic cascade resulting ultimately in the dismantling of the cell (Figure 5.2). This schematic representation of some apoptosis mechanisms does not cover all relevant known apoptotic steps because that would exceed the scope of this chapter. It is known that evolution employs numerous modes of action that are precisely controlled in apopotosis. For example, BCL2 is an anitapoptotic agent in the outer mitochonrial membrane controlling permeability and preventing the release of cytochome c. However, Bax and BID are apoptotic producing agents that promote prermealility and can enable cytochome c release into the cytoplasm. The caspases are cysteine-aspartic acid proteases. The caspases play essential roles in apoptosis (programmed cell death), necrosis, and inflammation. As Figures 5.1 and 5.2 heuristically demonstrate, the control of these precursors is posttranslational in a proteolytic cascade of events. Actuation of these proteolytic events can begin by stimulating apoptotic membrane receptors, such as tumor necrosis factor alpha (TNFα) or TRAIL (TNF-related apoptosis-inducing ligand), and other triggering event types can come from direct connections with local cells (via either gap junctions or juxtacrine receptors) or N-killer cells that is followed by phagocytic absorbtion and complete digestion of the apoptotic cell as suggested in Figure 5.2.
5.3. CELL BIOLOGY OF CANCER
111
Once past the tipping point in initial apoptosis, the cell is thought to immediately swell because all reactions are concentration-dependent and are disrupted by hyperosmolality (Criollo et al. 2007). The cell then undergoes morphological changes, which include plasma membrane blebbing, cell asymmetry, membrane detachment, reversal of size, cell shrinkage, asymmetric margination of mitochondria, nuclear chromosome fragmentation with formation of hemispherical condensed chromatin structures attached to the nucleolemma, and finally complete DNA fragmentation (Lassus et al. 2002; Mauri et al. 2008; Soengas et al. 1999) whereupon the whole cell is engulfed by a phagocyte, leaving no intercellular debris. All the biochemical building blocks of the “apoptosed cell” are recycled to conserve energy. There are two major pathways of caspase activation in mammalian cells: the extrinsic or receptor-mediated pathway and the intrinsic or mitochondria-dependent pathway (Figures 5.1 and 5.2). The extrinsic pathway involves the binding to their corresponding cell-surface receptors of extracellular death ligands such as Fas, TNF, or TRAIL (Kaufmann and Steensma 2005). This begins apoptosis and is followed by recruitment of initiator caspases (procaspase-8 and/or procaspase-10) in a multiprotein complex concerted at the plasma membrane and then activation of downstream effecter caspases, including procaspase-3 (Fadeel et al. 2008). See Figure 5.2 for a cellular schematic of some apoptotic components and mechanisms acting in programmed cell death. Candidates for inducing normal apoptosis are hormones, growth factors, nitric oxide, and cytokines. Improper dosages or blockage of these signals can prevent apoptosis and lead to cancer. The intrinsic pathway is characterized by mitochondrial activation and permeabilization. The mitochondrial membrane potential decreases (−ΔΨM), and then cytochrome c and other proteins are leaked through an osmotically activated mitochondrial pore (apoptosis-induced channel or MAC) into the cytoplasm (Figure 5.1; Criollo et al. 2007). Cytochrome c serves as a cofactor with the apoptosis protease activating factor, called Apaf-1, and by means of oligimerization and activation of procaspase-9 (an initiator of the cascade) forms a stable oligomeric protein complex, the apoptosome (Fadeel et al. 2008). This molecular platform activates downstream caspases, including procaspase-3, and eventually leads to death of the cell (Figures 5.1 and 5.2). Both the extrinsic and intrinsic pathways involve a cascade of proteolytic activity within the cell, downstream of either the plasma membrane or mitochondria, and converge on the activation of the same operating apoptotic caspases that cause cell ablation. Moreover, cross talk between these two pathways seems to occur by caspase-8, an initiator in the extrinsic pathway, cleaving and activating a pro-apoptotic Bcl-2 family member to promote cytochrome c release. This convergence amplifies the initial death stimulus of a cell. Further evidence has indicated caspase activation in disruption of mitochondria with carcinogen exposure (Lassus et al. 2002). When responding to different cell death stimuli, the various studies support the mechanistic argument that mitochondria seem to always act as amplifiers of the caspase protease cascade rather than being the initiators of the caspase activation. Apoptosis resistance most always becomes a feature of developing cancer cells (Hanahan and Weinberg 2000). It is suggested that in the caspase-dependent and caspase-independent apoptotic pathways, specific defects can contribute to tumor
112
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
development and progression. Chemotherapeutic drugs and irradiation treatment, at optimum doses, are thought to induce differential cell death of cancer cells by activation of the apoptosis pathway dependent on mitochondria (Kaufmann and Vaux 2003). The tumor suppressor gene p53 is often mutated (40% in human cancers) or inactivated in human cancer. A defective suppressor p53 gene may account for the observation of resistance to apoptosis in a large proportion of cancer patients because Apaf-1 is a target of the p53 gene product. The Myc gene codes for a transcription factor (TF, a protein with DNA affinity) that sets itself at the 5′-terminus of at least 15% of active genes. This common TF, when mutated, makes a product that fails to properly control the genes that are complementary to its protein product and thus transformation often occurs. In oncogenic transformation of myc-expressing cells, the apoptosome seen in Figure 5.1 might be controlling cancer development, because inactivation of either Apaf-1 or caspase-9 had the same result as inactivation of the suppressor p53 gene product in promoting cancer (Soengas et al. 1999). An Apaf-1 decrease occurs in human metastatic melanomas, which are often resistant to all anticancer treatments, and some melanomas lose all Apaf-1 protein expression altogether (Soengas et al. 2001). These authors considered epigenetic gene silencing to be the likely cause of Apaf-1 down-regulation because the Apaf-1 gene could be reactivated by exposing the in vitro malignant melanoma cells with inhibitors of DNA methylation or histone deacetylation (Dai et al. 2004). These inhibitor treatments also rescued the apoptotic inhibition and markedly enhanced the chemoresponsiveness of melanoma cells. The cell in progressed apoptotic removal shrinks and exhibits blebbing of the membrane. One causal agent of shrinkage is ROCK1, a protein serine/threonine kinase (Figure 5.2). ROCK1 is activated when bound to the GTP-bound form of Rho GTPase and is thus a downstream effector of Rho. ROCK1 phosphorylates and activates LIM kinase; this, in turn, phosphorylates cofilin, which, in turn, inhibits its actin-depolymerizing activity. Thus, ROCK1 contributes to actin-stability. However, as Figure 5.2 shows, when caspase 3, 6, and 7 interact with ROCK1 and block it, then cellular actin fibers are not protected and are deconstucted. Osmotic water shifts and pH changes also lead to the crenation of these cells in a manner typical of all cells in apoptotsis. Autophagy is a biological process that cleans up any debris left over from stress or local traumatic events involving apoptosis and/or necrosis (Galluzzi et al. 2008). Imperfect autophagy can lead to local protracted inflammation (cf. Section 5.3.5) and persistent trauma, or inflammation can lead to cancer (Balkwill and Mantovani 2001; Berenblum 1941; Deelman 1927; Triolo 1965). Any unnecessary or damaged proteins are engulfed in autophagy into membrane-coated autphagosomes, make their way to the lysosome, merge, and their contents are digested and conserved for reuse. A number of stressors can activate both apoptosis and autophagy pathways that seem to be linked. As in the case with apoptosis, autophagy is also controlled in part by p53 (Tasdemir et al. 2008). Defects in autophagy components caused by environmental chemicals can facilitate local cell transformations and cancer, but these autophagic mechanisms are still not well understood.
5.3. CELL BIOLOGY OF CANCER
113
5.3.3. Facilitation of Supporting Cells and Cell-to-Cell Communication Stromal fibroblasts play a supporting role in the development of epithelial cells and occur in high amounts among the cells that support the epithelium. Those fibroblasts that are altered by transformation seem to provide inadequate paracrine signaling to the epithelial cells, and stromal fibroblasts are often found in operable carcinomas (Karnoub et al. 2007; Orimo et al. 2005a; Orimo and Weinberg 2006, 2007b). Hepatocellular carcinomas (HCC) show excessive expression of fibroblast growth factor receptor 3 (FGFR3), which binds fibroblast growth factor (FGF). FGFR3 overexpression has been diagnosed as one causal basis for HCC. Because of this, increased FGFR3 expression may become a useful liver biomarker for correlating cellular changes that produce an HCC outcome (Qiu et al. 2005). Cancer-associated fibroblasts (CAFs) show the ability to inhibit cancer cell apoptosis, induce CAFs and epithelial cell proliferation, and stimulate angiogenesis. The latter comes into play when a tumor gets larger and more dissociated from its cellular environment and derives its own blood supply due to it altered metabolism (Orimo and Weinberg 2006). An increased number of myofibroblasts (main protein: smooth muscle actin) also appear in activated stroma at the chemical carcinogenesis site and also participate in the cancer process by supplying chemokines like stromal-derived factor (e.g., SDF-1) for tissue repair, growth, and neoangiogenesis (Orimo and Weinberg 2007b). The following myofibroblast biomarkers are present in the stroma: vimentin, FSP-1 (a fibroblast specific protein), chondroitin sulfate proteoglycan, and the enzyme prolyl4-hydroxylase. FSP-1 has been found to promote metastasis in breast tissue, but in knockout FSP-1 (−/−) mice a delayed and decreased tumor incidence was observed; this result was reversed if tested mice were coinjected with FSP-1 +/+ cells (GrumSchwensen et al. 2005; Orimo and Weinberg 2006; Schmidt-Hansen et al. 2004). This suggests FSP-1 might be one of the essential factors in the development of some types of breast cancer. Other stromal cell studies in prostate and breast cancers show that a small fraction of fibroblasts that have inactivated retinoblastoma gene product (pRb) function and either lack the p53 gene or have a mutated p53 (Hill et al. 2005a,b). The cell is a unit, but only a reproductive and biochemical unit, within the local functioning tissue or syncytium. A syncytium is defined here as an epithelium or tissue collection of cells in which there is cytoplasmic continuity and metabolic cooperation among the constituent cells. Tissues are integrated in part by means of the gap junction connections (GJs), cellular adhesion molecules (CAMs, e.g., E-cadherin), desmosomes, tight junctions, pores, and various specific channels. The local environment is electrically integrated by specific Na+, K+, Mg+2, Ca+2, and pH gradients (creating critical cellular membrane potentials) as well as biochemically linked at juxtracrine sites (physical docked cell-to-cell links) and paracrine controls (local hormonal links among a neighborhood of cells). The intercellular integration forms a connected functional array of cells in the local tissue. Such an integrated set of cells shall be referred to as the local tissue array (LTA) in the chapter. One of the first demonstrations of a specific gene that produced a phenotypic cancer-related factor came from an analysis of human colorectal tumors (Hedrick et al. 1994). A novel gene, DCC (deleted in colorectal cancer), was found located
114
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
on chromosome 18q and was found to act as a tumor suppressor gene (Fearon et al. 1990). The critical gene was a membrane-bound protein of the immunoglobulinCAM family, a cell-to-cell connector protein that was found by cell-surface immunohistochemical labeling and by sequencing cDNA clones. The protein is located in central and peripheral nervous system axons and in differentiated cell types of the intestine. Colorectal tumors, which lost their capacity to differentiate properly into mucus-producing cells, uniformly lacked this membrane protein which showed an inverse relationship between differentiation and tumorigenesis. The critical cell surface protein was identified as β-catenin, a subunit of the E-cadherin protein membrane complex that participates in the essential process of biological cellular adhesion, an intercellular process that is fundamental to all metazoans. E-cadherin, when suppressed, causes the progression of carcinogenesis and metastasis (Margulis et al. 2005). The localization of the human β-catenin gene on chromosome 3p22 links the adhesion gene to a region that is frequently altered in several human malignancies including basal cell carcinomas (Trent et al. 1995). When β-catenin binds certain transcription factors, the complex directs the membrane signaling of the Wnt pathway in organism development. Cell adhesion and orientation is essential to normal tissue-organized functional cells; but when spatial misalignment occurs it can often lead to transformation and then tumors. GJs are one example of an intercellular integration mechanism as shown in in Figure 5.3. GJs play major roles in physical and functional metabolic intercellular
Figure 5.3. Functional gap junction intercellular communication. Adjacent cells intercommunicate small molecules through GJ channels comprised of Connexon 43 (yellow cylinders). Two major signal transduction mechanisms, PC-PLC and Mekdependent, are also shown as involved in the inhibition of intercellular communication through cell-to-cell Cx43 gap junctional channels in response to tumor promoters. Cx43, connexin43; DAG, diacylglycerol; ERK, extracellular signal-regulated kinase; Mek, MAP/ERK kinase; PC-PLC, phosphatidylcholine-specific phospholipase C; RTK, receptor tyrosine kinase and DNA in dashed circles. [Slide kindly provided with permission by Dr, Brad Upham, Department of Pediactrics and Human Development, Michigan State University ([email protected]).] See insert for color representation of this figure.
5.3. CELL BIOLOGY OF CANCER
115
connections (Trosko 2005, 2007; Yotti et al. 1979). GJs can integrate regional cellular functions by allowing the rapid distributions of critical metabolites (e.g., ATP and cAMP, paracrine hormones) among cells in a cooperating local tissue array (LTA). If GJs are disrupted by chemicals with tumor promotion capabilities, then cells can become isolated. Cellular isolation is a main feature of cancer formation. The LTA normally averages cellular variances and capabilities and allows the integration of regional tissues into the design requirement for organ function. Cells also benefit from GJ-mediated sharing of essential metabolites because not all of the cells in a tissue possess the same metabolic capacity (Peterson 1983). Cells of lesser capacity benefit from those with ample capacity. If some cell fields are under stress and some cells are not, GJ sharing and other intercellular communications are beneficial to the tissue as a whole (Holder et al., 1993). Higher concentrations of reactive oxygen species (ROS) and reactive nitrogen species (RNS) can inhibit GJ function by connexon (Cx) inhibition, among other effects (Upham and Trosko, 2009). However, at lower doses of ROS or RNS, which do not damage cellular macromolecules, including the transmembrane connexons (Cxs), GJs act as conduits among a community of cells for secondary messengers that can be generated among the ROS and RNS (e.g., superoxide anion, hydrogen peroxide, nitric oxide). These chemicals can control specific cell signaling gene expressions in initiating the integrated cellular stress response (Upham and Trosko, 2009). The LTA may be a higher and more relevant physiological unit within each organ than the individual cells of the organ as a whole (Holder et al. 1993). At the organ level, the GJ interconnected cell fields are often heterogeneous with natural physiological barriers, such as fascia, bone, and various extracellular matrix elaborations. These create local variable environments in GJ-interconnected cell communities, each executing regional tissue-specific functions (Kam et al. 1986; Pitts et al. 1988). These regional adaptations allow for the integration and phenotypic expression of the cell collective which altogether create organ function. This organization can be disrupted in tumor promotion. The importance of GJs to normal functioning of cell-to-cell communication has been well-documented (Trosko 1988, 2007; Trosko et al. 2004). An important systemic GJ function is diluting xenochemicals among cells in an array of GJ-connected cells, thereby facilitating the metabolic dilution and steadystate elimination of their harmful effects. This dispersal is achieved by rapid, radiating intercellular movement of xenobiotic molecules from the entry point of exposure outward into the tissue with via passage through GJ channels. By diluting the toxicant in many cells, the elimination enzymes in the LTA tend to operate below their KM values, and thus the toxicant is eliminated closer to steady-state (Holder et al. 1993). Another GJ function is nourishing of sick or deprived cells by healthy neighboring cells (reciprocity). This is a concept that is basic to tissue homeodynamic equilibrium. This inuring process offers plasticity to cells during injurious processes and responses, such as those taking place in the I and P stages of chemical carcinogenesis. As long as all forms of cell-to-cell communication are not compromised and the toxicity threshold is not exceeded, the compromised cell is remediated with GJ-mediated assistance from healthy neighboring cells. Hence, an injured tissue or preneoplastic tissue can reversibly recover or remodel following a toxic insult
116
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
(Pitot et al. 1985). However, if toxicity threshold limits are exceeded, then affected cells may slow or even stop communicating because the GJ channels become disrupted (Figure 5.3). Disorganization within the LTA can then occur and can be followed by local morbidity, apoptosis, or necrosis. Remnant borderline-healthy cells are problematic. If cells do not fully correct to normalcy, they can remain in a quasi-dysfunctional state permissive to agents with tumor promoter activity (Barrett et al. 1986; Rosenkranz et al. 1997; Trosko et al. 1981, 1982; Warren et al. 1981). Early cell alterations cause toxicity, without necessarily mutating them, that can predispose these cells, or communities of cells, to a preneoplastic state. The alterations either repair in a timely manner or they do not (Vasiliev and Guelstein 1966). If not, with duration and sufficient dose, such toxic actions can lead to additional cell injury. Such injuries can either (1) force a cell mutator phenotype [i.e., the cell alterations cause cells to self-mutate much faster than background rates (Loeb et al. 2003, 2008)], (2) force a direct lethal mutation in the local cells (epithelial or stromal), or (3) alter a cell’s chromosomal epigenotypes, allowing the next step in I-stage fixation (abbreviated as I*) (Holliday 1987, 2005), or (4) cause a combination of these adverse responses to occur.
5.3.4.
Clonal Aspects of Carcinogenesis
It is established clinically that many tumor sites exhibit morphological and biochemical heterogeneity (pleomorphism) with variable tissue conversions and differentiations in different parts of a tumor (Fidler 1978; Hockel and Vaupel 2001; Orimo and Weinberg 2007b; Yamagiwa and Ichikawa 1977). Neoplastic phenotypes undergo changes with time; and their altered behavior begins early within the LTA, and then with time, in the organ as a whole, while becoming increasingly neoplastic and aggressive (Calabrese et al. 2007; Fidler et al. 2007). Progressive acquisition of novel phenotypic traits in neoplastic cells is the basis of the current established general model of chemical carcinogenesis which states that carcinogenesis occurs in successive requisite stages (Bernards and Weinberg 2002; Boutwell 1974; Foulds 1957; Furth 1953; Oliveira et al. 2007). It has been suggested that the individual steps within a stage can be activated randomly by different activators, at different times, morphospaces, or locations, but it is their resultant confluence that actuates the convergence of consummated steps in each stage of cancer progression. The differing modes of activators produces pleomorphism, which is often seen in neoplastic tissues. Each stage of carcinogenesis has its own characteristics, and each appears biochemically independent of the other stages (see Chapter 16) (reviewed in Slaga 1983, 1984a–c). What is biologically necessary for cancer is that all stages are executed. Hence, lifestyle factors can influence various stages at differing times during the life history of the organism, among many other factors (Foulds 1957; Malarkey and Maronpot 2005). Common clinical observations indicate that newly acquired phenotypic traits take more time to actuate and/or transact in some organ sites, as opposed to others. The order of their physiologic occurrence of cancer differs by life experiences, varying environmental factors, familial predisposition, and inheritance (Doll and Hill 1956, 2004; Lyman 1992; Wogan et al. 2004). The number of qualitatively independent stages from initiation through metastasis and then to distal colonization has been estimated to be at least 6–8 stages that are
5.3. CELL BIOLOGY OF CANCER
117
obligatory for cancer progression (Emmelot and Scherer 1977; Farber 1987; Gatenby and Gillies 2008; Marks et al. 2007; Rangarajan et al. 2004; Weinberg 1989). Each stage contains a number of steps; only some of these steps are defined (which are discussed in this chapter), others suspected indirectly, while others such as the episomal controls in chromatin and RNA controls are recognized but largely are uncharacterized at this time. Not only must all 6–8 essential stages be consummated in multistaged carcinogenesis, but also each stage must be driven to completion by a combination of duration and concentration of the causal agent(s) of the steps, host factors, genetic predisposition, lifestyle, stress, age, and son on. The stages of cancer are schematically represented in Figure 5.4, where some of the differences in the stages are indicated. The steps within a stage are not necessarily ordered and happen as results of different lifestyles and environments. The stages are an obligatory sequence for cancer to occur. It a common, but little understood, observation that some cancer organ sites transact these stages faster than others. The experimental study of “cancer stages” started in 1941 with the studies of Issack Berenblum but are still not fully understood nearly 70 years later, much less all the steps that comprise each of the stages (Berenblum 1954; Berenblum and Shubik, 1949; Farber 1984; Slaga 1983, 1984a–c). The following is an outline of eight stages of carcinogenesis that is thought to be consistent with the many documented reports on: (1) initiation stage (I stage) physical and/or chemical disruptive alterations in/on cells with eventual foci formation, competition, local adaptation, progressive loss of local communication in the LTA, and the beginning of altered glucose metabolism (cf. Section 5.3.9), (2) various phenotypic transformation processes occurring within the same organ from a common adverse stimulus, (3) critical DNA mutations and fixations (i.e., escape DNA or apoptotic removal) depicted by the transition: I (altered cell types) → I* ( fully transformed cell types and some are intiated), (4) the beginning of alternative differentiation pathways, with attributes of hyperplasia, hypertrophy, and dysplasia and anaplasia wherein affected cells are sensitive to mitosis and insensitive to apoptosis signals, (5) tumor promotion with untimely growth and often accompanied by misdirected differentiation with more loss of normal communication at various levels, hormone or system disintegration or sometimes with complete hormone dependency, and the tumor becomes recognizable or palpable in the tumor promotion stages, (6) selection of a subset of cells that acquire malignancy (cancer) characterized by I* (initiated cells) → I** (malignant cancer cells), (7) major adaptations producing heterogeneity and a complex tumor mass or masses, leading to a subset of I** cells that undergo metastatic conversion and enter the vascular system by intravasation, and (8) travel through blood or lymph, exit by extravasation at certain fertile distal organ sites where tumor cells readapt and regrow in a new environment (colonization). Not all organs are equally susceptible to I*-stage events; and according Fiedler ’s “seed and soil hypothesis,” it makes a difference where a metastasized cell migrates as to whether it will regrow at a distal site (Fidler et al., 2007). Because of the many requisite stages in the cancer process, it often takes quite some time to traverse all the stages. These process durations, and their varing probabilities of occurrence, likely account for the observed 15- to 30-year lag times from the carcinogenic stimulus to malignancy or metastasis (Aguirre-Ghiso 2007; Aslakson and Miller 1992). Because each stage is qualitatively and quantitatively different in their responses to the carcinogen, each has its own
118 Figure 5.4. Stages of carcinogenesis. Stages have necessary biochemical and cellular changes that are either directly or indirectly produced by the carcinogen.
5.3. CELL BIOLOGY OF CANCER
119
specific kinetics and carcinogen dose dependency (Druckrey et al. 1963; Emmelot and Scherer 1977; Slaga 1983, 1984a–c). The main question is whether the observed tissue phenotypic heterogeneity in tumors, as discussed by Fidler in 2007, is compatible with a monoclonal tumor origin. The very existence of humans as complex metazoans (multicellular, differentiated organisms) is itself the best example of a large-scale occurrence of differentiated but controlled heterogeneity from zygote to maturity despite identical DNA content in each cell type. Differential, controlled expression of genes and gene set combinations in chromatin (cf. Section 5.3.7) can achieve almost any level of phenotypic variability. Analogously, there are many variations of morphological and functional heterogeneity of tumor cell populations. These variations evolve via acquired genomic differences through the somatic mutations in daughter cells when the neoplastic clone expands (Elsasser 1984; Rubin 2007). Because there are more mitotic events in tumors, the genome is more susceptible to mutations because DNA is more often exposed in the S-phase and thus is more susceptible to mutation or other chromatin alterations. Within numerous different clones, there exist karyotypic heterogeneities (Heim et al. 1989; Hoglund et al. 2005). This variance has been observed to be a regular feature of many advanced neoplasms in both human and animal tumor systems (Fadl-Elmula et al. 1999; Hoglund et al. 2001). The detection of microscopically visible disturbances of the mitotic processes during carcinogenesis gave the main impetus to the formulation of the somatic mutation theory of cancer (Klein and Klein 1985). Mutability based on chromosomal instability has been reported to increase with tumor progression (Cifone and Fidler 1981). When a reactive compound enters an organ from the blood, it may react within a number of proximal cells in the local regions of the organ in a chemically stochastic manner. If not remodeled or removed by apoptosis, this can produce many types of altered focal cells (AFC) (Kitchin et al. 1994; Pitts et al. 1988). Limited foci evaluations have been done in short-term and mid-term rat liver models using a variety of histological stains and morphology for potential carcinogenicity (Popp and Goldsworthy 1989). Though most of the early focal cells are merely phenotypically altered, some foci may contain transformed cells and exhibit increases in the size (hypertrophy) and/or the number of foci (hyperplasia) long before any tumor appearance (Pitot et al. 1985, 1996). These foci are believed to contain the precursors of organ tumors, and possibly the precursors to metastatic cells, but the latter is far from a settled issue (Bernards and Weinberg 2002; Cairns 1975; Nowell 1986; Ting et al. 2006; Weinberg 2008b). A quantitative dosimetry study by Kitchin et al. (1994) was done on a number of known experimental tumor promoter chemicals: 12-O-tetradecanoylphorbol-13-acetate (TPA) (in mouse skin), 2,3,7,8tetrachlorodibenzo-p-dioxin (TCDD, dioxin) (in rat liver), phenobarbital (in rat liver), chloroform (in rat liver), and Clophen A50 (a polychorinated biphenyl) (in rat liver). All these promoters exhibited (1) second-order dose–response curves (y = a + bx + cx2) in producing increased foci in liver or papillomas in skin and (2) at lower picomolar doses of tumor promoters the focal responses decreased to the same response (or less) as control levels. This indicates a threshold for these promoters in the exposure range of picomoles/kg b.w. or femtomoles/animal (Kitchin et al. 1994). The null point (no response) varied for each chemical in the low femtomoles/
120
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
animal range. At dose–responses statistically greater than control focal responses, the number scored was in excess of 2000 up to 5000 foci/liver. Foci observations have indicated the following: (1) There are lower doses in a restricted range below which promoters do not produce the precursor foci for tumors, (2) tumor promoters show drug-typical dose–response curvature (initial concavity at the low end of the dose range), (3) dose ranges spanned from 1 to 3 orders of dose, and (4) for each chemical, the lowest “responding doses” produced thousands of foci that had varied phenotypes and presumably many genotypes among them. The prior discussion suggests the likelihood of poor or no communication between the tumorigenic precursor cells and their surrounding altered cells in the focus within a microenvironment of a tissue array (Barrett et al. 1986; Klein and Klein 1985; Upham et al. 1994, 2007). This early step commences “I-cell” formation in foci and their isolation from normal feedback controls of mitosis and apoptosis in accordance to the biological severity of the toxic event forming the focus (Trosko et al. 1982, 1984). Chemical structure and functional groups determine the size, number, nature of the phenotype, and conversion rate of foci to tumors in a particular LTA (Kitchin et al. 1994). No single morphological marker appears capable of identifying or characterizing all organ foci because of their dynamic phenotypic heterogeneity. The current problem is identifying which cardinal I*-stage effects should be measured; biological biomarkers for the I* stage are being considered, and some are already in development (D’Alessio et al. 2007; He et al. 2007a,b; Oliveira et al. 2007; Pitot et al. 1996; Scheel et al. 2007; Szyf 2007; Weinberg 1997; Wolters and MacKeigan 2008). For the P stage of cancer progression, biomarker detection is improving. There are new insights to intranuclear preservation of specific genetic and epigenetic controls of gene expression. Some biomarker examples of the P stage are: increases in DNA, RNA, and protein syntheses, decreases in apoptosis, increases in necrosis via the transcription factor nuclear factor kappa-light-chain enhancer of activated B cells (NF-κB), and overexpression of cyclooxygenase-2 (COX-2), which can in turn produce excessive prostaglandin E2 (Adler and Chang 2006; Albor and Kulesz-Martin 2007; Boutwell 1976; Cheng and Lai 2003; Feinberg et al. 2006; Makunin et al. 2007; Marks et al. 2007; Pereira and Stoner 1985). Another example seen in the DMBA/TPA two-stage classic chemical carcinogenesis I–P protocol for hyperproliferation of skin epithelium is dependence on tumor necrosis factor, TNFα (Moore et al. 1999). Cancer is widely perceived as a heterogeneous group of aberrant growths potentially arising from different organ sites each with markedly different biological properties. In medical practice, cancer growths, with their attrition of normal ontogenesis and diverted development in the affected LTA, initially become noticeable as palpable or being visible but are also discovered by manifested signs of organ dysfunction. Tumors have been thought to originate from a series of clonally selected genetic changes in key tumor-suppressor genes and oncogenes that originate from one malevolent cell in the I* stage that is critically genetically mutated. For reference to some of the known mutations that may be observed, refer to Meuth (1990) as well as to Nowell (1976) and Fialkow (1976) for their proposed rationale for the hypothesis of {one mutated cell} → {one mutated and transformed clone} → {one tumor type}. Various approaches have suggested monoclonality of certain tumor types—for example, neoplasms in women, who are constitutionally heterozygous at the glucose-
5.3. CELL BIOLOGY OF CANCER
121
6-phosphate dehydrogenase (G6PD) DNA locus and express only one G6PD variant in the tumor. Examples of tumor DNA heteromorphic variants that possess X-linked restriction fragment polymorphisms have yielded comparable results among tumor restriction fragments. Loss of heterozygosity (LOH) in an X-chromosome gene should produce identical descendents in the soma because they are copies of the original X inactivated gene. B-cell neoplasms express only one type of immunoglobulin on the cell surface. Heim et al. (1988) suggested an argument for “the monoclonal composition of most tumors” from cytogenetic investigations that showed chromosomal rearrangements from a tumor banded similarly. Monoclonality also makes sense based on Theodor Boveri’s original observations (Boveri 1929) as well as on subsequent rationale for the presumptive basis of the I–P protocol in further experimental investigations of carcinogenesis stages (Berenblum and Shubik 1949; Boutwell 1976). However, all the initial monoclonal assumptions were based on tumors of hematopoietic origin in contradistinction to observations of an early stage of tumorigenesis in solid tissues (Heim et al. 1988). Important too is the fact that histopathological and clinical evidence were based on epithelial carcinogenesis models. When, where, and what endpoints to investigate have been issues in I-stage investigations. In acute nonlymphocytic leukemias, certain G6PD localization isozyme studies have indicated that the tumors studied could be polyclonal rather than monoclonal (Fialkow 1976). Problematic to the clonality issue has been the difficulty of monitoring human solid tumor clonality longitudinally—that is, to follow directly in time the developing heterogeneity of the neoplastic cell populations (Fidler et al., 2007). A number of nongenetic events in the early I stage can cause poor intercellular integrations either temporarily or more long-term. Cardinal carcinogenic event(s) can be (1) transformative, (2) genetic, (3) nongenetic, or (4) a combination of all three events. Transformed cells are merely altered cells that exhibit morphological and/or biochemical alterations. These cells may be repaired, removed, or compartmentalized (e.g., in a cystic structure); but if they persevere through reactive immune responses and/or apoptosis, these imperfect surviving cells are often candidates for preneoplastic cell population formations. Many factors can later affect these cells toward irreversible transformations in the course of a normal lifestyle (Boutwell 1976; Feinberg 2005; Flanagan 2007; Furth 1953; Holliday 2004). The subset of tumor cells that increasingly becomes variable and preneoplastic are those transformants (i.e., I cells) that in time, and with further sufficient provocation, finally irreversibly acquire the potential to become cancers. That is, the initiation transition to a cancer-prone cell is designated as an I-cell → I*-cell conversion (Figure 5.4, Stage 2, Step 2). The I → I* conversion is considered by many to be an irreversible stage while consummating the stage of tumor initiation. Tumor promotion progresses to the I* cell and by definition is a viable group of expanding cell numbers and cell size, showing more deviations than those previous and become irreversibly altered while gaining autonomous characteristics (Figure 5.4, Stages 4 and 5). Essential in carcinogenesis is the imbalance between (1) the tendency of neoplastic cells to divide and spread throughout the body and (2) the capacity of the organism to regulate and restrain such growth. Cell disintegration can begin with anoikis as an example step in the P stage. Anoikis is where cells loosen or detach from the surrounding extracellular matrix (ECM), such as loss of E-cadherin, which
122
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
participates in cell-to-cell adhesion (Gort et al. 2008; Liu et al. 2008a; Woods et al. 2007). Because of essential ion, nutrient, and paracrine gradients, these loosened cells tend to remain in the vicinity where they were born, nurtured, and developed but nonetheless are not properly integrated in the tissue. Later, in tumor promotion these loosened cells can be freed mechanically from the linkages to each other and the basement membrane. Autonomous physical separation is thought to be an early step in the metastatic stage that can then lead to intravasation (i.e., entrance into the blood compartment) (Fidler 2002; Talmadge et al. 1982). Therefore, initiation of carcinogenesis is seen as the commencement of faulted localized cellular interactions, rather than some kind of system failure of proliferation control (Cheng and Lai 2003; Fearon et al., 1990). Tumor cells can give rise to other tumor cells with genetic defects, and the literature is replete with such observations, but cancer can also be nongenetically induced and evolved along aberrant differentiation pathways (Feinberg 2004, 2005; Holliday 1987; Jablonka 2004). That is, alterations of a significant nature can take place concerning when, where, how much, and the terminus of gene expression in lieu of altering the DNA of the gene or its controllers. It is suggested in this chapter that likely both produce carcinogenesis. It seems apparent that a local chemical carcinogenesis event not only alters epithelial cells but also interacts with the proximal organ cells, including the supporting stromal cells. Every solid tumor investigation to date is a “snapshot” that captures only an instantaneous picture of the life of the tumor at that particular time point. Whether the tumor cells actually evolve toward greater simplicity or toward greater complexity, when their genetics are assessed, has been conjectured (Elsasser 1984; Rubin 1984). The current polyclonal competition model states that phenotypically and genotypically “faulted clones” evolve and interact dysfunctionally among themselves and local normal cells. The faulted clones become more heterogeneous at the expense of normal cell organization. The cellular field effects become more chaotic in the P stage, compared with the I stage; and as a result of selective Darwinian pressure, these cancerous cells elaborate complex colonies sometimes exhibiting “tumors within tumors,” a rather common histological observation in oncology. If not remodeled, repaired, or removed, the tumor colonies evolve and finally converge to contain a few surviving, thriving, irreversible, autonomous and resistant clones that finally dominate the community of cells often by attrition (Heim et al. 1988). These are cancer cells and are referred to as I** cells; they are malignant but not necessarily metastatic. Metastasis conversion is a rare and complicated set of events of the all of the cancer stages (Figure 5.4, Stage 6). It evolves with further genomic instability and induction of specific transcription factors that stimulate an epithelial-to-mesenchymal cell transition (EMT) event (which is a recall of an earlier normal embryological event) allowing the highly changed cancer cells an entrance through vessel tunica and endothelial cells into the blood (Feinberg 2007; Fidler 1978; Weinberg 2008a,b). See Figure 5.4, Stage 6. A cell can be mutagenized by either overexposure to sunlight (UV) or adventitious environmental carcinogen exposure, or a cell may be spontaneously transformed by a nongenetic event that might give rise to the spontaneous tumors seen in organs in control groups of a two-year chemical carcinogenesis bioassay (Malarkey and Maronpot 2005; Rubin 1994). For the high doses used in standard cancer bioassays
5.3. CELL BIOLOGY OF CANCER
123
conducted by the US National Toxicology Program (NTP), there is quantitatively enough chemical to execute all of the inherently different and independent classic Initiation, Promotion, and Progression (I–P–P) stages to malignancy and sometimes to metastasis (Haseman et al. 1987). This has profound implications as far as cancer dosimetry in NTP cancer bioassays. Specific chemically reactive compounds often can act as total carcinogens (e.g., 2-AAF, DMN, nitrosourea, etc.). These potent carcinogens produce cancers even at low doses and shorten the time to the first observable tumor. Their oncogenicity often appears monoclonal or pseudomonoclonal because of the chemical activity at the high NTP test exposures along with their direct lethal effects on dying or incompetent cells in the process of apoptosis or necrosis. That is, normal interclonal competition that occurs in spontaneous tumors or lowpotency carcinogens is circumvented by the high-test exposures in the bioassay as currently designed into test bioassays. Some chemicals, by their nature, are not chemically or biologically active enough to execute all of the I–P–P stages at lower exposures within a lifetime. It seems likely for chemical carcinogenesis events at moderate to low environmental exposures that the development of cancer is not necessarily a one-cell event, in which case necessarily develops monoclonally to a full tumor, but rather is variable activations of a moderate number of altered cell types, some of which are I cells whose earliest appearance occur within altered foci or within foci-in-foci. Because these altered focal cells do not cooperate adequately with their neighboring focal cells, normal cellular homeodynamic equilibrium cannot be achieved in the niche of the altered I cell(s). Under these poor communicating circumstances, I cells can transform to become fixed I* cells and from hence progress to dysfunctional foci in the tissue and can then be promoted in P stages (Figure 5.4, Stages 4 and 5). As appealing as the single-cell evolution cancer model was (and still is for hemopoietic cancers, for high test doses of chemically active compounds, and for certain oncogenic viruses), numerous scientists now think multiple types of nongenetic alterations are required in the I stage before any critical genetic events take place and is the more common scenario (Feinberg 2007, 2008; Feinberg et al. 2006; Hanahan and Weinberg 2000; Holliday 1987; Trosko 1988). For cases of sufficient amounts of a total carcinogen or carcinogen plus promoter agents, the I*-stage events are followed by the promulgation of the P phase with additional types of genetic and nongenetic combinatorial changes necessary for tumor progression (Barrett and Ts’o 1978a; Klein and Klein 1985). Though only a teleological argument, it does not make evolutionary sense that the cell would evolve as “a reproductive construct” wherein a singular critical genetic element that, if altered, would inevitably lead all the way through all necessary cancer processes (I–P–P) (Jablonka and Lamb 2007). With such an Achilles heel, metazoans would not likely have survived long enough to evolve—obviously contrary to the facts. Biological maintenance and continuity necessitates a plethora of coordinated protective and corrective mechanisms to chaotic environmental forces and singular anomalous genomes, or epigenome restrictive or dead-end pathways (Anisimov 2007; Holliday 2004; Rubin 2007; Vicencio et al. 2008). Important cellular or body actions are generated and controlled by redundant means (Tononi et al., 1999). Redundancy is the rule in more complex organisms not the exception. The cell is protected by a number of systems and back-up systems to prevent stress-induced failure, precocious
124
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
aging, or untimely cancer (Anisimov 2003, 2008; Ben-Porath and Weinberg 2004; Holliday 2004; Klein et al. 2007).
5.3.5.
Biology of Inflammation and Cancer
The four basic aspects of the toxic response has been defined as: tumor (swelling), calor (heat), dolar (pain), and rubor (redness). As early as 1863, Rudolf Virchow found these toxic signs in neoplastic regions that consistently and concomitantly showed a lymphoreticular infiltrate mixed in with the altered local neoplastic cells. He asserted a connection between inflammation and cancer. The Nobel Laureate immunologist Mac Burnet stated: “in large, long-lived animals … inheritable changes must be common in somatic cells and a proportion of these changes will represent steps towards malignancy” (Burnet 1971). The causes of sporadic human cancer are seldom sought or recognized in clinical practice, but it is estimated that carcinogen exposure or chronic inflammation are two important underlying conditions for tumor development. Chronic inflammation accounts for at least 20% of all human cancers (Pikarsky et al. 2004). The main function of the mammalian immunodefense system (IDS) is to monitor tissue homeostasis, protect against invading or infectious pathogens, eliminate damaged cells or tumor cells. Recent clinical studies and experimental mouse models of carcinogenesis have expanded our understanding of the complex relationship between immune cells and developing tumors (DeNardo et al. 2008). There is a paradoxical role of adaptive and innate leukocytes as crucial regulators of cancer development and spontaneous tumorigenesis. There is evidence to suggest that the inflammatory and immune systems can and do inhibit the development of cancer (de Visser et al. 2006). Cancer inhibition may occur by two recognition events: (1) The host IDS has a dedicated mechanism to perceive and eliminate transformed cells in tumor surveillance, and (2) the adaptive immune recognition of tumorassociated specific antigens is an important means by which the IDS limits the development of certain cancers (Smyth et al. 2006). If the first event is specifically inhibited or is saturated by overwhelming carcinogenic forces in the local tissue array (LTA), then the critical homeostasis is disrupted in that LTA and the immune defense system (IDS) can actually participate in the promotion of cancer development (Rakoff-Nahoum 2006). The latter might be generally considered an eccentric form of autoimmune disease, for which there are many examples. In fact, most autoimmune experiences exhibit errors in the delicate IDS balance. It is plausible that qualitative and quantitative relationships exist between the extremes—that is, IDS protection or P-stage participation. Epidemiological evidence points to a connection between long-term inflammation and the development of cancer that is characterized by dysplasia, hyperplasia, and sometimes irreversible cancer transformations. Nearly 15% of worldwide cancer incidence in humans is associated with microbial infection (Kuper et al. 2002). Chronic infections, with human papilloma virus or with hepatitis B or C viruses in immunocompetent human hosts, can lead to cervical and hepatocellular carcinomas, respectively. Infection with the human herpes virus can produce Kaposi’s sarcoma in the skin. Karposi’s cancers are seen more often in the IDS-compromised AIDS patients. After protracted inflammation, Helicobacter pylori can cause stomach irrita-
5.3. CELL BIOLOGY OF CANCER
125
tion and cancer. Long-standing inflammatory bowel disease caused by the intestinal microflora can lead to colon cancer (Balkwill and Mantovani 2001; Coussens and Werb 2002). Long-term exposure to irritants such as cigarette smoke, asbestos, coal dust, cotton fiber dust, or silica can also predispose humans to carcinogenesis via inflammatory mechanisms. Inflammatory signs of leukocyte infiltration at tumors infected with microbes or sites of chronic irritation have been observed. As observed by Virchow in the middle of the 19th century, many tumors—even tumors such as mammary adenocarcinomas, for which infection or irritation are not thought to be obligatory factors—do exhibit a “lymphoreticular infiltrate.” Many tumors not requiring frank inflammatory conditions have activated fibroblasts and macrophages along with a gene expression profile typical of an inflammatory signature (de Visser et al. 2006; Rakoff-Nahoum 2006). The quantitative aspects of wound repair and suppression of inflammatory gene expression have been found to correlate negatively with cancer progress. For example, nonsteroidal anti-inflammatory drugs prevent spontaneous tumor formation in patients with gastrointestinal familial adenomatous polyposis which offers more support for inflammation participation in carcinogenesis (Ulrich et al. 2006). In summary, cancer and inflammation have been correlated by observations in epidemiology, histopathology, inflammatory profiles, and the efficacy of anti-inflammatory drugs in prophylaxis. Groups of cells within a tumor that do not integrate properly usually fail to do so because of structural or functional lesions, and the LTA cells can become sufficiently deviated so as to activate local immune sentinel cells that attract myeloid cells of the IDS for repair. In this sense, the IDS is an evolved conservative system of resource maintenance by cellular repair rather that complete cell removal and resynthesis. As reviewed above, if the injury is too severe, these cells are removed by apoptosis or necrosis with phagocytosis. Most tissues have or have access to innate immune cells [macrophages, dendritic cells (DC), mast cells (MC), natural killer cells, neutrophils, basophils, and eosinophils]. Interacting immune cell types function by sensing, mobilizing, repairing, or adapting any tissue of the body that may be perturbed by the many internal or external environmental hazards. Regional cells are composed of macrophages, DC, and MC. These cells are always positioned in organs as sentinels that constantly monitor the integrity and well-being of the various organ environments (de Visser et al. 2006). When local injury or tumor inception occurs, the sentinel cells release signals (chemokines) that recruit other immune cells to assist in repair. Various myeloid cells are attracted to the site by the agency of the chemokines classes “CXC” and “CC.” CXC chemokines recruit lymphocytes and neutrophils, whereas CC chemokines recruit monocytes, eosinophils, dendritic cells, lymphocytes, and natural killer cells (Balkwill and Mantovani 2001). The last 20 years of research has established the concept that tumor development and consequent malignancy is the result of involving both the affected cancer cells themselves and a number of other local noncancer cells. When a neoplasm forms, stromal and epithelial cells interact (covered previously) and local LTA myeloid sentinel cells recruit immune cells (listed above). A further example of multicellular involvements in tumors and tumor environments is demonstrated by the generation of neo-angiogenesis for tumor growth and thus the contribution of the vascular endothelial cells. That is, the heterocellular LTA tumor site is an interacting complex of different cell types (Folkman 2002).
126
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
Neoplastic tumors that have transformed cells present as “wounded regions” to the corporate IDS (Deelman 1927; Eming et al. 2007; Süss et al. 1973). Though the sensing modes are not well understood, the local IDS actively senses these neoplastic regions as (1) foreign bodies (nonself) or (2) structures departed from a norm threshold, which information is thought to be retained in cellular memory (Schones et al. 2008; Wang et al. 2008). Hence, additional white cells are recruited to the region as an infiltrate. The tumor-associated macrophages in the infiltrate derive from myeloblasts to become monocyte precursors. As previously discussed in this chapter, the supportive stromal fibroblasts may also be incorporated into an epithelial cell tumor. The tumor becomes a complex array of cells. The problem with many evolving transformed cells is that they do not present membrane surface epitopes sufficient for cellular recognition, and removal of these cells go undetected. Genetic mutations more often delete functions than add new functions; this might be another reason for insufficient recognition and removal (Klein et al. 2007). Because some altered cells are more or less tolerated, the immune defense may be deceived well into the I or P stages before the altered cells become sufficiently recognized. Most aberrant events recede from this point because if they did not, we would experience many more cancers that we actually epidemiologically observe (Klein et al. 2007). Correction is the usual case in the healthy individual. Cancer is a commonly experienced disease in the United States, with 77% of diagnosed cases occurring in persons 55 and older, and the lifetime chances are 1 in 2 if male and about 1 in 3 if female of having some form of cancer (American Cancer Society 2008). The chances of a successful cancer, however, are rare-to-uncommon events compared with the total lifetime potential carcinogenic attempts or events occurring during an average individual’s lifetime (Burnet 1971; Ames et al. 1995; Hanahan and Weinberg 2000). Repair is the rule rather than the exception (Déry and Masson 2007). As George Klein has pointed out, the cell is highly protected against cancer by at least four surveillance systems: (1) a number of DNA repair system types, both in the transcriptasome but in the conserved DNA-control regions too (see below for more discussion), (2) preservation of epigenetic control of cell memory, DNA expression, timing of expression, and maintenance of gene restriction appropriate for that cell type, in a reversible manner, which is linked to and responds to the cell’s environment by specific outside to inside control signals, (3) intracellular repair by cell-cycle arrest, anoikis, autophagy, or apoptosis, and (4) intercellular control by influence of neighboring cells, cell-to-cell contact inhibition and juxtracrine effects, paracrine effects, and cell field effects (e.g., extracellular matrices, hyaluronidases, anatomical, organ-specific). Chronic inflammatory states associated with infection or xenobiotic chemical exposure from the environment can produce genomic lesions that, in time, can become initiated tumors. It is known that hosts do fight microbial infections by moderate production of various free radicals: reactive oxygen species (ROS) [e.g., hydroxyl radical (OH•) and the superoxide radical (O2− )] or reactive nitrogen species (RNS) [e.g., nitric oxide (NO•) and the strong oxidant, peroxynitrite (ONOO−)]. Within limits inflammatory signaling pathways of the host can control excessive free radical concentrations by means of enzymes such as NADPH oxidase, myeloperoxidase, nitric oxide synthase, and others (Federico et al. 2007; Rakoff-Nahoum 2006).
⋅
5.3. CELL BIOLOGY OF CANCER
127
Individuals who respond strongly to tissue injuries that are highly distressed may overproduce ROS and RNS. This is especially important in chemical carcinogenesis where often there exist continual or frequent chemical injuries—for example, in rodent bioassays for carcinogenicity (Klaunig and Kamendulis 2004). When in excess of radical quenching mechanisms, radical production would be additive whether coming from internal or external sources; and the longer the inflammation persists, the higher the cancer risk (Federico et al. 2007; Perwez and Harris 2007). During cancer progression, the generation of ROS or RNS free radicals, and their copious intermediates like lipid free-radical carriers, can occur persistently because of an altered cancer metabolism (Gillies and Gatenby 2007; Kondoh 2008; Nishikawa 2008). This can create a cancerous condition. Net covalent reactions of free radicals with critical macromolecules, which remain uncorrected, depend on the efficiency and capacity of the various radical quenching systems and the ability to repair chemical conjugates by the biological remediation mechanisms. During the resultant inflammatory process of cancer, excessive unquenched free radicals can produce net DNA mutations in the affected cells (Federico et al. 2007; Perwez and Harris 2007; Troll and Wiesner 1985). Free radicals produce other free radicals species by electron transfer reactions. This produces reactive soluble inflammatory mediators such as metabolites of arachidonic acid, cytokines and chemokines, and they act further to produce even more reactive species. These, in turn, strongly recruit more inflammatory cells in what becomes a vicious reactive cycle. ROS and RNS can directly oxidize or form adducts with DNA, DNA transcription factors, histones, and/or control RNAs (i.e., RNAi) that can directly interfere with mechanisms of gene expression or repression. These reactive substances may also rapidly react with cellular proteins, carbohydrates, and lipids, and the derivative products may indirectly induce perturbations in the intracellular and intercellular homeostasis enough that net DNA mutation ensues. Prostaglandins and cytokines are the main substances that propagate cancer inflammation due to oxidative or nitrosative stress. The direct agents that represent the redox imbalance (i.e., between pro-oxidant and antioxidant enzyme activities) are (a) the enzymes lipoxygenase, cyclooxygenase, phospholipid hydroperoxidase, and glutathione-peroxidase and (b) effectors such as hydroperoxides, lipoperoxides, aldehydes, and peroxynitrites (Wallace 1997). Direct evidence for a link between tumorigenesis and host defense involving tissue repair has come from a number of observations. Many molecules and pathways can play dual roles in tumorigenesis and in homeostatic tissue repair. For example, the Wnt/β-catenin pathway plays a critical role in both the maintenance of the steady-state proliferative compartment and tumorigenesis of tissues (CoyleRink et al. 2002; Kim et al. 2005; Tejpar et al. 2005). The COX enzymes are involved in the synthesis of prostaglandins PGE1 and PGE2, which mediate the tissue repair in injury or inflammation via the coupling with transmembrane receptor G-protein (Marks et al. 2007). The COX genes produce key rate-limiting cellular enzymes, which catalyze synthetic pathways for prostaglandins and thromboxanes from arachidonic acid. COX-1, COX-2, and COX-3 are three isoforms identified so far. COX-3 controls febrile and analgesic processes and will not be covered here. The COX-1 gene is constitutively expressed and acts as a housekeeping gene in most tissues, and it mediates maintenance of many
128
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
physiological responses. COX-2 levels are barely detected in normal adult tissues, but the COX-2 gene has been shown to be superexpressed in many tumors and possesses pro-angiogenic and anti-apoptotic properties (Federico et al. 2007; RakoffNahoum and Medzhitov 2007). It is revealing that COX-2 is synthesized constitutively during normal embryonic growth. Knockout studies have indicated there is a necessary relationship between COX -1 and COX-2 and skin and intestinal cancer development (Marks et al. 2007; Tiano et al. 2002). COX-2 is widely regarded as a potential pharmacological target for preventing and treating malignancies, and pharmacological agents such as nonsteroidal anti-inflammatory drugs (SharmaWalia et al. 2006). COX-2 is an early inducible gene product in tumor promotion whose cellular synthesis can be up-regulated by either mitogenic, inflammatory stimuli or positive feedback by other prostaglandins. COX-2 appears capable of mediating critical step(s) in the tumor promotion stage (Marks et al. 2007; Williams et al. 1999). It is known that COX-2 produces an important inflammatory prostaglandin (i.e., PGE2, [aka dinoprostone]), which is a potent immunoregulatory lipid mediator. PGE2 plays key roles in the regulation of a number of cellular processes, including (a) the acute and chronic inflammatory responses and (b) innate immune responses that are generally produced in response to cytokines, mitogens, bacterial lipopolysaccharide, and viral infections (Sharma-Walia et al. 2006). Supportive evidence connecting systemic inflammation-related processes and cancer also comes from studies showing that dedicated tissue injury (wounding) supports tumor growth and neoplastic progression. For example, injection of the Rous sarcoma virus into chickens leads to the growth of a sarcoma at the site of injection; sarcomas may form at other sites of the chicken if that site is wounded (e.g., mechanically) in the injected chicken (Dolberg et al. 1985). Moreover, these distal wound-related tumors can be inhibited by systemic injection of glucocorticoids, such as dexamethasone (Sieweke et al. 1989). The inhibition mechanism seems to be mediated by the transforming growth factor beta (TGFβ) and fibroblast growth factors (FGFs). Recent studies investigating the role of NF-κB (a family of transcription factors central to the induction of inflammation) in tumorigenesis has provided more detailed insights into the role of inflammation in tumor promotion. Because NF-κB activation has been found frequently in tumors, it is proposed as a primary biomarker for the inflammatory response because NF-κB may constitute a major link between inflammation and cancer (Pikarsky et al. 2004). An example is the intiation–promotion (I*–P) colitis cancer model of associated colon cancer. It can be induced by the intraperitoneal injection of the carcinogen azoxymethane (the initiator), followed by multiple rounds of inflammation and leukocyte infiltration caused by dosing with the colonic epithelial cell toxin or dextran sulfate sodium (DSS, a tumor promotor) (Greten et al. 2004). It is clear in the colon that chronic inflammation augments tumorigenesis because when one dose of azoxymethane is given alone without DSS cycling, then no tumors arise. It was found that inactivation of the classical NF-κB pathway in epithelial cells of the colon by conditional deletion of the IκB kinase β (IKKβ) protein resulted in a substantial decrease in the frequency of visible tumors in this I*–P colon model (Greten et al. 2004). IKKβ is a kinase in the colonic epithelium responsible for mediating epithelial cell survival in protection from both infectious and noninfectious injury and host defense pathways in intestinal
5.3. CELL BIOLOGY OF CANCER
129
epithelium (Hu et al. 1999; Kim et al. 2006). In epithelial cells NF-κB signaling inhibits apoptosis shortly after administration of one round of azoxymethane and DSS, perhaps by the induction of anti-apoptotic factors such as Bcl-XL, a member of the pro-apoptotic Bcl-2 family. Thus, upon intestinal epithelial injury and the application of a mutagen, NF-κB provides a survival signal to initiated cells. Increased I-cell prevalence increases the probability of carcinogenicity. A similar role for NF-κB in survival of initiated cells was demonstrated in a chronic inflammation model of HCC by Pilarsky, who studied multiple drug resistance in Mdr2 knockout mice that spontaneously develop cholestatic hepatitis that is followed by the development of HCC (Pikarsky et al. 2004). Hepatitis and cancer progression was followed in Mdr2 knockout mice, and it was found that the inflammatory process triggers hepatocyte NF-κB through up-regulation of tumor necrosis factor alpha (TNFα) in adjacent endothelial and inflammatory cells. Switching off NF-κB in mice from birth up to 7 months of age, by use of a hepatocyte-specific inducible IκB-super-repressor transgene, had no effect on the course of hepatitis, nor did it affect early phases of hepatocyte transformation. By contrast, suppressing NF-κB inhibition through anti-TNFα treatment or induction of IκBsuper-repressor in later stages of tumor development resulted in apoptosis of transformed hepatocytes and caused a failure to progress to HCC. In conclusion, these studies indicate that NF-κB is essential for promoting inflammation-associated cancer, and NF-κB is therefore a potential target for cancer prevention in chronic inflammatory diseases (Pikarsky et al. 2004). Tumor promotion engenders not only the survival of I cells but also their expansion (cf. Section 5.3.6). Many inflammatory mediators such as cytokines, chemokines, and eicanosoids (e.g., prostaglandins and thromboxanes, discussed previously) are capable of stimulating the proliferation of both untransformed cells and tumor cells (Balkwill and Mantovani 2001). TNFα-deficient mice have fewer skin tumors upon administration of the phorbol ester TPA and the mutagen DMBA (Moore et al. 1999). Investigation into how TNFα regulates tumor progression in this model suggested that this inflammatory mediator acts as a tumor promotor mediator, because after application of TPA/DMBA (I–P protocol in skin) the keratinocytes hyperproliferation was shown to be dependent on TNFα and NF-κB. These factors have recently also been shown to play a critical role in the production of in inflammatory mediator myeloid cell activations in (1) the azoxymethane/DSS colon model (above) and (2) a mutagen-induced hepatocellular carcinoma upon administration of DEN (Maeda et al. 2005). Considering these two models, when myeloid cells were defective in NF-κB activation, there was impaired production of inflammatory mediators and a lack of proliferation of dysplastic epithelium. These impairments of NF-κB activation correlated with a decrease in the frequency and size of tumors compared to those produced in wild-type mice (Maeda et al. 2005). Another finding of the DEN study was that when NF-κB was impaired in hepatocytes, there was increased epithelial cell death while exhibiting an increased tumor burden. Rakoff-Nahoum and Medzhitov (2007) suggest that this myeloid cell finding may lead to the differential proliferation of initiated cells and might be detected by measuring epithelial cell death. Thus, the premalignant early I-stage tumors are “wound-like” (Coussens and Werb 2002). The known cancer stages were reviewed in Section 5.2. In many ways,
130
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
tumors are similar to healing or desmoplastic (fibrous) tissues like a clot of activated platelets (Dvorak 1986; Mueller and Fusenig 2004). As described by Coussens and Hanahan, tumor growth may be immunologically characterized as biphasic (Coussens et al. 1999). In the first phase, the body treats early tumors as wounds. This first phase is characterized by tumor growth that is mediated by the actions of the indirect stromal control that is similar to normal tissue repair. For example, in murine models of skin and pancreatic carcinogenesis, bone-marrow-derived cells (including mast cells) are responsible for producing matrix metalloproteases (MMPs, which are essential in tumor progression in a number of steps), which convert vascular endothelial growth factor (also known as VEGF) into a biologically active form to stimulate the protumorigenic angiogenic switch (Coussens et al. 1999, 2000). However, later in tumor progression, MMPs (as pro-inf1ammatory factors) come under direct control of the tumors themselves, a sign of autonomy (Coussens et al. 1999). A similar functional transition in the regulation of inf1ammation by early versus late tumors might be seen in murine and human spontaneous tumorigenesis in the intestine. COX-2 is expressed in stromal cells in early tumor-associated wounding; but in larger tumors, COX-2 is expressed only by the dysplastic epithelium (Marks et al. 2007; Williams et al. 1999). This might be explained by stromal cell having an ample but limited capacity for wound healing. When this capacity for healing is exceeded, the inflammatory condition could be selected in tumor cells that can autonomously maintain ancillary healing processes and are no longer dependent on the stroma for repair. Eventually, the tumor-associated stroma may also undergo selective pressure. There have been recent reports of genetic changes in tumorassociated stroma and even loss of p53 in tumor-associated fibroblasts (Coussens et al. 1999; Elenbaas and Weinberg 2001; Orimo et al. 2005b; Orimo and Weinberg 2007a; Rakoff-Nahoum and Medzhitov 2007).
5.3.6.
Stem Cell Biology and Cancer
In order to understand the origin of carcinogenesis, the biology of stem cells requires consideration. Upon fertilization the sperm and ovum cells combine to form a totipotent cell, the zygote, which is generated following the chromosome sorting process of meiosis. During meiosis, crossing-over events occur and 5′-cytosine methylation of certain male or female genes occurs that imprints or marks specific genes. By silencing one allele but expressing the other allele of the zygotic pair, gene imprinting is inherited from one or the other parent. The zygote has a strand of DNA from each contributor and is a gene-unrestricted cell that can make all organs and structures in a timely and orderly set of growth patterns known as development. Being a totipotent embryonic stem cell, which responds to various specific and environmental influences, the zygote can give rise to any and all of the body’s structures. The zygote is the most dedifferentiated or unrestricted cell the body produces. The second most unrestricted cell types are the pluripotent stem cells that give rise in the developing conceptus to each of the three germinal layers: the endoderm, the mesoderm, and the ectoderm. If the dividing zygote implants normally in the uterus in the proper environment, blastula development commences and is maintained within maternal
5.3. CELL BIOLOGY OF CANCER
131
homeodynamic equilibrium. The ordered time and spatial processes of growth and differential development begin under placental protection to form a fetus. However, if on the way to the uterus the conceptus implants within the fallopian tubes or escapes into the abdominal fluid and attaches to an abdominal structure, the conceptus can develop into a teratoma in an ectopic pregnancy (∼2% of all gestations) where it chaotically forms the three germinal layers. A teratoma is a misguided differentiating benign mass because of its foreign, unnatural environment. A teratoma contains many differentiated structural elements such as hair, teeth, organ parts, and so on. The teratoma mass can further develop into a teratocarcinoma. There was a classic study involving the passages of a nonmetastatic teratocarcinoma for eight years in murine ascites fluid in cycles of (1) injection of teratocarcinoma cells into the murine hydroperitoneum, (2) growing the teratocarconoma cell number, (3) isolating the core cells of the mature cancer mass, (4) reinjecting those core cells into new recipient mice ascites, and then (5) repeating the cycle (Mintz and Illmensee 1975). Finally, these cells were then carefully injected into nascent mouse blastocysts in order to monitor the developmental capabilities of the passage cell’s under normal cell field and gestation conditions. At parturition, the examined mosaic X/Y sex (i.e., males) animals possessed both normal mouse markers and the passage cell markers. No abnormalities were observed and moreover the males had functional sperm. The male mosaic mice displayed neither excess tumors in their lifetimes nor was there any regrowth of teratocarcinomas (Illmensee and Mintz 1976). The reversal of carcinogenic characteristics of the ascites passage cells in a healthy host shows that the host could redirect the altered genome by means of normal environmental control factors, with normal cell and LTA contacts, back to the normal state. This is an unequivocal example of the environmental alteration of critical portions of the injected transformed genome from malignancy back to normalcy. On the other hand, Figure 5.4 summarizes the antipodal scenario where progressively altered cell contacts and altered surrounding cell influences can “promote” an aberrant cell to the cancerous state. Somatic stem cells are the master cells of the body’s organs and have the specific ability to grow into one of more than 200 specific tissue types. These tissues are originated and maintained by somatic multipotent stem cells that have a specific range of outcomes in the differentiation processes suitable for each tissue. From immature forms, multipotent stem cells usually develop into only one, or a few, line(s) of cells, to intermediate forms, and finally to the mature and functioning form(s). After their specific cell life of use(s), these cells senesce and then turn over by sloughing or apoptosis. The multipotent cells are more gene-restricted in interphase gene types of activity than their predecessor pluripotent cells that produce the germinal layers. They are gene-restricted by modified histone blocking and controls. They differentiate into de novo structures in morphospace with increasingly organspecific functionally in the mature forms. Each organ has sequestered and protected stem cells that maintain and reproduce upon need the organ’s particular pattern of gene restriction which confers the unique abilities to each specific tissue lineage. The development of organ stem cells to progenitor cells and then mature organ cells is shown in Figure 5.5). Such is the way of metazoan development; they specialize. Homology with differentiation produces lineage. Stem cells retain the ability to divide throughout human life (naturally immortal) in accordance with the specific
132
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
Stem Cell compartment
A
Symmetrical mitosis Intermediate state
Multiipotent Stem Cell Stem Cell Pool A
A
Limited in number
Asymmetrical mitosis
A
B
B
e f th t o ent u to m en art em omp v Mo m c ste
B
Organ Progenitor Cell Compartment
D i ffe re n t i a t i o n
C
Differentiated cells (mature)
C
Key: A - Stem Cells of Organ B - Progenitor Cells C - Fully Differentiated Cells
Figure 5.5. A stem cell compartment. Cartoon characterization of the compartment wherein the organ stem cell(s) contribute to progenitor cell pool and then to mature organ cells. There are relatively few stem cells in an organ, perhaps only 0.001%. Stem cells contribute to the developing progenitor cells in the progenitor portion of the compartment. The progenitor cells expand and begin to differentiate. Finally, progenitor cells mature into the different types of functional mature cells of the organ. Cancer stem cells can occur directly from the stem cell compartment but may also originate later from the progenitor compartment.
turnover of the tissue in which they reside. For a specific example of differentiation, the primitive bone marrow stem cells (hemocytoblasts or HSCs) give rise to the other types of blood and white cell types that are descended via intermediate developmental cells types (Figure 5.6). Erythrocytes [(i.e., red blood cells (RBCs)] are regenerated by the hormone erythropoietin (EPO) acting on HSCs → proerythroblasts → polychromatic erythroblasts → RBCs (Figure 5.7). In another route of pleiotropy, the primitive marrow cells (HSCs) → myeloblasts (the latter are excessive in leukemias and chronic inflammatory diseases) → progranulocyte, which then can form (with the appropriate stimulations) granulocytes, that is, basophils, eosinophils, or neutrophils (Figure 5.7). All of these myeloid cell types appear in the inflammation process which was discussed previously. In another major developmental pathway the HSCs can also make lymphoid cell stem cells = lymphoblasts (LSCs) → lymphocytes, or in yet another pathway HSCs → monoblasts → monocytes → agranulocytes, or another where HSCs →
5.3. CELL BIOLOGY OF CANCER
133
Erythrocytes BFU-Ederived colony (erythroid)
Stem Cell Go
Platelets Neutrophils
CFU-GMderived colony (myeloid)
Monocytes CFU-Megderived colony (megakaryocyte)
Eosinophils
Basophils
Stem cell
Progenitor cells
Precursor cells
Mature cells
Figure 5.6. Hematopoiesis stem cell (HSC) differentiation(s) from the bone marrow hemocytoblast(s) into the red blood cell (RBC) and white cell lineages. The hematopoietic tissue compartment contains cells with long-term and short-term regeneration capacities with committed multipotent, oligopotent, and unipotent progenitor cells. Transplantation experiments point toward a limited clonal diversity of hematopoietic cells. That is, the HSC compartment consists of a fixed number of different types of HSCs. Each type possesses different epigenetically preprogrammed behavior for the different cell lineages. This model contrasts with older models of HSC behavior that postulated a single type of HSC that can be continuously molded into all of the different subtypes of HSCs. HSCs constitute about 0.01% of cells in the bone marrow compartment. [Figure done by Virginia C. Broudy, M.D. for AHA Teaching Cases for the Society of Hematology and was provided with permission by D. Preciado of the Society ([email protected]).] See insert for color representation of this figure.
megakaryoblast → megakaryocyte → thrombocytes (Figure 5.7). Clearly, one pluripotent cell type (the HSC) in the bone marrow is aided by various specific differentiating key factors, hormones, and microenviromental field effects to elaborate many useful different cell forms. Bone marrow stem-cell transfusions (or transplants) can be administered to replace various types of blood cells in severe aplastic anemia. It is notable that during severe anemia, extramedullary hematopoiesis can occur in organs outside the bone marrow: spleen, liver, and lymph nodes because these are embryonic sites of hemoglobin synthesis. They can be reactivated in severe anemia. Even more interesting is the fact that de novo synthesized hemoglobin in severe anemia contains fetal and perinatal δ- and γ-chains in place of some of the adult β-chains that combine with the α-chain to make various hemoglobin types. This is a prime example of systems redundancy and fetal-form recapitulation of δ- and γ-chains in tumorigenesis that confer plasticity in gene expression to the system. Fetal hemoglobin (αγαγ) carries O2 more efficiently and releases CO2 better than mature hemoglobin (αβαβ) and thus aids the parasitic relationship of mother to her fetus. Because tumors are usually anoxic, a hemoglobin switch to fetal hemoglobin is an advantage to neoplasms. Clearly by adding recapitulation, it creates
134
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
Hematopoiesis Pluripotent Stem Cell Pluripotent Stem Cell multiplication B Cell
Lymphoid Stem Cell
T Cell
Basophilic erythroblast
Myeloblast/ monoblast
Eosinophilic myelocyte Eosinophil Basophilic myelocyte
Orthochromatic erythroblast
Monocyte/ granulocyte progenitor Reticulocyte
Basophil
Megakaryocyte
Neutrophil Erythrocyte (Red Blood Cells)
Monocyte Thrombocytes (Platelets)
Figure 5.7. Hematopoiesis differentiation of HSCs to specialized cells. The differentiation from HSCs (hematopoiesis stem cells) is a complex process which is shown here. The relationships of the intermediate cells to each of the final mature forms are shown from top to bottom in this figure. The development is completed with the generation and specific spatial placement of the various specialized mature functional cell forms. [This image was obtained from HEAL (Health Education Assets Library) whose goal to provide free digital resources for health education (http://www.healcentral.org).] See insert for color representation of this figure.
robustness and enhanced survival in evolution. With the same genotype, organisms have the capacity to vary in developmental pattern, in phenotype, or in behavior according to varying environmental conditions. The stem cell compartment as depicted in Figure 5.5 contains a generative cell of a particular compartment or niche. Stem cell model examples are: (1) hematopoiesis in the blood compartment and white blood cell (WBC) production as shown in Figures 5.6 and 5.7, (2) the epidermal generation of basal cells to keratinocytes in the skin, or (3) the crypts basal stem cell generation to intestinal mucosal cells. Note in some differentiation processes, erythrocytes (RBCs) or keratinocytes represent terminal cell forms because the mature cells are anucleated in the final stages of maturation. This makes RBCs unable to recreate HSCs. The integument sloughs skin cells, following anucleation and keratinization, with a half-life for keratinocytes of about 7 days whereas RBC half-life is about 7–8 days (depending on anatomical
5.3. CELL BIOLOGY OF CANCER
135
area of skin) and turnover of RBCs in the spleen is 120 days. In high-turnover compartments, like the crypts in the gut, the terminal forms can only exit the compartment by senescence and sloughing from the crypt into the lumen. These stem cells accordingly respond in order to replenish lost cells and maintain the homeodynamic equilibrium level or normal number of cells per organ or compartment, where N is the preset number for that compartment. The N value is an allometric function of body weight, surface area, and metabolism. The stem cells retain “N-size information” for maintainance. Mitosis and apopotosis rate balance is commensurate with those cells lost due to normal damage and senescence for that particular tissue or organ. All stem cells have a relatively slow asymmetric mitosis rate of stem cell replacement, and it is the mitosis rate of the progenitor cells that mostly adjust for N-size maintenance compensations in normal tissue compartments (Figure 5.5). One obligatory property of a stem cell is that it must renew itself (regenesis) in a controlled manner when depleted in order to maintain the proper size of the organ (Wicha et al. 2006; Figure 5.5). This may happen in a stimulate-to-grow-spurt wherein stem cells replenish themselves and is called a symmetrical mitotic division; that is, one stem cell (SC) becomes two stem cells (Figure 5.5). The resulting cells are identical. Another basic property of a stem cell is that it can undergo a special mitosis whereby the dividing stem cell produces one new cell that is destined to replace the original stem cell plus another cell, a progenitor cell, which receives a signal set to “turn off stem-ness,” and thus loses its immortal stem cell character, while contemporaneously turning “on” the genome, epigenome, and membrane receptors to differentiate, usually in identifiable histological cell stages, toward the tissue’s mature cells (Lahad et al. 2005; Lobo et al. 2007; Shipitsin and Polyak 2008). This is called an asymmetrical mitotic division where one stem cell becomes one resident stem cell plus a progenitor cell (PC), namely, 1 SC → 1 SC + 1PC (Figure 5.5). Critical properties of stem cells and their compartments are: (1) selfrenewal of stem cells, ability of stem cells to differentiate to mature specialized cell types, active telomerase expression, (2) maintenance of anti-apoptotic pathways, (3) increased membrane transporter activity, and (4) conferred migration ability of the PC. When a cancer stem cell (CSC) is generated within the stem compartment telomerase activation has been observed to occur with immortalization.The CSC leaves the stem cell compartment with its stem-ness intact, whereas it normally looses stem-ness at this point. The CSC can then potentially be promoted to cancer depending on its environment (Sun et al. 1999; Chang et al. 2001). (See ahead to the cancer growth example presented in Table 5.1). Stem cells in the breast, liver, pancreas, kidney, mesenchyme, and intestine have been shown to be susceptible targets to form CSCs (Tai et al. 2005; Wicha et al. 2006). The presence of the breast cancer 1 gene (BRCA1) causes significant increased lifetime breast and ovarian cancer incidence. A majority of the breast cancers are of the “basal-like phenotype” that is characterized by a lack of expression of estrogen receptor (ER) and progesterone receptor. The role of BRCA1 in human mammary stem cell fate was investigated where the cancer phenotype appears to resemble that of breast stem cells, and not mature breast cells (Liu et al. 2008b). It has been shown that expression of the BRCA1 gene is required for differentiation of ER-negative
136
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
stem/progenitor cells to ER-positive duct luminal cells. This is based on inactivating the BRCA1 gene in primary breast epithelial cells, leading to (1) an increase in cells displaying the stem/progenitor cell marker, aldehyde dehydrogenase isozyme (ALDH1) and (2) a decrease in cells expressing luminal epithelial markers or the ER. Unlike normal controls, breast tissues having female germ-line BRCA1 mutations contain entire lobules that appear histologically normal even though they are positive for ALDH1 expression and are ER-negative. This suggests carcinogenicity starts significantly before anatomical or clinical manifestation as characterized in Figure 5.4. Loss of heterozygosity (LOH) for BRCA1 was determined in these ALDH1-positive lobules, but LOH did not occur in adjacent ALDH1-negative (normal) lobules. Altogether, these studies show that the normal BRCA1 gene plays a critical role in the differentiation of ER-negative stem/progenitor cells to ERpositive luminal cells. Because the BRCA1 gene product also plays a role in DNA repair, the loss or mutation of BRCA1 can result in the accumulation of genetically unstable breast stem cells, providing prime targets for further carcinogenic events (Liu et al. 2008b). Radiobiological studies suggest a hierarchical stem cell compartment (i.e., actual and potential stem cells). Actual stem cells are well-protected and have intolerance to genotoxic damage and to chemotherapeutics, thus exhibiting chemoresistence and stability. Stem cells are usually long-lived and finally die via scheduled apoptosis. When asymmetrical mitosis occurs, there is a sorting of the old DNA strand and the new DNA strand upon division. The new strand is checked for DNA errors and is corrected by DNA repair systems, and the corrected strand stays with the remaining resident stem cell in the stem cell compartment. Hence, the everrenewing DNA strand in the resident stem cell is normally relatively error-free as one would expect of the stewardship of a long-lived, generative cell (Bach et al. 2000). The old DNA strand leaves the stem compartment with the emergent progenitor cell. The Paneth cells of the intestinal crypts sit juxtaposed to the stem cells and act like neutrophils. The Paneth cells protect the crypt stem cells from microbial attack. High genotoxic insensitivity, selective DNA strand sorting, local protective barriers, anatomical placement, reserve stem cells, and helper or protective cells all provide highly effective protective strategies against both unscheduled replication and random exogenous errors in the small intestine stem cell. This provides a reasonable explanation for their low incidence of discovered small intestine cancers (0.39%). Small intestinal cancer mortality is 0.21% of all fatal cancer types, whereas colon and rectum mortality is 10.4%. Given the small intestine undergoes 1000 mitotic divisions in a mouse lifetime, it stands to reason that well-controlled apoptosis accounts for considerable cell loss to the feces in order to maintain a balance of about N = 250 cells per normal crypt (Bach et al. 2000). Were it not for abrasion losses and apoptosis, one crypt stem cell could theoretically execute replenishment in 8 cycles, that is, 28 = 256 cells. However, steady-state modeling of the crypt indicates that there are 4–6 stem cells in steady state that compensate for luminal losses in the specific homeodynamic equilibrium of the small intestine. The kinetic conclusion is that stem cells undergo 6 generations of dividing through transit and mature crypt cells counterbalanced by luminal sloughing (Loeffler et al. 1993). Some of these stem cells are in reserve but exactly
5.3. CELL BIOLOGY OF CANCER
137
how many is unknown, although, if the need arises, all stem cells can be activated for quick synthesis. These considerations present a dynamic situation in a highturnover organ that well protects the working stem cells in the execution of an organ’s normal physiology. Human colonic disease occurs at higher mortality incidence (10.4%) than does small intestinal disease (0.21%). Colonic cancer appears to originate from a series of genetic mutations in a few colonic stem cells occurring over an extended period of time (Kinzler and Vogelstein 1996, 1998). These authors proposed that genetic mutations occur in a niche environment in the organ in three essential functional steps of cell maintenance: (1) gatekeeper functions—sensing and regulation of cell growth in G0 and G1 of the cell cycle; (2) caretaker functions—DNA mutation sensing, repair, and verification; (3) landscaper functions—extracellular matrix, stromal, and mesenchymal cell interactions. Thus, caretakers protect the genome against mutations, while gatekeepers induce cell death or cell cycle arrest of potentially tumorigenic cells. Well-integrated cells are maintained by the landscaper functions. The clonal composition of human colorectal tumors was studied by means of restriction fragment length polymorphisms (RFLPs) (Fearon et al. 1987). X-linked RFLPs were used to examine the pattern of X chromosome inactivation in colorectal tumors of females, and all tumors examined showed a pattern of monoclonal X-chromosome inactivation. This suggests one or a few competing clonal cells originated the 20 carcinomas + 30 adenomas = 50 tumors examined. Also, RFLPs of autosomes were used as clonal markers to detect the somatic loss or gain of specific chromosomal sequences in colorectal tumors, and it was found that somatic loss found in chromosome 17p sequences occurred in over 75% of the carcinomas, but such a loss was rare in adenomas. This disparity suggests that the 17p loss in the female colon might be associated with the malignant state. Allelic deletions also involved chromosome 18q occurring in more than 70% of colorectal cancers but not in normal colon tissue. Deletions in 18q included the DCC gene, which involves maintenance of cell-to-cell adhesion molecules that provide essential intercellular connections (Fearon et al. 1990; Martin et al. 2006). The concept that adult tissues contain embryonic remnants that can lie dormant, that can later be activated to become involved with cancer, was first formalized by Cohnheim in the late 1800s as the “embryological rest theory” (Sell 2004). Clonal selection was introduced by Fidler and Kripke (1977) based on their observations of many developing-cancer specimens. This powerful model provides insight into the progression of carcinogenesis. The clonal selection model of carcinogenesis asserts that at least one tumor population, among a number of competing altered clones over time, expresses a group of deregulated control genes in a subset of cells unable to prevent progression to metastasis. These transformed clones compete with each other locally, and at the margins they also compete with normal cells which are in excess. As these clones compete for resources in a Darwinian fashion, a heterogeneous phenotypic and genotypic set of clones arise, some of which might derive a mutator phenotype and become more chromosome-destabilized or more aggressive in growth and malignancy (Loeb et al. 2003, 2008). The related concepts of cancer latency, dormancy, and delayed reoccurrence have received recent attention (Aguirre-Ghiso 2007; Luzzi et al. 1998).
138
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
The parallel evolution model of carcinogenesis posits that metastatic events occur early in tumor progression and co-evolves with the primary tumors. The dynamic evolution model posits that the frequency of metastatic variants may become unstable perhaps due to increased telomerase activity (which also occurs in old age) (Finkel et al. 2007; Rangarajan et al. 2004). This type of chromosomal instability results in a dynamic vacillating equilibrium between cell generation and cell loss within the cancer or metastatic phenotype. This model describes stem-cell mutation followed by formation of transformed stem cells that propagate to CSCs. Later, in the same premutated stem cells or nascent progenitor cells, these cells sustain further alterations from extrinsic or intrinsic sources in the progression of carcinogenesis. Multiple combinations of carcinogenic foci occur in the altered LTA. This derives a poorly cooperating community of clones in various states of differential development. Some young clones tend to be monoclinic, some middle-aged, and some compromised cells that lead to coexistence of various clone sets, and some clones dynamically differentiate and senesce into old, bizarre, and chaotic clone forms. The clonal dominance model posits that once it is formed, a metastatic subclone will outcompete and dominate the tumor mass, resulting in similarities between the primary tumor and the metastatic foci. The stem cell model of carcinogenesis posits that tumors arise from rare organ stem cells to directly form cancer stem cells. By resisting apoptosis, CSCs are immortal occurring outside the stem cell compartment where the normal surrounding tissue cells are finite in lifetime. CSCs outlast the attrition of the organ progenitor and mature cells and thus increase by accretion. All these models have supportive data, but the model that operates depends on the carcinogenic agent, the concentration, and exposure duration (i.e., Haber ’s Law), at what age the I* stage began, the degree of P-stage pressure, health status, nutritional status, heredity, and more environmental factors. One might argue that all these cancer progression models are various subsets of Fiedler ’s clonal selection model. The clonal selection model as well as the other models (above) must be considered within the context of the life history of transformed foci to tumorigenesis to cancer to metastasis and not just the events following clinical diagnosis which is often the approach in oncology. The life history and kinetics from CSCs to lethal cancer mass may be considered in the following example in Table 5.1. The doubling times of certain primary breast tumors have been measured. Some breast clones grow as fast as a doubling time of 44 days, and some grow as slow as a doubling time of 1869 days, with an average DT of 212 days (von Fournier et al. 1980). For example, in Table 5.1, if one were to consider a rapidly growing tumor whose net doubling time is assumed to be 130 days and accounting for apoptosis, then exponential tumor mass growth is estimated in Table 5.1. The threshold clinical value for a typically discovered tumor by a physician clinically is about 1 cm in diameter, which has about 109 cells in the tumor. Table 5.1 shows that this tumor is welladvanced in oncogenicity at 30 mitotic divisions (or 3900 days) to reach such a mass. There is enough Darwinian clonal selection in 30 divisions, or 10.7 years in this example where doubling time is 130 days, to acquire a mutator phenotype and/ or enhanced malignancy. A tumor mass usually reaches a lethal limit on average at about 1012 cells at 40 divisions, which, depending on the tumor type of the mass, is about 500–575 g. This is only 10 more divisions than when tumor was first detected
5.3. CELL BIOLOGY OF CANCER
TABLE 5.1.
139
Cancer Growth
Mitotic Divisions, N Number of Tumor Cells, 2N Days: Years:
0
1
2
4
10
20
30
40
1
2
4
16
1.024 × 103
1.048 × 106
1.73 × 109a
1.10 × 1012b
0 0
130 0.36
260 0.71
520 1.42
1300 3.56
2600 7,1
3900 10.7
5200 14.2
For this example, doubling time is assumed to be 130 days. Typically, a tumor mass is not detected by medical examination until about 109 cells or when the mass is about 1 cm in diameter. This mass takes about 30 divisions.
Only 3.5 years left to eradicate tumor
a
Size about 1 cm3.
b
Size about 1000 cm3.
Source: Data adapted from Talmadge (2007).
at 30 divisions. The tumor is usually detected late in the life history of the tumor progression. Often this is too late. This makes it imperative to practice prophylaxis in chemical carcinogenesis whenever possible in toxicological risk assessment and management because it is often too late to react when cancer is discovered (as seen in Table 5.1). It is best to prevent cancer progression before it occurs. The necessary early detection will include a better understanding of the expression patterns of specific growth control gene sets employed in the biology of stem cells: the Polycomb gene group, sonic hedgehog (Shh), Wnt/β-catenin, or Notch gene groups. These controls are discussed in Section 5.3.7. The application of “omics” in toxicology is advancing the understanding of which expression patterns are of relevance for assisting with early prediction and accurate hazard assessments of stem-cell and multistaged cancer progressions. For further discussion on omics and toxicology, see Chapters 22 and 23.
5.3.7. Specific Biological Growth and Growth Control Gene Sets and Their Pathways Specific crucial growth and growth control gene sets and their pathways are employed in the directed biology of stem, developing, and mature cells. Some of the main hierarchal gene sets are the Polycomb group, sonic hedgehog (Shh), Wnt/β-catenin, Notch, or Oct-4 gene groups. Many of these pathways are originally used in an organism’s organogenesis portion of development but naturally attenuate in the adult stage of life, except for the maintenance of the long-lived, immortal stem cells. A genetic or epigenetic (see Section 5.3.8 for a description of epigenetic changes) change in these hierarchical control systems in the conceptus, embryo, or fetus can lead to terata or newborns with a variety of critical syndromes including cancer susceptibility. If critical changes in these systems occur after organogenesis
140
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
is completed, then organ(s) stem/progenitor cells can be adversely affected so as to lead to improper cellular development in a local tissue array (LTA) which includes tumor initiation and promotion. During development, the Polycomb group (PcG) works by actively restricting numerous specific genes in chromosomes (Lobo et al. 2007). These include (a) the homeobox genes, which maintain genetic imprinting, and (b) body patterning genes in developing metazoans. Furthermore, members of the PcG are now known to regulate epigenetic cellular memory, pluripotency, and stem cell self-renewal. Genes that belong to the PcG are epigenetic gene silencers (restriction) with a vital role in the maintenance of cell identity by what sequence PcG does not restrict. One PcG repressor is the protein Bmi-1, a transcriptional repressor; and when in excess, Bmi-1 is involved with normal leukemic stem cells (LSC) and maintaining B-cell lymphomas. Bmi-1 repression is important in the regulation of the primitive marrow stem cell (HSC) (Figures 5.6 and 5.7). Bmi-1 has been found also to regulate neuronal stem cells and mammary stem cells when in excess (Liu et al. 2005). The Sonic hedgehog (Shh) signaling pathway develops directional or spatial and timely differential growth in the embryo by exporting Shh-N extracellularly. Shh is synthesized as a 45 kDa nascent protein that is cleaved by a specifc Shh-Nencoded endonuclease into two strands: (1) a 25-kDa amino-end signaling portion with its carboxyl end group linked to a cholesterol molecule, which aids in trafficking out of the endoplasmic reticulum and through the plasma membrane, and (2) a 20-kDa N-terminal signaling domain linked to a palmatate group (that aids in specific transport). This completes the Shh-N morphogen signal. Shh-N is then exported extracellularly and binds to target cell membranes that contain a complementary “patch receptor” for it and principally produces signal gradients of hedgehog proteins in a field of cells which provides instruction for growth directions and patterns. By building specialized tissues, Shh plays a role in directing organogenesis including diverse functions of growth and shaping such as the digits and limbs while also developing the mid-line structure of the brain, along with the spinal cord, by providing guidance for axonal growth. Because of these characteristics, Shh is called a morphogen. Shh controls cell division of adult stem cells. Shh has been implicated in skin and brain cancers, and it has been found to participate in gastric cancer and rapid killing pancreatic cancer progressions (Katoh and Katoh 2005; Li et al. 2007). The canonical Wnt pathway regulates cell fate development, determination, and maintenance in many tissues (Liu et al. 2005). Wnt binds to the frizzled protein and then binds to the disheveled plasma-membrane inserted proteins and transmembrane liporeceptor proteins. The membrane complex also includes interior binding of the cytoskeleton protein actin while also being linked to a G-protein that can be phosphorylated by GTP. This large receptor complex protects against β-catenin degradation. β-catenin is a protein that has armadillo sequence repeats in the secondary structure, which makes it highly suitable for protein–protein bindings. β-catenin maintains a stable complex with the tumor suppressor adenomatous polyposis coli (APC) protein product and the AXIN1 gene protein product. AXIN1 protein is a negative feedback regulator of the Wnt pathway. Wnt signaling initiates the release of complexed β-catenin intracellularly while preventing its degradation. β-catenin thus builds up in cytoplasmic concentration and makes its way into the nucleus
5.3. CELL BIOLOGY OF CANCER
141
where it binds transcription factors such as T-cell factor (TCF) and causes TCF to be exported from the nucleus. Several studies show that Wnt signaling has direct influence on HSC (Figure 5.7), epidermal basal stem cells, and intestinal crypt stem cells as reviewed by Liu et al. (2005). Concerning intestinal stem cells, sporadic cases of colorectal cancer are primarily initiated by gene mutations in members of the canonical Wnt pathway. This leads to β-catenin stabilization and its excessive accumulation in adult tissue. Cells displaying nuclear β-catenin accumulation are nonrandomly distributed throughout the growing tumor mass. β-catenin appears to be localized along the invasive, migratory front where organ parenchymal cells are in direct contact with the stromal microenvironment (Le et al. 2008). There is a stromal cell role in regulating β-catenin intracellular accumulation in a paracrine fashion which extends β-catenin involvement not only within the cell producing it but also among the local cells of a different type in the LTA. As such, the local tumor microenvironment has the potential to reinforce the CSC phenotype in a subset of cells, thus mediating invasion and metastasis. β-catenin is known to connect between 60 and 70 essential biochemical basic control reactions [cf. Table 1 in Le et al. (2008)]. Therefore, β-catenin likely represents a central modus operandi where different signals converge and are subsequently coordinated to regulate tissue homeostasis under physiological conditions and cancer stem-ness in relation to the interacting stromal cells. Because many of the β-catenin partners are themselves regulated by extracellular stimuli, it follows that the subsequent effects on β-catenin activation and cancer stem cell production are modulated in a context-sensitive manner—that is, proportional to P-stage influences. The Notch signaling pathway is a highly conserved cell signaling set of genes present in most multicellular organisms. Notch possesses four different notch isoreceptors forms, Notch1–Notch4, which make a single pass through the plasma membrane with a large extracellular portion of the receptor. This portion associates in a Ca+2-dependent, noncovalent interaction with a smaller piece of another Notch protein on an adjacent cell composed of a short extracellular region, a single transmembrane pass, and a small intracellular region. The Notch protein sits like a trigger spanning the cell membrane, part of it inside and part outside. Ligand protein binding to the extracellular domain induces proteolytic cleavage and release of the intracellular domain; this part enters the cell nucleus and alters expression of gene sets. The Notch signaling pathway is important for cell-to-cell docking and juxtracrine communication and involves gene regulation mechanisms that control multiple cell differentiation processes during embryonic and adult life. Notch has been implicated to be a key event in mammary carcinogenesis (Liu et al. 2005). Octamer-binding transcription factor-4 from the Oct-4 gene is a member of the POU transcription factors. It is initially present as a maternal factor in the normal oocyte and is continually expressed by the embryo throughout the preimplantation period (Kehler et al. 2004). Oct-4 is also expressed in germ cell precursors of adult mice. Oct-4 expression correlates with an undifferentiated phenotype both in the embryo and in cell lines derived from it. Previous studies have shown that Oct-4 has an essential role in maintaining the pluripotency of cells of the inner cell mass of the embryo. Loss of Oct-4 function leads to apoptosis of primordial germ cells (PGCs) rather than to differentiation into a trophectodermal lineage as has been
142
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
described for Oct-4-deficient ICM cells (Kehler et al., 2004). These results suggest a function of Oct-4 is to maintain viability of mammalian embryo and germline cells. Not only has the Oct-4 gene been noted to be expressed in embryonic stem cells but also in tumor cells. Upon differentiation, normal cells lose Oct-4 expression. Using Oct-4 antibodies and PCR primers, Oct-4 expression was tested in normal tissues: human breast, liver, pancreas, kidney, mesenchyme, and gastric stem cells. Also tested were HeLa and MCF-7 cancer cell lines and human, dog, and rat tumors (Tai et al. 2005). The results collectively indicate that Oct-4 expression occurs in adult human stem cells, immortalized nontumorigenic cells, tumor cells, and cancer cell lines, but not in differentiated mature cells (Tai et al. 2005). Oct-4 is expressed in only a few cells found in the basal layer of human skin epidermis but not in maturing keratinocytes: presumably these few cells are epidermal stem cells. Oct-4-positive cells apparently have the property of immortality in common with CSCs based on data demonstrating that adult stem cells maintain their expression of Oct-4, as do CSCs, which is consistent with the stem cell hypothesis of carcinogenesis discussed above (Trosko 1988; Trosko et al. 2004; Trosko and Chang 1989). A biomarker strategy for targeting and precisely measuring Oct-4 could present a method to detect and follow CSCs and could allow early-stage hazard evaluation of carcinogenesis. When a developmental gene or a gene set (i.e., Polycomb group, Shh, Wnt and β-catenin, or Notch gene POU groups) is/are adversely affected by mutational and/ or epigenetic perturbations, these events can lead to cancer because of their central controlling roles in normal growth. The basic principles of correct gene restriction and adult gene-controlled expression are breeched in the I* and P stages of carcinogenesis. Often the characteristics of embryonic-like growth reappear in some carcinogenesis, which is often reffered to as anaplasia (Ben-Porath et al. 2008; Katoh and Katoh 2005). In summary, carcinogenesis usually includes the biological formation of altered foci, focal selection, retarding apoptosis, altered differentiation enhancing cell mitosis, enhanced cell motility and migration, and acquiring immortality and some rare cases metastasis (Hanahan and Weinberg 2000; Trosko et al. 2004; Figure 5.4).
5.3.8.
Epigenetic Biology and Nuclear Traffic
Epigenetic changes are cybernetic biochemical alterations that do not alter the primary DNA sequence as genetic mutations do but nonetheless can alter gene function. For a review on DNA alterations in genetic mutations, refer to Meuth (1990). Epigenetic events include reversible alterations in various nuclear elements that control or modulate DNA 2° structures and chromatin 3° and 4° 3-D structures. Mitochondrial DNA also is affected by epigenetic alterations. Some epigenetic events activate timely and selective gene expression, but some events cause specific gene nonexpression (called restriction), thereby suppressing the gene, and yet others involve the metabolism and utilization of “controlling RNAs” other than mRNAs [e.g., interfering RNAs (RNAi) and long noncoding RNAs (lncRNA)] (D’Alessio and Szyf 2006; Ponting et al. 2009; Schones and Zhao 2008). In chromatin, double-stranded DNA is wrapped left-handed around a protein-based core structure with the DNA having 145–147 bases per nucleosome unit of chromatin or 1²⁄³ turns of the helix (curiously
5.3. CELL BIOLOGY OF CANCER
143
close to the golden ratio φ = 1.6180339...). These nucleosomes are the repeating folding units of chromatin and are the pleiotropic structural building blocks of the packed nuclear material in higher organisms. Histones are the principal chromatin proteins of the nucleus. They act as spools around which most DNA winds. Histones play a role in gene regulation because of their organizing properties conferred to the DNA genes and other DNA and RNA elements which indirectly control the genes and other aspects of nuclear physiology. Without structural histones, the unwound DNA in chromosomes would be unstructured and very long. For example, each human cell has about 1.8 m DNA length, but wound on the histones is about 90 mm of chromatin, which, when duplicated and condensed during mitosis, result in about 120 μm of chromosomes. H1 is a specific histone protein in chromatin that acts as a linker within and among DNA–histone chromosome structures. H1 histone promotes higher folding and packing into condensed heterochromatin, which is opaque in light microcopy and can be viewed distinctly during metaphase whereas euchromatin (unfolded) cannot be similarly viewed. The nucleosome usually contains an octomer set of short globular histone proteins which are all basic proteins due to their high contents of lysine [K] and arginine [R] amino acids. Nucleosomes are usually comprised of the specific histone types: H2A, H2B, H3, and H4. One common combination is an H3–H4 tetramer with a dimer of H2A–H2B to form an octomer to complete the nucleosome unit. There are many other known combinations. All nucleosomes have extended histone short protein chains that jut and flex outward from the nucleosome globular units and can be combinatorially altered by certain chemical modifiers of histones: methyl-, acetyl- (usually binding histone lysines), UBI- (ubiquitin), SUMO- (small ubiquitin-like modifier), PO4- (phosphorylation), and poly (ADP-ribose) (Biel et al. 2005). In organizing the cell’s approximately 109 nucleotides, the various formations of histones H2A, H2B, H3, and H4 occur within each of the cell’s nucleosomes as paired composites and are combined with various combinations of the above chemical modifications, which conservatively produces >108 potential folded-sequence specific 3-D variations. This calculation does not include the many combinations also contributed by DNA 2° structures, such as local hairpin loops, and long-range DNA–DNA associations even across chromosomes (Clark 2007). The enormous number of total chromatin potential structures is of great interest because such immense structural diversity offers significantly more information than the 1° (primary sequence) of DNA, which heretofore has been the sole rationale for the genetic code and hence for gene expression. Many of these combinations may give rise to explaining the “if,” “when,” “why,” “how long,” and the timely termination of gene expression “not” (Biel et al. 2005; Wang et al. 2008). Not all chromatin structural combinations will necessarily be useful chemically, but the functional structures that do possess utility give rise to The Histone Code (Jenuwein and Allis 2001; Wynter 2006). Chromatin functional arrangements, selected from such a large number of combinatorial structures, determine the cell’s biological cybernetic vocabulary, maintains cellular memory, and gives rise to specialized transcription events in a timely manner required in physiology. Functional chromatin loci are most often associated with specific attractors: the situation-specific and often environmentally controlled specific regulatory transcription factor proteins and their co-factors (NAD+, GTP, etc.).
144
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
Another often-used epigenetic change is DNA methylation modification of cytosine bases at the 5′ ring position with the methyl group donated by Sadenosylmethionine (SAM). 5′-Me-C sequences in DNA structures attract certain specific protein transcription marker factors and co-factors. Many DNA “marks” are maintained throughout the life of a cell and can be passed on to progeny cells via the ovum, which makes the methylation a fundamental part of inheritance (Jablonka and Lamb 2007; Szyf 2007). Methylated cytosines also occur in imprinted genes which are gender-based heritable genes. The methylated cytosines can occur anywhere in DNA where there is a CpG island, but a particular DNA locus affected is at the 5′ terminus of the transcription promoter region of genes (Holliday 1987). Methylated gene 5′-promoter regions usually restrict the gene activity, but these genes can be reactivated with the nuclear enzymes histone demethyltransferases (HDMT) that remove the methyl groups. These HDMT enzymes are highly conserved across the taxa and are critical in maintaining the degree of CpG island methylation on a long-term basis, and this is one way a cell acquires and maintains its identity. Histone acetyltransferases (HATs) and histone deacetylases (HDACs) are susceptible to toxic agents and are targets for drug therapy in a number of diverse diseases (Szyf 2007). Some genes can be turned “on” by cytosine methylations or they can be attenuated or turned “off” in a dynamic process during the cell cycle. Acetylation binding to histone R and K amino acids usually prompts heterochromatin unfolding that spatially allows DNA transcription. Histone deacetylation by HDACs acts oppositely by inducing heterochromatin formation. The organspecific propitious combinatorial binding of methyl and acetyl groups in histones contributes to activated chromatin for the proper DNA transcription (Wang et al. 2008). By chemical direction of specific 3° and 4° chromatin structures, the nucleus efficiently packs about 2 m of DNA (≈109 bases) into a cell nucleus volume whose diameter is about 10 μM. In condensed chromatin (heterochromatin), the nucleosomes are highly folded regions and the nuclear elements in the heterochromatin are turned off because the inaccessibility of the DNA and RNA transcribing machinery. In folded and ordered chromatin, interrelating 3-D functional sites can be created in an activation process (Schones and Zhao 2008). Sometimes the sites are conjoined by long-range chromatin interactions of intrachromasomal complimentary loci and even between chromosomes, all of which are created by physical and chemical associations (Clark 2007). Another fundamental set of changes is the histone acetylation at lysine (K) loci by HAT, which tends to form euchromatin; and the acetyl group can be removed by HDAC, which tends to form heterochromatin. These dynamic states are in functional equilibrium, depending on need. Only a small proportion, about 5–10% of the whole human genome, is stabily transcribed into RNA. Of this transcribed RNA, only about 1% of the genome codes for proteins whereas the remainder (∼9%) is high-molecular-weight RNA types that likely participate in cellular control functions (Ponting et al. 2009). It’s been experimentally observed that transcription occurs in euchromatin regions of chromatin where the chromatin is open for transcription and not condensed (Ting et al. 2006). Euchromatin fosters open reading frames for the (1) chromosomal active genes
5.3. CELL BIOLOGY OF CANCER
145
(mRNAs) and (2) control RNA regions—that is, micro-RNAs (miRNA), interfering RNA (RNAi), Piwi interacting RNA (piRNA), and long nonprotein coding RNA (lncRNA or ncRNA) (Amaral et al. 2008; Ponting et al. 2009). An operative notion of cell biology is arising: Selective production of RNAs is the cell’s main cybernetic set of tasks. It has been long recognized since the 1960s that nuclear RNA is initially synthesized, from the DNA template, as long RNA pieces of about 6500 nucleotides known as heterogeneous nuclear RNA (hnRNA). About 10% of the hnRNA is processed to about the size needed for an average protein coding ≈ 1500 nucleotides. Because the remaining RNA quickly associates with nuclear proteins (and possibly DNA), it may be assumed that some of the non-mRNA DNA sequences function to produce control RNAs and not just “noise” RNA (Ponting et al. 2009). The reading and nonreading chromatin regions move about during the cell cycle presumably according to need. This is called position effect variegation (PEV) (Biel et al. 2005). Genes that encode products which promote transcription are enhancers of PEV and are called e(var), of which HATs are representatives, a large enzyme family from yeasts to humans. The genes that suppress PEV (and thus transcription) are collectively called su(var), of which the histone linker H1 is one example and the highly conserved HDAC gene is another (Moss and Wallrath 2007). An objective case can be made for the existence of biological nongenetic cancer mechanisms. A list of 54 chemicals based on positive carcinogenesis, but no genetic positive findings, makes this case that 54 bioassayed carcinogens are not knowingly mutagenic in their cancer causality (Tennant 1993). It remains that there could be genetic causes that are not measured by current genotoxicity protocols. Some tests, like DNA recombination, have been recorded for some nongenetic carcinogens (Schiestl 1993). Clearly, better biomarkers for mutagenic and nonmutagenic carcinogens are needed to assess the hazard of potential carcinogens. The known nongenotoxic carcinogen list is large, and it is worth noting that some of the most potent carcinogens do not demonstrate mutagenicity. It can be surmised from this observation that some chemicals cause chromatin disturbances, not by classical genetic means but rather by disruption or disorganization of the epigenetic code (Feinberg et al. 2006; Jones and Baylin 2007). Another objective case for epigenetic mechanisms can be found in a carcinogenesis study, which evaluated Swedish, Danish, and Finnish identical twins (Lichtenstein et al. 2000). This epidemiological study made observations of cancer incidences at 11 major organ sites. It showed that identical twins that are separated at birth have vastly different cancer site incidences at these 11 sites that were consistent with their respective rearing environments during development and adulthood. Control twins who remained together in rearing were similar in site and frequency. Because identical twins have identical DNA sequences, gene expressions and/or utilization of DNA in resisting cancer must be different in the separated twins. That is, their respective gene expressions relate to their particular environmental factors that influence their gene expressions. Moreover, it’s a common observation that siblings of the same set of paternal and maternal genes set can be more different phenotypically than random crossing-over events might predict. Environmental conditions in utero and externally for the mother vary from pregnancy to pregnancy; it is widely suspected that epigenetic control of plasticity and reversibility has a role in this sibling variance. Long-term phenotypic traits
146
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
inherited in families or within tribes can become altered or even lost upon social migration to a different environment where new traits are acquired that better accommodate the exigencies and vagaries of the new environment (Maresca and Schwartz 2006). It is becoming clear that biologic controls used in the cell’s cybernetics are more than the Neurospora-based hypothesis of “one gene produces one mRNA, which in turn produces one protein” (Beadle and Tatum 1941; Holliday 2006; Jablonka and Lamb 2007; Morange 2002). The central dogma was advanced by Francis Crick in 1958 when he posited that the sequence information to make a cell (or the whole soma) moves in one direction from the nuclear DNA to RNA to protein, irreversibly. The dogma may need amending in light of genetic findings since 1958 and the many new epigenomic and intra- and extranuclear RNA controls of gene expression (Crick 1958; Hertel 2008; Morange 2008; Ting et al. 2005; Wolters and MacKeigan 2008; Wynter 2006). The original model has indeed served well in the past for experimentation but does not incorporate the reversible and specific adaptability seen in the rapid (sometimes within one generation) construction or restructuring of biological tissues and a means to negotiate with or be responsive to the ever-changing environments in which the organism exists. Darwinian adaptation by classical genetic mechanisms is just too slow to explain the rapid fixed changes that happen within one or two generations. DNA sequence random variability with Darwinian selection of the “fittest” simply cannot produce such rapid and sometimes altruistic adaptations. Nor does the dogma of Crick account for the all the nonvertical transmission of dynamic phenotypic trait changes including ontogeny (Jablonka 2004; Jablonka and Lamb 2007; Szyf 2007). The human genome project found about 20,000–25,000 transcribed DNA sequences as potential human genes (Stein 2004). This ab initio indicated too few genes to carry out all the known phenotypic tasks executed in and among human cells according to the one gene → one protein sequence model. Also, Drosophila melanogaster has about one-half of the amount of coding DNA that humans have. This implies that humans employ DNA more efficiently (or the fly is very inefficient) than flies because of additional structural and developmental complexity, as well as the degree of encephalization that humans possess. This disjunctive observation suggests that humans, albeit suppositionally, make greater combinatorial uses of human epigenetic processes within their episome. This ability is likely true for all higher evolved organisms. Higher organisms utilize an additional mechanism that includes post-transcription RNA splicing (cis-splicing to remove introns and trans-splicing to ligate various mRNAs), thus forming multiple recombinations into various mRNAs from the same DNA sequence (1∼10 mRNAs/gene) (Hertel 2008; Holliday and Murray 1994; Nigro et al. 1991; Shepard and Hertel 2008). The episome of higher evolved organisms is the basis for more information of higher utility being stored in well-ordered but versatile chromatin structures. There is no doubt that DNA contains the blueprints for complete corporal construction and is the cynosure of the cell’s information repository and is the genesis of all cellular construction. However, it is the machinations of the episome that executes much of the control of DNA usage and transactions—that is, (1) which specific genes are expressed in euchromatin (gene activation), (2) when genes are individually expressed but coordinately to achieve
5.3. CELL BIOLOGY OF CANCER
147
a unitary purpose (synchronization), (3) how many times are genes are transcribed and protected from RNases (gene dosage), and (4) when and how genes can be attenuated, or shut “off,” or turned “off” permanently. All these actions are affected by the controls of genes. In summary, the DNA sequence proposes, the episome and its interactive alterations disposes. The aspect of how the episome, especially in the progression of cancer, links with the cell’s internal networks (Hahn and Weinberg 2002) and with the environment or external networks is a subject of systems biology currently in rapid development (Wang et al. 2007) and is covered elsewhere in this volume. Already this eclectic field promises to press the axiom that the whole is merely the sum of its parts, and it appears that interacting complex systems may generate new or varied traits.
5.3.9.
Biological Initiation of Chemical Carcinogenesis
Epigenetic changes are known to be involved in normal development, the evolutionary process, as well as with human diseases, and chemical-induced cellular changes (Holliday 1987, 1989, 2002; Feinberg 2007; Jablonka and Lamb 2002; Oliveira et al. 2007). Cancer development not only depends on the heretofore much published genetic alterations but also involves abnormal cellular memory, which maintains, executes, and passes genomic information to successor cells. As seen in the past, aberrant cell memory can be maintained by mutated DNA, but also altered cell memory can influence the highly articulated nucleosomal structures and their functions in the episome. Both altered memories are active in the carcinogenesis process during transmission of gene expressions, cytoskeletal patterns, and cell-tocell interactions in the neoplastic I stage and P stage (reviewed earlier in this chapter), malignant stage, then metastasis, and finally colonization at a distal site (see Figure 5.1; and also Bachman et al. 2006; Curtin et al. 2006; Feinberg 2005; Feinberg et al. 2006; Ohlsson et al. 2003; Vasiliev 2004; Verma et al. 2004). Recent data suggest that cancer is generated (i.e., “genesis of cancer” or carcinogenesis) by polyclonal disruption of stem cells and/or their progenitor cells that are found experimentally to be epigenically disrupted in specific tumor-progenitor genes in the stem cells of the niche interface to differentiating progenitor and mature cells [refer back to Figure 5.5; also see Feinberg (2004, 2005) and Jablonka (2006)]. Altered development leading to tumor cell heterogeneity has much to do with epigenetic variation among clones (cf. Section 5.3.4). Variation occurs in progenitor cells that devolve with time acting upon a select set of flawed I*-stage cells as well as other cells in the LTA. Plasticity is the normal ability of a tissue or organism to change and adapt, but epigenetic plasticity together with genetic lesions can drive tumor progression (Chan et al. 2008). The reversible early role for nongenetic epigenetic alterations in neoplasia is a necessary precancerous state that is a prelude to later epigenetic alterations that can substitute for, or cooperate with, induced crucial genetic variations in tumor progression (Figure 5.4). Therefore, non-neoplastic but epigenetically disrupted stem/progenitor cells might be a crucial target for cancer risk assessment, chemoprevention, and prophaxis (Feinberg et al. 2006; Marks et al. 2007; Tai et al. 2005; Trosko et al. 2004). Meaningful biomarkers for environmental detection could be derived from these disruptions, too.
148
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
These cell-altering epigenetic mechanisms occur in a number of ways that include the following: (1) In chemical carcinogenesis, causation of inappropriate levels or placements of histone methylations and acetylations allows restricted genes to be uncovered, or requires unrestricted genes to be restricted, which presents or removes in some cases an untimely or inappropriate set of activities; (2) chromatin epigenetic changes are also manifested in global hypomethylation in chemical carcinogenesis; (3) sometimes, however, in the LTA, chromosome sites exhibit hypermethylation usually located in transcription 5′-promoter regions; and (4) abnormal chromatin packaging or folding alterations can participate in the cancer process because structure and function are the yin and yang of cell being (Biel et al. 2005; Holliday 1987). It is increasingly apparent that cancer development depends not only on genetic alterations but also on abnormal cellular activities and abnormal cellular memory that dynamically interacts and fails to normally adapt with the environment (e.g., other like-cells; support cells such as stromal; external matrices; tissue fluids; etc.). These environments can influence horizontal and vertical heritable gene expression patterns critical for neoplastic initiation and progression (Scheel et al. 2007). In various diseases a unifying theme of epigenetics is the occurrence of defects in phenotypic plasticity—that is, an inability of the collective abilities of cells to adequately change their behaviors in response to internal or external environmental cues. A basic question in toxicological cancer prevention can be posed: What process starts this loss of homeostasis and initiates aberrant development so that the LTA, as an affected locus, disengages from its community and corporal controls? Over 80 years ago, Otto Warburg observed that cancer cells metabolized glucose (GLU) more than normal cells (Warburg 1925). He further noted that cancer cells metabolized GLU more by anaerobic glycolysis (his term “fermentation”) than by oxidative respiration. That is, cancer cells tended not to completely oxidize glucose to CO2 and H2O, as is done in the oxidative respiration in the TCA cycle, but rather excreted lactic acid, CH3CH2(OH)COOH (abbreviated LACA), a glycolysis byproduct that is produced from pyruvic acid, CH3CHO–COOH (abbreviated PYRA). This excess LACA is known as the Warburg Glycolytic Effect, shown in Figure 5.8. Normally, blood GLU comes from food sources or from glycogen and is specifically transported into the cell by insulin. GLU is then phosphorylated (GLU-P), which traps GLU in the cell, and then GLU-P is converted (by a series of glycolytic cytoplasmic enzymes) to PYRA, which in turn enters the TCA cycle, where, in the presence of respiratory O2, is absorbed into the mitochondrion, which executes complete PYRA oxidation to CO2 and H2O by the TCA cycle. Complete GLU aerobic metabolism to CO2 and H2O releases a total of 38 ATPs, the normal basic driver for energy in the cell. In the presence of ample O2, anaerobic glycolysis is usually depressed (the Pasteur Effect) in favor of the more efficient respiration metabolism. This is the normal state. The Warburg effect occurs in exhaustive exercise or when a tissue is hypoxic and the toxic PYRA accumulates because there is limited or not enough O2 present to drive TCA cycle. The cell has evolved lactic dehydrogenase A to act in hypoxia to convert PYRA to LACA. LACA is easily excreted from the metabolizing cell into extracellular environment. Warburg
5.3. CELL BIOLOGY OF CANCER
Anaerobic dissimilation
149
Inhibit cell-to-cell Interactions Alter chromatin structures H+ H+ + H+ H
1 Glucose
2 Lactic acid ΔpH 2 NAD
2 NADH2 Net yield: 2ATP
1 Glucose 2 Pyruvic acid
Glycolysis
Lack of oxygen condition increases as tumor grows away from arterial blood supply Fermentation
Insufficient Oxidation to CO2 and H2O Figure 5.8. The Warburg anaerobic glycolytic effect. Because of the early growth effects of foci and small neoplasms, the transformed cells separate from the local blood supply and become more anoxic as they expand. Cells at the lead edge show the most effects of O2 deprivation, and this deprivation switches metabolism control in these distal cells from the 38 ATP-rich TCA cycle to glycolysis (only 2 ATPs) and fermentation to lactic acid. Because of the reduced energy yield, the glyclosis pathway shown here cycles rapidly in neoplasms, thus depleting glycogen. As cells accumulate and excrete lactic acid, local acid effects occur which can dirupt pH control intra- and intercellularly. Some of the proposed effects of this hypothesis are shown. It is posited that mutagenesis is more likely under these altered conditions (Warburg 1956a,b).
hypothesized that these metabolic differences were the initial cause and not the result of carcinogenesis (Warburg 1956a,b). The cause of hypoxia can come early in the chemical carcinogenesis process where field effects in the LTA become perturbed by the entrance of a chaotropic xenochemical that produces reactive hyperplasia. Chaotropic agents are known to cause numerous types of physical and chemical molecular and cellular and intercellular alterations (e.g., disruption of actin and E-cadherin with the ECM), or alterations in Na+, K+, or H+ gradients which can in turn affect osmolality by shifting the cellular H2O balance in favor of uptake that leads to swollen cells (hypertrophy). Reactive hyperplasia can occur because of episomal responses and produces increases the compartment’s number of cells beyond the normal N-cell number. This is especially true if exposure to the chaotropic chemical is frequent and of long duration, thus preventing reparation. These disruptive cellular reactions have been characterized by a number of investigators as being some of the earliest of responses even before transformed foci form (Boutwell 1976; Feinberg and Tycko 2004; Foulds 1957; Rubin 1994). Hypertrophic and reactive hyperplastic responses are the tissue’s strategies of quickly diluting and shielding the xenochemical insult while preparing
150
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
TABLE 5.2.
Ion Λ0
Specific Electrical Conductances of Physiological Ions (Λ0)a
H+ 350
K+ 73
Na+ 50
Li+ 39
Ca2+ 119
Mg2+ 57
a
Extrapolated to infinite dilution to minimize chemical activity interactions, and the temperature of the aqueous solution is normalized to 25 °C with specific aqueous conduction of ions Λ0 and the unitis: 10−4 m2·A·s/mole, where m = meters, A = amps, and s = seconds.
the tissue for chemokine injury signaling (Coussens and Werb 2002; Federico et al. 2007; Maronpot et al. 1989; Pitot et al. 1985; Porta et al. 2007). Hypoxia takes place when the cells swell and are increasing in number such that the expanding cells become ever more distanced from their local blood supply. More new cell displacement causes less available O2 because these expanded regions exceed the passive diffusion limit of O2, which is about 100 μM from the blood supply (Gatenby and Gillies 2004). The hyperplastic new cells become hypoxic, causing the tissue to favor the hypoxic state, and LACA excretion commences (the Warburg anaerobic glycolysis effect; see Figure 5.6). That the H+ ions from LACA ionization can cause such pleiotropy should come as no surprise because the specific electrical conductances of physiologic ions seen in Table 5.2 clearly show that the H+ is by far the most mobile ion of cellular ions. It is a proton with a very large charge/mass ratio. This means that a modest change in pH will make substantial structural changes by interacting with various anionic groups in enzymes, chromatin, DNA, and mitochondrial and plasma membranes. The acid (from LACA) in the extracellular fluid leads to even more environmental perturbance and more reactive hyperplasia, along with activation of sentinel white cells, and finally the affected cells enter into a condition of regional acidosis. The penumbral region of distal cells exhibits the most hypoxicity and, in time, have been observed to become premalignant foci that are increasingly resistant to apoptosis and express increased membrane transporters in order to maintain intracellular pH (Gatenby and Gillies 2004; Klaassen and Lu 2008; Pikarsky et al. 2004). All is still reversible if the chemical is removed or otherwise metabolized in a timely manner. If not, the region proceeds to distort from homeodynamic equilibrium and develops early competing clones for resources such as glucose and O2. Oxygen metabolism and its balance can have much influence on tumor promotion (Troll and Wiesner 1985). In a thoughtful review by DeBerardinis, the authors provide a discussion on reprogramming of local metabolism in hypoxia and in establishing anaerobic glycolysis which can become sustained by the induction of hypoxia-inducible factor (HIF-1) expression and other factors (DeBerardinis et al. 2008). Many reviewers of today not cited here (due to space) still think that Warburg’s idea might be the capital effect of the I stage (Kondoh 2008). The biological sequela to form competing clones of foci is a prelude to oncogenesis and has been covered earlier in this chapter. Proof that Warburg’s hypothesis of GLU mismanagement is provided by the discovery that part of this hypoxia mechanism causes cachexia. This wasting condition is common in about one-half of all cancer patients and is the reason why
5.3. CELL BIOLOGY OF CANCER
151
one-fourth of cancer patients die. Besides gaining anorexia, the cancer patients acquire both GLU intolerance and insulin insensitivity (Lelbach et al. 2007). Metabolic effects analogous to the Warburg Effect start in the earliest part of the I stage. Warburg claims that it precedes clinical discovery only to worsen in advanced stages of cancer (Warburg 1956b). Patient muscular activity leads to the release of epinephrine (adrenaline), which causes the breakdown of glycogen in the muscles to GLU. That is, the muscle-stored glycogen is cleaved from the nonreducing polysaccharide ends of the carbohydrate chain by the enzyme glycogen phosphorylase to produce glucose-1-phosphate that is converted to glucose 6-phosphate which cannot leak out of the cell. The breakdown of glycogen is for the production of ATP that is consumed during muscular activity. Continued activity creates the demand for more ATP. In the beginning of the cancer disease, glycolysis produces PYRA that is converted to acetyl CoA, which is metabolized in the citric acid cycle to make ATP via aerobic metabolism. Later in the disease, however, O2 becomes scarce (see above and Figure 5.6) and anaerobic glycolysis becomes more dominant over TCA cycle or aerobic GLU oxidation with time. An effect called the Cori cycle sets up between the muscles and the liver (Figure 5.9). First, the LACA that the muscle produces is released into the blood, circulated, and then taken up by the liver; and by gluconeogenesis, which consumes 6 ATPs, it converts LACA → PYRA → GLU, which is then released from the liver into the blood. Subsequently, GLU is transported from the blood into the muscle and by the glycolysis process converts GLU → PYRA → LACA while releasing 2 ATPs. This completes the muscle–liver Cori cycle with a net deficit of −4 ATP/cycle. This
The Cori Cycle Blood Glucose
More Oxygenated
Glucose 2 ATP
6 ATP
2 Pyruvate
2 Pyruvate 2 Lactate H+
Liver
H+
2 Lactate
Blood Less Oxygenated
H+
H+
Muscle
Figure 5.9. The Cori cycle. Note that muscle and other tissues can produce excessive lactic acid in early cancer. Some lactate ionizes to H+ ion and thus lowers local pH. These acid effects are thought to disturb cell-to-cell interactions by cellular water and pH imbalances with chromatin/DNA rearrangements and hence is proposed to initiate neoplastic environments according to the Warburg cancer theory. The Cori cycle repeats many times in advancing cancer; and because of the net loss of energy by 2 ATP − 6 ATP = −4 ATPs per each Cori cycle, there is loss in the ability maintain tissue. Hence, there is a wasting of tissue over time. Cachexia is seen in a high percentage of terminal cancer patients.
152
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
progressive ATP or energy deficit produces increasing malaise and weakness usually seen in terminal patients, and it establishes a one-way catabolism of glycogen, body fat through usage of acetyl-CoA, and finally protein muscle mass: the cachexia syndrome (Tisdale 1997). It is not proved, but is a consistent observation, that LACA production in the Cori cycle is most likely correlated with adverse effects of essential structures because of excessive acid (see above) and the creation of a persistent ATP deficit. These effects of hypoxia and ATP deficits are essential properties inherent in cancer disease progression, and their origins exist from the beginning of carcinogenesis. This metabolic disturbance is a rational candidate for the incipient fatal events in chemical carcinogenesis.
5.4. SOME FINAL THOUGHTS ON BIOLOGY AND CANCER Cancer is a rare event among all cellular dystrophies; but because the number of events and adverse environments are so numerous over a lifetime, the chances are one in four of experiencing some form of cancer. Adults lose about 50–70 billion cells per day and execute about 2 × 1015 apoptoses/lifetime. Of necessity, this must be close to the cell replacement rate given good health at dynamic equilibrium. Because some cells rarely turn over, the total sum for a septuagenarian must be somewhat greater than the 2 × 1015 apoptoses/lifetime. The total number of mitoses/ person has been estimated to be about 1016 cell cycles/lifetime/adult body (Weinberg 1997). For all of this mitotic activity and gene expression (of DNA and chromatin), cell physiology exhibits extremely high fidelity in maintenance and passage of correct information compared to any other process we humans experience. Most of the time, our cells are corrected by various combinations of independent protection mechanisms: immune surveillance, numerous systems of genetic and chromatin repair, intracellular and epigenetic surveillance, and intercellular surveillance (Klein et al. 2007). It is indeed the reason why we do not succumb sooner to the chaos of thermodynamic entropy which acts on all processes. On average, there occur relatively few age-corrected diseases or cancers until old age. Heritable defects, adverse lifestyles, and adventitious exposures to chemical carcinogens are examples that can accelerate defect rates beyond the cell’s normal potential to correct mistakes, lesions, and injuries (cf. Section 5.3.5). The preemption or ablation of pesticide excessive exposures can avoid I-stage events or interrupt early P-stage events and positively affect cancer prophylaxis. If caught early enough, in fact, some mistakes or lesions can reverse or remodel as was reviewed previously, and be eliminated by apoptosis or by cell senescence (cf. Section 5.3.2). If the intervention is not imposed soon enough, however, the progression of carcinogenesis can proceed far enough in the P stage such that the cancer mass can pass a “point of no return” (Table 5.1). This point is characterized by the acquisition of multiple and irreversible critical stage steps that in time become permanent (Figure 5.4). The tumor becomes cancerous and usually grows at the physiologic expense of the affected site. Only through a new set of metastatic signals can the localized cancer become delocalized and spread in metastasis, often with fatal results (Weinberg 2008b).
5.4. SOME FINAL THOUGHTS ON BIOLOGY AND CANCER
153
Stress can have biologically significant effects on carcinogenesis (Godbout and Glaser 2006). Stress has two connotations: There can be chemically induced stress reactions and/or psychologically induced stresses. When an organ is under chemical stress, there are induced heat shock proteins like HSP27, HSP 70, and HSP 90 (Calderwood et al. 2006; Calderwood and Ciocca 2008). These protein factors are highly conserved in evolution and are part of a highly effective immunological response to counter chemical toxicity events. However, one factor of this family (HSP 1) seems to support carcinogenesis (Dai et al. 2007). There are many reviews on reactions to chemical stressors such as carcinogens (Zhang and Vande Woude 2007). The reader is also referred to Chapter 6 in this volume. Less known, but not necessarily less important, is the induction of the stress reaction by unremitting psychological or perceived adverse pressure from psychosocial factors that produce signaling that links pathologic stress responses with chemical carcinogenesis (Murakami et al. 2007). Aging was once thought to be the stochastic accumulation of errors and the statistical eventuality that the individual had reached the design limits of the body. More recent evidence suggests that there may be a specific aging program that, when activated, starts an ordered species-specifc decline (Holliday 2004). We postulate here that both stochastic and determinative programmatic theories may be true. If an “aging program” for higher organisms is operative, one might see biochemical reactions favoring a programmed decline or gradual shutdown. More age-related event biomarkers are needed to establish whether this is the fact. One of the fundamental properties of cells in culture was discovered by Leonard Hayflick in 1965, when he demonstrated that normal human cells in an in vitro culture divide about 52 times before autonomously entering a senescence phase. One reason for this is that each mitotic cycle allows the enzyme telomerase to shorten the ends of chromosomes, called telomeres (Ben-Porath and Weinberg 2004). When telomeres become too short or are ablated, chromosomes become unstable structurally and this leads to cell death. This mechanism acts as a countdown clock recording the number times the cell transacts mitosis (Finkel et al. 2007; Stewart and Weinberg 2006). The longer a cell exists, the more chances there are to accumulate errors within the cellular DNA and/or the histone code because these repair systems collectively have a finite error rate. Apoptosis and rebuilding a rigorous new cell replacement is believed to have been selected in evolution in order to protect the body from creating old defect-ridden cells that could lead to cancer before reproduction is achieved. After a time, it is thermodynamically more precise and accurate to start over by generating a new cell from a well-protected stem cell as a template. Short telomeres have often been observed in the P stage of carcinogenesis. Hence, a cell is born from asymmetrical stem cell mitosis, matures, replicates ↔ functions, and finally reaches the cell time limit set for that organ. Cell senescence does not happen abruptly, as once was thought, but instead cells undergo senescence during the last cycles of the aging cells (Weinberg 1997). There is evidence that the self-recognizing immunologic reactions, established early by 2–3 years of age, begin to break down in old age, giving rise to more defects in normal cells. With advanced age, individuals become more susceptible to disease, including cancer, while also establishing “autoimmune types” of problems (Anisimov 2007). However, there is
154
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
evolutionary evidence to suggest that long-life animals may have more repair mechanisms that are more efficient, thus accounting for their extended age (Holliday 2004). One plausible theory of aging is that somatic stem cells begin to show their age by their accumulated defects: As the organ generative cells go, so go the organs. There may be incorrect control of autophagy in old age, thus causing organelle damages that would, over time, produce faltering cells (Finkel et al. 2007). With so many known evolved control mechanisms in embryos, neonates, and puberty, it is reasonable to assume that mechanisms might also exist to end life in order to maintain and replenish a rigorous gene pool. If true, hazard evaluation must realistically take into account this essential aspect of human design. This may explain the control incidences of various organ cancers observed in control mice and rats in the U.S. NTP’s cancer bioassays (Maronpot 2007; Finkel et al. 2007; Holliday 2004). Examination of an aging cohort presents excellent opportunities to study the senescence of mechanisms and pathways and possibly adding to quality-of-life issues that the elderly face. The goal of this chapter was to cover many of the essential biological reactions and pathways that can participate in carcinogenesis. There are many protective evolved mechanisms in humans that ward off cancer and other diseases. Section 5.3.4 showed that it is an aberrant I cell in a field of physicochemically disrupted normal cells that provides for poor physiological communication in the local tissue array (LTA). This sets the stage for premalignancy but is reversible as shown in Figure 5.4. The participation of sentinel cells, cytokine-recruited cells, and some distal cells as in the Warburg Effect and the Cori cycle (Figures 5.8 and 5.9) demonstrate that multiple cell types can co-participate in cancer initiation and progression. The number of qualitatively independent stages from initiation to malignancy to metastasis and then to distal colonization has been estimated by many authors to be at least 6–8 separately acting, principal stages that are obligatory for cancer progression (Emmelot and Scherer 1977; Farber 1987; Gatenby and Gillies 2008; Marks et al. 2007; Rangarajan et al. 2004; Weinberg 1989). The overall process is not stochastic but has deterministic characteristics too. Cancer manifestation depends on the completion of biochemically different steps within each stage and the orderdependent 6–8 stages. Each stage contains a number of separate steps; and at this time, only some of these steps are recognized and understood. Unique steps should provide unique markers for each the stages. Not only must all 6–8 essential stages be satisfied in multistaged carcinogenesis, but each stage must also be driven to completion by a combination of duration and concentration of that step’s causal agent(s), host factors, genetic predisposition, lifestyle, stress, age, and so on. It has also been reviewed here that the I-stage duration may be as long as the last stage of colonization, but the middle P-stage and malignancy acquisition can go fast by comparison under sufficient carcinogenic pressure. The time versus cancer event curve is likely bell-shaped response. Preliminary biological modeling suggests that the accumulative dose–response curve of the 6–8 multistages is likely a sigmoid curve (whatever the probabilities of the individual steps are). That is, the dose– cancer response would start slow at low exposures but would show higher exponential rates at higher accumulated exposures. In the latter case, high exposures will force all stage steps to react faster, which allows less time for clonal competition,
REFERENCES
155
repair, and remodeling. At a low dose of chemical carcinogen, the competition among clones is less and chemical pressure is less. Either the malignant clone arises on a long time course or it does not arise at all because it can be within the range of detection, detoxification, repair, remodeling (e.g., cyst formation), or resorption. With less clonal remediation time, the pressure of high-doses force genetic and epigenetic alterations would favor malignant clones with more heterogeneity. The response would rise fast with time but would slow down in the latter stages of malignancy formation, metastasis, and colonization which are respectively less frequent and less probable events. This would complete the plateau region of the sigmoid curve. It has been proposed that malignant regrowth after cancer surgery may not always mean that some residual malignant cells were not removed, but rather there is a systematic and ongoing organ dysfunction within the episome or “conditioned cell memory” that is persistent in all the cells of the organ such that tumor regrowth is inevitable (Ruggiero and Bustuoabad 2006). That is, unless this episomal lesion in this LTA is fixed, the organ cells can self-initiate to produce the I* state—that is, mutagenize themselves because of improper controls. For the particulars reviewed here, we currently have a paucity of verified specific biomarkers to monitor cancer progress. This is especially relevant in light of the new cell advances in recent years. There are reports that miRNA can act as specific transcription factors and RNAi that specifically turn off steps in cancerous processes (He et al. 2007a,b; Ma and Weinstein 2008; Makunin et al. 2007). Functional RNAs, such as miRNA and RNAi, promise not only to be rich areas for biomarkers in toxicology, but also of immense value to society in our war against cancer (Wynter 2006). Because cancer development is varied in the various organs, it is suggested that organ cancer specifics be studied so as to better understand the interplay of biology and cancer: bladder (Luis et al. 2007); brain (Calabrese et al. 2007); gastric (Humar and Guilford, 2008); myeloma (Caers et al. 2008); thyroid (Reisco-Eizaguirre and Santisteban 2007); pancreas (Li et al. 2007); lymphatics (Allan et al. 2006); colon (Hedrick et al. 1994); prostate (Mimeaut and Batra 2006); Prins et al. 2008); and breast (Polyak 2007).
REFERENCES Adler, A. S., and Chang, H. Y. (2006). From description to causality: mechanisms of gene expression signatures in cancer. Cell Cycle 5(11), 1148–1151. Aguirre-Ghiso, J. A. (2007). Models, mechanisms and clinical evidence for cancer dormancy. Nat Rev Cancer 7(11), 834–846. Albor, A., and Kulesz-Martin, M. (2007). Novel initiation genes in squamous cell carcinomagenesis: A role for substrate-specific ubiquitylation in the control of cell survival. Mol Carcinog 46(8), 585–590. Allan, A., George, R., Vantyghem, S., Lee, M. Hodson, N. Engel, J. Holliday, R., Girvan, D., Scott, L., et al. (2006). Role of integrin binding protein osteopontin in lymphatic metastasis of breast cancer. Am J Pathol 169(1), 233–245. Amaral, P. P., Dinger, M. E., Mercer, T. R., and Mattick, J. S. (2008). The eukaryotic genome as an RNA machine. Science 319(5871), 1787–1789. Americal Cancer Society (2008). Cancer Facts and Figures 2008. Accessed at http://www.cancer.org/ docroot/home/index.asp, viewed on 2-23-2009.
156
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
Ames, B. N., Gold, L. S., and Willet, W. C. (1995). The causes and prevention of cancer. Proc Natl Acad Sci 92, 5258–5265. Angel, J. M., and DiGiovanni, J. (1999). Genetics of skin tumor promotion. Prog Exp Tumor Res 35, 143–157. Anisimov, V. N. (2003). The relationship between aging and carcinogenesis: A critical appraisal. Crit Rev Oncol Hematol 45(3), 277–304. Anisimov, V. N. (2007). Biology of aging and cancer. Cancer Control 14(1), 23–31. Anisimov, V. N. (2008). Carcinogenesis and aging 20 years after: Escaping horizon. Mech Ageing Dev 130(1–2), 105–121. Aslakson, C. J., and Miller, F. R. (1992). Selective events in the metastatic process defined by analysis of the sequential dissemination of subpopulations of a mouse mammary tumor. Cancer Res 52(6), 1399–1405. Bach, S. P., Renehan, A. G., and Potten, C. S. (2000). Stem cells: The intestinal stem cell as a paradigm. Carcinogenesis 21(3), 469–476. Bachman, A. N., Curtin, G. M., Doolittle, D. J., and Goodman, J. I. (2006). Altered methylation in genespecific and GC-rich regions of DNA is progressive and nonrandom during promotion of skin tumorigenesis. Toxicol Sci 91(2), 406–418. Balkwill, F., and Mantovani, A. (2001). Inflammation and cancer: Back to Virchow? Lancet 357(9255), 539–545. Baltzer, F. (1964). Theodor Boveri. Science 15, 809–815. Barrett, J. C., Kakunaga, T., Kuroki, T., Neubert, D., Trosko, J. E., Vasiliev, J. M., Williams, G. M., and Yamasaki, H. (1986). Short-term assays to predict carcinogenicity. Mammalian cell transformation in culture. IARC Sci Publ (83), 267–286. Barrett, J. C., and Ts’o, P. O. (1978a). Evidence for the progressive nature of neoplastic transformation in vitro. Proc Natl Acad Sci USA 75(8), 3761–3765. Barrett, J. C., and Ts’o, P. O. (1978b). Relationship between somatic mutation and neoplastic transformation. Proc Natl Acad Sci USA 75(7), 3297–3301. Beadle, G. W., and Tatum, E. L. (1941). Genetic control of biochemical Reactions in Neurospora. Proc Natl Acad Sci USA 27(11), 499–506. Ben-Porath, I., Thomson, M. W., Carey, V. J., Ge, R., Bell, G. W., Regev, A., and Weinberg, R. A. (2008). An embryonic stem cell-like gene expression signature in poorly differentiated aggressive human tumors. Nat Genet 40(5), 499–507. Ben-Porath, I., and Weinberg, R. A. (2004). When cells get stressed: an integrative view of cellular senescence. J Clin Invest 113(1), 8–13. Berenblum, I. (1941). The mechanism of carcinogensis. Cancer Res 1, 804–814. Berenblum, I. (1954). Carcinogenesis and tumor pathogenesis. Adv Cancer Res 2, 129–175. Berenblum, I., and Shubik, P. (1949). An experimental study of the initiating state of carcinogenesis, and a re-examination of the somatic cell mutation theory of cancer. Br J Cancer 3(1), 109–118. Bernards, R., and Weinberg, R. A. (2002). A progression puzzle. Nature 418(6900), 823. Berwald, Y., and Sachs, L. (1965). In vitro transformation of normal cells to tumor cells by carcinogenic hydrocarbons. J Natl Cancer Inst 35(4), 641–661. Biel, M., Wascholowski, V., and Giannis, A. (2005). Epigenetics—An epicenter of gene regulation: histones and histone-modifying enzymes. Angew Chem Int Ed Engl 44(21), 3186–3216. Boreiko, C., Mondal, S., Narayan, K. S., and Heidelberger, C. (1980). Effect of 12-O-tetradecanoylphorbol13-acetate on the morphology and growth of C3H/10T1/2 mouse embryo cells. Cancer Res 40(12), 4709–4716. Boutwell, R. K. (1964). Some biological aspects of skin carcinogenesis. Prog Exp Tumor Res 4, 207–250. Boutwell, R. K. (1974). The function and mechanism of promoters of carcinogenesis. CRC Crit Rev Toxicol 2(4), 419–443. Boutwell, R. K. (1976). The biochemistry of preneoplasia in mouse skin. Cancer Res 36(7 PT 2), 2631–2635. Boutwell, R. K. (1985). Tumor promoters in human carcinogenesis. In Important Advances in Oncology. Lippincott Williams & Wilkins, Philadelphia, pp. 16–27. Boveri, T. (1929). The Origin of Malignant Tumors, The William & Wilkin Company, Baltimore, pp. 1–119. Burnet, F. M. (1971). Immunological surveillance in neoplasia. Transplant Rev 7, 3–25.
REFERENCES
157
Burns, F., Albert, R., Altshuler, B., and Morris, E. (1983). Approach to risk assessment for genotoxic carcinogens based on data from the mouse skin initiation–promotion model. Environ Health Perspect 50, 309–320. Caers, J., Valckenborgh, E., Menu, E., Van Camp, B., and Vanderkerken, K. (2008). Unraveling the biology of multiple myeloma disease: Cancer stem cells, acquired intracellular changes and interactions with surrounding micro-environment. Bull Cancer 95(3), 301–313. Cairns, J. (1975). Mutation selection and the natural history of cancer. Nature 255(5505), 197–200. Calabrese, C., Poppleton, H., Kocak, M., Hogg, T., Fuller, C., Hamner, B., Oh, E. Gaber, M. W., Finklestein, D., Allen, M., et al. (2007). A perivascular niche for brain tumor stem sells. Cancer Cell 11, 69–82. Calderwood, S. K., and Ciocca, D. R. (2008). Heat shock proteins: stress proteins with Janus-like properties in cancer. Int J Hyperthermia 24(1), 31–39. Calderwood, S. K., Khaleque, M. A., Sawyer, D. B., and Ciocca, D. R. (2006). Heat shock proteins in cancer: Chaperones of tumorigenesis. Trends Biochem Sci 31(3), 164–172. Campbell, H. A., Xu, Y. D., Hanigan, M. H., and Pitot, H. C. (1986). Application of quantitative stereology to the evaluation of phenotypically heterogeneous enzyme-altered foci in the rat liver. J Natl Cancer Inst 76(4), 751–767. Casto, B. C., Janosko, N., and DiPaolo, J. A. (1977). Development of a focus assay model for transformation of hamster cells in vitro by chemical carcinogens. Cancer Res 37(10), 3508–3515. Chan, T. A., Glockner, S., Yi, J. M., Chen, W., Van, N. L., Cope, L., Herman, J. G., Velculescu, V., Schuebel, K. E., Ahuja, N., and Baylin, S. B. (2008). Convergence of mutation and epigenetic alterations identifies common genes in cancer that predict for poor prognosis. PLoS Med 5(5), e114. Chang, C. C., Sun, W., Cruz, A., Saitoh, M., Tai, M. H., and Trosko, J. E. (2001). A human breast epithelial cell type with stem cell characteristics as target cells for carcinogenesis. Radiat Res 155 (1 Pt 2), 201–207. Cheng, L., and Lai, M. D. (2003). Aberrant crypt foci as microscopic precursors of colorectal cancer. World J Gastroenterol 9(12), 2642–2649. Cifone, M. A., and Fidler, I. J. (1981). Increasing metastatic potential is associated with increasing genetic instability of clones isolated from murine neoplasms. Proc Natl Acad Sci 78(11), 6949–6952. Clark, S. J. (2007). Action at a distance: epigenetic silencing of large chromosomal regions in carcinogenesis. Hum Mol Genet 16(Spec No 1), R88–R95. Cook, J. W., Heiger, I., Kennaway, E. L., and Mayneord, W. V. (1932). The production of cancer by pure hydrocarbons—Part 1. R Soc Proc 111(Part B), 455–484. Coussens, L. M., Raymond, W. W., Bergers, G., Laig-Webster, M., Behrendtsen, O., Werb, Z., Caughey, G. H., and Hanahan, D. (1999). Inflammatory mast cells up-regulate angiogenesis during squamous epithelial carcinogenesis. Genes Dev 13(11), 1382–1397. Coussens, L. M., Tinkle, C. L., Hanahan, D., and Werb, Z. (2000). MMP-9 supplied by bone marrowderived cells contributes to skin carcinogenesis. Cell 103(3), 481–490. Coussens, L. M., and Werb, Z. (2002). Inflammation and cancer. Nature 420(6917), 860–867. Coyle-Rink, J., Del, V. L., Sweet, T., Khalili, K., and Amini, S. (2002). Developmental expression of Wnt signaling factors in mouse brain. Cancer Biol Ther 1(6), 640–645. Crick, F. H. (1958). On protein synthesis. Symp Soc Exp Biol 12, 138–163. Criollo, A., Galluzzi, L., Maiuri, M. C., Tasdemir, E., Lavandero, S., and Kroemer, G. (2007). Mitochondrial control of cell death induced by hyperosmotic stress. Apoptosis 12(1), 3–18. Curtin, G. M., Hanausek, M., Walaszek, Z., Zoltaszek, R., Swauger, J. E., Mosberg, A. T., and Slaga, T. J. (2006). Short-term biomarkers of cigarette smoke condensate tumor promoting potential in mouse skin. Toxicol Sci 89(1), 66–74. D’Alessio, A. C., and Szyf, M. (2006). Epigenetic tete-a-tete: The bilateral relationship between chromatin modifications and DNA methylation. Biochem Cell Biol 84(4), 463–476. D’Alessio, A. C., Weaver, I. C., and Szyf, M. (2007). Acetylation-induced transcription is required for active DNA demethylation in methylation-silenced genes. Mol Cell Biol 27(21), 7462–7474. Dai, C., Whitesell, L., Rogers, A. B., and Lindquist, S. (2007). Heat shock factor 1 is a powerful multifaceted modifier of carcinogenesis. Cell 130(6), 1005–1018. Dai, D. L., Martinka, M., Bush, J. A., and Li, G. (2004). Reduced Apaf-1 expression in human cutaneous melanomas. Br J Cancer 91(6), 1089–1095.
158
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
de Visser, K. E., Eichten, A., and Coussens, L. M. (2006). Paradoxical roles of the immune system during cancer development. Nat Rev Cancer 6(1), 24–37. DeBerardinis, R. J., Lum, J. J., Hatzivassiliou, G., and Thompson, C. B. (2008). The biology of cancer: Metabolic reprogramming fuels cell growth and proliferation. Cell Metab 7(1), 11–20. Deelman, H. T. (1927). The part played by injury and repair in the development of cancer. Br Med J 1, 872. DeNardo, D. G., Johansson, M., and Coussens, L. M. (2008). Immune cells as mediators of solid tumor metastasis. Cancer Metast Rev 27(1), 11–18. Déry, U., and Masson, J. Y. (2007). Twists and turns in the function of DNA signalling and repair proteins by post-translational modifications. Dolberg, D. S., Hollingsworth, R., Hertle, M., and Bissell, M. J. (1985). Wounding and its role in RSVmediated tumor formation. Science 230(4726), 676–678. Doll, R., and Hill, A. B. (1956). Lung cancer and other causes of death in relation to smoking; a second report on the mortality of British doctors. Br Med J 2(5001), 1071–1081. Doll, R., and Hill, A. B. (2004). The mortality of doctors in relation to their smoking habits: A preliminary report. 1954. Br Med J 328(7455), 1529–1533. Dragan, Y. P., Campbell, H. A., Xu, X. H., and Pitot, H. C. (1997). Quantitative stereological studies of a “selection” protocol of hepatocarcinogenesis following initiation in neonatal male and female rats. Carcinogenesis 18(1), 149–158. Druckrey, H. (1967). Quantitative aspects in chemical carcinogenesis. In Potential Carcinogenic Hazards from Drugs, Springer-Verlag, Berlin, pp. 60–78. Druckrey, H., Schildbach, A., Schmaehl, D., Preussmann, R., and Ivankovic, S. (1963). Quantitative analysis of the carcinogenic effect of diethylnitrosamine. Arzneimittelforschung 13, 841–851. Dvorak, H. F. (1986). Tumors: Wounds that do not heal. Similarities between tumor stroma generation and wound healing. N Engl J Med 315(26), 1650–1659. Elenbaas, B., and Weinberg, R. A. (2001). Heterotypic signaling between epithelial tumor cells and fibroblasts in carcinoma formation. Exp Cell Res 264(1), 169–184. Elsasser, W. M. (1984). Outline of a theory of cellular heterogeneity. Proc Natl Acad Sci USA 81(16), 5126–5129. Eming, S. A., Krieg, T., and Davidson, J. M. (2007). Inflammation in wound repair: Molecular and cellular mechanisms. J Invest Dermatol 127(3), 514–525. Emmelot, P., and Scherer, E. (1977). Multi-hit kinetics of tumor formation, with special reference to experimental liver and human lung carcinogenesis and some gneral conclusions. Cancer Res 37(6), 1702–1708. Enomoto, K., and Farber, E. (1982). Kinetics of phenotypic maturation of remodeling of hyperplastic nodules during liver carcinogenesis. Cancer Res 42(6), 2330–2335. Fadeel, B., Ottosson, A., and Pervaiz, S. (2008). Big wheel keeps on turning: Apoptosome regulation and its role in chemoresistance. Cell Death Differ 15(3), 443–452. Fadl-Elmula, I., Gorunova, L., Mandahl, N., Elfving, P., Lundgren, R., Mitelman, F., and Heim, S. (1999). Cytogenetic monoclonality in multifocal uroepithelial carcinomas: Evidence of intraluminal tumour seeding. Br J Cancer 81(1), 6–12. Farber, E. (1984). The multistep nature of cancer development. Cancer Res 44(10), 4217–4223. Farber, E. (1987). Pathogenesis of experimental liver cancer: Comparison with humans. Arch Toxicol Suppl 10, 281–288. Fearon, E. R., Cho, K. R., Nigro, J. M., Kern, S. E., Simons, J. W., Ruppert, J. M., Hamilton, S. R., Preisinger, A. C., Thomas, G., Kinzler, K. W., et al. (1990). Identification of a chromosome 18q gene that is altered in colorectal cancers. Science 247(4938), 49–56. Fearon, E. R., Hamilton, S. R., and Vogelstein, B. (1987). Clonal analysis of human colorectal tumors. Science 238(4824), 193–197. Federico, A., Morgillo, F., Tuccillo, C., Ciardiello, F., and Loguercio, C. (2007). Chronic inflammation and oxidative stress in human carcinogenesis. Int J Cancer 121(11), 2381–2386. Feinberg, A. P. (2004). The epigenetics of cancer etiology. Semin Cancer Biol 14(6), 427–432. Feinberg, A. P. (2005). A genetic approach to cancer epigenetics. Cold Spring Harb Symp Quant Biol 70, 335–341.
REFERENCES
159
Feinberg, A. P. (2007). Phenotypic plasticity and the epigenetics of human disease. Nature 447(7143), 433–440. Feinberg, A. P. (2008). Epigenetics at the epicenter of modern medicine. JAMA 299(11), 1345–1350. Feinberg, A. P., Ohlsson, R., and Henikoff, S. (2006). The epigenetic progenitor origin of human cancer. Nat Rev Genet 7(1), 21–33. Feinberg, A. P., and Tycko, B. (2004). The history of cancer epigenetics. Nat Rev Cancer 4(2), 143–153. Fialkow, P. J. (1976). Clonal origin of human tumors. Biochim Biophys Acta 458(3), 283–321. Fidler, I. J. (1978). Tumor heterogeneity and the biology of cancer invasion and metastasis. Cancer Res 38(9), 2651–2660. Fidler, I. J. (2002). Critical determinants of metastasis. Semin Cancer Biol 12(2), 89–96. Fidler, I. J., Kim, S. J., and Langley, R. R. (2007). The role of the organ microenvironment in the biology and therapy of cancer metastasis. J Cell Biochem 101(4), 927–936. Fidler, I. J., and Kripke, M. L. (1977). Metastasis results from preexisting variant cells within a malignant tumor. Science 197(4306), 893–895. Finkel, T., Serrano, M., and Blasco, M. A. (2007). The common biology of cancer and ageing. Nature 448(7155), 767–774. Flanagan, J. M. (2007). Host epigenetic modifications by oncogenic viruses. Br J Cancer 96(2), 183–188. Folkman, J. (2002). Role of angiogenesis in tumor growth and metastasis. Semin Oncol 29(6 Suppl 16), 15–18. Foulds, L. (1957). Tumor progression. Cancer Res 17(5), 355–356. Furth, J. (1953). Conditioned and autonomous neoplasms: A review. Cancer Res 13(7:1), 477–492. Galluzzi, L., Vicencio, J. M., Kepp, O., Tasdemir, E., Maiuri, M. C., and Kroemer, G. (2008). To die or not to die: That is the autophagic question. Curr Mol Med 8(2), 78–91. Gatenby, R. A., and Gillies, R. J. (2004). Why do cancers have high aerobic glycolysis? Nat Rev Cancer 4(11), 891–899. Gatenby, R. A., and Gillies, R. J. (2008). A microenvironmental model of carcinogenesis. Nat Rev Cancer 8(1), 56–61. Gillies, R. J., and Gatenby, R. A. (2007). Adaptive landscapes and emergent phenotypes: Why do cancers have high glycolysis? J Bioenerg Biomembr 39(3), 251–257. Godbout, J. P., and Glaser, R. (2006). Stress-induced immune dysregulation: Implications for wound healing, infectious disease and cancer. J Neuroimmune Pharmacol 1(4), 421–427. Gort, E. H., Groot, A. J., van der Wall, E., van Diest, P. J., and Vooijs, M. A. (2008). Hypoxic regulation of metastasis via hypoxia-inducible factors. Curr Mol Med 8(1), 60–67. Greten, F. R., Eckmann, L., Greten, T. F., Park, J. M., Li, Z. W., Egan, L. J., Kagnoff, M. F., and Karin, M. (2004). IKKbeta links inflammation and tumorigenesis in a mouse model of colitis-associated cancer. Cell 118(3), 285–296. Grum-Schwensen, B., Klingelhofer, J., Berg, C. H., El-Naaman, C., Grigorian, M., Lukanidin, E., and Ambartsumian, N. (2005). Suppression of tumor development and metastasis formation in mice lacking the S100A4(mts1) gene. Cancer Res 65(9), 3772–3780. Hahn, W. C., and Weinberg, R. A. (2002). Modelling the molecular circuitry of cancer. Nat Rev Cancer 2(5), 331–341. Hanahan, D., and Weinberg, R. A. (2000). The hallmarks of cancer. Cell 100(1), 57–70. Haseman, J. K., Huff, J. E., Zeiger, E., and McConnell, E. E. (1987). Comparative results of 327 chemical carcinogenicity studies. Environ Health Perspect 74, 229–235. He, L., He, X., Lim, L. P., de, S. E., Xuan, Z., Liang, Y., Xue, W., Zender, L., Magnus, J., Ridzon, D., Jackson, A. L., Linsley, P. S., Chen, C., Lowe, S. W., Cleary, M. A., and Hannon, G. J. (2007a). A microRNA component of the p53 tumour suppressor network. Nature 447(7148), 1130–1134. He, X., He, L., and Hannon, G. J. (2007b). The guardian’s little helper: MicroRNAs in the p53 tumor suppressor network. Cancer Res 67(23), 11099–11101. Hedrick, L., Cho, K. R., Fearon, E. R., Wu, T. C., Kinzler, K. W., and Vogelstein, B. (1994). The DCC gene product in cellular differentiation and colorectal tumorigenesis. Genes Dev 8(10), 1174–1183. Heim, S., Mandahl, N., and Mitelman, F. (1988). Genetic convergence and divergence in tumor progression. Cancer Res 48(21), 5911–5916.
160
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
Heim, S., Mertens, F., Jin, Y. S., Mandahl, N., Johansson, B., Biorklund, A., Wennerberg, J., Jonsson, N., and Mitelman, F. (1989). Diverse chromosome abnormalities in squamous cell carcinomas of the skin. Cancer Genet Cytogenet 39(1), 69–76. Hertel, K. J. (2008). Combinatorial control of exon recognition. J Biol Chem 283(3), 1211–1215. Hill, R., Song, Y., Cardiff, R. D., and Van Dyke, T. (2005a). Heterogeneous tumor evolution initiated by loss of pRb function in a preclinical prostate cancer model. Cancer Res 65(22), 10243–10254. Hill, R., Song, Y., Cardiff, R. D., and Van, D. T. (2005b). Selective evolution of stromal mesenchyme with p53 loss in response to epithelial tumorigenesis. Cell 123(6), 1001–1011. Hockel, M., and Vaupel, P. (2001). Biological consequences of tumor hypoxia. Semin Oncol 28 (2 Suppl 8), 36–41. Hoglund, M., Frigyesi, A., Sall, T., Gisselsson, D., and Mitelman, F. (2005). Statistical behavior of complex cancer karyotypes. Genes Chromosomes Cancer 42(4), 327–341. Hoglund, M., Sall, T., Heim, S., Mitelman, F., Mandahl, N., and Fadl-Elmula, I. (2001). Identification of cytogenetic subgroups and karyotypic pathways in transitional cell carcinoma. Cancer Res 61(22), 8241–8246. Holder, J. W., Elmore, E., and Barrett, J. C. (1993). Gap junction function and cancer. Cancer Res 53(15), 3475–3485. Holliday, R. (1987). The inheritance of epigenetic defects. Science 238(4824), 163–170. Holliday, R. (1989). DNA methylation and epigenetic mechanisms. Cell Biophys 15(1–2), 15–20. Holliday, R. (2002). Epigenetic comes of age in the twenty first century. Indian Acad Sci 81(1), 1–4. Holliday, R. (2004). The multiple and irreversible causes of aging. J Gerontol A Biol Sci Med Sci 59(6), B568–B572. Holliday, R. (2005). DNA methylation and epigenotypes. Biochemistry (Moscow) 70(5), 500–504. Holliday, R. (2006). Epigenetics: A historical overview. Epigenetics 1(2), 76–80. Holliday, R., and Murray, V. (1994). Specificity in splicing. Bioessays 16(10), 771–774. Hu, M. C., Wang, Y., Qiu, W. R., Mikhail, A., Meyer, C. F., and Tan, T. H. (1999). Hematopoietic progenitor kinase-1 (HPK1) stress response signaling pathway activates IkappaB kinases (IKK-alpha/ beta) and IKK-beta is a developmentally regulated protein kinase. Oncogene 18(40), 5514–5524. Humar, and Guilford (2008). Hereditary diffuse gastric cancer and lost cell polarity: A short path to cancer. Future Oncology 4(2), 229–239. Illmensee, K., and Mintz, B. (1976). Totipotency and normal differentiation of single teratocarcinoma cells cloned by injection into blastocysts. Proc Natl Acad Sci USA 73(2), 549–553. Jablonka, E. (2004). Epigenetic epidemiology. Int J Epidemiol 33(5), 929–935. Jablonka, E. (2006). Commentary: Induction and selection of variations during cancer development. Int J Epidemiol 35(5), 1163–1165. Jablonka, E., and Lamb, M. J. (2002). The changing concept of epigenetics. Ann NY Acad Sci 981, 82–96. Jablonka, E., and Lamb, M. J. (2007). Precis of evolution in four dimensions. Behav Brain Sci 30(4), 353–365. Jenuwein, T., and Allis, C. D. (2001). Translating the histone code. Science 293(5532), 1074–1080. Jones, and Baylin (2006). The epigenomics of cancer. Cell 128(4), 683–692. Kam, E., Melville, L., and Pitts, J. D. (1986). Patterns of junctional communication in skin. J Invest Dermatol 87(6), 748–753. Karnoub, A. E., Dash, A. B., Vo, A. P., Sullivan, A., Brooks, M. W., Bell, G. W., Richardson, A. L., Polyak, K., Tubo, R., and Weinberg, R. A. (2007). Mesenchymal stem cells within tumour stroma promote breast cancer metastasis. Nature 449(7162), 557–563. Katoh, Y., and Katoh, M. (2005). Hedgehog signaling pathway and gastric cancer. Cancer Biol Ther 4(10), 1050–1054. Kaufmann, S. H., and Steensma, D. P. (2005). On the TRAIL of a new therapy for leukemia. Leukemia 19(12), 2195–2202. Kaufmann, S. H., and Vaux, D. L. (2003). Alterations in the apoptotic machinery and their potential role in anticancer drug resistance. Oncogene 22(47), 7414–7430. Kaufmann, W. K., MacKenzie, S. A., and Kaufman, D. G. (1985). Quantitative relationship between hepatocytic neoplasms and islands of cellular alteration during hepatocarcinogenesis in the male F344 rat. Am J Pathol 119(2), 171–174.
REFERENCES
161
Kehler, J., Tolkunova, E., Koschorz, B., Pesce, M., Gentile, L., Boiani, M, et al. (2004). Oct-4 is required for primodial germ cell survival. EMBO Rep 5(11), 1078–1083. Kiba, T., Koyama, K., Ishiki, Y., Kimura, S., and Fukushima, T. (1960). On the mechanism of the development of multiple-drug-resistant clones of Shigella. Jpn J Microbiol 4, 219–227. Kim, H. J., Hawke, N., and Baldwin, A. S. (2006). NF-kappaB and IKK as therapeutic targets in cancer. Cell Death Differ 13(5), 738–747. Kim, Y., Sills, R. C., and Houle, C. D. (2005). Overview of the molecular biology of hepatocellular neoplasms and hepatoblastomas of the mouse liver. Toxicol Pathol 33(1), 175–180. Kinzler, K. W., and Vogelstein, B. (1996). Lessons from hereditary colorectal cancer 1. Cell 87(2), 159–170. Kinzler, K. W., and Vogelstein, B. (1998). Landscaping the cancer terrain. Science 280(5366), 1036–1037. Kitchin, K. T., Brown, J. L., and Setzer, R. W. (1994). Dose–response relationship in multistage carcinogenesis: Promoters. Environ Health Perspect 102(Suppl 1), 255–264. Klaassen, C. D., and Lu, H. (2008). Xenobiotic transporters: Ascribing function from gene knockout and mutation studies. Toxicol Sci 101(2), 186–196. Klaunig, J. E., and Kamendulis, L. M. (2004). The role of oxidative stress in carcinogenesis. Annu Rev Pharmacol Toxicol 44, 239–267. Klein, G., Imreh, S., and Zabarovsky, E. R. (2007). Why do we not all die of cancer at an early age? Adv Cancer Res 98, 1–16. Klein, G., and Klein, E. (1985). Evolution of tumours and the impact of molecular oncology. Nature 315(6016), 190–195. Kondoh, H. (2008). Cellular life span and the Warburg effect. Exp Cell Res 314(9), 1923–1928. Kulesz-Martin, M. F., Koehler, B., Hennings, H., and Yuspa, S. H. (1980). Quantitative assay for carcinogen altered differentiation in mouse epidermal cells. Carcinogenesis 1(12), 995–1006. Kuper, H., Boffetta, P., and Adami, H. O. (2002). Tobacco use and cancer causation: Association by tumour type. J Intern Med 252(3), 206–224. Lahad, J. P., Mills, G. B., and Coombes, K. R. (2005). Stem cell-ness: A “magic marker” for cancer. J Clin Invest 115(6), 1463–1467. Lassus, P., Opitz-Araya, X., and Lazebnik, Y. (2002). Requirement for caspase-2 in stress-induced apoptosis before mitochondrial permeabilization. Science 297(5585), 1352–1354. Le, N. H., Franken, P., and Fodde, R. (2008). Tumour–stroma interactions in colorectal cancer: Converging on beta-catenin activation and cancer stemness. Br J Cancer 98(12), 1886–1893. Lelbach, A., Muzes, G., and Feher, J. (2007). Current perspectives of catabolic mediators of cancer cachexia. Med Sci Monit 13(9), RA168–RA173. Li, C., Heldt, D., Dalerba, P., Burant, C., Zhang, L., Adsay, V., Wicha, M., et al. (2007). Identification of pancreatic cancer stem cells. Cancer Res 67(3), 1030–1037. Lichtenstein, P., Holm, N. V., Verkasalo, P. K., Iliadou, A., Kaprio, J., Koskenvuo, M., Pukkala, E., Skytthe, A., and Hemminki, K. (2000). Environmental and heritable factors in the causation of cancer—analyses of cohorts of twins from Sweden, Denmark, and Finland. N Engl J Med 343(2), 78–85. Liu, G., Meng, X., Jin, Y., Bai, J., Zhao, Y., Cui, X., Chen, F., and Fu, S. (2008a). Inhibitory role of focal adhesion kinase on anoikis in the lung cancer cell A549. Cell Biol Int 32(6), 663– 670. Liu, S., Dontu, G., and Wicha, M. S. (2005). Mammary stem cells, self-renewal pathways, and carcinogenesis. Breast Cancer Res 7(3), 86–95. Liu, S., Ginestier, C., Charafe-Jauffret, E., Foco, H., Kleer, C. G., Merajver, S. D., Dontu, G., and Wicha, M. S. (2008b). BRCA1 regulates human mammary stem/progenitor cell fate. Proc Natl Acad Sci USA 105(5), 1680–1685. Lobo, N. A., Shimono, Y., Qian, D., and Clarke, M. F. (2007). The biology of cancer stem cells. Annu Rev Cell Dev Biol 23, 675–699. Loeb, L. A., Bielas, J. H., and Beckman, R. A. (2008). Cancers exhibit a mutator phenotype: Clinical implications. Cancer Res 68(10), 3551–3557. Loeb, L. A., Loeb, K. R., and Anderson, J. P. (2003). Multiple mutations and cancer. Proc Natl Acad Sci USA 100(3), 776–781.
162
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
Loeffler, M., Birke, A., Winton, D., and Potten, C. (1993). Somatic mutation, monoclonality and stochastic models of stem cell organization in the intestinal crypt. J Theor Biol 160(4), 471–491. Luis, N., Lopez-Knowles, E., Real, F. (2007). Molecular biology of bladder cancer. Clin Transl Oncol 9(1), 5–12. Luzzi, K. J., MacDonald, I. C., Schmidt, E. E., Kerkvliet, N., Morris, V. L., Chambers, A. F., and Groom, A. C. (1998). Multistep nature of metastatic inefficiency: Dormancy of solitary cells after successful extravasation and limited survival of early micrometastases. Am J Pathol 153(3), 865–873. Lyman, G. H. (1992). Risk factors for cancer. Prim Care 19(3), 465–479. Ma, L., and Weinstein, R. A. (2008). MicroRNAs in malignant progression. Cell Cycle 7(5), 570–572. Maeda, S., Kamata, H., Luo, J. L., Leffert, H., and Karin, M. (2005). IKKbeta couples hepatocyte death to cytokine-driven compensatory proliferation that promotes chemical hepatocarcinogenesis. Cell 121(7), 977–990. Magee, P. N., and Barnes, J. M. (1967). Carcinogenic nitroso compounds. Adv Cancer Res 10, 163–246. Makunin, I. V., Pheasant, M., Simons, C., and Mattick, J. S. (2007). Orthologous microRNA genes are located in cancer-associated genomic regions in human and mouse. PLoS ONE 2(11), e1133. Malarkey, D. E., and Maronpot, R. R. (2005). Carcinogenesis. In Encyclopedia of Toxicology, 2nd edition Wexler, P., ed., Elsevier, New York, pp. 445–466. Maresca, B., and Schwartz, J. H. (2006). Sudden origins: A general mechanism of evolution based on stress protein concentration and rapid environmental change. Anat Rec B New Anat 289(1), 38–46. Margulis, A., Zhang, W., lt-Holland, A., Crawford, H. C., Fusenig, N. E., and Garlick, J. A. (2005). E-cadherin suppression accelerates squamous cell carcinoma progression in three-dimensional, human tissue constructs. Cancer Res 65(5), 1783–1791. Marks, F., Furstenberger, G., and Muller-Decker, K. (2007). Tumor promotion as a target of cancer prevention. Recent Results Cancer Res 174, 37–47. Maronpot, R. R. (2007). Cancer bioassays. In Cancer Handbook, 2nd edition Alison, M., ed., John Wiley & Sons., West Sussex, England, pp. 1–25. Maronpot, R. R., Pitot, H. C., and Peraino, C. (1989). Use of rat liver altered focus models for testing chemicals that have completed two-year carcinogenicity studies. Toxicol Pathol 17(4 Pt 1), 651–662. Martin, M., Simon-Assmann, P., Kedinger, M., Martin, M., Mangeat, P., Real, F. X., and Fabre, M. (2006). DCC regulates cell adhesion in human colon cancer derived HT-29 cells and associates with ezrin1. Eur J Cell Biol 85(8), 769–783. Mauri, F., McNamee, L. M., Lunardi, A., Chiacchiera, F., Del Sal, G., Brodsky, M. H., and Collavin, L. (2008). Modification of Drosophila p53 by SUMO modulates its transactivation and pro-apoptotic functions. J Biol Chem 283(30), 20848–20856. Meuth, M. (1990). The structure of mutation in mammalian cells. Biochim Biophys Acta 1032(1), 1–17. Mimeaut, M., and Batra, S. (2006). Recent Advances on multiple tumorigenic cascades involved in prostatic cancer progression and targeting therapies. Carcinogenesis 27(1), 1–22. Mintz, B., and Illmensee, K. (1975). Normal genetically mosaic mice produced from malignant teratocarcinoma cells. Proc Natl Acad Sci USA 72(9), 3585–3589. Mondal, S., and Heidelberger, C. (1980). Inhibition of induced differentiation of C3H/10T 1/2 clone 8 mouse embryo cells by tumor promoters. Cancer Res 40(2), 334–338. Moore, R. J., Owens, D. M., Stamp, G., Arnott, C., Burke, F., East, N., Holdsworth, H., Turner, L., Rollins, B., Pasparakis, M., Kollias, G., and Balkwill, F. (1999). Mice deficient in tumor necrosis factor-alpha are resistant to skin carcinogenesis. Nat Med 5(7), 828–831. Morange, M. (2002). The relations between genetics and epigenetics: A historical point of view. Ann NY Acad Sci 981, 50–60. Morange, M. (2007). The field of cancer research: An indicator of present transformations in biology. Oncogene 26(55), 7607–7610. Morange, M. (2008). What history tells us XIII. Fifty years of the central dogma. J Biosci 33(2), 171–175. Moss, T. J., and Wallrath, L. L. (2007). Connections between epigenetic gene silencing and human disease. Mutat Res 618(1–2), 163–174.
REFERENCES
163
Mueller, M. M., and Fusenig, N. E. (2004). Friends or foes—Bipolar effects of the tumour stroma in cancer. Nat Rev Cancer 4(11), 839–849. Murakami, S., Noguchi, T., Takeda, K., and Ichijo, H. (2007). Stress signaling in cancer. Cancer Sci 98(10), 1521–1527. Murphy, J. B., and Sturm, E. (1925). Primary lung tumors in mice following the cutaneous application of coal tar. J Exp Med 42, 693–700. Musracchi, P., and Shimkin, M. B. (1956). Gendron’s enquiries into the nature, knowledge, and cure of cancers. Cancer 9(4), 645–647. Nakayama, J., Yuspa, S. H., and Poirier, M. C. (1984). Benzo(a)pyrene-DNA adduct formation and removal in mouse epidermis in vivo and in vitro: Relationship of DNA binding to initiation of skin carcinogenesis. Cancer Res 44(9), 4087–4095. Nigro, J. M., Cho, K. R., Fearon, E. R., Kern, S. E., Ruppert, J. M., Oliner, J. D., Kinzler, K. W., and Vogelstein, B. (1991). Scrambled exons. Cell 64(3), 607–613. Nishikawa, M. (2008). Reactive oxygen species in tumor metastasis. Cancer Lett 266(1), 53–59. Nowell, P. C. (1976). The clonal evolution of tumor cell populations. Science 194(4260), 23–28. Nowell, P. C. (1986). Mechanisms of tumor progression. Cancer Res 46(5), 2203–2207. Ohlsson, R., Kanduri, C., Whitehead, J., Pfeifer, S., Lobanenkov, V., and Feinberg, A. P. (2003). Epigenetic variability and the evolution of human cancer. Adv Cancer Res 88, 145–168. Oliveira, P. A., Colaco, A., Chaves, R., Guedes-Pinto, H., De-La-Cruz, P. L., and Lopes, C. (2007). Chemical carcinogenesis. An Acad Bras Cienc 79(4), 593–616. Orimo, A., Gupta, P. B., Sgroi, D. C., Renzana-Seisdedos, F., Delaunay, T., Naeem, R., Carey, V. J., Richardson, A. L., and Weinberg, R. A. (2005a). Stromal fibroblasts present in invasive human breast carcinomas promote tumor growth and angiogenesis through elevated SDF-1/CXCL12 secretion. Cell 121(3), 335–348. Orimo, A., Gupta, P. B., Sgroi, D. C., Renzana-Seisdedos, F., Delaunay, T., Naeem, R., Carey, V. J., Richardson, A. L., and Weinberg, R. A. (2005b). Stromal fibroblasts present in invasive human breast carcinomas promote tumor growth and angiogenesis through elevated SDF-1/CXCL12 secretion. Cell 121(3), 335–348. Orimo, A., and Weinberg, R. A. (2006). Stromal fibroblasts in cancer: A novel tumor-promoting cell type. Cell Cycle 5(15), 1597–1601. Orimo, A., and Weinberg, R. A. (2007a). Heterogeneity of stromal fibroblasts in tumors. Cancer Biol Ther 6(4), 618–619. Orimo, A., and Weinberg, R. A. (2007b). Heterogeneity of stromal fibroblasts in tumors. Cancer Biol Ther 6(4), 618–619. Peraino, C., Fry, R. J., Staffeldt, E., and Christopher, J. P. (1975). Comparative enhancing effects of phenobarbital, amobarbital, diphenylhydantoin, and dichlorodiphenyltrichloroethane on 2-acetylaminofluorene-induced hepatic tumorigenesis in the rat. Cancer Res 35(10), 2884–2890. Pereira, M. A., and Stoner, G. D. (1985). Comparison of rat liver foci assay and strain A mouse lung tumor assay to detect carcinogens: A review. Fundam Appl Toxicol 5(4), 688–699. Perwez, H. S., and Harris, C. C. (2007). Inflammation and cancer: An ancient link with novel potentials. Int J Cancer 121(11), 2373–2380. Peterson, J. A. (1983). The widespread nature of phenotypic variability in hepatomas and cell lines, in the form of a geometric series. J Theor Biol 102(1), 41–53. Pikarsky, E., Porat, R. M., Stein, I., Abramovitch, R., Amit, S., Kasem, S., Gutkovich-Pyest, E., UrieliShoval, S., Galun, E., and Ben-Neriah, Y. (2004). NF-kappaB functions as a tumour promoter in inflammation-associated cancer. Nature 431(7007), 461–466. Pitot, H. C., Dragan, Y. P., Teeguarden, J., Hsia, S., and Campbell, H. (1996). Quantitation of multistage carcinogenesis in rat liver. Toxicol Pathol 24(1), 119–128. Pitot, H. C., Goldsworthy, T. L., Moran, S., Kennan, W., Glauert, H. P., Maronpot, R. R., and Campbell, H. A. (1987). A method to quantitate the relative initiating and promoting potencies of hepatocarcinogenic agents in their dose–response relationships to altered hepatic foci. Carcinogenesis 8(10), 1491–1499. Pitot, H. C., Grosso, L. E., and Goldsworthy, T. (1985). Genetics and epigenetics of neoplasia: Facts and theories. Carcinog Compr Surv 10, 65–79. Pitts, J. D., Finbow, M. E., and Kam, E. (1988). Junctional communication and cellular differentiation. Br J Cancer Suppl 9, 52–57.
164
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
Polyak, K. (2007). Breast cancer: Origins and evolution. J Clin Invest 117(11), 3155–3163. Ponting, C. P., Oliver, P. L., Relk, W. (2009). Evolution and functions of long noncoding RNAs. Cell 136(4), 629–641. Popp, J. A., and Goldsworthy, T. L. (1989). Defining foci of cellular alteration in short-term and mediumterm rat liver tumor models. Toxicol Pathol 17(4 Pt 1), 561–568. Porta, C., Subhra, K. B., Larghi, P., Rubino, L., Mancino, A., and Sica, A. (2007). Tumor promotion by tumor-associated macrophages. Adv Exp Med Biol 604, 67–86. Prins,G., Tung, W., Belmont, J., Ho, S. (2008). Perinatal Exposure to oestradiol and bisphenol A alters the prostate epigenome and increases susceptibility to carcinogenesis. Basic Clin Pharmacol Toxicol 102, 134–138. Punnett, R. C. (1907). Mendelism, Bowes and Bowes, London, pp. 1–84. Qiu, W. H., Zhou, B. S., Chu, P. G., Chen, W. G., Chung, C., Shih, J., Hwu, P., Yeh, C., Lopez, R., and Yen, Y. (2005). Over-expression of fibroblast growth factor receptor 3 in human hepatocellular carcinoma. World J Gastroenterol 11(34), 5266–5272. Rakoff-Nahoum, S. (2006). Why cancer and inflammation? Yale J Biol Med 79(3–4), 123–130. Rakoff-Nahoum, S., and Medzhitov, R. (2007). Prostaglandin-secreting cells: A portable first aid kit for tissue repair. J Clin Invest 117(1), 83–86. Rangarajan, A., Hong, S. J., Gifford, A., and Weinberg, R. A. (2004). Species- and cell type-specific requirements for cellular transformation. Cancer Cell 6(2), 171–183. Renehan, A. G., Booth, C., and Potten, C. S. (2001). What is apoptosis, and why is it important? Br Med 322(7301), 1536–1538. Reisco-Eizaguirre, G., and Santisteban, P. (2007). Molecular biology of thyroid cancer initiation. Clin Transl Oncol, 9(11), 686–693. Rosenkranz, M., Rosenkranz, H. S., and Klopman, G. (1997). Intercellular communication, tumor promotion and non-genotoxic carcinogenesis: Relationships based upon structural considerations. Mutat Res 381(2), 171–188. Rubin, H. (1984). Early origin and pervasiveness of cellular heterogeneity in some malignant transformations. Proc Natl Acad Sci USA 81(16), 5121–5125. Rubin, H. (1992). Cancer development: the rise of epigenetics. Eur J Cancer 28(1), 1–2. Rubin, H. (1994). Incipient and overt stages of neoplastic transformation. Proc Natl Acad Sci USA 91(25), 12076–12080. Rubin, H. (1999). Cell damage, aging and transformation: a multilevel analysis of carcinogenesis. Anticancer Res 19(6A), 4877–4886. Rubin, H. (2001). Multistage carcinogenesis in cell culture. Dev Biol (Basel) 106, 61–66. Rubin, H. (2007). Ordered heterogeneity and its decline in cancer and aging. Adv Cancer Res 98, 117–147. Ruggiero, R., and Bustuoabad, O. (2006). The biological sense of cancer: A hypothesis. Theor Bio Med Modelling 3(43), 1–14. Satzinger, H. (2008). Theodor and Marcella Boveri: Chromosomes and cytoplasm in heredity and development. Nat Rev Genet 9(3), 231–238. Scheel, C., Onder, T., Karnoub, A., and Weinberg, R. A. (2007). Adaptation versus selection: The origins of metastatic behavior. Cancer Res 67(24), 11476–11479. Schiestl, R. H. (1993). Nonmutagenic carcinogens induce intrachromosomal recombination in dividing yeast cells. Environ Health Perspect 101(Suppl 5), 179–184. Schleiden, M., and Schwann, T. (1839). Beitrage zur Phytogenesis. Translated along with Schwann’s Mikroskopische Untersuchungen by H. Smith for the Sydenham Society, London, in 1947. Schmidt-Hansen, B., Klingelhofer, J., Grum-Schwensen, B., Christensen, A., Andresen, S., Kruse, C., Hansen, T., Ambartsumian, N., Lukanidin, E., and Grigorian, M. (2004). Functional significance of metastasis-inducing S100A4(Mts1) in tumor–stroma interplay. J Biol Chem 279(23), 24498–24504. Schones, D. E., and Zhao, K. (2008). Genome-wide approaches to studying chromatin modifications. Nat Rev Genet 9(3), 179–191. Schones, D. E., Cui, K., Cuddapah, S., Roh, T. Y., Barski, A., Wang, Z., Wei, G., and Zhao, K. (2008). Dynamic regulation of nucleosome positioning in the human genome. Cell 132(5): 887–898. Sell, S. (2004). Stem cell origin of cancer and differentiation therapy. Crit Rev Oncol Hematol 51(1), 1–28.
REFERENCES
165
Sharma-Walia, N., Raghu, H., Sadagopan, S., Sivakumar, R., Veettil, M. V., Naranatt, P. P., Smith, M. M., and Chandran, B. (2006). Cyclooxygenase 2 induced by Kaposi’s sarcoma-associated herpesvirus early during in vitro infection of target cells plays a role in the maintenance of latent viral gene expression. J Virol 80(13), 6534–6552. Shepard, P. J., and Hertel, K. J. (2008). Conserved RNA secondary structures promote alternative splicing. RNA 14(8), 1463–1469. Shimkin, M. B. (1980). Some Classics of Experimental Oncology, National Institutes of Health (NIH), pp. 1–739. Shimkin, M. B., and Triolo, V. A. (1969). History of chemical carcinogenesis: Some prospective remarks. Prog Exp Tumor Res 11, 1–20. Shipitsin, M., and Polyak, K. (2008). The cancer stem cell hypothesis: In search of definitions, markers, and relevance. Lab Invest 88(5), 459–463. Sieweke, M. H., Stoker, A. W., and Bissell, M. J. (1989). Evaluation of the cocarcinogenic effect of wounding in Rous sarcoma virus tumorigenesis. Cancer Res 49(22), 6419–6424. Slaga, T. J. (ed.). (1983). Tumor Promotion in Internal Organs, Volume I of Mechanisms of Tumor Promotion, CRC Press, Boca Raton, FL, pp. 1–179. Slaga, T. J. (ed.). (1984a). Tumor Promotion and Skin Carcinogenesis, Volume II of Mechanisms of Tumor Promotion, CRC Press, Boca Raton, FL, pp. 1–205. Slaga, T. J. (ed.). (1984b). Tumor Promotion and Carcinogensis in Vitro, Volume III of Mechanisms of Tumor Promotion, CRC Press, Boca Raton, FL, pp. 1–192. Slaga, T. J. (ed.). (1984c). Tumor Promotion in Internal Organs, Volume IV of Mechanisms of Tumor Promotion, CRC Press, Boca Raton, FL, pp. 1–154. Smela, M. E., Currier, S. S., Bailey, E. A., and Essigmann, J. M. (2001). The chemistry and biology of aflatoxin B(1): From mutational spectrometry to carcinogenesis. Carcinogenesis 22(4), 535–545. Smyth, M. J., Dunn, G. P., and Schreiber, R. D. (2006). Cancer immunosurveillance and immunoediting: The roles of immunity in suppressing tumor development and shaping tumor immunogenicity. Adv Immunol 90, 1–50. Soengas, M. S., Alarcon, R. M., Yoshida, H., Giaccia, A. J., Hakem, R., Mak, T. W., and Lowe, S. W. (1999). Apaf-1 and caspase-9 in p53-dependent apoptosis and tumor inhibition. Science 284(5411), 156–159. Soengas, M. S., Capodieci, P., Polsky, D., Mora, J., Esteller, M., Opitz-Araya, X., McCombie, R., Herman, J. G., Gerald, W. L., Lazebnik, Y. A., Cordon-Cardo, C., and Lowe, S. W. (2001). Inactivation of the apoptosis effector Apaf-1 in malignant melanoma. Nature 409(6817), 207–211. Solt, D. B., Cayama, E., Tsuda, H., Enomoto, K., Lee, G., and Farber, E. (1983). Promotion of liver cancer development by brief exposure to dietary 2-acetylaminofluorene plus partial hepatectomy or carbon tetrachloride. Cancer Res 43(1), 188–191. Stein, L. D. (2004). Human genome: End of the beginning, Nature 431, 915–916. Stewart, S. A., and Weinberg, R. A. (2006). Telomeres: cancer to human aging. Annu Rev Cell Dev Biol 22, 531–557. Sun, W., Kang, K. S., Morita, I., Trosko, J. E., and Chang, C. C. (1999). High susceptibility of a human breast epithelial cell type with stem cell characteristics to telomerase activation and immortalization. Cancer Res 59(24), 6118–6123. Süss, R., Kinzel, V., and Scribner, J. D. (1973). Cancer: Experiment and Concepts, Springer-Verlag, New York, pp. 1–285. Szyf, M. (2007). The dynamic epigenome and its implications in toxicology. Toxicol Sci 100(1), 7–23. Tai, M. H., Chang, C. C., Kiupel, M., Webster, J. D., Olson, L. K., and Trosko, J. E. (2005). Oct4 expression in adult human stem cells: Evidence in support of the stem cell theory of carcinogenesis. Carcinogenesis 26(2), 495–502. Talmadge, J. E. (2007). Clonal selection of metastasis within the life history of a tumor. Cancer Res 67(24), 11471–11475. Talmadge, J. E., Wolman, S. R., and Fidler, I. J. (1982). Evidence for the clonal origin of spontaneous metastases. Science 217(4557), 361–363. Tasdemir, E., Maiuri, M. C., Galluzzi, L., Vitale, I., Djavaheri-Mergny, M., D’Amelio, M., Criollo, A., Morselli, E., Zhu, C., Harper, F., Nannmark, U., Samara, C., Pinton, P., Vicencio, J. M., Carnuccio, R., Moll, U. M., Madeo, F., Paterlini-Brechot, P., Rizzuto, R., Szabadkai, G., Pierron, G., Blomgren,
166
CHAPTER 5 THE INTERPLAY OF CANCER AND BIOLOGY
K., Tavernarakis, N., Codogno, P., Cecconi, F., and Kroemer, G. (2008). Regulation of autophagy by cytoplasmic p53. Nat Cell Biol 10(6), 676–687. Tejpar, S., Michils, G., Denys, H., Van, D. K., Nik, S. A., Jadidizadeh, A., and Cassiman, J. J. (2005). Analysis of Wnt/Beta catenin signalling in desmoid tumors. Acta Gastroenterol Belg 68(1), 5–9. Tennant, R. W. (1993). A perspective on nonmutagenic mechanisms in carcinogenesis. Environ Health Perspect 101(Suppl 3), 231–236. Tiano, H. F., Loftin, C. D., Akunda, J., Lee, C. A., Spalding, J., Sessoms, A., Dunson, D. B., Rogan, E. G., Morham, S. G., Smart, R. C., and Langenbach, R. (2002). Deficiency of either cyclooxygenase (COX)-1 or COX-2 alters epidermal differentiation and reduces mouse skin tumorigenesis. Cancer Res 62(12), 3395–3401. Ting, A. H., McGarvey, K. M., and Baylin, S. B. (2006). The cancer epigenome—Components and functional correlates. Genes Dev 20(23), 3215–3231. Ting, A. H., Schuebel, K. E., Herman, J. G., and Baylin, S. B. (2005). Short double-stranded RNA induces transcriptional gene silencing in human cancer cells in the absence of DNA methylation. Nat Genet 37(8), 906–910. Tisdale, M. J. (1997). Biology of cachexia. J Natl Cancer Inst 89(23), 1763–1773. Tononi, G., Sporns, O., and Edelman, G. (1999). Measures of degeneracy and redundancy in biological networks. Proc Natl Acad Sci 96, 3257–3263. Trent, J. M., Wiltshire, R., Su, L. K., Nicolaides, N. C., Vogelstein, B., and Kinzler, K. W. (1995). The gene for the APC-binding protein beta-catenin (CTNNB1) maps to chromosome 3p22, a region frequently altered in human malignancies. Cytogenet Cell Genet 71(4), 343–344. Triolo, V. A. (1965). Nineteenth century foundations of cancer research advances in tumor pathology, nomenclature, and theories of oncogenesis. Cancer Res 25, 75–106. Troll, W., and Wiesner, R. (1985). The role of oxygen radicals as a possible mechanism of tumor promotion. Annu Rev Pharmacol Toxicol 25, 509–528. Trosko, J. E. (1988). A failed paradigm: Carcinogenesis is more than mutagenesis. Mutagenesis 3(4), 363–366. Trosko, J. E. (2005). The role of stem cells and cell–cell communication in radiation carcinogenesis: Ignored concepts. BJR Suppl 27, 132–138. Trosko, J. E. (2007). Gap junctional intercellular communication as a biological “Rosetta stone” in understanding, in a systems biological manner, stem cell behavior, mechanisms of epigenetic toxicology, chemoprevention and chemotherapy. J Membr Biol 218(1–3), 93–100. Trosko, J. E., and Chang, C. C. (1989). Stem cell theory of carcinogenesis. Toxicol Lett 49(2–3), 283–295. Trosko, J. E., Chang, C. C., Upham, B. L., and Tai, M. H. (2004). Ignored hallmarks of carcinogenesis: Stem cells and cell–cell communication. Ann NY Acad Sci 1028, 192–201. Trosko, J. E., Dawson, B., and Chang, C. C. (1981). PBB inhibits metabolic cooperation in Chinese hamster cells in vitro: Its potential as a tumor promoter. Environ Health Perspect 37, 179–182. Trosko, J. E., Jone, C., and Chang, C. C. (1984). The use of in-vitro assays to study and to detect tumour promoters. IARC Sci Publ 56, 239–252. Trosko, J. E., Yotti, L. P., Warren, S. T., Tsushimoto, G., and Chang, C. (1982). Inhibition of cell–cell communication by tumor promoters. Carcinog Compr Surv 7, 565–585. Ulrich, C. M., Bigler, J., and Potter, J. D. (2006). Non-steroidal anti-inflammatory drugs for cancer prevention: promise, perils and pharmacogenetics. Nat Rev Cancer 6(2), 130–140. Upham, B. L., Guzvic, M., Scott, J., Carbone, J. M., Blaha, L., Coe, C., Li, L. L., Rummel, A. M., and Trosko, J. E. (2007). Inhibition of gap junctional intercellular communication and activation of mitogen-activated protein kinase by tumor-promoting organic peroxides and protection by resveratrol. Nutr Cancer 57(1), 38–47. Upham, B. L., Masten, S. J., Lockwood, B. R., and Trosko, J. E. (1994). Nongenotoxic effects of polycyclic aromatic hydrocarbons and their oxygenation by-products on the intercellular communication of rat liver epithelial cells. Fundam Appl Toxicol 23(3), 470–475. Upham, B. L., and Trosko, J. E. (2009). Oxidative-dependent integration of signal transduction with intercellular gap junctional communication in the control of gene expression. Antioxidants and Redox Signalling 11(2), 1–11. Vasiliev, J. M. (2004). Cytoskeletal mechanisms responsible for invasive migration of neoplastic cells. Int J Dev Biol 48(5–6), 425–439.
REFERENCES
167
Vasiliev, J. M., and Guelstein, V. I. (1966). Local cell interactions in neoplasms and in the foci of carcinogenesis. Prog Exp Tumor Res 8, 26–65. Verma, M., Maruvada, P., and Srivastava, S. (2004). Epigenetics and cancer. Crit Rev Clin Lab Sci 41(5–6), 585–607. Vicencio, J. M., Galluzzi, L., Tajeddine, N., Ortiz, C., Criollo, A., Tasdemir, E., Morselli, E., Ben, Y. A., Maiuri, M. C., Lavandero, S., and Kroemer, G. (2008). Senescence, apoptosis or autophagy? When a damaged cell must decide its path—A mini-review. Gerontology 54(2), 92–99. von Fournier, F. D., Weber, E., Hoeffken, W., Bauer, M., Kubli, F., and Barth, V. (1980). Growth rate of 147 mammary carcinomas. Cancer 45(8), 2198–2207. Wallace, K. B. (book editor) (1997). Free Radical Toxicology, Target Organ Toxicology series, Hayes A. W., Thomas, J. A., Garner, D. E., series editors, Taylor & Francis, Washington, D.C., pp. 1–442. Wang, E., Lenferink, A., and O’Connor-McCourt, M. (2007). Cancer systems biology: Exploring cancerassociated genes on cellular networks. Cell Mol Life Sci 64(14), 1752–1762. Wang, Z., Zang, C., Rosenfeld, J. A., Schones, D. E., Barski, A., Cuddapah, S., Cui, K., Roh, T. Y., Peng, W., Zhang, M. Q., and Zhao, K. (2008). Combinatorial patterns of histone acetylations and methylations in the human genome. Nat Genet 40(7), 897–903. Warburg, O. (1925). Über de Stoffwechsel der Carcinomzelle. Klin Wochenschr 4, 534–536. Warburg, O. (1956a). On respiratory impairment in cancer cells. Science 124(3215), 269–270. Warburg, O. (1956b). On the origin of cancer cells. Science 123(3191), 309–314. Warren, S. T., Yotti, L. P., Moskal, J. R., Chang, C. C., and Trosko, J. E. (1981). Metabolic cooperation in CHO and V79 cells following treatment with a tumor promoter. Exp Cell Res 131(2), 427–430. Watabe, T. (1983). Metabolic activation of 7,12-dimethylbenz[alpha]anthracene (DMBA) and 7-methylbenz[alpha]anthracene (7-MBA) by rat liver P-450 and sulfotransferase 1. J Toxicol Sci 8(2), 119–131. Weinberg, R. A. (1989). Oncogenes, antioncogenes, and the molecular bases of multistep carcinogenesis. Cancer Res 49(14), 3713–3721. Weinberg, R. A. (1997). The cat and mouse games that genes, viruses, and cells play. Cell 88(5), 573–575. Weinberg, R. A. (2008a). Coevolution in the tumor microenvironment. Nat Genet 40(5), 494–495. Weinberg, R. A. (2008b). Mechanisms of malignant progression. Carcinogenesis 29(6), 1092–1095. Wicha, M. S., Liu, S., and Dontu, G. (2006). Cancer stem cells: An old idea—A paradigm shift. Cancer Res 66(4), 1883–1890. Williams, C. S., Mann, M., and DuBois, R. N. (1999). The role of cyclooxygenases in inflammation, cancer, and development. Oncogene 18(55), 7908–7916. Wogan, G. N., Hecht, S. S., Felton, J. S., Conney, A. H., and Loeb, L. A. (2004). Environmental and chemical carcinogenesis. Semin Cancer Biol 14(6), 473–486. Wolters, N. M., and MacKeigan, J. P. (2008). From sequence to function: Using RNAi to elucidate mechanisms of human disease. Cell Death Differ 15(5), 809–819. Woods, N. T., Yamaguchi, H., Lee, F. Y., Bhalla, K. N., and Wang, H. G. (2007). Anoikis, initiated by Mcl-1 degradation and Bim induction, is deregulated during oncogenesis. Cancer Res 67(22), 10744–10752. Wynter, C. V. (2006). The dialectics of cancer: A theory of the initiation and development of cancer through errors in RNAi. Med Hypotheses 66(3), 612–635. Yamagiwa, K., and Ichikawa, K. (1918). Experimental study of the pathogeneisis of carcinoma. J Cancer Res 3, 1–29. Yamagiwa, K., and Ichikawa, K. (1977). Experimental study of the pathogenesis of carcinoma. CA Cancer J Clin 27(3), 174–181. Yotti, L. P., Chang, C. C., and Trosko, J. E. (1979). Elimination of metabolic cooperation in Chinese hamster cells by a tumor promoter. Science 206(4422), 1089–1091. Yuspa, S. H., Hawley-Nelson, P., Koehler, B., and Stanley, J. R. (1980). A survey of transformation markers in differentiating epidermal cell lines in culture. Cancer Res 40(12), 4694–4703. Yuspa, S. H., and Poirier, M. C. (1988). Chemical carcinogenesis: From animal models to molecular models in one decade. Adv Cancer Res 50, 25–70. Zhang, Y. W., and Vande Woude, G. F. (2007). Mig-6, signal transduction, stress response and cancer. Cell Cycle 6(5), 507–513.
CH A P TE R
6
CHEMICAL CARCINOGENESIS: A BRIEF HISTORY OF ITS CONCEPTS WITH A FOCUS ON POLYCYCLIC AROMATIC HYDROCARBONS Stephen Nesnow
6.1. A BRIEF HISTORY OF CHEMICAL CARCINOGENESIS As discussed in Chapter 5, the first reported example of a carcinogenic exposure that led to human cancer is ascribed to Sir John Percival Pott.* Pott was born in 1714 and became a respected surgeon who practiced at St. Bartholomew’s Hospital, London, Great Britain. In his practice, Dr. Pott observed “sores” on the scrotums of chimney sweeps in London. While other surgeons presumed that these were the results of venereal disease, Dr. Pott realized that they were some kind of skin cancer. He surmised that the cause of the cancers was “a lodgement of soot in the ruggae of the scrotum.” In 1775 he reported these findings in “Chirugical Observations Relative to the Cataract, the Polypus of the Nose, the Cancer of the Scrotum, the Different Kinds of Ruptures, and the Mortification of Toes and Feet.” This publication was the first in epidemiology that related external exposures of coal tar/soot to a human cancer (Pott 1775). Originally termed Pott’s cancer, it is more commonly referred to as chimney-sweep’s cancer. One hundred and forty years later, the first experimental evidence that coal tar was carcinogenic came from two Japanese pathologists, Katsusaburo Yamagiwa and *This manuscript has been reviewed by the National Health and Environmental Effects Research Laboratory at the U.S. Environmental Protection Agency and approved for publication. The views expressed in this chapter are those of the authors and do not necessarily represent the views or policy of the U.S. Environmental Protection Agency. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
168
6.2. JAMES A. AND ELIZABETH C. MILLER AND THEIR THEORY OF METABOLIC ACTIVATION
169
Koichi Ichikawa at the University of Tokyo in 1915 (Yamagiwa and Ichikawa 1915). This was also the first example of experimental chemical carcinogenesis. Professors Yamagiwa and Ichikawa were testing the irritation theory of cancer proposed by the Danish pathologist, Johannes Andreas Grib Fibiger, based on the earlier writings of Julius Vogel (1814–1880) and Rudolph Virchow (1821–1902). Fibiger achieved the first controlled induction of cancer in laboratory animals in 1913 by feeding mice cockroaches infected with the worm Gongylonema neoplasticum (Fibiger 1913). The larvae of a worm induced a chronic inflammation of stomach tissue, eventually inducing gastric tumors. Fibiger received the Nobel Prize for this research in 1926. Yamagiwa and Ichikawa repeatedly painted coal tar on the ears of rabbits and succeeded in producing multiple squamous cell carcinomas, a range of benign and malignant hyperplastic lesions, and inflammatory changes in the painted areas. They were also the first to describe the complexity and progressive nature of the carcinogenic process as they identified the conversion of less malignant to more malignant tumor cells as well as the regression of benign tumors (Yamagiwa and Ichikawa 1915). Years later the search for the active carcinogenic components in coal tar began in the laboratory of Ian Heiger at the Institute for Cancer Research in Great Britain, who isolated carcinogenic polycyclic aromatic hydrocarbons (PAHs) from coal tar (Cook et al. 1933). Using two tons of coal tar pitch, they isolated several components, one of which was highly carcinogenic on mouse skin. It was identified as benzo[a]pyrene (B[a]P). B[a]P was then synthesized, and the synthetic material was also found to be highly tumorigenic on mouse skin (Kennaway 1955). Since then, B[a]P has remained the archetypical PAH used for the study of the mechanisms of chemical carcinogenesis and is the most widely and thoroughly studied PAH in this chemical class. Because B[a]P is a product of the incomplete combustion of fossil fuels, it is pervasive in the environment. Structurally, B[a]P is a fused pentacyclic PAH and has been found to be tumorigenic in almost every species tested by many different routes of exposure: mice (dermal, subcutaneous, intraperitoneal, feed), rats (subcutaneous, inhalation, intratracheal), hamsters (intratracheal), rabbit (dermal), fish (water), and dogs (endobronchial). The target organs for B[a]P-induced neoplasia include: skin, forestomach, lung, liver, mammary gland, esophagus, and tongue (see references for details) (Grimmer et al. 1987). Based on the preponderance of mechanistic, experimental, and epidemiological data, the International Agency for Research on Cancer (IARC) has recently classified both B[a]P and the occupation of chimney sweeping as human carcinogens (Straif et al. 2005).
6.2. JAMES A. AND ELIZABETH C. MILLER AND THEIR THEORY OF METABOLIC ACTIVATION By the mid-20th century it was known that a diverse group of chemicals were either experimental carcinogens or were associated with human neoplasia from epidemiological investigations of workplace exposures. The variety of chemical structures with this group of carcinogens defied a unifying, common mechanism that could
170
CHAPTER 6 CHEMICAL CARCINOGENESIS
explain their tumorigenic activities and a common unifying theory which explained their mechanisms of action. The concept of metabolic activation and electrophilicity—now a well-accepted mechanism of action of many, if not all, genotoxic carcinogens—was first described by the late James A. Miller (1915–2000) and Elizabeth C. Miller (1920–1987) based on a number of studies conducted in the late 1950s at the McArdle Laboratory for Cancer Research in Madison, Wisconsin. It had been reported that many classes of chemical carcinogens (PAHs, nitrosamines, aflatoxins, aromatic amines) covalently bound to cellular macromolecules, DNA, RNA, and protein in target tissues based on radiometric techniques. However, none of these classes of carcinogens covalently bound to these isolated macromolecules in the test tube. The Millers reasoned that in cells and tissues there must be enzymes that metabolize these agents to chemically reactive forms which then bind to the most likely macromolecular target for cancer, namely, DNA. It is instructive to review a brief example of their work on aromatic amines. Aromatic amines were first associated with human bladder cancer in 1895 based on observations of aniline dye workers (Rehn 1895). It was later determined that a series of related chemicals were also associated with this cancer: benzidine and β-naphthylamine (2-naphthylamine) (Case et al. 1954). Critical research on metabolic activation carried out by the Millers used 2-acetylaminofluorene (AAF), a potent carcinogen in the liver, bladder, intestine, and mammary gland. They found that AAF was N-hydroxylated to its proximate carcinogenic form N-hydroxy-AAF, which was converted into it ultimate carcinogenic form, a sulfate ester by hepatic 3′-phosphoadenosine 5′-phosphosulfate (PAPS) sulfotransferases (Figure 6.1). This sulfate ester could form an incipient nitrenium ion that bound covalently to DNA, and this initiated the cancer process. Using the model N-acetoxyAAF, the C8 of deoxyguanosine was the target in DNA yielding the N-(deoxyguanosin-8-yl)-AAF adduct. Other AAF adducts have been identified: the nonacetylated N-(deoxyguanosine-8-yl)-AF adduct as well as the 3-(deoxyguanosine-N2-yl)-AAF adduct arising from the nitrenium ion activation of the C3 carbon of AAF (Beland and Kadlubar 1985). In 1960, the Millers reported that a metabolite (N-hydroxy-AAF) proved to be much more carcinogenic than its parent compound (AAF) and produced tumors in tissues including the site of administration (Miller et al. 1960). This research demonstrated that for many carcinogens, the initiation of cancer depended on metabolic activation of parent molecules to electrophiles, a major unifying concept of their research. The conclusions of their research on metabolic activation were as follows: (1) Chemical carcinogens that are not themselves chemically reactive must be metabolically converted into a chemically reactive form. (2) The activated metabolite must be an electrophilic form in order to bind to DNA. (3) The covalent adducts to DNA that formed can initiate the process of carcinogenesis. Moreover, a common feature of many diverse chemical carcinogens is that their reactive forms were electrophiles (Heidelberger 1975). The Millers also found that many of the enzymes that participated in the metabolic activation process were the Phase I microsomal mixed function oxidases and Phase II enzymes whose functions were to detoxify and to make xenobiotics more water soluble so they could be excreted (Miller and Miller 1979). The Millers laid the foundation for our current understanding of chemical carcinogenesis.
171
6.2. JAMES A. AND ELIZABETH C. MILLER AND THEIR THEORY OF METABOLIC ACTIVATION
Ac NH
P450
Ac N
Proximate carcinogen OH
PAPS sulfotransferase
Ac
Ultimate carcinogen
N
+
O-
SO 3
O
O
Ac N
N
NH
N N
-
N
NH
N
NH2
N
dR
dR
N-(deoxyguanosin-8-yl)-AAF
NH HN
Ac
3-(deoxyguanosin-N2-yl)-AAF
O N
NH
NH N
N
NH2
dR
N-(deoxyguanosin-8-yl)-AF
Figure 6.1. The metabolic activation of AAF as described by the Millers and further refined by Beland and Kadlubar (1985).
The concept of metabolic activation developed with AAF was then applied to PAHs, aflatoxins, nitrosamines, nitrosoureas, hydrazines, urethane, and vinyl chloride. Several metabolic activation schemes are presented in Figure 6.2. In each case a highly reactive electrophilic carbocation is formed. We now know that the concept of metabolic activation applies to many genotoxic carcinogens and helps to explain
172
CHAPTER 6 CHEMICAL CARCINOGENESIS O
O
O O
O
O
O
P450 O
O O
O
CH3
O
N
CH3
Aflatoxin B1-2,3-epoxide
Aflatoxin B1 H3C
O
OH
O
N
P450
NH
O
NH
H3C
H3C
Dimethylnitrosamine O N N
CH3
H2O
OH
N N
H3C H2N
CH3
Methyl carbonium ion
O
Methylnitrosourea P450
H3C
N
CH3
O
P450
NH NH
N
H3C
Dimethylhydrazine
N CH3
N
-
+
H3C
Azomethane
O
OH
Methylazoxymethanol
O
O
P450 H2N
+
O
CH3
H2N
Ethyl carbamate (urethane)
O
O
P450 CH2
H2N
Vinyl carbamate
O
Vinyl carbamate epoxide
O
Cl
P450
Cl
H2C Vinyl chloride
Vinyl chloride epoxide
Figure 6.2. Metabolic activation schemes for a group of genotoxic carcinogens. In each example the parent carcinogen is converted into a more reactive electrophilic form that alkylates DNA through carbocation formation.
their mechanisms of action. There are other classes of chemical carcinogens that do not require metabolic activation, namely, the nongenotoxic or epigenetic carcinogens consisting of cytotoxicants, mitogens, peroxisome proliferators, and endocrine disruptors. These chemicals do not bind directly to DNA but generally induce DNA mutations by indirect methods (Williams 2001).
6.2. JAMES A. AND ELIZABETH C. MILLER AND THEIR THEORY OF METABOLIC ACTIVATION
6.2.1.
173
Metabolic Activation of PAH and Tumorigenesis
Probably the most intensive and detailed efforts at delineating metabolic activation mechanisms and their relationships to tumorigenesis have been centered in the study of PAH carcinogenesis. To date, there are four major and several minor theories of the metabolic activation of PAHs. The four major theories will be discussed further in detail using B[a]P as an example, where appropriate. The major hypotheses are the bay- and fjord-region diol epoxide metabolic activation mechanism (Figure 6.3A,B), the radical cation mechanism (Figure 6.3C), the o-quinone/reactive oxygen species (ROS) mechanism (Figure 6.3D), and the cyclopenta-ring oxidation mechanism (Figure 6.3E). 6.2.1.1. Bay- and Fjord-Region Diol Epoxide Metabolic Activation Mechanism. The generalized diol epoxide mechanism was developed from the bay region theory proposed by Jerina et al. (1976) and was based on the earlier observations of the nature of PAH metabolites identified by Boyland and Sims (1964) and the results from a quantum mechanical model. This theory recognized that angular benzo ring fusions on PAHs created a topological indentation on the polycyclic ring structure, called the bay region. For B[a]P the bay region encompasses four carbons (carbons 10, 10a, 10b, and 11) and three carbon–carbon bonds (Figure 6.3A; see Figure 6.4 for carbon numbering). In the example of B[a]P, metabolism by the cytochrome P450 isozymes at the carbon 7-carbon 8 aromatic double bond disrupts the aromatic nucleus by saturating that carbon–carbon bond and creates an arene oxide, B[a]P-7,8-oxide (Figure 6.4). The stereospecific and regiospecific metabolizing activity of each cytochrome P450, in combination with the capability of carbons to form chiral centers through metabolism, can create multiple forms of many PAH metabolites. Therefore, due to the chirality of carbon, two stereoisomeric (enantiomeric) forms of B[a]P-7,8-oxide are created: (−)-B[a]P-7S,8R-oxide and (+)-B[a]P-7R,8S-oxide (Figure 6.4). The (+) and (−) terminology refers to the ability of stereoisomers to rotate polarized light in a clockwise or counterclockwise direction, and the R and S terminology refers to their absolute stereochemistry. B[a]P-7,8-oxide is hydrated by epoxide hydrolase to a form two trans-dihydrodiols (diol): (+)-B[a]P-7R,8R-diol and (−)-B[a]P-7S,8Sdiol (Figure 6.4). The B[a]P-7,8-diols are further metabolized (epoxidized) by the cytochrome P450 isozymes at the carbon 9–carbon 10 double bond to give four bay region diol epoxides (BPDE): (+)-anti-7R,8S,9S,10R-BPDE and its enantiomer (−)-anti-7S,8R,9R,10S-BPDE, as well as (+)-syn-7R,8S,9R,10S-BPDE and its enantiomer (−)-syn-7S,8R,9S,10R-BPDE (Figure 6.4). The anti and syn terminology refers to the spatial relationship between the hydroxyl group on carbon 7 and the oxide on carbons 9 and 10. These diol epoxides possess an inherent activity to undergo carbon–oxygen bond scission or ring opening to form a reactive carbonium ion on carbon 10 (i.e., a positive-charged carbon atom). Carbonium ions are highly reactive electrophilic species that react with nucleophiles such as DNA and proteins to form covalent adducts. Theoretically, each of the four BPDEs can covalently bind to specific atoms (almost always nitrogen) on the DNA bases to give two adducts based on the mechanism of the epoxide ring opening yielding both cis and trans
174
CHAPTER 6 CHEMICAL CARCINOGENESIS
A Bay region O P450/EH
P450 HO
HO
OH
OH trans-B[a]P-7,8-dihydrodiol
B[a]P
anti-trans-B[a]P-7,8-dihydrodiol -9,10-epoxide
B Fjord region
O P450/EH
P450
HO
HO OH
DB[a,l]P
OH
trans-DB[a,l]P-11,12-dihydrodiol
anti-trans-DB[a,l]P-11,12dihydrodiol-13,14-epoxide
C P450 +
CH B[a]P
B[a]P radical cation
D AKR/O2 HO
O OH
O B[a]P-7,8-quinone
trans-B[a]P-7,8-dihydrodiol
O
E P450
Cyclopenta[cd]pyrene
Cyclopenta[cd]pyrene-3,4-oxide
Figure 6.3. The metabolic activation of PAH through diol epoxide, radical cation, o-quinone, and arene oxide activation mechanisms.
adducts, thus giving a potential total of 16 unique BPDE-DNA stereoisomeric adducts for each site on the nucleic acid base. In practice, far fewer metabolically formed adducts are observed because the anti-7R,8S,9S,10R-BPDE tends to be the major diol epoxide formed. For example, only one BPDE adduct was detected in mouse lungs treated with B[a]P, the (+)-anti-trans-7R,8S,9R,10S-BPDE-
6.2. JAMES A. AND ELIZABETH C. MILLER AND THEIR THEORY OF METABOLIC ACTIVATION 12
175
1
11
2
10 3
1 4
8 7
6
5
B[a]P
O
O
(-)-B[a]P-7S,8R-oxide
(+)-B[a]P-7R,8S-oxide
HO
HO
OH
OH
(-) B[a]P-7S,8S-diol
(+) B[a]P-7R,8R-diol
O
O
HO
HO
OH
OH
(+)-anti-7R,8S,9S,10R-BPDE
(-)-anti-7S,8R,9R,10S-BPDE +
+ O
O
HO
HO OH
(-)-syn-7R,8S,9R,10S-BPDE
OH
(+)-syn-7S,8R,9S,10R-BPDE
Figure 6.4. The metabolic activation of B[a]P to BPDE including a complete description of all of the possible stereoisomers.
deoxyguanosine (dGuo) adduct (Figure 6.5, upper left structure) (Ross et al. 1995). In vitro, human bronchoalveolar adenocarcinoma H358 cells treated with (±)anti-BPDE also formed the (+)-anti-trans-7R,8S,9R,10S-BPDE-dGuo adduct as the major adduct, with minor amounts of three other adducts (Figure 6.5) (Ruan et al. 2006).
176
CHAPTER 6 CHEMICAL CARCINOGENESIS
N
N N dR
O
N dR
O
N
HN
N
HN
NH
NH
HO
HO
HO
HO OH
OH
(+)-anti-trans-7R,8S,9R,10S-BPDE-dGuo
(-)-anti-trans-7S,8R,9S,10R-BPDE-dGuo
N
N N dR
O N
HN
N dR
O N
HN NH
NH HO
HO
HO
HO OH (+)-anti-cis-7R,8S,9R,10R-BPDE-dGuo
OH (-)-anti-cis-7S,8R,9S,10S-BPDE-dGuo
Figure 6.5. Structures of BPDE–deoxyguanosine adducts. The BPDE adduct (upper left) is the major DNA adduct found in most mammalian tissues after exposure to B[a]P.
One of the postulated quantitative measures of the reactivity of diol epoxides is ΔEdeloc/β, which is based on perturbational molecular orbital calculations that predict the ease of carbonium ion formation. The greater the ΔEdeloc/β value, the more reactive the carbonium ion and greater values were associated with the PAHs exhibiting higher tumorigenic activities (Jerina et al. 1976). This theory was expanded to include PAH structures with deeper peripheral indentations in their structure, those containing a fjord region. An example of a fjord region PAH is the extremely potent PAH, dibenzo[a,l]pyrene (DB[a,l]P) (Figure 6.3B). The fjord region encompasses five carbons and four carbon–carbon bonds, and in some cases the steric interactions between hydrogen atoms within the fjord region of the PAH forces the PAH ring system out of planarity (Katz et al. 1998). Some PAH fjord region diol epoxides are nonplanar (Lewis-Bevan et al. 1995), and these nonplanar fjord region PAH diol epoxides possess even higher reactivities and tumorigenic activities (presumably due to their nonplanarity) than that predicted by ΔEdeloc/β alone (Xue and Warshawsky 2005). The enzymes primarily responsible for Phase I metabolism of PAHs are (a) the cytochrome P450s (CYPs) CYP1A1, CYP1A2, and CYP1B1, (b) NADPH
6.2. JAMES A. AND ELIZABETH C. MILLER AND THEIR THEORY OF METABOLIC ACTIVATION
177
cytochrome P450 reductase, which converts PAHs to a series of arene oxides, and (c) epoxide hydrolase, which catalyzes the addition of water to the arene oxides to form trans diols. PAH phenols are also formed either by rearrangement of arene oxides or by direct oxygen insertion into a carbon–hydrogen bond, while quinones are formed by further oxidation of phenols or by the enzymatic action of aldo–keto reductases on PAH diols. The Phase II enzymes, UDP-glucuronyl transferase, PAPS sulfotransferase, and glutathione S-transferases conjugate PAH diols, phenols, and epoxides to glucuronic acid, sulfate, and glutathione, respectively (Osborne and Crosby 1987). One of the basic tenets of the theory of metabolic activation is that the proximate carcinogen should have greater biological activity compared to the parent molecule it was derived from. Similarly, the ultimate metabolite of the carcinogen should have greater biological activity than the proximate metabolite from which it was derived. For the bay region or fjord region diol epoxide mechanism, the PAH is metabolically activated in a sequence through the diol to the diol epoxide. This process creates intermediates that generally possess greater biological activities than their precursors. This effect is amply demonstrated in the case of benz[a]anthracene (B[a]A). One of the anti-diol epoxides of B[a]A (anti-B[a]A-3,4-diol-1,2-oxide) possesses greater activity as a mouse skin tumorigen or mouse lung tumorigen compared to its precursor diol (B[a]A-3,4-diol), which in turn possesses greater activity compared to the parent PAH, B[a]A (Levin et al. 1978; Wislocki et al. 1979). While this effect is observed for many PAHs, it is not universal for all PAHs that are metabolized to diols and diol epoxides due to a number of confounding factors (e.g., reactivity with water and biological constituents, or cytotoxicity). Also, an important observation to note is that the formation of a bay region PAH diol epoxide by itself does not confer a tumorigenic potential to that PAH. This is the case for phenanthrene because both phenanthrene and its bay region diol epoxide (anti-phenanthrene-3,4-diol-1,2 oxide) are inactive as tumorigens in newborn mice (Buening et al. 1979). Bay-region and fjord-region diol epoxides possess many biological activities, one of the most important being their ability to form covalent stable adducts with DNA. The nature and sequence specificity of these DNA adducts are based, in part, on the absolute configuration, molecular conformation, and stereochemistry of the diol epoxide, the specific purine (or pyrimidine) base being adducted, the site of adduction, and the nature and sequence of the DNA being adducted (Jerina et al. 1986). PAH–DNA adducts represent a type of DNA damage that can be converted into heritable mutations by misrepair or faulty DNA syntheses (Rodriguez and Loechler 1995; Watanabe et al. 1985). Bay- or fjord-region diol epoxide–DNA adducts can be repaired by nucleotide excision repair (Geacintov et al. 2002). There are numerous examples of bay- and fjord-region PAH diol epoxides that are mutagenic in bacteria and that can induce mutations, damage DNA, and chromosomal damage in rodent and human cells in culture. Many diol epoxides are tumorigenic in mice, thereby inducing skin, lung, and liver tumors. Concomitant with these findings are complementary observations that the parent PAHs and the diol metabolites of these PAHs also induced gene mutation, DNA damage, or chromosomal damage in these bioassay systems and were tumorigenic in mice. Furthermore, PAHs or their
178
CHAPTER 6 CHEMICAL CARCINOGENESIS
bay- or fjord-region diol epoxides induced mutations in critical genes associated with carcinogenesis such as proto-oncogenes (Chakravarti et al. 1998; Prahalad et al. 1997) and tumor suppressor genes (Ruggeri et al. 1993; Ramet et al. 1995). A strong relationship exists between the nature of the diol epoxide DNA adducts and the type of ras proto-oncogene mutations observed in DNA from tumors induced by the PAHs. In general, PAHs that form DNA adducts at deoxyguanosine primarily induce mutations in the ras gene at codons 12 or 13, while those that form DNA adducts at deoxyadenosine induce mutations in the ras gene at codon 61. Those PAHs that form adducts at both purine bases induced both types of mutations (Ross and Nesnow 1999). In addition to their genotoxic effects, some bay- or fjord-region diol epoxides are reported to induce apoptosis and cell cycle arrest in mammalian cells (Chramostova et al. 2004). PAH diol epoxide–DNA adducts have not only been detected in rodent tissues in experimental systems after PAH exposure, but have also been identified in (a) populations exposed to complex mixtures containing PAHs, (b) foundry workers (Perera et al. 1988; Hemminki et al. 1988), (c) coke oven workers (Rojas et al. 1995; Pavanello et al. 1999), (d) cigarette smokers (Lodovici et al. 1998; Rojas et al. 1995), (e) chimney sweeps (Pavanello et al. 1999), and (f) populations exposed to smoky coal combustion mixtures (Mumford et al. 1993). Some bay- or fjord-region diol epoxides form DNA adducts in the human p53 tumor suppressor gene at sites that are hotspots for lung cancer (Smith et al. 2000). There are several variants on the diol epoxide mechanism. Bis-diol epoxide– DNA adducts were formed from dibenz[a,h]anthracene (Platt and Schollmeier 1994; Nesnow et al. 1994a) and dibenz[a,j]anthracene (Vulimiri et al. 1999). A bis-diol epoxide was proposed as a mechanism for carcinogenesis for dibenz[a,h]anthracene, while for dibenz[a,j]anthracene its biological significance is unknown. A phenolic diol epoxide–DNA adduct was formed from benz[b]fluoranthene and was proposed to contribute to the biological activity of benz[b]fluoranthene (Weyand et al. 1993). Finally, a phenolic oxide–DNA adduct of B[a]P has also been described with unknown biological significance (Fang et al. 2001). 6.2.1.2. Radical Cation Mechanism. A radical cation is formed when a single electron is removed from the π electron system of a PAH by an oxidation process. Some of the processes that have been described that can perform this oxidation of PAHs are iodine, electrochemical, horseradish peroxidase, and cytochrome P450 (Hanson et al. 1998; RamaKrishna et al. 1992; Cavalieri et al. 1988, 1990). Radical cations are electrophiles and bind to DNA bases to form covalent adducts. The structures of these adducts are dependent on the charge localization of the PAH radical cation and the charge density on specific atoms within the nucleic acid bases. Charge localization of PAHs of radical cations favors the meso positions of PAHs, and for B[a]P the charge localization favors the radical cation on carbon 6 (Cremonesi et al. 1992). The maximum charge density of guanine is found on N7, while those of adenine are on N7 and N3. B[a]P radical–cation DNA adducts have been characterized in vitro after metabolic activation with horseradish peroxidase, or with 3-methylcholanthrene-induced microsomes, and in vivo from the skin of mice treated with B[a]P (Chen et al. 1996). The major B[a]P radical cation–DNA adducts
6.2. JAMES A. AND ELIZABETH C. MILLER AND THEIR THEORY OF METABOLIC ACTIVATION
O
179
N N
HN N
H2N
N N
N
H2N
N
BP-6-N7-Gua
BP-6-N7-Ade
O N
HN H2N
N
N H
N
N N
N NH2
BP-6-C8-Gua
Figure 6.6.
BP-6-N3-Ade
Structures of B[a]P radical cation DNA adducts from Chen et al. (1996).
obtained by horseradish peroxidase were 7-(benzo[a]pyrene-6-yl)-guanine (B[a] P-6-N7Gua), B[a]P-6-C8Gua, B[a]P-6-N7Adenine (Ade), and B[a]P-6-N3Ade (Figure 6.6). In contrast, microsomal activation of B[a]P gave B[a]P-6-N7Ade, B[a] P-6-N7Gua, and B[a]P-6-C8Gua; these depurinating adducts were also identified in the mouse skin treated with B[a]P. Each of these adducts was formed from an intermediary charged unstable covalent B[a]P–DNA adduct that has undergone bond scission at the glycosidic bond to give a depurinating adduct. The result of this bond scission is an apurinic site in the DNA molecule and a released B[a]P–DNA adduct. The role of depurinating adducts and apurinic sites in the PAH-induced cancer process is controversial and has yet to be fully elucidated. There are lines of evidence that both support and refute this theory. In support of this theory, the levels of depurinating adducts of B[a]P correlated with mutations in the H-ras oncogene in DNA isolated from mouse skin papillomas initiated by this compound (Chakravarti et al. 1995). It is well known that the initiation of skin tumors in mice is associated with the formation of mutations in the H-ras gene [reviewed by Ross and Nesnow (1999)]. DB[a,l]P treatment of mouse skin forms papillomas which contain the H-ras codon 61 (CAA to CTA) mutation. These same mutations were induced in early preneoplastic skin within one day after DB[a,l]P treatment and appear to be related to DB[a,l]P-Ade-depurinating adducts. Studies have shown that apurinic
180
CHAPTER 6 CHEMICAL CARCINOGENESIS
sites generated from depurination may undergo error-prone excision repair in preS-phase cells to induce these mutations. The initiated cells carrying specific oncogenic mutations, formed presumably by misrepair, undergo rapid clonal expansion and regression (Chakravarti et al. 2000, 2001). Other investigators studying PAHs such as B[a]P, 7,12-dimethylbenz[a]anthracene (DMBA), and DB[a,l]P have found results not in accord with unstable DNA adduct formation and apurinic site formation. Studies were conducted in cytochrome P450-expressing mammary carcinoma MCF-7 cells and in leukemia HL-60 cells, which produce a high peroxidase activity but no cytochrome P450-mediated activity. The results from these studies demonstrated that metabolic activation of B[a]P, DMBA, and DB[a,l]P was primarily mediated by the cytochrome P450 enzymes leading to diol epoxides that form predominantly stable DNA adducts. Because only low levels of AP sites were detected, the radical cation pathway was not a major contributor to the metabolic activation scenario (Melendez-Colon et al. 1997, 1999a,b, 2000). Additional studies need to be conducted to resolve the issues surrounding the role of apurinic sites in the PAH carcinogenesis process. 6.2.1.3. o-Quinone/Reactive Oxygen Species Mechanism. PAHs are metabolized to trans-dihydrodiols by CYP1A1 and epoxide hydrolase. Aldo–keto reductases catalyze the NAD(P)-dependent oxidation of non-K-region transdihydrodiols to o-quinones of many PAHs, including phenanthrene, chrysene, 5-methylchrysene, DB[a,l]P, and B[a]P (see Figure 6.3A,C) (Smithgall et al. 1986). Aldo–keto reductases are present in many mammalian species. In humans, AKR1C1– AKR1C4 and AKR1A1 are capable of activating trans-dihydrodiols by converting them to redox-active o-quinones (Palackal et al. 2001). The conversion of dihydrodiols to quinones requires the formation of catechols that undergo air oxidation via two sequential one electron events to yield the o-quinones. Each of these steps produces ROS (Penning et al. 1996). The o-quinones can be reduced back to the catechols by nonenzymatic means such as by NADPH and begin the oxidative cycling process again yielding further quantities of ROS (Burczynski and Penning 2000) (Figure 6.7). ROS forms have been implicated in cytotoxic, mutagenic, and
O2
ROS
AKR HO
HO OH
trans-B[a]P-7,8-dihydrodiol
O HO
O B[a]P-7,8-quinone
7,8-dihydroxy-B[a]P NAD(P)
+
NAD(P)H
Figure 6.7. The metabolic activation of trans-B[a]P-7,8-diol to B[a]P-7,8-quinone by AKR and the generation of ROS from Burczynski and Penning (2000).
6.2. JAMES A. AND ELIZABETH C. MILLER AND THEIR THEORY OF METABOLIC ACTIVATION
181
tumorigenic processes through DNA, lipid, and protein oxidation. ROS can induce DNA breaks or can oxidize DNA to form DNA adducts, the most common being 8-oxo-dG (Park et al. 2005). Not only can ROS damage DNA, it can also alter important cell signaling pathways that are involved in cell proliferation. For instance, ROS have been reported to alter protein kinase C. This receptor is susceptible to oxidative modification at the N-terminal regulatory domain containing a zincbinding cysteine-rich motif. When oxidized by ROS, protein kinase C activity is stimulated and signals downstream to c-fos and c-jun (Gopalakrishna and Jaken 2000). Other stress MAPK kinases such as the JNK/AP1 pathway have been altered by ROS (Benhar et al. 2002). It should be noted that to date the o-quinone/ROS mechanism has only been described in vitro. There are no reports that that PAH o-quinones are formed in vivo from B[a]P and are stable enough to redox cycle and induce ROS. B[a]P-7,8quinone has been the most intensely studied o-quinone and has been found to adduct to DNA in vitro (Balu et al. 2004, 2006), but not in vivo (Nesnow et al. 2005). B[a] P-7,8-quinone induces DNA breaks (Park et al. 2008) and induces ROS in vitro (Flowers-Geary et al. 1996). Current thought is that B[a]P-7,8-quinone mediates its in vitro biological effects through ROS formation in vitro. 6.2.1.4. Cyclopenta-Ring Oxidation Mechanism. The cyclopenta-ring oxidation mechanism involves arene oxide formation at a highly electron-rich isolated double bond located at a five-membered cyclopenta-ring within a cyclopenta-PAH. The cyclopenta ring is an external five-membered carbocyclic ring situated on a carbocyclic hexameric fused ring system. For example, a cyclopenta-ring derivative of benz[a]anthracene is benz[j]aceanthrylene while that of pyrene is cyclopenta[c,d] pyrene (Figure 6.3E). In general, cyclopenta ring derivatives of PAH are more mutagenic than their unsubstituted counterparts. For example, anthracene is nonmutagenic while its cyclopenta-ring counterpart, aceanthrylene, is highly mutagenic (Kohan et al. 1985). Similarly cyclopenta-ring derivatives of PAH are generally more tumorigenic than their unsubstituted counterparts. For example, pyrene is not tumorigenic while cyclopenta[c,d]pyrene is highly tumorigenic (Nesnow et al. 1998). Since the cyclopenta-ring is usually the region of highest electron density, it is a major site of oxidation by the cytochrome P450 isozymes (Nesnow et al. 1984, 1988). Rat and mouse liver preparations, human and rodent cells in culture, human CYP1A1, CYP1A2, and CYP3A4, human liver microsomes, and rats in vivo metabolize cyclopenta-fused PAHs at the cyclopenta ring double bond to give cyclopenta ring oxides and diols (Gold and Eisenstadt 1980; Mohapatra et al. 1987; Kwon et al. 1992; Nyholm et al. 1996; Johnsen et al. 1998a,b; Hegstad et al. 1999). Cyclopenta-ring oxides are reactive intermediates and bind to DNA to form DNA adducts in vitro and in vivo mainly at deoxyguanosine (Beach and Gupta 1994; Hsu et al. 1999). Cyclopenta[c,d]pyrene, a mouse lung tumorigen, formed cyclopentaring–deoxyguanosine–DNA adducts in lung tissues and induced mutations at the Ki-ras protooncogene in lung tumors of treated mice (Nesnow et al. 1994b). Cyclopenta-ring oxides like their parent cyclopenta-PAHs are mutagenic in bacterial and mammalian cells and can morphologically transform immortalized cells in culture (Bartczak et al. 1987; Nesnow et al. 1991). Cyclopenta-ring oxides are
182
CHAPTER 6 CHEMICAL CARCINOGENESIS
hydrated by epoxide hydrolase to diols. Some cyclopenta-ring diols are conjugated to sulfate esters by PAPS-sulfotransferase, and these reactive intermediates are mutagenic and can bind to DNA to form adducts (Surh et al. 1993). 6.2.1.5. Summary of the Mechanisms of Metabolic Activation of PAH. For B[a]P, based on the wealth of data, the diol epoxide metabolic activation mechanism seems to be the dominant mechanism in the induction of lung carcinogenesis in rodents and humans. This conclusion is based on mechanistic data obtained from experimental animal and human biomarker studies (Straif et al. 2005). Both the radical cation and diol epoxide metabolic activation mechanisms can explain the mouse skin tumorigenic activities of B[a]P, but their relative contributions to this tumorigenesis process needs to be defined. To date, the PAH o-quinone/ROS metabolic mechanism has been described only in vitro, and no in vivo studies using PAH o-quinones are available that describe ROS formation, DNA damage, or tumorigenic effects, prerequisites for validating this mechanism. Further research is needed to define the relevance of this mechanism of PAH metabolic activation. For cyclopentaring PAHs, arene oxide formation at the cyclopenta ring is currently accepted as the major route of metabolic activation for this class of PAHs.
6.3. THE CONCEPTS OF INITIATION, PROMOTION, AND PROGRESSION: THE ORIGIN OF MULTISTAGE CARCINOGENESIS Much currently accepted theory of chemical carcinogenesis has evolved from studies using the mouse skin model of tumorigenesis using PAH as model compounds. The terms initiation and promotion essentially were first described by Rous and Kidd based on the application of coal tar to the ears of rabbits (initiate) following this treatment with physical wounding (promote) giving rise to tumors (Rous and Kidd 1941). Note that the development and growth of tumors in general involves three distinct stages: (1) initiation, (2) promotion, and (3) progression. The ideas that chemicals could alter DNA and induce mutations came from a series of investigators in the 19th century leading to a book authored by Karl Heinrich Bauer in 1924 titled Mutational Theory on the Origin of Cancers [see review by Edler and Kopp-Schneider (2005)]. Many of these ideas were codified by Berenblum and Shubik for dividing chemical carcinogenesis into two discrete stages, initiation and promotion (Berenblum and Shubik 1949). Initiation was defined as resulting from the single administration of an agent such as B[a]P at a dose that would not induce cancer. Promotion was defined as subsequent repeated administration of an agent [e.g., croton oil or its active component, tetradecanoyl phorbol acetate (TPA)] such as an irritant that by itself would not induce significant numbers of tumors. However, the combined administration of an initiator and a promoter was effective at tumor induction. These concepts were experimentally studied in depth by Boutwell and his colleagues at the McArdle Institute for Cancer Research in Madison, Wisconsin, U.S.A. (Boutwell 1974). This led to a series of characteristics that defined the two stages. Initiation was found to be irreversible, additive, and dose
6.3. THE ORIGIN OF MULTISTAGE CARCINOGENESIS
A X
No Tumors
B X
Tumors
C X
Tumors
D
X
183
No Tumors
E No Tumors F X
No Tumors
Time X = Application of Initiator
= Application of Promoter
Figure 6.8. A series of experiments using the mouse skin model that demonstrated the concepts of initiation and promotion adapted from Pitot (1986).
responsive. Mice treated with a single low dose of an initiator, such as a PAH, did not develop tumors after one year (Figure 6.8A). It has been shown that these mice had covalent DNA adducts that could lead to mutagenic events. Initiated cells in mouse skin could be promoted either immediately (generally twice weekly) after initiation (Figure 6.8B) or one year after initiation (Figure 6.8C), both protocols yielding tumors. Mice first treated with repeated doses of a promoter and then a single dose of an initiator did not develop tumors (Figure 6.8D). Promotion was reversible, nonadditive and possessed a threshold and saturation (Pitot 1986). Mice treated only with repeat doses of a promoter exhibited either no tumors or a low frequency of papillomas (Figure 6.8E). Promotion has been attributed to the clonal expansion of single initiated cells through changes in gene expression, altered gap junction intercellular communication, modified key receptor interactions, and altered cell proliferative responses. This requires the continued presence of the promoting agent because it is reversible upon withdrawal of the promoting agent. This was shown by treating mice with an initiator and then a promoter once every 4 weeks. No tumors were found (Figure 6.8F). The promotion response is assumed to fix the mutations induced by the tumor initiators. Tumor promotion in mouse skin was further stratified into two stages by Slaga et al. (1982, 1996). Based on the theory of multistage carcinogenesis, chemicals could be classified into initiators, promoters, and complete carcinogens [reviewed by Nesnow et al. (1983)]. A complete carcinogen is a chemical that is able to induce cancer itself; that is, it possesses properties of (a) initiation and promotion or (b) initiation, promotion, and progression. While the majority of studies on tumor promotion have focused on mouse skin, there are cancers in other organs that are applicable to this model. Bladder
184
CHAPTER 6 CHEMICAL CARCINOGENESIS
cancer (Hicks 1980), colon cancer (Pitot et al. 2000), and liver cancer (Pitot and Sirica 1980) are the major examples. The concepts of multistage carcinogenesis based on initiation, promotion, and progression have led in part to the current view that cancer is a multistage process due to series of mutations in oncogenes, tumor suppressor genes, and genetic instability genes. This was described by Fearon and Vogelstein, who proposed a model of colorectal carcinogenesis that correlated specific genetic events with changes in tissue morphology from normal tissue to evolving polyp formation and finally to an invasive cancer (Fearon and Vogelstein 1990). Tumor progression can be defined as a stage of neoplastic development that is characterized by additional mutations in tumor oncogenes and tumor suppressor genes leading to karyotype alterations, increased malignancy, metastases, and tumor aggressiveness (Pitot 2001). The most succinct summary of mouse skin initiation, promotion, and progression is found in Hennings et al. (1993) (Figure 6.9). Carcinogenesis in mouse skin can be divided into three distinct stages: initiation, promotion, and progression (malignant conversion). Initiation, induced by a single exposure to a genotoxic carcinogen, can result from a mutation in a single critical gene (e.g., rasHa), apparently in only a few epidermal cells. The change is irreversible. Promotion, resulting in the development of numerous benign tumors (papillomas), is accomplished by the repeated application of a nonmutagenic tumor promoter. The effects of single applications of tumor promoters are reversible since papillomas do not develop after insufficient exposure of initiated skin to promoters or when the interval between individual promoter applications is increased sufficiently. The reversibility of promotion suggests an epigenetic mechanism. Promoter treatment provides an environment that allows the selective clonal expansion of foci of initiated cells. The conversion of squamous papillomas to carcinomas (termed progression or malignant conversion) occurs spontaneously at a low frequency. The rate of progression to malignancy can be significantly increased by treatment of papilloma-bearing mice with certain genotoxic agents. These progressor agents or converting agents are likely to act via a second genetic change in papillomas already bearing the initiating mutation. Progression in the skin is characterized by genetic changes that result in several distinct changes in the levels or activity of structural proteins, growth factors, and proteases.” A more complex model of multistage carcinogenesis was proposed by Hanahan and Weinberg in their paper “The Hallmarks of Cancer” (Hanahan and Weinberg 2000). They proposed a series of discrete steps: (1) self-sufficiency in growth signals (one of the key characteristics of the tumor cell is its capacity for proliferation without dependence on external growth factors), (2) insensitivity to antigrowth signals (antigrowth signals must be avoided for cancer cells to survive and replicate), (3) tissue invasion and metastasis (cancer cells colonize distant sites to form metastases and overcome the normal suppressors of invasion), (4) limitless potential for
Initiation
Promotion
Progression
Figure 6.9. Multistage carcinogenesis describing the conversion of an initiated cell into malignant tumors adapted from Pitot (1986).
REFERENCES
185
replication (tumor cells must become immortal), (5) sustained angiogenesis (angiogenic ability must be acquired to increase tumor growth and size), and (6) evading apoptosis (cancer cells acquire resistance to apoptosis is to maintain proliferation). In summary, the development and maturation of the concepts of multistep and multistage chemical carcinogenesis have been intertwined with the study of PAH exposures, the concepts of metabolic activation, PAH tumorigenesis, and mechanisms of multistep mouse skin tumorigenesis. Much has been accomplished from the early beginnings of the chimney sweep epidemiological studies of Percival Pott in 1714, to our current understanding of the carcinogenesis process over the last 294 years. The mechanisms of chemical carcinogenesis and the development of human cancer are still, relatively speaking, “black boxes.” However, in the past 294 years these “black boxes” have been shrunk remarkably as new molecular techniques have been applied to these questions. New techniques such as Q-real time RT-PCR, genomics, proteomics, and metabolomics in combination with results from the human genome project will hopefully, in the next decades, remove these “black boxes” and give us a complete understanding of these important processes.
REFERENCES Balu, N., Padgett, W. T., Lambert, G. R., Swank, A. E., Richard, A. M., and Nesnow, S. (2004). Identification and characterization of novel stable deoxyguanosine and deoxyadenosine adducts of benzo[a]pyrene-7,8-quinone from reactions at physiological pH. Chem Res Toxicol 17, 827–838. Balu, N., Padgett, W. T., Nelson, G. B., Lambert, G. R., Ross, J. A., and Nesnow, S. (2006). Benzo[a] pyrene-7,8-quinone-3′-mononucleotide adduct standards for 32P postlabeling analyses: Detection of benzo[a]pyrene-7,8-quinone-calf thymus DNA adducts. Anal Biochem 355, 213–223. Bartczak, A. W., Sangaiah, R., Ball, L. M., Warren, S. H., and Gold, A. (1987). Synthesis and bacterial mutagenicity of the cyclopenta oxides of the four cyclopenta-fused isomers of benzanthracene. Mutagenesis 2, 101–105. Beach, A. C., and Gupta, R. C. (1994). DNA adducts of the ubiquitous environmental contaminant cyclopenta[cd]pyrene. Carcinogenesis 15, 1065–1072. Beland, F. A., and Kadlubar, F. F. (1985). Formation and persistence of arylamine DNA adducts in vivo. Environ Health Perspect 62, 19–30. Benhar, M., Engelberg, D., and Levitzki, A. (2002). ROS, stress-activated kinases and stress signaling in cancer. EMBO Rep 3, 420–425. Berenblum, I., and Shubik, P. (1949). A new quantitative approach to the study of the stages of chemical carcinogenesis in mouse skin. Br J Cancer 1, 383–390. Boutwell, R. K. (1974). The function and mechanism of promoters of carcinogenesis. CRC Crit Rev Toxicol 2, 419–443. Boyland, E., and Sims, P. (1964). Metabolism of polycyclic compounds. 24. The metabolism of benz[alpha]anthracene. Biochem J 91, 493–506. Buening, M. K., Levin, W., Karle, J. M., Yagi, H., Jerina, D. M., and Conney, A. H. (1979). Tumorigenicity of bay-region epoxides and other derivatives of chrysene and phenanthrene in newborn mice. Cancer Res 39, 5063–5068. Burczynski, M. E., and Penning, T. M. (2000). Genotoxic polycyclic aromatic hydrocarbon orthoquinones generated by aldo–keto reductases induce CYP1A1 via nuclear translocation of the aryl hydrocarbon receptor. Cancer Res 60, 908–915. Case, R. A., Hosker, M. E., Mc, D. D., and Pearson, J. T. (1954). Tumours of the urinary bladder in workmen engaged in the manufacture and use of certain dyestuff intermediates in the British chemical industry. I. The role of aniline, benzidine, alpha-naphthylamine, and beta-naphthylamine. Br J Ind Med 11, 75–104.
186
CHAPTER 6 CHEMICAL CARCINOGENESIS
Cavalieri, E., Rogan, E., and Sinha, D. (1988). Carcinogenicity of aromatic hydrocarbons directly applied to rat mammary gland. J Cancer Res Clin Oncol 114, 3–9. Cavalieri, E. L., Rogan, E. G., Devanesan, P. D., Cremonesi, P., Cerny, R. L., Gross, M. L., and Bodell, W. J. (1990). Binding of benzo[a]pyrene to DNA by cytochrome P-450 catalyzed one-electron oxidation in rat liver microsomes and nuclei. Biochemistry 29, 4820–4827. Chakravarti, D., Mailander, P., Franzen, J., Higginbotham, S., Cavalieri, E. L., and Rogan, E. G. (1998). Detection of dibenzo[a,l]pyrene-induced H-ras codon 61 mutant genes in preneoplastic SENCAR mouse skin using a new PCR-RFLP method. Oncogene 16, 3203–3210. Chakravarti, D., Mailander, P. C., Cavalieri, E. L., and Rogan, E. G. (2000). Evidence that error-prone DNA repair converts dibenzo[a,l]pyrene-induced depurinating lesions into mutations: Formation, clonal proliferation and regression of initiated cells carrying H-ras oncogene mutations in early preneoplasia. Mutat Res 456, 17–32. Chakravarti, D., Mailander, P. C., Li, K. M., Higginbotham, S., Zhang, H. L., Gross, M. L., Meza, J. L., Cavalieri, E. L., and Rogan, E. G. (2001). Evidence that a burst of DNA depurination in SENCAR mouse skin induces error-prone repair and forms mutations in the H-ras gene. Oncogene 20, 7945–7953. Chakravarti, D., Pelling, J. C., Cavalieri, E. L., and Rogan, E. G. (1995). Relating aromatic hydrocarboninduced DNA adducts and c-H-ras mutations in mouse skin papillomas: The role of apurinic sites. Proc Natl Acad Sci USA 92, 10422–10426. Chen, L., Devanesan, P. D., Higginbotham, S., Ariese, F., Jankowiak, R., Small, G. J., Rogan, E. G., and Cavalieri, E. L. (1996). Expanded analysis of benzo[a]pyrene-DNA adducts formed in vitro and in mouse skin: Their significance in tumor initiation. Chem Res Toxicol 9, 897–903. Chramostova, K., Vondracek, J., Sindlerova, L., Vojtesek, B., Kozubik, A., and Machala, M. (2004). Polycyclic aromatic hydrocarbons modulate cell proliferation in rat hepatic epithelial stem-like WBF344 cells. Toxicol Appl Pharmacol 196, 136–148. Cook, J. W., Hewitt, C. L., and Heiger, I. (1933). The isolation of a cancer producing hydrocarbon from coal tar. J Chem Soc, 395–405. Cremonesi, P., Rogan, E., and Cavalieri, E. (1992). Correlation studies of anodic peak potentials and ionization potentials for polycyclic aromatic hydrocarbons. Chem Res Toxicol 5, 346–355. Edler, L., and Kopp-Schneider, A. (2005). Origins of the mutational origin of cancer. Int J Epidemiol 34, 1168–1170. Fang, A. H., Smith, W. A., Vouros, P., and Gupta, R. C. (2001). Identification and characterization of a novel benzo[a]pyrene-derived DNA adduct. Biochem Biophys Res Commun 281, 383–389. Fearon, E. R., and Vogelstein, B. (1990). A genetic model for colorectal tumorigenesis. Cell 61, 759–767. Fibiger, J. A. G. (1913). Untersuchungen über eine Nematode Spiroptera sp. n.) und deren Fähigkeit papillomatöse und carcinomatöse Geschwulstbildungen im Magen der Ratte hervorzurufen. Z Krebsforschung 13, 217–280. Flowers-Geary, L., Bleczinki, W., Harvey, R. G., and Penning, T. M. (1996). Cytotoxicity and mutagenicity of polycyclic aromatic hydrocarbon ortho-quinones produced by dihydrodiol dehydrogenase. Chem Biol Interact 99, 55–72. Geacintov, N. E., Broyde, S., Buterin, T., Naegeli, H., Wu, M., Yan, S., and Patel, D. J. (2002). Thermodynamic and structural factors in the removal of bulky DNA adducts by the nucleotide excision repair machinery. Biopolymers 65, 202–210. Gold, A., and Eisenstadt, E. (1980). Metabolic activation of cyclopenta(cd)pyrene to 3,4-epoxycyclopenta(cd) pyrene by rat liver microsomes. Cancer Res 40, 3940–3944. Gopalakrishna, R., and Jaken, S. (2000). Protein kinase C signaling and oxidative stress. Free Radic Biol Med 28, 1349–1361. Grimmer, G., Brune, H., Deutsch-Wenzel, R., Dettbarn, G., Jacob, J., Naujack, K. W., Mohr, U., and Ernst, H. (1987). Contribution of polycyclic aromatic hydrocarbons and nitro-derivatives to the carcinogenic impact of diesel engine exhaust condensate evaluated by implantation into the lungs of rats. Cancer Lett 37, 173–180. Hanahan, D., and Weinberg, R. A. (2000). The hallmarks of cancer. Cell 100, 57–70. Hanson, A. A., Rogan, E. G., and Cavalieri, E. L. (1998). Synthesis of adducts formed by iodine oxidation of aromatic hydrocarbons in the presence of deoxyribonucleosides and nucleobases. Chem Res Toxicol 11, 1201–1208.
REFERENCES
187
Hegstad, S., Lundanes, E., Holme, J. A., and Alexander, J. (1999). Characterization of metabolites of benz(j)aceanthrylene in faeces, urine and bile from rat. Xenobiotica 29, 1257–1272. Heidelberger, C. (1975). Chemical carcinogenesis. Annu Rev Biochem 44, 79–121. Hemminki, K., Perera, F. P., Phillips, D. H., Randerath, K., Reddy, M. V., and Santella, R. M. (1988). Aromatic deoxyribonucleic acid adducts in white blood cells of foundry and coke oven workers. Scand J Work Environ Health 14(Suppl 1), 55–56. Hennings, H., Glick, A. B., Greenhalgh, D. A., Morgan, D. L., Strickland, J. E., Tennenbaum, T., and Yuspa, S. H. (1993). Critical aspects of initiation, promotion, and progression in multistage epidermal carcinogenesis. Proc Soc Exp Biol Med 202, 1–8. Hicks, R. M. (1980). Multistage carcinogenesis in the urinary bladder. Br Med Bull 36, 39–46. Hsu, C. H., Skipper, P. L., and Tannenbaum, S. R. (1999). DNA adduct formation by secondary metabolites of cyclopenta[cd]pyrene in vitro. Cancer Lett 136, 137–141. Jerina, D. M., Lehr, R., Yagi, H., Hernandez, O., Dansette, P., Wislocki, P. G., Wood, A. W., Chang, R. L., Levin, W., and Conney, A. H. (1976). Mutagenicity of benzo(a)pyrene derivatives from the description of a quantum mechanical model which predicts the ease of carbonium ion formation from diol epoxides. In In Vitro Metabolic Activation and Mutagenesis Testing, (deSerres, F. J., Fouts, J. R., Bend, J. R., and Philpot, R. M., eds., Elsevier/North-Holland Biomedical, Amsterdam, pp. 159–177. Jerina, D. M., Sayer, J. M., Agarwal, S. K., Yagi, H., Levin, W., Wood, A. W., Conney, A. H., PruessSchwartz, D., Baird, W. M., Pigott, M. A., et al. (1986). Reactivity and tumorigenicity of bay-region diol epoxides derived from polycyclic aromatic hydrocarbons. Adv Exp Med Biol 197, 11–30. Johnsen, N. M., Brunborg, G., Haug, K., Scholz, T., and Holme, J. A. (1998a). Metabolism and activation of cyclopenta polycyclic aromatic hydrocarbons in isolated human lymphocytes, HL-60 cells and exposed rats. Chem Biol Interact 114, 77–95. Johnsen, N. M., Nyholm, S. H., Haug, K., Scholz, T., and Holme, J. A. (1998b). Metabolism and activation of cyclopenta polycyclic aromatic hydrocarbons in liver tissue from rats and humans. Chem Biol Interact 113, 217–237. Katz, A. K., Carrell, H. L., and Glusker, J. P. (1998). Dibenzo[a,l]pyrene (dibenzo[def,p]chrysene): fjord-region distortions. Carcinogenesis 19, 1641–1648. Kennaway, E. (1955). Identification of carcinogenic compound in coal tar. Br Med J 2, 749–752. Kohan, M. J., Sangaiah, R., Ball, L. M., and Gold, A. (1985). Bacterial mutagenicity of aceanthrylene: a novel cyclopenta-fused polycyclic aromatic hydrocarbon of low molecular weight. Mutat Res 155, 95–98. Kwon, H., Sahali, Y., Skipper, P. L., and Tannenbaum, S. R. (1992). Oxidation of cyclopenta[cd]pyrene by human and mouse liver microsomes and selected cytochrome P450 enzymes. Chem Res Toxicol 5, 760–764. Levin, W., Thakker, D. R., Wood, A. W., Chang, R. L., Lehr, R. E., Jerina, D. M., and Conney, A. H. (1978). Evidence that benzo(a)anthracene 3,4-diol-1,2-epoxide is an ultimate carcinogen on mouse skin. Cancer Res 38, 1705–1710. Lewis-Bevan, L., Little, S. B., and Rabinowitz, J. R. (1995). Quantum mechanical studies of the structure and reactivities of the diol epoxides of benzo[c]phenanthrene. Chem Res Toxicol 8, 499–505. Lodovici, M., Akpan, V., Giovannini, L., Migliani, F., and Dolara, P. (1998). Benzo[a]pyrene diolepoxide DNA adducts and levels of polycyclic aromatic hydrocarbons in autoptic samples from human lungs. Chem Biol Interact 116, 199–212. Melendez-Colon, V. J., Luch, A., Seidel, A., and Baird, W. M. (1999a). Cancer initiation by polycyclic aromatic hydrocarbons results from formation of stable DNA adducts rather than apurinic sites. Carcinogenesis 20, 1885–1891. Melendez-Colon, V. J., Luch, A., Seidel, A., and Baird, W. M. (1999b). Comparison of cytochrome P450- and peroxidase-dependent metabolic activation of the potent carcinogen dibenzo[a,l]pyrene in human cell lines: formation of stable DNA adducts and absence of a detectable increase in apurinic sites. Cancer Res 59, 1412–1416. Melendez-Colon, V. J., Luch, A., Seidel, A., and Baird, W. M. (2000). Formation of stable DNA adducts and apurinic sites upon metabolic activation of bay and fjord region polycyclic aromatic hydrocarbons in human cell cultures. Chem Res Toxicol 13, 10–17. Melendez-Colon, V. J., Smith, C. A., Seidel, A., Luch, A., Platt, K. L., and Baird, W. M. (1997). Formation of stable adducts and absence of depurinating DNA adducts in cells and DNA treated
188
CHAPTER 6 CHEMICAL CARCINOGENESIS
with the potent carcinogen dibenzo[a,l]pyrene or its diol epoxides. Proc Natl Acad Sci USA 94, 13542–13547. Miller, E. C., and Miller, J. A. (1979). Milestones in chemical carcinogenesis. Semin Oncol 6, 445–460. Miller, J. A., Cramer, J. W., and Miller, E. C. (1960). The N- and ringhydroxylation of 2acetylaminofluorene during carcinogenesis in the rat. Cancer Res 20, 950–962. Mohapatra, N., MacNair, P., Bryant, B. J., Ellis, S., Rudo, K., Sangaiah, R., Gold, A., and Nesnow, S. (1987). Morphological transforming activity and metabolism of cyclopenta-fused isomers of benz[a] anthracene in mammalian cells. Mutat Res 188, 323–334. Mumford, J. L., Lee, X., Lewtas, J., Young, T. L., and Santella, R. M. (1993). DNA adducts as biomarkers for assessing exposure to polycyclic aromatic hydrocarbons in tissues from Xuan Wei women with high exposure to coal combustion emissions and high lung cancer mortality. Environ Health Perspect 99, 83–87. Nesnow, S., Easterling, R. E., Ellis, S., Watts, R., and Ross, J. (1988). Metabolism of benz[j]aceanthrylene (cholanthrylene) and benz[l]aceanthrylene by induced rat liver S9. Cancer Lett 39, 19–27. Nesnow, S., Lasley, J., Curti, S., Ross, J., Nelson, G., Sangaiah, R., and Gold, A. (1991). Morphological transformation and DNA adduct formation by benz[j]aceanthrylene and its metabolites in C3H10T1/2CL8 cells: Evidence for both cyclopenta-ring and bay-region metabolic activation pathways. Cancer Res 51, 6163–6169. Nesnow, S., Leavitt, S., Easterling, R., Watts, R., Toney, S. H., Claxton, L., Sangaiah, R., Toney, G. E., Wiley, J., Fraher, P., et al. (1984). Mutagenicity of cyclopenta-fused isomers of benz(a)anthracene in bacterial and rodent cells and identification of the major rat liver microsomal metabolites. Cancer Res 44, 4993–5003. Nesnow, S., Mass, M. J., Ross, J. A., Galati, A. J., Lambert, G. R., Gennings, C., Carter, W. H., Jr., and Stoner, G. D. (1998). Lung tumorigenic interactions in strain A/J mice of five environmental polycyclic aromatic hydrocarbons. Environ Health Perspect 106(Suppl 6), 1337–1346. Nesnow, S., Padgett, W., Balu, N., Nelson, G., Winnik, W., Lambert, G., George, M. A., and Ross, J. S. (2005). Benzo[a]pyrene-7,8-quinone forms covalent-DNA adducts in vitro but none were detected in the lungs or livers of strain A/J mice in vivo. In ISPAC 16, Trondheim, NO. Nesnow, S., Ross, J., Beck, S., Lasley, J., Nelson, G., Lambert, G., Platt, K. L., and Agarwal, S. C. (1994a). Morphological transformation and DNA adduct formation by dibenz[a,h]anthracene and its metabolites in C3H10T1/2CL8 cells. Carcinogenesis 15, 2225–2231. Nesnow, S., Ross, J. A., Nelson, G., Wilson, K., Roop, B. C., Jeffers, A. J., Galati, A. J., Stoner, G. D., Sangaiah, R., Gold, A., et al. (1994b). Cyclopenta[cd]pyrene-induced tumorigenicity, Ki-ras codon 12 mutations and DNA adducts in strain A/J mouse lung. Carcinogenesis 15, 601–606. Nesnow, S., Triplett, L. L., and Slaga, T. J. (1983). Mouse skin tumor initiation-promotion and complete carcinogenesis bioassays: Mechanisms and biological activities of emission samples. Environ Health Perspect 47, 255–268. Nyholm, S. H., Alexander, J., Lundanes, E., Frandsen, H., Andersson, R., Grivas, S., Nesnow, S., and Holme, J. A. (1996). Biotransformation of the cyclopenta-fused polycyclic aromatic hydrocarbon benz[j]aceanthrylene in isolated rat liver cell: Identification of nine new metabolites. Carcinogenesis 17, 1111–1120. Osborne, M. R., and Crosby, N. T. (1987). Benzopyrenes, Cambridge University Press, Cambridge, England. Palackal, N. T., Burczynski, M. E., Harvey, R. G., and Penning, T. M. (2001). The ubiquitous aldehyde reductase (AKR1A1) oxidizes proximate carcinogen trans-dihydrodiols to o-quinones: Potential role in polycyclic aromatic hydrocarbon activation. Biochemistry 40, 10901–10910. Park, J. H., Gopishetty, S., Szewczuk, L. M., Troxel, A. B., Harvey, R. G., and Penning, T. M. (2005). Formation of 8-oxo-7,8-dihydro-2′-deoxyguanosine (8-oxo-dGuo) by PAH o-quinones: Involvement of reactive oxygen species and copper(II)/copper(I) redox cycling. Chem Res Toxicol 18, 1026–1037. Park, J. H., Mangal, D., Tacka, K. A., Quinn, A. M., Harvey, R. G., Blair, I. A., and Penning, T. M. (2008). Evidence for the aldo–keto reductase pathway of polycyclic aromatic trans-dihydrodiol activation in human lung A549 cells. Proc Natl Acad Sci USA 105, 6846–6851.
REFERENCES
189
Pavanello, S., Favretto, D., Brugnone, F., Mastrangelo, G., Dal Pra, G., and Clonfero, E. (1999). HPLC/ fluorescence determination of anti-BPDE–DNA adducts in mononuclear white blood cells from PAHexposed humans. Carcinogenesis 20, 431–435. Penning, T. M., Ohnishi, S. T., Ohnishi, T., and Harvey, R. G. (1996). Generation of reactive oxygen species during the enzymatic oxidation of polycyclic aromatic hydrocarbon trans-dihydrodiols catalyzed by dihydrodiol dehydrogenase. Chem Res Toxicol 9, 84–92. Perera, F. P., Hemminki, K., Young, T. L., Brenner, D., Kelly, G., and Santella, R. M. (1988). Detection of polycyclic aromatic hydrocarbon–DNA adducts in white blood cells of foundry workers. Cancer Res 48, 2288–2291. Pitot, H. C. (1986). Fundamentals of Oncology, Marcel Dekker, New York. Pitot, H. C. (2001). Pathways of progression in hepatocarcinogenesis. Lancet 358, 859–860. Pitot, H. C., Hikita, H., Dragan, Y., Sargent, L., and Haas, M. (2000). Review article: The stages of gastrointestinal carcinogenesis—Application of rodent models to human disease. Aliment Pharmacol Ther 14(Suppl 1), 153–160. Pitot, H. C., and Sirica, A. E. (1980). The stages of initiation and promotion in hepatocarcinogenesis. Biochim Biophys Acta 605, 191–215. Platt, K. L., and Schollmeier, M. (1994). Bisdihydrodiols, rather than dihydrodiol oxides, are the principal microsomal metabolites of tumorigenic trans-3,4-dihydroxy-3,4-dihydrodibenz[a,h]anthracene. Chem Res Toxicol 7, 89–97. Pott, P. (1775). Chirugical Observations Relative to the Cataract, the Polypus of the Nose, the Cancer of the Scrotum, the Different Kinds of Ruptures, and the Mortification of Toes and Feet, L. Hawkes, W. Clarke, & R. Collins, London. Prahalad, A. K., Ross, J. A., Nelson, G. B., Roop, B. C., King, L. C., Nesnow, S., and Mass, M. J. (1997). Dibenzo[a,l]pyrene-induced DNA adduction, tumorigenicity, and Ki-ras oncogene mutations in strain A/J mouse lung. Carcinogenesis 18, 1955–1963. RamaKrishna, N. V., Gao, F., Padmavathi, N. S., Cavalieri, E. L., Rogan, E. G., Cerny, R. L., and Gross, M. L. (1992). Model adducts of benzo[a]pyrene and nucleosides formed from its radical cation and diol epoxide. Chem Res Toxicol 5, 293–302. Ramet, M., Castren, K., Jarvinen, K., Pekkala, K., Turpeenniemi-Hujanen, T., Soini, Y., Paakko, P., and Vahakangas, K. (1995). p53 protein expression is correlated with benzo[a]pyrene–DNA adducts in carcinoma cell lines. Carcinogenesis 16, 2117–2124. Rehn, L. (1895). Ueber Blasentrumoren bei Fruchsinalbeitern. Arch Kind Chir 50, 588. Rodriguez, H., and Loechler, E. L. (1995). Are base substitution and frameshift mutagenesis pathways interrelated? An analysis based upon studies of the frequencies and specificities of mutations induced by the (+)-anti diol epoxide of benzo[a]pyrene. Mutat Res 326, 29–37. Rojas, M., Alexandrov, K., Auburtin, G., Wastiaux-Denamur, A., Mayer, L., Mahieu, B., Sebastien, P., and Bartsch, H. (1995). Anti-benzo[a]pyrene diolepoxide–DNA adduct levels in peripheral mononuclear cells from coke oven workers and the enhancing effect of smoking. Carcinogenesis 16, 1373–1376. Ross, J. A., Nelson, G. B., Wilson, K. H., Rabinowitz, J. R., Galati, A., Stoner, G. D., Nesnow, S., and Mass, M. J. (1995). Adenomas induced by polycyclic aromatic hydrocarbons in strain A/J mouse lung correlate with time-integrated DNA adduct levels. Cancer Res 55, 1039–1044. Ross, J. A., and Nesnow, S. (1999). Polycyclic aromatic hydrocarbons: Correlations between DNA adducts and ras oncogene mutations. Mutat Res 424, 155–166. Rous, P., and Kidd, J. G. (1941). Conditional neoplasma and subthreshold neoplastic states: A study of tar tumors in rabbits. J Exp Med 73, 365–389. Ruan, Q., Kim, H. Y., Jiang, H., Penning, T. M., Harvey, R. G., and Blair, I. A. (2006). Quantification of benzo[a]pyrene diol epoxide DNA-adducts by stable isotope dilution liquid chromatography/tandem mass spectrometry. Rapid Commun Mass Spectrom 20, 1369–1380. Ruggeri, B., DiRado, M., Zhang, S. Y., Bauer, B., Goodrow, T., and Klein-Szanto, A. J. (1993). Benzo[a] pyrene-induced murine skin tumors exhibit frequent and characteristic G to T mutations in the p53 gene. Proc Natl Acad Sci USA 90, 1013–1017. Slaga, T. J., Budunova, I. V., Gimenez-Conti, I. B., and Aldaz, C. M. (1996). The mouse skin carcinogenesis model. J Investig Dermatol Symp Proc 1, 151–156.
190
CHAPTER 6 CHEMICAL CARCINOGENESIS
Slaga, T. J., Fischer, S. M., Weeks, C. E., Klein-Szanto, A. J., and Reiners, J. (1982). Studies on the mechanisms involved in multistage carcinogenesis in mouse skin. J Cell Biochem 18, 99–119. Smith, L. E., Denissenko, M. F., Bennett, W. P., Li, H., Amin, S., Tang, M., and Pfeifer, G. P. (2000). Targeting of lung cancer mutational hotspots by polycyclic aromatic hydrocarbons. J Natl Cancer Inst 92, 803–811. Smithgall, T. E., Harvey, R. G., and Penning, T. M. (1986). Regio- and stereospecificity of homogeneous 3-alpha-hydroxysteroid-dihydrodiol dehydrogenase for trans-dihydrodiol metabolites of polycyclic aromatic hydrocarbons. J Biol Chem 261, 6184–6191. Straif, K., Baan, R., Grosse, Y., Secretan, B., El Ghissassi, F., and Cogliano, V. (2005). Carcinogenicity of polycyclic aromatic hydrocarbons. Lancet Oncol 6, 931–932. Surh, Y. J., Kwon, H., and Tannenbaum, S. R. (1993). Sulfotransferase-mediated activation of 4hydroxy- and 3,4-dihydroxy-3,4-dihydrocyclopenta[c,d]pyrene, major metabolites of cyclopenta[c,d] pyrene. Cancer Res 53, 1017–1022. Vulimiri, S. V., Baer-Dubowska, W., Harvey, R. G., Zhang, J. T., and DiGiovanni, J. (1999). Analysis of highly polar DNA adducts formed in SENCAR mouse epidermis following topical application of dibenz[a,j]anthracene. Chem Res Toxicol 12, 60–67. Watanabe, M., Maher, V. M., and McCormick, J. J. (1985). Excision repair of UV- or benzo[a]pyrene diol epoxide-induced lesions in xeroderma pigmentosum variant cells is “error free.” Mutat Res 146, 285–294. Weyand, E. H., Cai, Z. W., Wu, Y., Rice, J. E., He, Z. M., and LaVoie, E. J. (1993). Detection of the major DNA adducts of benzo[b]fluoranthene in mouse skin: Role of phenolic dihydrodiols. Chem Res Toxicol 6, 568–577. Williams, G. M. (2001). Mechanisms of chemical carcinogenesis and application to human cancer risk assessment. Toxicology 166, 3–10. Wislocki, P. G., Buening, M. K., Levin, W., Lehr, R. E., Thakker, D. R., Jerina, D. M., and Conney, A. H. (1979). Tumorigenicity of the diastereomeric benz[a]anthracene 3,4-diol-1,2-epoxides and the (+)- and (−)-enantiomers of benz[a]anthracene 3,4-dihydrodiol in newborn mice. J Natl Cancer Inst 63, 201–204. Xue, W., and Warshawsky, D. (2005). Metabolic activation of polycyclic and heterocyclic aromatic hydrocarbons and DNA damage: A review. Toxicol Appl Pharmacol 206, 73–93. Yamagiwa, K., and Ichikawa, K. (1915). Experimentelle Studie über die Pathogenese der Epithelialgeschwülste. Mitt Med Fac Kaiserl Univ Tokyo 15, 295–344.
CH A P TE R
7
HORMESIS AND CANCER RISKS: ISSUES AND RESOLUTION Paolo F. Ricci Edward J. Calabrese
7.1.
INTRODUCTION
This chapter aims to discuss of cancer risks and hormesis within the U.S. regulatory risk assessment practices, which legally require the characterization of causal relations between exposures or doses and responses to quantify the probability of cancer response. Causal reasoning characterizes environmental decision-making (EPA 2005; statement in square brackets added) because: The extent of health protection provided to the public ultimately depends upon what risk managers decide is the appropriate course of regulatory action. … When there are alternative procedures having significant biological support [one of which being the existence of hormetic mechanisms], the Agency encourages assessments to be performed using these alternative procedures, if feasible, in order to shed light on the uncertainties in the assessment, recognizing that the Agency may decide to give greater weight to one set of procedures than another in a specific assessment or management decision.
Managing risks, and deciding on the appropriate causal model, is an essential component of both public and private risk decision-making. It follows that current cancer risk assessments must reflect the state-of-science and rely as little as possible on conjecture. Specifically, the U.S. Environmental Protection Agency (EPA) (EPA 2005) adds that: Encouraging risk assessors to be receptive to new scientific information, NRC discussed the need for departures from default options when a “sufficient showing” is made. It called on EPA to articulate clearly its criteria for a departure so that decisions to depart from default options would be “scientifically credible and receive public acceptance”. … It was concerned that ad hoc departures would undercut the scientific credibility of a risk assessment. U.S. National Research Council (NRC) envisioned that principles for choosing and departing from default options would balance several objectives, including “protecting the public health, ensuring scientific validity,
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
191
192
CHAPTER 7 HORMESIS AND CANCER RISKS: ISSUES AND RESOLUTION
minimizing serious errors in estimating risks, maximizing incentives for research, creating an orderly and predictable process, and fostering openness and trustworthiness.” …
Although guidelines are unenforceable, they are mountains very few attempt to climb because of cost and the truism that one cannot fight City Hall—unless one is able and willing to spend years in litigation. A well-supported reason for recalibrating the credibility and soundness of decisions can be provided in the public interest and address the question, What is a sufficient evidentiary showing? A premise of this discussion is that cancer is a multifactorial disease that has several endpoints—not just the tumor growth itself (primary or metastatic). For example, although cancer can be a solid tumorigenic cellular mass (e.g., a carcinoma), its effects on a living organism go well beyond the abnormal cellular mass (e.g., the tumor can cause cachexia). The justification for any default regulatory model must be based on the weight of the evidence, understood as the collection of outcomes related to exposure and that manifest themselves as adverse endpoints associated with a specific cancer, and not merely with the observation of the tumor itself. Sufficiency (in the context of necessary and sufficient logical expressions) is demonstrated by empirical and theoretical arguments. The EPA arguments about default causal models (i.e., the linear, nonthreshold at low doses cancer dose–response model) is a logical fallacy: Scientific conjecture trumps facts. This chapter deals with correcting the use of conjectures as defaults in regulatory policy, in the context of experimental evidence of hormesis and causation and alternative probabilistic cancer models. Specifically, we summarize how the combination of mode-of-action and weight-of-evidence supports both J-shaped and U-shaped, rather than the linear, no-threshold (LNT) models. The EPA uses the terms nonlinear for the threshold model and low-dose-linear for the LNT models (meaning that the slope is greater than zero at zero dose), which is well-approximated by a straight line, at very low doses and beginning from zero dose (EPA 2005). The totality of the scientific evidence for a causal default—a fundamental dose–response model, given the state-of-science—now discounts conjectural arguments (the linear, at low-dose, nonthreshold model) or arbitrary ones, such as those based on extrapolation (the threshold model) because both of them eliminate a very large number of experimentally observed health benefits. According to the EPA, the use of defaults is a subjective choice (EPA 2005). As the EPA states: Generally, if a gap in basic understanding exists or if agent-specific information is missing, a default option may be used. If agent-specific information is present but critical analysis reveals inadequacies, a default option may also be used. If critical analysis of agent-specific information is consistent with one or more biologically based models as well as with the default option, the alternative models and the default option are both carried through the assessment and characterized for the risk manager. In this case, the default model not only fits the data, but also serves as a benchmark for comparison with other analyses.
But subjectivism raises the paradox: Poorly understood causation requires more, not less, knowledge. Thus, when default dose–response models become the “benchmark” for comparing results with other models, how can the decision be
7.1. INTRODUCTION
193
properly calibrated? Default-based reasoning is not justifiable—contrary to assertions about being precautionary—when experimentally valid alternatives such as the J-shaped cancer model is available and is superior to default-based argumentations: It puts the cart before the horse. A conjecture cannot serve as a benchmark, particularly when that conjecture cannot be demonstrated at the low dose levels generally of concern to regulators [where the individual lifetime excess cancer risks (probability of response) are between 10−4 and 10−6]. Cancer risks are regulated though guidelines (EPA 2005, citations omitted) where: [the u]se health protective risk assessment procedures … means that estimates, while uncertain, are more likely to overstate than understate hazard and/or risk. NRC (1994) reaffirmed the use of default options as “a reasonable way to cope with uncertainty about the choice of appropriate models or theory.”
This combination of scientific analyses and choices (appropriate models) with policy judgment (reasonable way) is fraught with danger. First, causal defaults are conjectured models: For risk management, they are based either on a single model of causation, such as the linear no-threshold (LNT) dose–response model, or on other models that are linear at low dose and originate at the (0, 0) point on the dose– response axis. Second, their combination affects the optimality of risk management choices by inducing unknown and serious differences between the actual and conjectured risks. The LNT conjecture has the distinct disadvantage of negating any potential health benefits; and so does the threshold model. Society pays heavy economic direct and indirect costs from actions designed to avoid a nonexistent danger that, paradoxically, is a benefit. A seemingly plausible reason for default options to be a reasonable way to cope with uncertainty about the choice of appropriate models lies within the proper concern with meeting the ethical principle that it is better to be safe rather than sorry. It is preferable to be conservative when uncertainties are large and when the magnitude of the potential harm is great or dreaded. For example, according to the U.S. Office of Management and Budget (OMB) (2003): The United States employs precautionary approaches throughout the process of risk assessment and management so that the overall level of precaution in a given regulatory decision is appropriate … When analysts assess risks, they frequently use “conservative” or “default” assumptions or explicitly add safety margins or uncertainty factors to characterize a “plausible” upper bound.
But this unimpeachable precautionary approach, as applied, is illogical. It relies on two models that deny proven direct benefits and fails to meet the very reason for its formulation. Moreover, while fundamentally agreeing with the ethical basis of managing cancer risks by being precautionary, the OMB begs the issue that lies within its statement: Is the causal default scientifically sound, given the best and current state of science? In other words, what if conservative defaults cause more harm than good, when applied to regulate exposure? To answer these questions, we first develop a causal network for risk analysis and management. Figure 7.1 depicts the relationship between possible risk management acts that can decrease the incidence of cancer, taking into account other risk factors in the population.
194
CHAPTER 7 HORMESIS AND CANCER RISKS: ISSUES AND RESOLUTION
act → Δ(exposure) → Δ(cancer incidence) ↑ ↑ [behaviors] [susceptibility] Figure 7.1. A causal graph for risk analysis. The model depicted in this figure can be formalized using a Bayesian network (Ricci et al. 2006): A probabilistic framework interprets the model described in this figure as a Bayesian belief network or causal graph model. Each variable with inward-pointing arrows is interpreted as a random variable with a conditional probability distribution that depends only on the values of the variables that point into it. The essence of this approach to modeling and evaluating uncertain risks is to sample successively from the (often conditional) distribution of each variable, given the values of its predecessors. Algorithms exist to identify and validate possible causal structures.
Figure 7.1 also depicts changes via behaviors, such as occupation, ambient exposure, and predisposition, such as genetic. Logically, it is correct regardless of the shape of the dose–response model. At low dose or at environmental (ambient) exposures, cancer risk assessment models used in regulatory law are either linear or linearized; that is, each is a cumulative distribution function of lifetime cancer risk and thus is a monotonic function. Hormetic cancer dose–response models are also probabilistic; however, they are nonmonotonic (they are relations). The EPA summarizes the reasons for using statistical and probabilistic methods in risk assessment as follows (EPA 2005): The main aim of statistical evaluation is to determine whether exposure to the test agent is associated with an increase of tumor development …. A statistically significant response may or may not be biologically significant and vice versa. The selection of a significance level is a policy choice based on a trade-off between the risks of false positives and false negatives. A result with a significance level of greater or less than 5% (the most common significance level) is examined to see if the result confirms other scientific information ….
This argument leads to a central point of this chapter. That is, what is the evidence necessary and sufficient for a finding of regulatory causation, in the context of the LNT and the hormetic cancer dose–response models? Legal answers to this question were given by Ricci and Molton (1981) and then placed in the context of international tort and environmental law by Ricci and Gray (1998). The following sections focus narrowly on scientific evidence in the context of regulatory science.
7.2. EVIDENCE FOR REGULATORY CANCER RISK ASSESSMENT In U.S. regulatory science, the scientific evidence most relevant to assessing risky outcomes consists of results from animal and human studies. Although there are many other tests, such as in vitro tests using cell lines, the results from those studies are not generally used by regulatory agencies for causal arguments leading to either
7.2. EVIDENCE FOR REGULATORY CANCER RISK ASSESSMENT
195
a guideline or a standard. Regarding empirical studies and causation, the EPA states the following (EPA 2005): Determining whether an observed association (risk) is causal rather than spurious involves consideration of a number of factors. Sir Bradford Hill (Hill 1965) developed a set of guidelines for evaluating epidemiologic associations that can be used in conjunction with the discussion of causality ….
Thus, the EPA opts for a causal analysis (that parallels and updates Hill’s initial work consisting of nine criteria) for judging causation (EPA 2005): (a) Consistency of the Observed Association. An inference of causality is strengthened when a pattern of elevated risks is observed across several independent studies. The reproducibility of findings is one of the strongest arguments for causality. If there are discordant results among investigations, possible reasons such as differences in exposure, confounding factors, and the power of the study are considered. (b) Strength of the Observed Association. The finding of large, precise risks increases confidence that the association is not likely due to chance, bias, or other factors. (c) Specificity of the Observed Association. Based on our current understanding that many agents cause cancer at multiple sites, and many cancers have multiple causes, this is now considered one of the weaker guidelines for causality. Thus, although the presence of specificity may support causality, its absence does not exclude it. (d) Temporal Relationship of the Observed Association. A causal interpretation is strengthened when exposure is known to precede development of the disease. This is among the strongest criteria for an inference of causality. (e) Biological Gradient (Exposure–Response Relationship). A clear exposure– response relationship (e.g., increasing effects associated with greater exposure) strongly suggests cause and effect, especially when such relationships are also observed for duration of exposure (e.g., increasing effects observed following longer exposure times). (f) Biological Plausibility. An inference of causality tends to be strengthened by consistency with data from experimental studies or other sources demonstrating plausible biological mechanisms. A lack of mechanistic data, however, is not a reason to reject causality. (g) Coherence. An inference of causality may be strengthened by other lines of evidence that support a cause-and-effect interpretation of the association. Information is considered from animal bioassays, toxicokinetic studies, and short-term studies. The absence of other lines of evidence, however, is not a reason to reject causality. (h) Experimental Evidence (from Human Populations). Strong evidence of causality can be provided when a change in exposure brings about a change in disease frequency—for example, the decrease in the risk of lung cancer that follows cessation of smoking.
196
CHAPTER 7 HORMESIS AND CANCER RISKS: ISSUES AND RESOLUTION
(i) Analogy. Information on mode of action for a chemical, as one of many structural analogues, can inform decisions regarding likely causality. None of these nine criteria (individually and in their aggregate) is antithetic to either the J- or inverted J-shaped dose–response model that characterizes hormetic cancer models and toxic effects related to cancer. These, as all criteria of causation we are familiar with, fully support the empirical and biological basis of hormetic causation. But, they do not and cannot support regulatory standards or guidelines based on conjectures. This view is beginning to take hold. For example, the EPA states (Draft in review, Arsenic, Advisory on EPA’s Assessments of Carcinogenic Effects of Organic and Inorganic Arsenic: An Advisory Report of the EPA Science Advisory Board, Dec. 27, 2005) the following (citations omitted for brevity): One cannot dismiss the possibilities of hormesis effects in humans exposed to low-dose arsenic or the essentiality of arsenic to humans. Evidence for essentiality of arsenic has been reported for a number of mammalian species as well as for chickens. These may explain some of the apparent low-dose benefits seen in a variety of systems …. Low concentrations fuel angiogenesis, while higher concentrations injure endothelial cells and promote the vessels dysfunction seen in ischemic diseases and peripheral vascular diseases …. However, arsenic at high doses has been used to destroy the tumor vasculature. If arsenic is essential for humans and/or if epidemiological data could be strengthened at the low-dose range to demonstrate either a low-dose benefit or no effect at low dose, then a threshold is certain. However, at this time, the data are lacking or problematic with regard to low-dose effects. This is an extremely important issue and should be investigated.
The concept of hormesis has been substantially documented with many thousands of studies having passed peer review in numerous journals over multiple decades. A concerted effort has been made to subject each of the possible examples of a hormetic response to the same rigorous a priori evaluative criteria. Those articles passing the a priori hormesis review criteria have been entered into an extensive database (Calabrese and Blain 2005), with many becoming integratively synthesized into comprehensive biomedical and toxicological reviews, including mutagens and carcinogens (Calabrese and Baldwin 1999), toxic metals (Calabrese and Baldwin 2003a; Calabrese and Blain 2004), chemotherapeutics (Calabrese and Baldwin 2003b), reproductive toxins (Calabrese and Baldwin 2000), neuroprotective agents (Calabrese 2008a), growing neurons (Calabrese 2008b), pain (Calabrese 2008c), memory-enhancing agents (Calabrese 2008d), stress (Calabrese 2008e), pglycoprotein membrane efflux systems (Calabrese 2008f), stroke medications (Calabrese 2008g), anxiolytic drugs (Calabrese 2008h), anti-seizure drugs (Calabrese 2008i), chemical and radiation immune stimulatory responses (Calabrese 2005), chemo-attractants and the effects of numerous natural synthetic agonists—for example, estrogens (Calabrese 2001a), androgens (Calabrese 2001b), dopamine (Calabrese 2001c), serotonin (Calabrese 2001d), nitric oxide (Calabrese 2001e), opiates (Calabrese 2001f), prostaglandins (Calabrese 2001g), and adrenergic agonists (Calabrese 2001h), and many others. These findings reveal that the hormetic concept is generalizable, being independent of biological model, endpoint measured and chemical class, and/or physical stressor agent.
7.2. EVIDENCE FOR REGULATORY CANCER RISK ASSESSMENT
197
The data supporting the hormetic dose–response therefore not only far exceeds normal “proof of concept” criteria but have been employed in the development of drugs for numerous human conditions, thereby satisfying a stronger “proof” of application requirements. In head-to-head direct comparisons the hormesis model far outcompeted the threshold and linear at low dose models (Calabrese and Baldwin 2001, 2003c; Calabrese et al. 2006). In fact, while the threshold model was shown to poorly predict below threshold responses the linear at low-dose models, the LNT models cannot be practically validated in either moderate or large-scale studies. For example, in the mid-1970s, the U.S. Food and Drug Administration (FDA) conducted a long-term study with over 24,000 mice exposed to the carcinogen 2-acetylaminofluorene (AAF) to determine the nature of the dose–response in the low-dose zone (because of its magnitude, this study became known as the megamouse experiment). Despite the very large (and yet to be matched in size) number of animals, the estimated cancer risk was only sensitive for a risk of 1/100, far less than the 1 in 1,000,000 to 1 in 100,000 range currently used by regulatory agencies to set tolerable doses or exposures in the United States. The failure to validate risk predictions below 1 in 100 is a serious limitation of the linear at low dose risk assessment because it makes predictions of risk at the very low doses used in regulatory law solely model-dependent and unverifiable. Importantly, the U.S. Society of Toxicology (SOT) 14-member expert panel reviewed the results of the mega-mouse study and reported that it supported a hormetic dose–response model, when the analysis included a time component based on interim sacrifices (Bruce et al. 1981). The SOT indicated that the 2-AAF induced a J-shaped dose–response for bladder cancers that was consistent in each of the six separate rooms in which the large number of animals were housed, thereby relying on an internal replication of the hormetic findings. This study points to an obvious problem: It is practically impossible to demonstrate dose–response behaviors below an excess risk of about 1/100. Although it is possible to develop bioassays that involve several additional dose– response groups in the low-dose region of the experiment, it is the overwhelming evidence across multiple species and endpoints that should demonstrate the superiority of the J- and inverted J-shaped models. Similar findings of J-shaped dose responses in predictive cancer bioassays designed to test hormetic hypotheses have been reported for several epigenetic liver carcinogens by Japanese investigators (Kang et al. 2006; Kinoshita et al. 2006; Puatanachokchai et al. 2006; Fukushima et al. 2005a,b; Hoshi et al. 2004; Sukata et al. 2002; Masuda et al. 2001). Of particular interest have been detailed findings with the banned pesticide and carcinogen dichloro-diphenyl-trichloroethane (DDT). At high doses the Japanese investigators reported that DDT causes dose-dependent increases in liver foci of the Fischer 344 rat (Fukushima et al. 2005a,b; Sukata et al. 2002). However, at lower doses, decreased frequencies of liver foci have been reliably observed, supporting a hormetic interpretation. Detailed mechanistic studies have provided evidence concerning underlying factors that explain high-dose tumor enhancement and low-dose tumor protection caused by exposure to the DDT. These findings are consistent with a broad range of cancer bioassays supporting the hormetic biphasic dose–responses in fish (Brown-Peterson et al. 1999) and rodent models (Kopp-Schneider and Lutz 2001; Teeguarden et al. 2000). Of particular
198
CHAPTER 7 HORMESIS AND CANCER RISKS: ISSUES AND RESOLUTION
further interest is that low doses of a large number of chemical mutagens act via J-shaped dose responses at low dose (Maki-Paakkanen and Hakulinen 2008; Wilms et al. 2008; Demsia et al. 2007; Lacoste et al. 2006; Pu et al. 2006; Hoshi et al. 2004; Jagetia et al. 2003; Knasmuller et al. 2002; Sasaki et al. 2002; Hartmann et al. 2001; Kirkland and Muller 2000) further supporting the observations of hormetic dose– responses within cancer bioassays. Other evidence indicates that low doses of various xenobiotics enhance cell-to-cell communication function reducing possible risks of tumor promotion (Rivedal and Witz 2005; Jeong et al. 2001; Rivedal et al. 2000; Mikalsen and Sanner 1994; Mercier et al. 1993; Mikalsen et al. 1992; LochCaruso et al. 1984; Kurata et al. 1982). Furthermore, an extensive literature is available indicating that low doses of numerous immune system active agents display biphasic dose–responses enhancing parameters related to immune surveillance at low concentrations further supporting an hormetic interpretation (Calabrese 2005). The arguments justifying hormetic models as the regulatory defaults rely on fundamental scientific reasoning and meet Hill’s nine criteria and the more current causal criteria. Rather than relying on a conjecture that is not provable and denies true benefits, and thus avoids costly societal errors, the regulatory default should be based on the overall evidence for hormetic behaviors and the resulting estimates obtained by the J-model for cancer, or by its inverted form for toxicological endpoints.
7.3. HORMESIS AND CANCER RISK ASSESSMENT: MODELS The hormetic dose–response model for cancer is J-shaped. It accounts for and resolves several of the issues that cannot be resolved by the practical use of its regulatory alternatives. A direct way to test the validity of a hormetic statement is to assess if there is scientifically sound evidence that demonstrates adaptive, nonadverse, or beneficial events that (a) meet good scientific practices, causal criteria, peer reviews, and open discussions, (b) avoid secrecy, and (c) are independent of funding sources and other determining factors. This suggests framing policy science by answering the following two questions: 1. Theoretical/Empirical Question: Since the LNTs negate any protective response at low dose rates, what is the appropriate science policy to overcome this limitation? 2. Corollary Question: Since the J-shaped hormetic (or biphasic) cancer dose– response model yields empirically demonstrated protective (stimulatory) effects at low doses in one or more species, is biologically plausible, and describes a damaging relationship at higher dose that is consistent with the LNT, which of the two is the logical and prudential default model? Although the answer to the first question is legal, and thus beyond the scope of this chapter, the answer to the second question falls well within our framework. We can begin to frame the answers by a limited review of current well-known cancer dose–response models. The hormetic J-shaped model is depicted in Figure 7.2.
7.3. HORMESIS AND CANCER RISK ASSESSMENT: MODELS
199
Adverse response (%) 100 Experimental results (dots are enlarged for exposition
LNT model Hormetic model Experimental control response
Experimental region of adverse responses, common to LNT and hormetic models
Dose rate
Protection optimized
Figure 7.2. Biphasic (hormetic) dose–response model for cancer incidence (the percent response in the controls must be nonzero). Protection is considered “optimized” because it represents the greatest degree of protection at a dose range furthest away from an adverse effect.
The LNT-based dose–response model for cancer, being a cumulative distribution function, begins at zero and is proportional to doses (i.e., is linear at low doses, resulting in the LNT hypothesis). The early form of the LNT model is the one-hit model: Pr ( D ) = 1 − exp [ − ( qD )] In this model, Pr(D) is the lifetime probability of cancer death from lifetime exposure to dose D (often expressed in units of mg/kg-day, consistent with animals’ exposures). The multistage model is Pr ( D ) = 1 − exp [ − ( ∑ i qi D i )] where the same notation used for the single-hit model applies. This model can account for a threshold, but cannot account for any beneficial effect of exposure. A more recent model is the Moolgalvkar–Dewanjii–Venzon (MVK) model (Moolgalvkar et al. 1988), which is a two-stage stochastic model that accounts for cell growth, death, and differentiation. Figure 7.3 depicts the two-stage MVK cellular process in which two adverse and irreparable events must occur for normal cells to become malignant. The events may be mutations or other effects inherited by the cells. The cellular process consists of two stages (excluding the stage in which cells are normal) and the following transition rates, [cells/time]−1, as follows:
200
CHAPTER 7 HORMESIS AND CANCER RISKS: ISSUES AND RESOLUTION
Normal cells: N1
μ1
Preneoplastic cells: N2
β1
α1
β2
Malignant cells: N3
μ2
α2
β3
α3
Dead or differentiated cells
Figure 7.3.
Cellular biology of the two-stage MVK model.
1. Cellular death or differentiation, βi, (a carcinogen decreases it) 2. Cellular division into a normal or premalignant cell, μi, (a carcinogen increases it) 3. Cellular division into two normal cells, αi, the mitotic rate, (a carcinogen can increase it) The parameters of the MVK model can be dose-dependent. As Figure 7.3 depicts, the MVK model can represent cell proliferation due to exposure to a chemical that aids such proliferation and can account for different cell division rates. The assumptions include the following: (1) Cancer is a two-stage process, (2) cellular transformations are independent, and (3) once a cell becomes malignant, potentially cancerous cells proliferate independently of the normal cells resulting in a detectable cancer. Despite their conceptual value and practical successes, these stochastic transition models leave some important phenomena unexplained. These include (Cox and Ricci 2005): (a) Importance of Proliferation of Normal Cells in Increasing Cancer Risks. Many chemical carcinogens were found to increase tumor rates in experimental animals only in situations that also cause cytotoxicity and regenerative hyperplasia or compensating proliferation of apparently normal cell populations in response to the toxic injury. Examples include chloroform, diesel exhaust, formaldehyde, and many others. When such compensating proliferation is a prerequisite for chemically induced carcinogenesis, traditional linearized multistage modeling may overestimate risks at low concentrations or predict significant risks at low concentrations even if none truly exists. Thus, dose– response models that better account for the role of normal stem cell proliferation and kinetics following cytotoxic damage may be needed to obtain more realistic risk estimates for some chemicals. (b) Carcinogenic Thresholds in Dose Rate and/or Duration of Exposure Are Arbitrary. A generalization of these stochastic models allows for more than one possible sequence of events (e.g., somatically heritable transformations,
7.3. HORMESIS AND CANCER RISK ASSESSMENT: MODELS
201
epigenetic events) leading from normalcy to malignancy, while keeping the feature that some events must be completed before others can occur. A general process of this type can be described as an event tree, explicitly enumerating the possible sequences of events that take a normal stem cell to a malignant one. Each node in the tree corresponds to the sequence of events that have occurred so far, and the successors of each node are the events that can occur next, corresponding to branches out of the current node. The branch probabilities at each node complete the specification of the process. A directed acyclic graph (DAG) indicating allowed transitions among events, along with their probabilities, [i.e., a stochastic transition network (STN)], often provides a more concise representation of the same information. However, the tree provides a clear, useful conceptual model of multiple alternative paths and precedence partial ordering constraints among events.
7.3.1.
Answers to Our Question
The answer to the question we posed earlier in this essay is clear: The time is ripe for including hormesis as the principal regulatory model because it is not conjectural and is based on data consistent with all criteria put forth in the EPA’s causal arguments. Specifically, this answer is justified by the following findings: • The hormetic dose response can be tested because its low-dose response starts immediately to the left (in the dose–response space) of any hypothetical threshold. Recollecting that the threshold model is the linearized form of the S-shaped toxicological cumulative distribution of responses, this response is generally not within the observations (it is an extrapolation via a probit transformation from the experimental results to a dose intercept). On the other hand, the hormetic dose–response can be either validated or rejected with normal testing protocols, provided that a sufficient number of experimental results are available (five or more). • The hormetic dose–response can predict harm below or above the toxicological threshold, and thus it is consistent with positive and negative outcomes, unlike the LNT or the S-shaped models. • The hormetic dose–response model can predict the occurrence of beneficial responses below the toxicological threshold. This can be seen with endpoints such as enhanced longevity, decreased disease incidence, and improved cognition, unlike the threshold and linear at low-dose (LNT) models. • Chemical interactions can be accounted for. While threshold dose response model can only deal with chemical interactions for responses that exceed a threshold, the hormetic model also does this. These models differ where the interaction occurs in the hormetic stimulatory zone. In the case of the hormetic chemical interactions, the maximum response is still constrained to 30–60% above the control value a characteristic that the threshold and linear at lowdose models do not have.
202
CHAPTER 7 HORMESIS AND CANCER RISKS: ISSUES AND RESOLUTION
• A hormetic dose–response model has the same dose–response for all biological models, endpoints, and chemical or physical agents. This means that hormesis can harmonize risk assessment procedures for both carcinogens and noncarcinogens. • The hormetic dose–response can account for different biological mechanisms, in different species, and for different endpoints. This is relevant to the EPA “mode of action,” which is a weak form “mechanism of action” (EPA 2005). Hormetic dose–response models thus present a significant new challenge and opportunity for regulatory agencies because they permit those agencies to address the question of what is an “optimized” societal acceptable exposure or dose. That is, it answers the question, What exposure standard yields the greatest overall (societally optimal) health benefits? For example, a dose that may provide a beneficial effect in the normal population may be a harmful to those in a high subgroup. Conversely, a dose that provides a beneficial effect in a high-risk group may not have biological impact on the normal population. Because the normal population may be 95% of the entire population, with the high-risk segment the remainder, the total number of increased years of life for the members of society may occur if the exposure standard were established to maximally protect normal individuals. With the use of the threshold and linear at low-dose models, this situation cannot be assessed. Federal agencies, such as the EPA, typically state that an environmental standard is set at a dose that will protect all normal- and most high-risk members of the population. The hormetic model actually allows the possibility of avoiding hazardous exposures and increasing health benefits with the challenge of estimating the optimal overall response for society. The hormesis databases developed by Calabrese and his colleagues are supportive of the EPA mode of action and weight of evidence. For example, they are consistent (in fact, essential) to meet the EPA’s requirement of a weight-of-evidence narrative that should describe and be intelligible to risk managers and nonexpert readers (EPA 2005) regarding: • The quality and quantity of the data • All key decisions and the basis for these major decisions • Data, analyses or assumptions that are unusual for or new to EPA Specifically, this narrative should include (EPA 2005) the following: • Conclusions about human carcinogenic potential (choice of descriptor(s), …) • A summary of the key evidence supporting these conclusions (for each descriptor used), including information on the type(s) of data (human and/or animal, in vivo and/or in vitro) used to support the conclusion(s) • Available information on the epidemiologic or experimental conditions that characterize expression of carcinogenicity (e.g., if carcinogenicity is possible only by one exposure route or only above a certain human exposure level) • A summary of potential modes of action and how they reinforce the conclusions
ACKNOWLEDGMENTS
203
• Indications of any susceptible populations or lifestages, when available • A summary of the key default options invoked when the available information is inconclusive The mode of action in the assessment of potential carcinogens is a main focus of these cancer guidelines (EPA 2005). Its aspects involve a general acceptance criterion that establishes a particular sequence of biological key events and processes beginning at the cellular level and ending with the tumor. The important characteristic of the mode of action concept is that those key events are empirically observable precursors and thus are necessary, but perhaps not sufficient, stages in the development of cancer. The EPA also defines a mechanism of action, which has an even higher granularity than the mode of action: Molecular events fall under this rubric. The mode of action is, according to the EPA, a data-rich assessment. We add that it is the full description of the biological process being investigated, regardless of the endpoint under study. If so, the compelling empirical evidence in the hormetic databases fulfills the data-rich aspect of any empirical and theoretical analysis of hormetic behaviors. The same cannot be said for the LNT or the threshold models: The former is conjectural; the latter is an extrapolation that disregards the complete biological process.
7.4.
CONCLUSIONS
The first conclusion is that the factual and theoretical evidence points to replacing the classical causal regulatory defaults used to deal with low dose–response, the linear no-threshold, and the linear at low-dose–response models, or monotonic functions, with the J- and inverse J-shaped models—or relations. These models have been demonstrated to apply to toxicological and cancer outcomes for a very wide range of substances and diseases. The classical defaults may still be applicable on a case-by-case basis. The reasons for changing the defaults include the fact that the J-shaped class of models quantifies a wide set of health benefits that are completely excluded from estimations that use monotonic models. We conclude that replacing both a conjecture and an arbitrary model with two theoretically and empirically sound ones leads to rational decision and does not exclude actually demonstrable benefits. Overall, the sum is positive for society.
ACKNOWLEDGMENTS The effort of Edward J. Calabrese was sponsored by the Air Force Office of Scientific Research, Air Force Materiel Command, USAF, under grant number FA9550-07-10248. The U.S. Government is authorized to reproduce and distribute for governmental purposes notwithstanding any copyright notation thereon. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsement, either expressed or implied, of the Air Force Office of Scientific Research or the U.S. Government.
204
CHAPTER 7 HORMESIS AND CANCER RISKS: ISSUES AND RESOLUTION
REFERENCES Brown-Peterson, N., Krol, R. M., Zhu, Y., and Hawkins, W. E. (1999). N-Nitrosodiethylamine initiation of carcinogenesis in Japanese Medaka (Oryzias latipes): Hepatocellular proliferation, toxicity, and neoplastic lesions resulting from short term, low level exposure. Toxicol Sci 50, 186–194. Bruce, R. D., Carlton, W. W., Ferber, K. H., Hughes, D. H., Quast, J. F., Salsburg, D. S., Smith, J. M. (members of the Society of Toxicology ED01 Task Force), Brown, W. R., Cranmer, M. F., Sielken, J. R., Van Ryzin, J., and Barnard, R. C. (1981). Re-examination of the ED01 study whey the society of toxicology became involved. Fund Appl Toxicol 1, 26–128. Calabrese, E. J. (2001a). Estrogen and related compounds: Biphasic dose responses. Crit Rev Toxicol 31, 503–515. Calabrese, E. J. (2001b). Androgens: Biphasic dose responses. Crit Rev Toxicol 31, 517–522. Calabrese, E. J. (2001c). Dopamine: Biphasic dose responses. Crit Rev Toxicol 31, 563–583. Calabrese, E. J. (2001d). 5-Hydroxytryptamine (serotonin): Biphasic dose responses. Crit Rev Toxicol 31, 553–561. Calabrese, E. J. (2001e). Nitric oxide: Biphasic dose responses. Crit Rev Toxicol 31, 489–501. Calabrese, E. J. (2001f). Opiates: Biphasic dose responses. Crit Rev Toxicol 31, 585–604. Calabrese, E. J. (2001g). Prostaglandins: Biphasic dose responses. Crit Rev Toxicol 31, 475–487. Calabrese, E. J. (2001h). Adrenergic receptors: Biphasic dose responses. Crit Rev Toxicol 31, 523–538. Calabrese, E. J. (2005). Hormetic dose–response relationships in immunology: Occurrence, quantitative features of the dose response, mechanistic foundations, and clinical implications. Crit Rev Toxicol 35, 89–295. Calabrese, E. J. (2008a). Dose–response features of neuroprotective agents: An integrative summary. Crit Rev Toxicol 38, 253–348. Calabrese, E. J. (2008b). Pharmacological enhancement of neuronal survival. Crit Rev Toxciol 38, 349–389. Calabrese, E. J. (2008c). Pain and U-shaped dose responses: Occurrence, mechanisms and clinical implications. Crit Rev Toxicol 38, 579–590. Calabrese, E. J. (2008d). Alzheimer ’s disease drugs: An application of the hormetic dose–response model. Crit Rev Toxicol 38, 419–451. Calabrese, E. J. (2008e). Stress biology and hormesis: The Yerkes–Dodson law in psychology—A special case of the hormesis dose response. Crit Rev Toxicol 38, 453–462. Calabrese, E. J. (2008f). P-glycoprotein efflux transporter activity often displays biphasic dose–response relationships. Crit Rev Toxicol 38, 473–487. Calabrese, E. J. (2008g). Drug therapies for stroke and traumatic brain injury often displays U-shaped dose responses: Occurrence, mechanisms, and clinical implications. Crit Rev Toxicol 38, 557–577. Calabrese, E. J. (2008h). An assessment of anxiolytic drug screening tests: Hormetic dose responses predominate. Crit Rev Toxicol 38, 489–542. Calabrese, E. J. (2008i). Modulation of the epileptic seizure threshold: Implications of biphasic dose responses. Crit Rev Toxicol 38, 543–556. Calabrese, E. J., and Baldwin, L. A. (1999). Can the concept of hormesis be generalized to carcinogenesis. Regul Toxicol Pharmacol 28, 230–241. Calabrese, E. J., and Baldwin, L. A. (2000). Reproductive toxicity and hormetic responses. In Toxicology in Risk Assessment, Salem, H., ed., Taylor & Francis, Philadelphia, p. 106. Calabrese, E. J., and Baldwin, L. A. (2001). The frequency of U-shaped dose-responses in the toxicological literature. Toxicol Sci 62, 330–338. Calabrese, E. J., and Baldwin, L. A. (2003a). Inorganics and hormesis. Crit Rev Toxicol 33, 215–304. Calabrese, E. J., and Baldwin, L. A. (2003b). Chemotherapeutics and hormesis. Crit Rev Toxicol 33, 305–353. Calabrese, E. J., and Baldwin, L. A. (2003c). The hormetic dose response model is more common than the threshold model in toxicology. Toxicol Sci 71, 246–250. Calabrese, E. J., and Blain, R. (2004). Metals and hormesis. J Environ Monit 6, 14N–19N. Calabrese, E. J., and Blain, R. (2005). The occurrence of hormetic dose responses in the toxicological literature, the hormesis database: An overview. Toxicol Appl Pharmacol 202, 289–301.
REFERENCES
205
Calabrese, E. J., Staudenmayer, J. W., Stanek, E. J., and Hoffmann, G. R. (2006). Hormesis outperforms threshold model in NCI anti-tumor drug screening data. Toxicol Sci 94, 368–378. Cox, L. A., and Ricci, P. F. (2005). Causation in risk assessment and management: Models, inference, biases, and a microbial risk–benefit case study. Environ Int 31, 377–397. Demsia, G., Vlastos, D., Goumenou, M., and Matthopoulos, D. P. (2007). Assessment of the genotoxicity of imidaclorpid and metalaxyl in cultured human lymphocytes and rat bone-marrow. Mutat Res 634, 32–39. EPA (2005). Guidelines for carcinogen risk assessment, EPA/630/P-03/001F. Fukushima, S., Kinoshita, A., Puatanachokchai, R., Kushida, M., Wanibuchi, H., and Morimura, K. (2005a). Hormesis and dose–response-mediated mechanisms in carcinogenesis: Evidence for a threshold in carcinogenicity of non-genotoxic carcinogens. Carcinogenesis 26, 1835–1845. Fukushima, S., Wanibuchi, H., Morimura, K., Nakae, D., Tsuda, H., Imaida, K., Shirai, T., Tatematsu, M., Tsukamoto, T., Hirose, M., and Furukawa, F. (2005b). Lack of potential of low dose Nnitrosodimethylamine to induce preneoplastic lesions, glutathione S-transferase placental form-positive foci, in rat liver. Cancer Lett 222, 11–15. Hartmann, A., Kiskinis, E., Fjallman, A., and Suter, W. (2001). Influence of cytotoxicity and compound precipitation on test results in the alkaline comet assay. Mutat Res 497, 199–212. Hill, A. B. (1965). The environment and disease: Association or causation? Proc R Soc Med 58, 295–300. Hoshi, M., Morimura, K., Wanibuchi, H., Wei, M., Okochi, E., Ushijama, T., Takaoka, K., and Shoji, F. (2004). No-observed effect levels for carcinogenicity and for in vivo mutagenicity of a genotoxic carcinogen. Toxicol Sci 81, 273–279. Jagetia, G. C., Venkatesh, P., and Baliga, M. S. (2003). Evaluation of the radioprotective effect of Aegle marmelos (l.) Correa in cultured human peripheral blood lymphocytes exposed to different doses of gamma-radiation: A micronucleus study. Mutagenesis 18, 387–393. Jeong, S.-H., Cho, M.-H., and Cho, J.-H. (2001). Effects of cadmium on gap junctional intercellular communication in WB-F344 rat liver epithelial cells. Hum Exp Toxicol 20, 577–583. Kang, J. S., Wanibuchi, H., Morimura, K., Totsuka, Y., Yoshimura, I., and Fukushima, S. (2006). Existence of a no effect level for MelQx hepatocarcinogenicity on a background of thioacetamideinduced liver damage in rats. Cancer Sci 97, 453–458. Kinoshita, A., Wanibuchi, H., Wei, M., and Fukushima, S. (2006). Hormesis in carcinogenicity of nongenotoxic carcinogens. J Toxicol Pathol 19, 111–122. Kirkland, D. J., and Muller, L. (2000). Interpretation of the biological relevance of genotoxicity test results: The importance of thresholds. Mutat Res 464, 137–147. Knasmuller, S., Steinkellner, H., Majer, B. J., Nobis, E. C., Scharf, G., and Kassie, F. (2002). Search for dietary antimutagens and anticarcinogens: Methodological aspects and extrapolation problems. Food Chem Toxicol 40, 1051–1062. Kopp-Schneider, A., and Lutz, W. K. (2001). J-shaped dose–response relationship for tumor induction by caffeic acid in the rat forestomach, modeled by non-monotonic dose response for DNA damage and cell proliferation. HERA 7, 921–931. Kurata, M., Hirose, K., and Umeda, M. (1982). Inhibition of metabolic cooperation in Chinese hamster cells by organochlorine pesticides. Gann 73, 217–221. Lacoste, S., Castonguay, A., and Drouin, R. (2006). Formamidopyrimidine adducts are detected using the comet assay in human cells treated with reactive metabolites of 4-(methylnitrosamino)-1-(3pyridyl)-1-butanone (NNK). Mutat Res 600, 138–149. Loch-Caruso, R., Trosko, J. E., and Corcos, I. A. (1984). Interruption of cell-to-cell communication in Chinese hamster V79 cells by various alkyl glycol ethers: Implications for teratogenicity. Environ Health Perspect 57, 119–123. Maki-Paakkanen, J., and Hakulinen, P. (2008). Assesssment of the genotoxicity of the rat carcinogen 3-chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanone (MX) in rat liver epithelial cells in vitro. Toxicol in Vitro 22, 535–640. Masuda, C., Wanibuchi, H., Otori, K., Wei, M., Yamamoto, S., Hiroi, T., Imaoka, S., Funae, Y., and Fukushima, S. (2001). Presence of a no-observed effect level for enhancing effects of development of the α-isomer of benzene hexachloride (α-BHC) on diethylnitrosamine-initiatied hepatic foci in rats. Cancer Lett 163, 179–185.
206
CHAPTER 7 HORMESIS AND CANCER RISKS: ISSUES AND RESOLUTION
Mercier, T., Honikman-Leban, E., Chaumontet, C., Martel, P., and Shahin, M. M. (1993). Studies on the modulating effects of retinoic acid and retinol acetate using dye transfer and metabolic cooperation assays. Fundam Appl Toxicol 21, 270–276. Mikalsen, S.-O., Rivedal, E., and Sanner, T. (1992). Heavy metal ions, cytotoxicity and gap junctional intercellular communication in Syrian hamster embryo cells. ATLA 20, 213–217. Mikalsen, S.-O., and Sanner, T. (1994). Increased gap junctional intercellular communication in Syrian hamster embryo cells treated with oxidative agents. Carcinogenesis 15, 381–387. Moolgalvkar, S. H., Dewanjii, A., and Venzon, D. J. (1988). A stochastic two-stage model for cancer risk assessment, 1: The hazard function and probability of tumor. Risk Anal 8, 383–392. National Research Council (NRC). (1994). Committee on Risk Assessment of Hazardous Air Pollutants, US National Academy of Science, Science and Judgment in Risk Assessment, National Academies Press, Washington, D.C. Office of Management and Budget (OMB). (2003). Proposed Risk Assessment Bulletin. http://www. whitehouse.gov/omb/inforeg/proposed_risk_assessment_bulletin_010906.pdf. Pu, X., Kamendulis, L. M., and Klaunig, J. E. (2006). Acrylonitrile-induced oxidative DNA damage in rat astrocytes. Environ Mol Mutagen 47, 631–638. Puatanachokchai, R., Morimura, K., Wanibuchi, H., Oka, M., Kinoshita, A., Mitsuru, F., Yamaguchi, S., Funae, Y., and Fukushima, S. (2006). Alpha-benzene hexachloride exerts hormesis in preneoplastic lesion formation of rat hepatocarcinogenesis with the possible role of hepatic detoxifying enzymes. Cancer Lett 240, 102–113. Ricci, P. F., Cox, L. A. Jr., and MacDonald, T. (2006). Science-policy in environmental and health risk assessment: If we cannot do without, can we do better? Hum Exp Toxicol 25, 29–43. Ricci, P. F., and Gray, N. (1998). Towards a new way to deal with toxic torts: Risks in toxic tort law. Part I: Probabilistic causation and legal causation. Univ New South Wales Law J 21, 787–806. Ricci, P. F., and Molton, L. (1981). Risk and benefits in environmental law. Science 214, 1096. Rivedal, E., Mikalsen, S.-O., and Sanner, T. (2000). Morphological transformation and effect on gap junction intercellular communication in Syrian hamster embryo cells as screening tests for carcinogens devoid of mutagenic activity. Toxicol in Vitro 14, 185–192. Rivedal, E., and Witz, G. (2005). Metabolites of benzene are potent inhibitors of gap–junction intercellular communication. Arch Toxicol 79, 303–311. Sasaki, Y. F., Kawaguchi, S., Kamaya, A., Ohshita, M., Kabasawa, K., Iwama, K., Taniguchi, K., and Tsuda, S. (2002). The comet assay with 8 mouse organs: results with 39 currently used food additives. Mutat Res 519, 103–119. Sukata, T., Uwagawa, S., Ozaki, K., Ogawa, M., Nishikawa, T., Iwai, S., Kinoshita, A., Wanibuchi, H., Imaoka, S., Punae, Y., Okuno, Y., and Fukushima, S. (2002). Detailed low-dose study of 1,1-B IS (p-chlorophenyl)-2,2,2-trichloroethane carcinogenesis suggests the possibility of a hormetic effect. Int J Cancer 99, 112–118. Teeguarden, J. G., Dragan, Y., and Pitot, H. C. (2000). Hazard assessment of chemical carcinogens: The impact of hormesis. J Appl Toxicol 20, 113–120. Wilms, L. C., Kleinjans, J. C. S., Moonen, E. J. C., and Briede, J. J. (2008). Discriminative protection against hydroxyl and superoxide anion radicals by quercetin in human leucocytes in vitro. Toxicol in Vitro 22, 301–307.
CH A P TE R
8
THRESHOLDS FOR GENOTOXIC CARCINOGENS: EVIDENCE FROM MECHANISM-BASED CARCINOGENICITY STUDIES Shoji Fukushima Min Wei Anna Kakehashi Hideki Wanibuchi
8.1.
OVERVIEW
In this chapter, the results of a medium-term rat carcinogenicity bioassay for rapid in vivo detection of carcinogenic potential are presented to examine the carcinogenicity of low doses of five genotoxic carcinogens: 2-amino-3,8-dimethylimidazo [4,5-f ] quinoxaline (MeIQx), a heptocarcinogen contained in seared fish and meat; N-nitrosodiethylamine (DEN) and N-nitrosodimethylamine (DMN), heptocarcinogens synthesized in the stomach through the reaction of secondary amines and nitrites; 2-amino-1-methyl-6-phenylimidazo[4,5-b] pyridine (PhIP), a colon carcinogen contained in seared meat and fish; and potassium bromate, a kidney carcinogen that is a contaminate of tap water and also used as a food additive in some countries. DNA damage, gene mutation, and surrogate endpoints for carcinogenicity were examined: Carcinogenic endpoints were glutathione S-transferase placental form (GST-P) positive foci in the liver, a well-known preneoplastic lesion marker in rat hepatocarcinogenesis, and altered crypt foci (ACF), a well-known surrogate marker of preneoplastic lesions in the colon. Low doses of MeIQx induced formation of DNA-MeIQx adducts; somewhat higher doses caused elevation of 8-hydroxy-2′deoxyquanosine (8-OHdG) levels; at still higher doses, gene mutations occurred; and the very highest dose of MeIQx induced formation of GST-P positive foci. Similarly, only the highest doses of DEN and DMN caused an increase in the number of GST-P positive foci in the liver; the lower doses had no effect. Similar results were obtained with the colon carcinogen PhIP. PhIP–DNA adduct formation was
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
207
208
CHAPTER 8 THRESHOLDS FOR GENOTOXIC CARCINOGENS
observed after treatment with low doses, while only high doses were found to induce ACF. Finally, in experiments with potassium bromate, 8-OHdG formation, GC-toTA transversions, and gene mutations in the rat kidney were observed only after administration of the highest doses of KBrO3; histopathological changes related to carcinogenicity were not observed at any dose used. These data support the existence of thresholds for the genotoxic carcinogens examined in this chapter.
8.2.
INTRODUCTION
Compounds known to be carcinogenic to humans have primarily been identified by epidemiological methods—for instance, cancer development in factory workers (occupational cancer) (see Chapter 15). Epidemiological data, however, are usually not suitable to establish risk from exposure to different levels of human carcinogens. Also, epidemiological data are available only after exposed humans develop cancers. Carcinogen risk assessment aspires to identify and assess risk from exposure to carcinogens prior to extensive human exposure. Identification and assessment of most carcinogens is done using two-year carcinogenicity tests performed in rodents, particularly rats and mice (see Chapter 14). To assess risk in humans, carcinogenic response curves obtained from these tests are used. Importantly, the carcinogenicity of low doses of carcinogenic compounds is generally extrapolated from the carcinogenicity data obtained using high doses; to obtain statistically acceptable data, carcinogens are generally used in rodent carcinogenicity tests at high doses, including the maximum tolerated dose. The principal method of assessing risk posed to humans by exposure to genotoxic carcinogens uses nonthreshold approaches to model experimental data: The curves generated by nonthreshold approach modeling are S-shaped or linear low-dose straight lines that reach zero (see Chapter 24). This “nonthreshold concept” of genotoxic carcinogenicity reflects the idea that a single event caused by a genotoxic carcinogen can have a positive influence on cancer development in humans. However, the physiology of living organisms suggests that, in practical terms, thresholds can exist, even for genotoxic carcinogens. Most chemical carcinogens must be metabolized within the cell to their active forms, known as the ultimate carcinogen, before they are able to exert their carcinogenic activity. The ultimate carcinogen formed from most genotoxic compounds binds covalently to DNA, forming an adduct. These adducts can interfere with normal DNA metabolism, leading to DNA mutations and carcinogenicity. However, these DNA adducts are efficiently repaired by the cell. Still, for any particular adduct there is the possibility of misrepair or replication of damaged DNA resulting in fixation of a mutation into the cell’s genome. Therefore, there is a finite risk of mutation arising from a single adduct. Next, at the level of DNA mutation, carcinogen–DNA adduct formation is essentially random in the euchromatic DNA (DNA that is not highly condensed); consequently, only a minute fraction of the mutations arising from these adducts will actually occur in a gene and have an affect on the cell, and only a very small fraction of these will be carcinogenic. In practical terms then, only a minute fraction of adducts actually give rise to DNA mutations and only a minute fraction of these mutations will affect the cell. As the number of adducts increases, however, the possibility of mutations occurring increases and mutated cells eventually begin
8.3. LOW-DOSE CARCINOGENICITY OF MEIQX IN THE RAT LIVER
209
to appear (Kuraoka 2008). This is especially relevant as the number of adducts becomes greater than the capacity of the cell to repair this damage. Most damaged or mutated cells will die due to metabolic dysfunctions or be eliminated by irreversible senescence or apoptosis, but it is possible that some will survive and be viable. In the two-stage chemical carcinogenesis model, this sequence of events is thought to occur during the initiation stage. Cell proliferation enhances the ability of initiated cells to form preneoplastic lesions and to develop into tumors, benign and then malignant. Evidence indicates that, before developing into tumors, most preneoplastic lesions disappear spontaneously, presumably due, at least in part, to elimination by the immune system. The development from initiated cells into tumors is the promotion stage of the two-stage chemical carcinogenesis model. Therefore, in a finite population, if physiological functions such as DNA repair, induction of senescence or apoptosis, and immune surveillance are effective, there will be levels of exposure to genotoxic carcinogens below which induction of carcinogenesis is effectively zero (see Chapter 26). For detection of carcinogenicity, the standard method is long-term carcinogenicity testing in two rodent species, such as mice and rats (≥50 animals/sex/group), with at least three dose levels, and in-life study termination at 18 months for mice and 24 months for rats (OECD 1981). However, such tests are extremely timeconsuming, laborious, and expensive. This is particularly true when examining the effects of low doses of suspected carcinogens since many more animals are required to reliably determine whether the low doses used are in fact able to induce an increase in tumor formation. In practical terms, it is currently impossible to examine the carcinogenicity of all suspect compounds using long-term rodent assays. Therefore, recently, an alternative method to long-term carcinogenicity testing in which preneoplastic lesions are accepted as endpoint markers for the assessment of carcinogenicity has been proposed (Tsuda et al. 2003). Results are obtained from this in vivo medium-term bioassay system of carcinogens in a matter of weeks rather than, as with long-term testing, many months. The presence or absence of a threshold will determine the reliability of carcinogenic risk assessment when extrapolated from high-dose rodent testing. Therefore, it is essential to verify scientifically whether the nonthreshold concept is valid. Herein, we provide data from low-dose carcinogenicity studies for genotoxic carcinogens using a medium-term bioassay for carcinogens. In addition to determining no-effect doses for carcinogenicity, we also examined markers that cells typically acquire as they move through the initiation and promotion stages of carcinogenesis. Analysis of all the data strongly support the existence of thresholds for the carcinogenetic effects of the five genotoxic carcinogens examined.
8.3. LOW-DOSE CARCINOGENICITY OF 2-AMINO3,8-DIMETHYLIMIDAZO[4,5-F]QUINOXALINE (MEIQx) IN THE RAT LIVER MeIQx is a heterocyclic amine contained in fried meat and fish. MeIQx at doses of 100–400 ppm in the diet is carcinogenic in the rat liver (Kato et al. 1988). To investigate the effect of exposure to low doses of MeIQx, 1145 21-day-old male F344
210
CHAPTER 8 THRESHOLDS FOR GENOTOXIC CARCINOGENS
rats were divided into seven groups and administered MeIQx in the diet at doses of 0, 0.001, 0.01, 0.1, 1, 10 ppm (low-dose groups) and 100 ppm (high-dose group) for 4–32 weeks (Fukushima et al. 2002). The data on the induction of the GST-P positive foci after treatment with various doses of MeIQx for 16 weeks is presented in Table 8.1 (GST-P positive foci is a preneoplastic lesion in rat hepatocarcinogenesis and the endpoint marker in the rat liver medium-term carcinogenicity bioassay). The numbers of GST-P positive foci were not significantly elevated in the 0.001– 10 ppm MeIQx groups, but a statistically significant increase was detected in the 100 ppm group. The same results were observed when the treatments with MeIQx were continued for 32 weeks (Figure 8.1). MeIQx is metabolized in liver cells to an ultimate carcinogen capable of covalently binding DNA. In contrast to GST-P foci induction, the formation of
TABLE 8.1. Induction of GST-P positive Foci in the Liver of Rats Treated with MeIQx for 16 Weeks
Group 1 2 3 4 5 6 7
MeIQx Dose (ppm)
Number of Rats
0 0.001 0.01 0.1 1 10 100
150 150 150 150 150 50 50
Size Distribution of GST-P Positive Foci (No./cm2) 2–4 Cells
5–10 Cells
≥11 Cells
Total
0.12 ± 0.17 0.12 ± 0.18 0.13 ± 0.21 0.14 ± 0.20 0.16 ± 0.20 0.35 ± 0.33 13.86 ± 5.11a
0.05 ± 0.17 0.02 ± 0.06 0.03 ± 0.07 0.04 ± 0.08 0.04 ± 0.08 0.10 ± 0.12 8.85 ± 3.23a
0.02 ± 0.09 0.01 ± 0.05 0.01 ± 0.05 0.02 ± 0.10 0.02 ± 0.07 0.01 ± 0.05 6.51 ± 4.06a
0.18 ± 0.35 0.15 ± 0.19 0.16 ± 0.24 0.19 ± 0.25 0.21 ± 0.24 0.47 ± 0.35 29.2 ± 10.99a
p > 0.01 (vs. group 1).
a
No. of GST-P positive foci (no./cm2)
100
*
10
1
0.1
0
0.01
0.1
1
10
100
MeIQx (ppm, in diet)
Figure 8.1. GST-P positive foci in the livers of F344 rats treated with MeIQx at various doses for 32 weeks. Asterisk (*) indicates p < 0.01 versus 0 ppm group.
211
8.3. LOW-DOSE CARCINOGENICITY OF MEIQX IN THE RAT LIVER
(/107nds) 100
MeIQx-DNA adduct
A
(/105dG) 10
8-OHdG
B
#
*
10
#
1 #
1 0.1
0.01
0.1
0.001
0
0.1 1 0.001 0.01 MeIQx (ppm, in diet)
10
100
0
0.001
0.01 0.1 1 MeIQx (ppm, in diet)
10
100
Figure 8.2. MeIQx–DNA adduct formation (A) and 8-OHdG formation levels (B) in the liver of F344 rats treated with MeIQx at various doses for 4 weeks. Asterisk (*) p < 0.01 versus 0.01 ppm group; number symble (#) indicates p < 0.01 versus 0 ppm group.
MeIQx–DNA adducts at week 4 was induced by administration of 0.01 ppm and higher doses of MeIQx, and induction was dose-dependent and statistically significant in the 100 ppm dose group (Figure 8.2); adduct formations in the 0 and 0.001 ppm MeIQx groups were below the limit of detection. Similar results were obtained after 16 weeks of MeIQx administration. DNA is subject to constant oxidative damage from endogenous oxidants. 8-Hydroxy-2′-deoxyguanosine (8-OHdG) is a marker for oxidative DNA damage, and 8-OHdG levels rise as a cell becomes more metabolically active. 8-OHdG levels at week 4 were unaffected by treatment with 0.001, 0.01, or 0.1 ppm MeIQx, but became statistically significantly elevated after treatment with MeIQx at doses of 1, 10, and 100 ppm (Figure 8.2). Similar results were obtained after 16 weeks of MeIQx administration. Finally, mutation of the H-ras gene, whose role in rat hepatocarcinogenesis is still unclear, was statistically significantly increased in the liver of rats treated with MeIQx for 2 weeks at 10 and 100, but did not differ at 0.001, 0.01. 0.1 and 1 ppm compared to the control value (unpublished data). We also examined mutation of the lacI gene and induction of GST-P positive foci in the livers of Big Blue® rats (Hoshi et al. 2004). Forty male Big Blue® rats were divided into 7 groups and administered MeIQx in the diet at doses of 0, 0.001, 0.01, 0.1, 1, 10, and 100 ppm for 16 weeks. A statistically significant elevation of lacI gene mutation level was detected in the 10 and 100 ppm groups (Figure 8.3). On the other hand, formation of GST-P positive foci was statistically significantly induced by administration of 100 ppm but not 10 ppm or less MeIQx (Figure 8.3). The results obtained from the experiments described above demonstrate the existence of a no-effect level (the highest dose of MeIQx at which there is no effect) for MeIQx mutagenicity and carcinogenicity. Since there is a no-effect level of MeIQx for gene mutagenicity, the initiation activity of MeIQx was examined in a two-stage carcinogenesis model using phenobarbital as a promoter of hepatocarcinogenesis (Fukushima et al. 2003). A total of 850 21-day-old male F344 rats were
212
CHAPTER 8 THRESHOLDS FOR GENOTOXIC CARCINOGENS
(No./10 6)
A
Mutation frequencies
(No. /cm2)
B
GST-P positive foci
100
*
1000
*
10 100
* 1
10
0.1
1 0
0.001
0.01
0.1
1
MeIQx (ppm, in diet)
10
100
0
0.001
0.01
0.1
1
10
100
MeIQx (ppm, diet)
Figure 8.3. Lac I gene mutation frequencies (A) and GST-P positive foci (B) in the liver of Big Blue® rats treated with MeIQx at various doses for 16 weeks. Asterisk (*) indicates p < 0.001 versus 0 ppm group.
divided into seven groups and administered MeIQx at doses of 0, 0.001, 0.01, 0.1, 1, 10, and 100 ppm for 4 weeks. This was followed by administration of 500 ppm phenobarbital in the diet. The numbers of GST-P positive foci were not elevated in the 0.001–1 ppm MeIQx groups, but statistically significant increases in GST-P positive foci formation were detected in the 10 and 100 ppm MeIQx groups. These results indicate the existence of a no-effect level for MeIQx initiation activity and are consistent with the existence of a no-effect level for MeIQx mutagenicity. Little is known about differences in the low dose–response relationship of genotoxic carcinogens among different strains of rat. Therefore, we examined MeIQx hepatocarcinogenicity using GST-P positive foci in both F344 and BN strains, with a total of 180 in each group. The background level of GST-P positive foci in the nontreated F344 rats was statistically significantly lower than that of BN rats, and the numbers of GST-P positive foci in the livers of MeIQx-treated F344 rats were statistically significantly lower in nearly all treatment groups compared with the corresponding BN strain groups (Wei et al. 2006). However, the results of MeIQx induction of GST-P positive foci in these two strains was the same: Lower doses of MeIQx, 0.1–10 ppm, had no statistically significant effect on the number of GST-P positive foci compared to the corresponding controls, while a statistically significant increase was detected at 100 ppm in both strains compared to the respective control groups (Table 8.2). Finally, we examined the carcinogenicity MeIQx in damaged livers (Kang et al. 2006). A total of 280 male F344 rats were divided into 14 groups. Liver damage was induced in 7 of these groups by administration of 0.03% thioacetamide (TAA), a well-known hepatotoxin, in their drinking water for 12 weeks. After cessation of TAA treatment, the rats received 0, 0.001, 0.01, 0.1, 1, 10, and 100 ppm MeIQx in the diet for 16 weeks. In both TAA-treated and untreated groups, the lower doses of MeIQx had no effect on the number of GST-P positive foci, but a statistically
8.3. LOW-DOSE CARCINOGENICITY OF MEIQX IN THE RAT LIVER
213
TABLE 8.2. Development of GST-P Positive Foci in the Livers of BN and F344 Rats Treated with Various Doses of MeIQx
Group
MeIQx (ppm)
BN Rat 1 0 2 0.1 3 1 4 5 5 10 6 100 F344 Rat 7 0 8 0.1 9 1 10 5 11 10 12 100
Size Distribution of GST-P Positive Foci (No./cm2)
Number of Rats
2–4 Cells
5–10 Cells
≥11 Cells
Total
30 30 30 30 30 30
0.16 ± 0.21 0.14 ± 0.23 0.12 ± 0.21 0.23 ± 0.33 1.17 ± 0.98 13.26 ± 7.07a
0.06 ± 0.12 0.03 ± 0.08 0.04 ± 0.09 0.11 ± 0.19 0.42 ± 0.57 7.37 ± 4.78a
0.02 ± 0.10 0.03 ± 0.19 0.04 ± 0.14 0.02 ± 0.09 0.06 ± 0.14 4.25 ± 3.88a
0.24 ± 0.29 0.19 ± 0.30 0.20 ± 0.33 0.36 ± 0.49 1.64 ± 1.43 24.88 ± 14.67a
30 30 30 30 30 30
0.01 ± 0.05c 0.03 ± 0.08c 0.07 ± 0.15 0.08 ± 0.16c 0.29 ± 0.49c 3.60 ± 2.22b,c
0c 0 0c 0.01 ± 0.05c 0.04 ± 0.12c 1.83 ± 1.33b,c
0 0.01 ± 0.05 0 0 0c 0.99 ± 1.01b,c
0.01 ± 0.05c 0.04 ± 0.10c 0.07 ± 0.15 0.08 ± 0.18c 0.33 ± 0.59c 6.41 ± 4.04b,c
p < 0.01 (vs. group1).
a
p < 0.01 (vs. group7).
b
p < 0.01 (vs. corresponding BN rat group).
c
significant increase was observed in the 100 ppm MeIQx groups (Figure 8.4). Using the method of maximum likelihood to model this data, the numbers of GST-P positive foci, with and without TAA treatment, fitted the hockey stick regression model; that is, no statistically significant differences in foci number were observed in the 0–10 ppm MeIQx groups, whereas a statistically significant increase in foci number was observed in the 100 ppm MeIQx group. In contrast, a linear dosedependent increase of MeIQx–DNA adduct formation was evident from 0.1 to 100 ppm; adduct formation in the 0.001 and 0.01 ppm MeIQx groups were below the limit of detection (Figure 8.4). The formation of MeIQx-DNA adducts was virtually identical in undamaged and damaged livers. These results are consistent with the previous results and support the existence of a no-effect level for MeIQx hepatocarcinogenicity, even on a background of liver damage. A summary of the results obtained in our experiments is presented in Figure 8.5. The formation of DNA–MeIQx adducts was observed at very low doses of MeIQx. Due to limitations in detection of these adducts, we were unable to determine whether a threshold dose of MeIQx was required for MeIQx–DNA adduct formation. Increasing the dose of MeIQx next resulted in an elevation of 8-OHdG formation, then gene mutation and the appearance of initiation activity, and, finally, at the highest dose used, an increase in the endpoint marker for carcinogenicity (GST-P positive foci). Notably, these data demonstrate that increased doses
214
CHAPTER 8 THRESHOLDS FOR GENOTOXIC CARCINOGENS
A
B GST-P positive foci
(No. /cm2) 100
MeIQx-DNA adduct
(x10-7nds) 100
TAA → MeIQx MeIQx
*
# ##
10
**
10
# ##
1 #
1
##
0.1 TAA → MeIQx MeIQx
0.1 0
0.001
0.01
0.1
1
10
100
0.01
0
0.001
0.01
MeIQx (ppm, in diet)
0.1
1
10
100
MeIQx (ppm, in diet)
Figure 8.4. GST-P positive foci (A) and formations of MeIQx-DNA adduct (B) in the liver of F334 rats treated with MeIQx with or without thioacetamide. Asterisk (*) indicates p < 0.01 versus TAA intiation alone group; double asterisk (**) indicates p < 0.01 versus nontreatment group; Number symbol (#) indicates p < 0.01 versus 0.1 ppm MeIQx without TAA intiation; double number symbol (##) indicates p < 0.01 versus 0.1 ppm MeIQx with TAA intiation.
Response Liver cancer 8-OHdG H-ras mutation lacI mutation Initiation activity
GST-P positive foci
MeIQx-DNA adduct
Control level MeIQx doses Figure 8.5. Risk of liver cancer: Reaction curves for carcinogenesis markers are dependent on the dose of MeIQx.
of MeIQx were required as MeIQx-mediated effects moved from simple adduct formation to cellular metabolic changes (possibly due in part to increased DNA repair) to gene mutation and cancer initiation to carcinogenesis. These results argue strongly for the existence of a threshold, at least a practical threshold, for MeIQx hepatocarcinogenicity in the rat. In support of this conclusion, our 2-year carcinogenicity test of MeIQx in rats showed no hepatocarcinogenicity at low doses (Murai et al. 2008).
8.5. LOW-DOSE CARCINOGENICITY OF PHIP IN THE RAT COLON
215
8.4. LOW-DOSE HEPATOCARCINOGENICITY OF N-NITROSO COMPOUNDS N-nitroso compounds such as diethylnitroamine (DEN) and dimethylnitrosamine (DMN) are synthesized in the stomach through the reaction of secondary amines and nitrites in the diet. They are also found as contaminants of a variety of manufactured food products. Peto et al. (1991) investigated the carcinogenicity of DEN using 2040 male and 2040 female Colworth rats. DEN at doses of 0.033–16.896 ppm was administered to the rats in their drinking water, induction of liver tumors was found to be dependent on the applied dose of DEN, and at the lower doses a linear dose–tumor incidence relationship was observed (Peto et al. 1991). Therefore, it was concluded that DEN had no threshold for its carcinogenicity in the rat liver. We have reexamined the carcinogenic influence of low doses of DEN (Fukushima et al. 2002). Approximately 2000 21-day-old male F344 rats were administered DEN at doses ranging from 0.0001 to 10 ppm in their drinking water for 16 weeks. No increase in the number of GST-P positive foci was found at DEN doses of 0.0001–0.01 ppm; however, the number of GST-P positive foci was statistically significantly elevated at 0.1 and 1 ppm DEN. In the 10 ppm group, the numbers of GST-P positive foci were so numerous that quanitation was not possible. Therefore, we conclude that there is a no-effect level for DEN hepatocarcinogenicity in the rat. Low-dose carcinogenicity experiments were also performed with DMN (Fukushima et al. 2005). The carcinogen was applied to 540 21-day-old F344 rats at doses ranging from 0.001 to 10 ppm in their drinking water for 16 weeks. No induction of GST-P positive foci was found at doses of 0.001 to 0.1 ppm; however, statistically significant increases in the number of GST-P positive foci were observed at 1 and 10 ppm. Therefore, similarly to DEN, we concluded that there is a no-effect level for DMN hepatocarcinogenicity in the rat.
8.5. LOW-DOSE CARCINOGENICITY OF 2-AMINO-1METHYL-6-PHENYLIMIDAZO[5,6-B]PYRIDINE (PHIP) IN THE RAT COLON The heterocyclic amine PhIP is a carcinogen contained in seared meat and fish, and it exerts its carcinogenicity in the rat colon. We investigated the carcinogenicity of PhIP in the rat colon when applied at doses of 0.001– 400 ppm (Fukushima et al. 2004). A total of 1759 6-week-old F344 male rats were administered PhIP in their diet for 16 weeks. The development of aberrant cell foci (ACF), the surrogate marker of preneoplastic lesions in the colon, was not altered by PhIP administration at 0.001–10 ppm; however, at doses of 50 – 400 ppm, statistically significant increases in ACF were observed (Figure 8.6). Like MeIQx, DEN, and DMN, PhIP is a genotoxic compound and is metabolized in cells to an ultimate carcinogen capable of covalently binding DNA.
216
CHAPTER 8 THRESHOLDS FOR GENOTOXIC CARCINOGENS
A
B
ACF
Total ACF/rat 10
* #
PhIP-DNA adduct
(/107nds) 100
*
* 10
*
1
* 1
0.1
*
0.1
*
*
*
0.01
0.01 0
0.001
0.01
0.1
1
PhIP (ppm, in diet)
10
50 100 400
0
0.001
0.01
0.1
1
10
50 100 400
PhIP (ppm, in diet)
Figure 8.6. Aberrant crypt foci (A) and formations PhIP–DNA adduct (B) in the colons of F344 rats treated with PhIP at various doses for 16 weeks. Number symbol (#) indicates p < 0.05 versus 0 ppm group. Asterisk (*) indicates p < 0.01 versus 0 ppm group. Note that PhIP–DNA adduct levels were also statistically significantly increased in the same manner at week 4.
Statistically significant increases in the formation of PhIP–DNA adduct levels were found in the groups treated with 0.01 ppm and higher doses of PhIP at 16 weeks (Figure 8.6). Thus, similarly to MeIQx, DNA adduct formation is observed after administration of low doses of PhIP while doses required to induce ACF are much higher (approximately 50,000 times higher) than that needed for PhIP–DNA adduct formation. These results argue for a no-effect level and a threshold dose for PhIP colon carcinogenicity in the rat. Finally, we assessed the effect of low doses of PhIP in the progression of colon tumors (Doi et al. 2005). A total of 192 6-week-old male F344 rats were subcutaneously injected twice with the colon carcinogen azoxymethane (AOM) with a 1-week interval, and then the animals were continuously fed PhIP at doses ranging from 0.001 to 200 ppm for 16 weeks. Lower doses (0.001–10 ppm) of PhIP had no significant effect on AOM-initiated colon carcinogenesis; higher doses (50–200 ppm) of PhIP caused a statistically significantly enhancement of AOM-initiated colon carcinogenesis (Table 8.3). Results obtained from this initiation–promotion model show a no-effect level of 10 ppm for PhIP promotion of colon carcinogenesis and again argue for a threshold dose for PhIP colon carcinogenicity in the rat.
8.6. LOW-DOSE CARCINOGENICITY OF POTASSIUM BROMATE, KBRO3 IN THE RAT KIDNEY Potassium bromate is a rodent renal carcinogen which can be found as a contaminant of tap water and which is used as a dough conditioner and food additive in some countries. It is a genotoxic carcinogen that is reduced in renal proximal tubular cells to yield bromine oxides and radicals, which are the ultimate carcinogens that specifically cause guanine oxidation, leading to renal mutagenesis and carcinogenesis.
TABLE 8.3.
Induction of Tumors in the Colon of F344 Rats Treated with Azoxymethane Followed by PhIP
Incidences (%) Histologic Findings
0 ppm (n = 16)
0.001 ppm (n = 16)
0.01 ppm (n = 16)
0.1 ppm (n = 16)
1 ppm (n = 16)
10 ppm (n = 16)
50 ppm (n = 16)
200 ppm (n = 14)
Adenoma Adenocarcinoma Totala
2 (12.5) 8 (50) 9 (56.3)
3 (18.8) 7 (43.8) 10 (62.5)
1 (6.3) 10 (62.5) 10 (62.5)
5 (31.3) 5 (31.3) 8 (50)
3 (18.8) 9 (56.3) 11 (68.8)
2 (12.5) 8 (50) 8 (50)
14 (87.5)b 14 (87.5) 16 (100)c
14 (100)d 14 (100)c 14 (100)c
a
Total of adenoma and adenocarcinoma. p < 0.005 (vs. 0 ppm).
b c
p < 0.05 (vs. 0 ppm). p < 0.0001 (vs. 0 ppm).
d
217
218
CHAPTER 8 THRESHOLDS FOR GENOTOXIC CARCINOGENS
Thus, the genotoxic mechanism of potassium bromate is different from that of MeIQx, DEN, DMN, and PhIP. The studies described to this point indicate that the genotoxic compounds MeIQx, DEN, DMN, and PhIP have no-effect levels for induction of various carcinogenesis markers and strongly suggest that a threshold dose exists for induction of carcinogenesis by these compounds. In the following experiments, we investigated the relationship between potassium bromate dose and induction of gene mutation, one of the markers of carcinogenesis. A total of 40 male Big Blue® rats were divided into 8 groups and administered potassium bromate in their drinking water at doses of 0, 0.02, 0.2, 2, 8, 30, 125, and 500 ppm for 16 weeks (Yamaguchi et al. 2008). No significant induction of lacl gene mutation was observed in the 0.02–125 ppm groups, but a statistically significant increase in lacl gene mutation was observed in the 500 ppm group (Figure 8.7). Similarly, statistically significantly elevated 8-OHdG levels and GC to TA transversions, a mutation known to occur as a result of 8-OHdG adduct formation, also occurred only at a potassium bromate dose of 500 ppm (Figure 8.7). No preneoplastic or neoplastic lesions were detected in the kidney in these experiments. Therefore, we concluded that there is a no-effect level for potassium bromate-induced 8-OHdG formation and mutagenicity in the rat kidney. Finally, the renal carcinogenicity of potassium bromate was examined using a two-stage carcinogenesis model. A total of 240 male Wistar rats were treated with N-ethyl-N-hydroxyethylnitrosamine for the initiation of kidney carcinogenesis and were thereafter administered potassium bromate at doses of 0, 0.02, 0.2, 2, 8, 30, 125, and 500 ppm in their drinking water for 16 weeks (Wei et al. 2009): Due to
A
Qxidative DNA damage (8-OHdG)
(/105dG)
*
1.6 1.2 0.8 0.4 0
0
0.02
0.2
2
8
30
125
500
KBrO3 (ppm, in drinking water)
B
C
Total lacI mutation frequency
lacI mutation frequency (GC to TA)
(/106 plaques)
(/106 plaques) 80
*
*
30
60 20 40 10
20 0
0
0
0.02
0.2
2
8
30
KBrO3 (ppm, in drinking water)
125
500
0
0.02
0.2
2
8
30
125
500
KBrO3 (ppm, in drinking water)
Figure 8.7. 8-OHdG formation levels (A) and LacI gene mutation frequencies (B, C) in the kidney of Big Blue rats treated with KBrO3 for 16 weeks. (*) p < 0.05 versus 0 ppm group.
ACKNOWLEDGMENTS
219
toxicity, the highest dose, 500 ppm, was reduced to 250 ppm from week 12. Enhancement of a preneoplastic lesion, an atypical tubular hyperplasia, and enhancement of tumorigenesis in the kidney was observed only in the highest dosed group. The results of these two sets of experiments support the conclusion that there is a no-effect level and threshold dose for potassium bromate renal carcinogenicity in the rat.
8.7.
CONCLUSION
For the genotoxic carcinogens examined, the no-effect doses for initiation markers (i.e., DNA adduct formation, 8-OHdG formation, and gene mutation) were much lower than the no-effect doses for promotion marker (i.e., GST-P positive foci and ACF); and, generally, induction of promotion markers occurred at doses of carcinogen which did not induce carcinogenesis. These results strongly suggest that processes such as DNA repair, irreversible senescence, apoptosis, and immune system function operate to inhibit the effects of genotoxic carcinogens and that the inhibition is significant. Therefore, we conclude that there are thresholds, at least practical thresholds, for the carcinogens examined in this study. The genotoxic carcinogens examined in this study can be classified into two types from the viewpoint of mechanism (Hengstler et al. 2003). In one type, the carcinogen is metabolized by the cell to an ultimate carcinogen, which binds covalently to the DNA to form DNA adducts. In the second type, the compound is metabolized by the cell to an ultimate carcinogen, which causes oxidative damage to the DNA. The first type of genotoxic carcinogen encompasses heterocyclic amines (e.g., MeIQx and PhIP) and N-nitrosocompounds (e.g., DEN and DMN). The second type of genotoxic carcinogen is represented by potassium bromate. Notably, the first type of genotoxic carcinogen induces formation of DNA adducts at low doses but higher doses are required for gene mutation, while the second type of genotoxic carcinogen causes DNA damage and gene mutation at equivalent doses. This undoubtedly reflects the different mechanisms by which these two types of genotoxic compounds cause DNA damage. Nevertheless, both types of genotoxic compounds clearly have no-effect doses for initiation, which are lower than the no-effect doses for promotion and carcinogenicity. It is probable, therefore, that other (perhaps most or even all) genotoxic carcinogens also have this pattern of no-effect dose and, consequently, do have thresholds for carcinogenicity.
ACKNOWLEDGMENTS The authors would like to acknowledge the help of Masao Hirose (Division of Pathology, National Institute of Health Sciences), Yoichi Konishi (Department of Oncological Pathology, Cancer Center, Nara Medical University), Dai Nakae (Tokyo Metropolitan Institute of Public Health), Shuzo Otani (Department of Biochemistry, Osaka City University Graduate School of Medicine), Tomoyuki Shirai (Department Pathology, Nagoya City University Graduate School of Medicine), Michihito
220
CHAPTER 8 THRESHOLDS FOR GENOTOXIC CARCINOGENS
Takahashi (Div. Pathology, National Institute of Health Sciences), Masae Tatematsu (Division of Oncological Pathology, Aichi Cancer Center Research Institute), Hiroyuki Tsuda (Department of Molecular Toxicology, Nagoya City University Graduate School of Med.), and Keiji Wakabayashi (Cancer Prevention Research Division, National Cancer Center Research Institute). The authors would also like to acknowledge the encouragement of Dr. Nobuyuki Ito (Professor Emeritus, Nagoya City University Medical School, Nagoya, Japan) and Dr. Tomoyuki Kitagawa (Institute Director Emeritus, Japanese Foundation for Cancer Research, Tokyo, Japan). These studies were supported by a grant from the Japan Science and Technology Corporation, included in the Project of Core Research for Evolutional Science and Technology (CREST), and by a grant from the Ministry of Education, Culture, Sports, Science and Technology of Japan.
REFERENCES Doi, K., Wanibuchi, H., Salim, E. I., Morimura, K., Kinoshita, A., Kudoh, S., Hirata, K., Yoshikawa, J., and Fukushima, S. (2005). Lack of large intestinal carcinogenicity of 2-amino-1-methyl-6phenylimidazo[4,5-b]pyridine at low doses in rats initiated with azoxymethane. Int J Cancer 115, 870–878. Fukushima, S., Wanibuchi, H., Morimura, K., Wei, M., Nakae, D., Konishi, Y., Tsuda, H., Uehara, N., Imaida, K., Shirai, T., Tatematsu, M., Tsukamoto, T., Hirose, M., Furukawa, F., Wakabayashi, K., and Totsuka, Y. (2002). Lack of a dose–response relationship for carcinogenicity in the rat liver with low doses of 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline or N-nitrosodiethylamine. Jpn J Cancer Res 93, 1076–1082. Fukushima, S., Wanibuchi, H., Morimura, K., Wei, M., Nakae, D., Konishi, Y., Tsuda, H., Takasuka, N., Imaida, K., Shirai, T., Tatematsu, M., Tsukamoto, T., Hirose, M., and Furukawa, F. (2003). Lack of initiation activity in rat liver of low doses of 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline. Cancer Lett 191, 35–40. Fukushima, S., Wanibuchi, H., Morimura, K., Iwai, S., Nakae, D., Kishida, H., Tsuda, H., Uehara, N., Imaida, K., Shirai, T., Tatematsu, M., Tsukamoto, T., Hirose, M., and Furukawa, F. (2004). Existence of a threshold for induction of aberrant crypt foci in the rat colon with low doses of 2-amino-1-methyl6-phenolimidazo[4,5-b]pyridine. Toxicol Sci 80, 109–114. Fukushima, S., Wanibuchi, H., Morimura, K., Nakae, D., Tsuda, H., Imaida, K., Shirai, T., Tatematsu, M., Tsukamoto, T., Hirose, M., and Furukawa, F. (2005). Lack of potential of low dose Nnitrosodimethylamine to induce preneoplastic lesions, glutathione S-transferase placental form-positive foci, in rat liver. Cancer Lett 222, 11–15. Hengstler, J. G., Bogdanffy, M. S., Bolt, H. M., and Oesch, F. (2003). Challenging dogma: thresholds for genotoxic carcinogens? The case of vinyl acetate. Annu Rev Pharmacol Toxicol 43, 485–520. Hoshi, M., Morimura, K., Wanibuchi, H., Wei, M., Okochi, E., Ushijima, T., Takaoka, K., and Fukushima, S. (2004). No-observed effect levels for carcinogenicity and for in vivo mutagenicity of a genotoxic carcinogen. Toxicol Sci 81, 273–279. Kang, J. S., Wanibuchi, H., Morimura, K., Totsuka, Y., Yoshimura, I., and Fukushima, S. (2006). Existence of a no effect level for MeIQx hepatocarcinogenicity on a background of thioacetamideinduced liver damage in rats. Cancer Sci 97, 453–458. Kato, T., Ohgaki, H., Hasegawa, H., Sato, S., Takayama, S., and Sugimura, T. (1988). Carcinogenicity in rats of a mutagenic compound, 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline. Carcinogenesis 9, 71–73. Kuraoka, I. (2008). Effects of DNA lesions on transcription elongation by RNA polymerases. Genes Environ 30, 63–70. Murai, T., Mori, S., Kang, J. S., Morimura, K., Wanibuchi, H., Totsuka, Y., and Fukushima, S. (2008). Evidence of a threshold-effect for 2-amino-3,8-dimethylimidazo-[4,5-f]quinoxaline liver carcinogenicity in F344/DuCrj rats. Toxicol Pathol 36, 472–477.
REFERENCES
221
OECD (1981). Carcinogenicity Studies. OECD Guideline for Testing of Chemicals 451, 1–17. Peto, R., Gray, R., Brantom, P., and Grasso, P. (1991). Effects on 4080 rats of chronic ingestion of Nnitrosodiethylamine or N-nitrosodimethylamine: A detailed dose–response study. Cancer Res 51, 6415–6451. Tsuda, H., Fukushima, S., Wanibuchi, H., Morimura, K., Nakae, D., Imaida, K., Tatematsu, M., Hirose, M., Wakabayashi, K., and Moore, M. A. (2003). Value of GST-P positive preneoplastic hepatic foci in dose–response studies of hepatocarcinogenesis: Evidence for practical thresholds with both genotoxic and nongenotoxic carcinogens. A review of recent work. Toxicol Pathol 31, 80–86. Wei, M., Hori, T. A., Ichihara, T., Wanibuchi, H., Morimura, K., Kang, J. S., Puatanachokchai, R., and Fukushima, S. (2006). Existence of no-observed effect levels for 2-amino-3,8-dimethylimidazo[4,5-f ] quinoxaline on hepatic preneoplastic lesion development in BN rats. Cancer Lett 231, 304–308. Wei, M., Hamoud, A.S., Yamaguchi, T., Kakehashi, A., Morimura, K., Doi, K., Kushida, M., Kitano, M., Wanibuchi, H., and Fukushima, S. (2009). Potassium bromate enhances N-ethyl-Nhydroxyethylnitrosamine-induced kidney carcinogenesis only at high doses in Wistar rats: indication of the existence of an enhancement threshold. Toxicol Pathol 37, 983–991. Yamaguchi, T., Wei, M., Hagihara, N., Omori, M., Wanibuchi, H., and Fukushima, S. (2008). Lack of mutagenic and toxic effects of low dose potassium bromate on kidneys in the Big Blue rat. Mutat Res 652, 1–11.
PART
III
GENETIC TOXICOLOGY, TESTING GUIDELINES AND REGULATIONS, AND NOVEL ASSAYS
CH A P TE R
9
DEVELOPMENT OF GENETIC TOXICOLOGY TESTING AND ITS INCORPORATION INTO REGULATORY HEALTH EFFECTS TEST REQUIREMENTS Errol Zeiger
9.1.
INTRODUCTION
Genetic toxicology testing—the testing for the ability of substances to produce mutations or chromosome aberrations, or otherwise damage DNA—has been central to the safety evaluation of chemicals since the mid- to late 1970s. Concern for induction of genetic damage began with concern for heritable gene and chromosomal germ cell mutations in the offspring of exposed individuals. This concern was reflected in the early guidance documents (see, e.g., Crow 1968; DHEW 1969; Drake et al. 1975; EPA 1975, 1979, 1980; Flamm et al. 1977; NRC 1983) that were produced by various agencies and scientific societies. However, with the accumulating evidence that mutagenesis was an early step in the development of a tumor and that carcinogenic chemicals were mutagenic (Ames 1971; Ames et al. 1973; McCann et al. 1975; Sugimura et al. 1976; Purchase et al. 1978), the genetic toxicology testing emphasis switched from heritable mutations to carcinogenesis. This concern for heritable effects, although valid, but will not be addressed here. The continuing concern for heritable mutations in addition to carcinogenesis is reflected in the US Environmental Protection Agency’s Office of Prevention, Pesticides and Toxic Substances (EPA OPPTS) testing schemes (Auletta et al. 1993) and in the international Organization for Economic Co-operation and Development (OECD) international harmonized test guidelines (OECD 2001), which classifies mutagens based on their potential for causing heritable mutations in humans. One reason behind this shift in concern is because there are a number of demonstrated human carcinogens but no demonstrated human germ cell mutagens. Another reason is that the chemicals
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
225
226
CHAPTER 9 DEVELOPMENT OF GENETIC TOXICOLOGY TESTING
that are known germ cell mutagens tend to also be in vivo somatic cell mutagens and carcinogens. Therefore, although the public health concerns for germ cell mutagenicity is high and taken into consideration by the regulatory authorities, it is easier and more practical to regulate mutagens based on their potential carcinogenicity. As a consequence of testing compilations first published in the 1970s which showed high cancer predictivities for the in vitro genetic toxicity tests, the United States and other regulatory authorities began requiring premarket genetic toxicity testing for chemicals and drugs. The in vitro genetic toxicity assays used internationally for regulatory approval of chemicals are the bacterial (Salmonella; E. coli), mammalian cell mutagenicity (L5179Y mouse lymphoma cells; CHO cells), and/or mammalian cell chromosome damage (L5178Y, CHO, CHL cells, or human lymphocytes) assays (see Chapter 11). In vivo testing uses primarily the rodent bone marrow cell chromosome aberration or micronucleus assay (Chapter 12). Substances that are positive in the in vitro tests are considered to be of the most concern for inducing cancer or genetic mutations in rodents and, by extension, in humans. As a consequence of these early studies, formal guidelines for conducting the tests were developed and recommended by U.S. and international organizations (EPA 2008; FDA 2000; OECD 2008; ICH 2008). These in vitro positives are then tested in rodents to determine if they have the capability of inducing genetic damage in the animal. In vivo genetic toxicity testing currently is also a prerequisite for identifying germ cell mutagens—that is, those that have the potential to mutate sperm or egg cells resulting in offspring either expressing or only carrying a mutant gene (see Chapter 10). The apical endpoint, cancer or germ cell mutagenicity, is currently demonstrated and quantified by extensive animal experiments because the genetic toxicity assays and the structure–activity relationship models are not sufficiently accurate predictors of these effects, or of their dose–responses, to support human health and safety decisions. In many cases, the high cost and lengthy nature of the follow-up in vivo tests, coupled with the high probability that the chemical would be tumorigenic or produce germ cell mutations, will lead companies to drop the chemical from further consideration without performing the confirmatory in vivo test. This chapter is designed to present a brief overview of the development of genetic toxicity testing for regulatory purposes; specifically the identification and characterization of carcinogens. It also presents supporting rationales for the types of tests mandated and the use of the data, and it identifies scientific and practical issues that will need to be resolved in the near future. More detailed descriptions of the tests, testing strategies, and decision processes are addressed elsewhere in this volume.
9.2.
DEFINITIONS AND USAGE
The terms mutagenic and genotoxic are often used interchangeably, although they are not the same, so that it is important to clarify the distinction between genotoxicity and mutagenicity. Mutagenicity includes gene mutations (either point mutations or deletions), chromosome breaks and rearrangements, and aneuploidies. A mutagenic
9.3. THE HISTORICAL DEVELOPMENT OF GENETIC TOXICITY TESTING
227
event is, by definition, heritable and will be passed to daughter (F1) cells (somatic cell tests) or to the offspring (germ cell tests). This means that genetic damage that is not compatible with cell survival or reproduction will not lead to a mutant organism. In contrast, genotoxicity is a broad term that includes mutagenicity, but also includes interaction with or damage to DNA, adduct formation, interference with the DNA replication or repair processes, and other nonspecific DNA-related effects. Genotoxic events do not always lead to mutagenicity and, if they are not toxic, may have no noticeable or lasting effects on the cell. Point mutations—that is, changes in single DNA bases or intragenic deletions and rearrangements—are considered to be heritable effects because they are typically measured in the post-treatment generation cells. Chromosome breakage (clastogenicity) is typically measured in the treated cells. These effects can be heritable and are the cause of many genetic diseases, although most clastogenicity seen in genetic toxicology tests is not compatible with cell survival and would therefore not result in a heritable effect. However, for testing purposes, the presence of chromosome breaks or rearrangements is evidence that the substance will cause heritable effects even though the test, itself, does not measure whether the effects seen will allow the cell to divide. An exception to this is the measurement of micronuclei (MN), which can be the effect of chromosome breaks or aneuploidy and are measured in the post-treatment-generation cells. Mutagens can also be classified as direct or indirect. This approach to classification has become an area of great interest in the regulatory agencies and industry. Direct mutagens are those that directly interact with and damage DNA, either as the parent compound or as a metabolite. Indirect mutagens act via two different means. They act either through the generation of intermediate molecules, such as active oxygen species, that subsequently react with DNA, or by interfering with the cell’s replicative proteins (either DNA synthesis or repair) or with the mitotic spindle. The determination of whether or not a substance is mutagenic as a result of its direct interaction with DNA, or as a secondary effect of other cellular interactions or reactions, is central to the regulation of chemicals causing mutation and cancer—that is, whether the cancer dose extrapolation should be linear, nonlinear, or threshold—and is addressed further elsewhere in this volume. In practice, in the absence of positive human epidemiological studies, the cancer response in the rat or mouse by a specific chemical is considered to be definitive for the identifying the chemical as a presumptive human carcinogen and for determining the relative carcinogenic potency in the exposed individuals. The exceptions to this practice are situations where it can be shown that the rodent carcinogenicity occurs by a mechanism(s) that is not operative in humans.
9.3. THE HISTORICAL DEVELOPMENT OF GENETIC TOXICITY TESTING A history of the development of genetic toxicity testing can be found in Zeiger (2004). Briefly, in the early 1950s, chemicals were tested for mutagenicity in E. coli using suspension and plate tests (Demerec et al. 1951; Hemmerly and Demerec
228
CHAPTER 9 DEVELOPMENT OF GENETIC TOXICOLOGY TESTING
1955). Subsequently, Szybalski (1958) reported on the screening of more than 400 chemicals for mutagenicity by applying the chemicals on a filter paper disk in petri dishes containing an E. coli strain spread on the agar. The E. coli strain used at that time responded only to a single base change and was relatively insensitive. This was also prior to the use of liver enzyme systems to provide mammalian metabolism. Ames (1971) adopted the spot-test method for screening mutagens using Salmonella typhimurium histidine mutant strains, which was followed by the development of in vitro metabolic activation systems and the currently used plate test. Protocols for performing the test were subsequently published (Ames et al. 1973; Maron and Ames 1983; Mortelmans and Zeiger 2000). The initial publication by McCann et al. (1975) that approximately 90% of all rodent carcinogens and noncarcinogens could be predicted by the Salmonella test was followed by similar studies and values in other laboratories (e.g., Purchase et al. 1978; Sugimura et al. 1976). Some of the differences in responses among the different compilations can be attributed to different classes of chemicals tested. The early (i.e., pre-1985) results led to the incorporation of this, as well as mammalian cell mutagenicity tests, into regulatory and industrial decision making, although similar data were not always available for the mammalian cell tests. Later studies of the effectiveness of the Salmonella and the mammalian cell tests produced lower predictivity values (e.g., Dunkel et al. 1985; Tennant et al. 1987; Zeiger et al. 1990; Zeiger 1998; Kirkland et al. 2005), but not low enough to remove them from their status as cancer-predictive tests. Unfortunately, a number of the compilations used datasets with very high frequencies of carcinogens, so that tests tending to be positive appeared to be highly effective, although high proportions of noncarcinogens were also detected as positive; that is, the specificities were low. The in vitro mammalian cell tests currently in use were also developed and/ or refined in the late 1960s to early 1970s (Zeiger 2004), and they comprised tests for gene mutation, chromosome aberration, and sister chromatid exchanges (SCE). The SCE tests were initially viewed as an alternative to the chromosome aberration assays because they were easier and less expensive to perform. However, they subsequently dropped out of favor based primarily on their performance in National Toxicology Program (NTP) validation studies (Tennant et al. 1987; Zeiger et al. 1990) and because of questions concerning their relevance to heritable genetic effects and cancer initiation. More recently, the in vitro MN test in mammalian cells has been proposed as an alternative measure of chromosome damage to the aberration test because, like the SCE test, it is easier and less expensive to perform than the chromosome aberration test, but it can also be used to distinguish between MN caused by chromosome aberrations and nondisjunction (Kirsch-Volders et al. 2000; Parry et al. 1996).
9.4.
TYPES OF AVAILABLE TESTS
By the late 1970s a large number of diverse tests had been developed or adapted for carcinogen screening in the hope that they would be useful as a replacement or adjunct to the Salmonella test (now called the Ames test) which had become the
9.5. TESTING APPROACHES
229
benchmark. A compilation by Hollstein et al. (1979) identified 119, while a later compilation (IARC 1987) identified 173, test systems or endpoints and included plant, insect, microbial, and mammalian in vitro and in vivo tests. Some of these tests were fairly widely used, with reports of the testing of many chemicals, while most had seen limited use (often only in the laboratory of the test developer) and had a database of relatively few chemicals which tended to be potent mutagens and clastogens. A number of other mammalian or microbial cell lines, and in vivo systems, have been proposed since then, and the endpoints measured have extended to new molecular effects. It has been estimated (Zeiger, unpublished) that there are 200–300 such test systems available at the present time, or reported in the literature. The majority of these test systems have not been systematically examined for their ability to discriminate between carcinogens and noncarcinogens. The currently used test systems for genetic toxicity for health effects testing for regulatory submissions, along with their EPA and OECD Test Guidelines, where they exist, are listed in Table 9.1. The selection of these tests does not necessarily signify that they are the best, or the only ones available for the particular endpoints, but was made based on information available in the 1970s–1980s and familiarity with their use and was encouraged by the reputations or persistence of the individual scientists or agencies advocating the tests. With time, with the exception of the Salmonella and E. coli tests, the plant tests and nonmammalian tests (e.g., yeast, Drosophila) were considered to be less relevant for human health prediction than were the mammalian tests and are no longer performed for regulatory submissions.
9.5.
TESTING APPROACHES
In order to make sense of the large number of available genetic toxicity tests and to simplify their use for supporting regulatory decisions, Bridges (1973) proposed a tier testing scheme for identifying potential carcinogens and germ cell mutagens, which was further elaborated on by Flamm (1974) and Bridges (1976). This tier approach forms the basis for the majority of the current regulatory testing schemes. In its early form, the initial tier would comprise in vitro tests for gene mutation and chromosome damage that are highly sensitive so as not to miss any potential in vivo mutagens. One requirement for the initial testing tier was that it not produce too many false negatives, with the consideration that the higher, in vivo, tiers would be capable of distinguishing between the “true” and the “false” positives. The second tier would consist of in vivo mammalian tests for the same endpoints to confirm the in vitro positive findings and/or to ensure that high exposure substances that were negative in vitro would also be negative in vivo. These first two tiers would be used to provide qualitative data on potential mutagenicity or clastogenicity in somatic and/or germ cells. The third, and final, tier would comprise apical in vivo rodent germ cell tests that could be used for quantitative genetic risk assessment of chemicals that were positive in tier 1 and/or 2 but considered sufficiently valuable for further development or study despite their potential mutagenicity and carcinogenicity. It is worth noting that these testing schemes were proposed at the time that the predictivity of the short term in vitro tests for carcinogenicity was believed to be
230
CHAPTER 9 DEVELOPMENT OF GENETIC TOXICOLOGY TESTING
TABLE 9.1. Tests Currently Used for Genetic Toxicity Screening and for Regulatory Approval of Commercial Chemicalsa
Test Guideline No. Test
Example(s)
Effect Measured
OECD
U.S. EPA
Gene mutations
471
870.5100
DNA damage repair Gene mutations
476
870.5500 870.5300
Chromosome damage; nondisjunction
473 487b 487b
870.5375 — —
DNA damage
482 —c —
870.5550 — —
475 474 — —
870.5385 870.5395 — —
486d
—
Chromosome damage
483
870.5380
Chromosome damage incompatible with embryo survival Heritable (to F1) chromosome rearrangements Gene or chromosome damage in F1
478
870.5450
485
870.5460
In Vitro—Bacterial and Mammalian Cells Bacterial mutagenicity Bacterial DNA damage Mammalian cell mutation Mammalian cell cytogenetics
Mammalian cell DNA damage
Ames (Salmonella) test; E. coli test SOS test Mouse lymphoma test; CHO-hprt test CHO, CHL, or human lymphocyte chromosome aberration or MN test UDS; comet assay; adduct formation
In Vivo—Rodent Somatic Cells Bone marrow cytogenetics Transgenic rodent gene mutation DNA damage
Aberrations; micronuclei; aneuploidy BigBlue mouse; MutaMouse Liver UDS; comet assay; DNA adducts
Chromosome damage; nondisjunction Gene mutations in various tissues DNA damage leading to strand breaks
In Vivo—Rodent Germ Cells Male germ cell cytogenetics Sperm cell chromosome damage Heritable sperm cell chromosome damage Heritable gene mutations
Spermatogonial, spermatocyte cytogenetics Dominant lethal assay
Heritable translocation test Mouse-specific locus test
—
870.5195; 870.5200
a
This listing is not exhaustive, but includes the test systems currently addressed by formal test guidelines, or which may be recommended for health effects screening or subsequent testing. The use of these tests for regulatory submissions is addressed in more detail elsewhere in this volume. b
OECD Guideline (No. 487) for in vitro MN and aneuploidy tests is being developed, but is not approved at the time of this writing. It is not anticipated to be formalized before 2010.
c
—, No Test Guideline available.
d
Test Guideline only for liver UDS test.
9.5. TESTING APPROACHES
231
approximately 90%. Although the performance of germ cell mutagenicity tests was not as well quantified, this endpoint was considered to be of equal importance to carcinogenicity. Such a tier system is based on a number of premises about the relationships among the different tests and endpoints and cancer. The basic premises derived from the above publications, along with their explicit or implicit rationales, can be summarized as follows: Premise #1. The Salmonella mutation test is a necessary component of genetic toxicity testing schemes. Gene mutations are a necessary, if not sufficient, inducer of the tumorigenic process. The test is mechanistically simple and the easiest to perform, and it has been validated more extensively than the other tests. It is also less susceptible than the in vitro mammalian cell tests to artifactual positive results. Despite the fact that the bacterial chromosome is structurally and functionally different from the mammalian chromosome, substances that directly damage or adduct nucleotides in the DNA helix would be expected to act similarly in both chromosome types. E. coli mutation tests are performed in addition to Salmonella for some regulatory needs (Gatehouse et al. 1994). Premise #2. Tests for chromosome aberrations in mammalian cells are needed in addition to gene mutation tests. Chromosome aberrations are the classical genotoxic response, are involved in the tumor initiation and development processes, and are associated with a large proportion of human genetic diseases. Chromosome aberrations can also be used as a biomonitor of exposure; thus, such results in test systems can be correlated with chromosome damage events in humans. Additionally, there are genotoxic and carcinogenic chemicals that produce chromosome aberrations but not gene mutations, which would be not be identified if only gene mutation tests were used. Premise #3. A mammalian cell mutagenicity test is needed to confirm or complement the Salmonella mutation test. Mammalian cell tests are considered to be more relevant for mammalian carcinogenesis than are microbial or other nonmammalian tests because of the similarity of mammalian chromosomes and DNA repair and replication processes across species. Positive results in mammalian cells for a substance that is positive in bacterial cells ensure that the result seen in bacteria was not unique to the bacterial chromosome or bacterial metabolism. However, a negative result in mammalian cell tests would not necessarily negate the implications of a positive response in the bacterial test. Premise #4. An in vivo test is needed to confirm a positive in vitro test. In vivo tests are more relevant than in vitro tests because they integrate the relevant factors of test chemical absorption, distribution, metabolism, and excretion. As a result of these considerations, and because they use the animal’s metabolism rather than a surrogate metabolic system (i.e., a liver homogenate with cofactors),
232
CHAPTER 9 DEVELOPMENT OF GENETIC TOXICOLOGY TESTING
in vivo tests should be less likely than in vitro tests to produce “false” positives. The ability to administer the test substance by the routes and doses relevant to human exposures allows for extrapolation of the results to humans. In addition, in vivo tests provide the potential to identify genetic damage in all tissues of interest as well as in germ cells, which cannot be adequately studied in vitro. Premise #5. Results from test batteries have a higher predictivity for cancer induction than results from the individual component tests. Gene mutations, chromosome aberrations, and nondisjunction are initial steps in the initiation of cancer; therefore chemicals causing any of these events are also capable of initiating cancer. Some chemicals induce only one type of relevant genetic damage—that is, gene mutations, chromosome aberrations, or nondisjunction—and therefore tests are needed for all endpoints. In order to be effective as a test battery, the combination of tests should be more effective than the individual tests for identifying chemicals of concern, and the tests must be complementary. That is, they must measure different effects and not duplicate each other to any large extent. In an ideal test battery, “false” negatives in one test will be correctly identified by one or more of the other tests. The accumulated data from the above testing schemes has shown that the in vitro tests are not complementary. The chemicals that are positive in the Salmonella test tend to also be positive in the mammalian cell tests, regardless of whether they are true positives (i.e., carcinogens) or “false” positives (i.e., noncarcinogens) (Zeiger 1998, 2001). Even more relevant to the interpretation of the test battery results has been the showing that the in vitro mammalian cell tests produce a high rate of false positives (Zeiger et al. 1990; Zeiger 1998, 2001; Kirkland et al. 2005). Every test has its false positives and false negatives; these values are generally quantified during validation tests and are partly a function of the chemicals being tested. When a battery of tests is used, each test that is added to the battery brings along its own, unique, true and false positives. Therefore, the more tests that are added to the battery to fill in the “gaps” left by the other tests, the more true positives that will be detected. This comes with a price: A higher proportion of false positives will also be detected. The additional false positives have the potential to overwhelm the number of true positives in a screening situation where the tests have a high sensitivity and the majority of test chemicals are not anticipated to be positive; that is, there is a low prevalence of noncarcinogens in the tested population.
9.6.
WHERE ARE WE NOW?
The current regulatory testing schemes are based on tier or battery testing. The initial tests used are the in vitro bacterial and mammalian cell gene mutation and mammalian cell cytogenetics tests. Chemicals that are negative at this level are typically not tested further. Chemicals that produce genetic effects in vitro are generally tested in short-term in vivo somatic cell tests to address the simple question of whether
9.6. WHERE ARE WE NOW?
233
or not the in vitro genetic toxicity can be translated to the animal and therefore, presumably, be more of a risk for carcinogenicity or heritable damage than chemicals that are not detected in the in vivo tests. However, the in vivo tests (i.e., bone marrow cytogenetics, in vivo/in vitro unscheduled DNA synthesis, and the transgenic mouse mutation test) generally used tend to be less sensitive than the in vitro tests. As a consequence, carcinogenic, DNA-reactive chemicals that are readily detected in vitro are often not detected in the in vivo assays, which is why a negative in vivo test is not sufficient, by itself, to negate the implications of the positive in vitro test. Similar difficulties arise when attempting to confirm in vitro positives as germ cell mutagens. In general, it is assumed that chemicals that do not produce somatic cell mutations or chromosome aberrations in vivo will not produce germ cell genetic effects. This is why a negative in vivo somatic cell test is sufficient to conclude that the substance will not be positive in germ cell tests. As a result, the potential carcinogenicity of the in vivo chemicals often dominates the implications of heritable mutations. There are a number of potential reasons for this lack of concordance between in vitro and in vivo responses and the relatively high rate of positives in the in vitro tests compared to the in vivo tests, primary of which are: The in vivo doses to the target cells are often lower than are reached in vitro, or the active metabolite may not be sufficiently stable to reach the target cells. Other possibilities include chemicals that may be uniquely positive in bacteria as a result of bacterial metabolic pathways not found in mammalian cells (e.g., sodium azide; Owais et al. 1979), differences in activation (or inactivation) activities, and metabolite profiles between the in vitro S9 preparation and in vivo metabolism (Ku et al. 2007). Additionally, in vitro mammalian cell systems, specifically those measuring chromosome damage, can produce artifactual positives as a secondary effect of high toxicity, high osmolality, or changes in pH (Brusick 1986; Galloway et al. 1987). A number of studies since the early 1990s have shown that the testing approach directed by the above-mentioned premises is not as effective as originally thought for identifying potential carcinogens. The reduced concordance of the genetic toxicity tests with rodent carcinogenesis is not unexpected because these genetic toxicology tests measure gene mutations, chromosome aberrations, and other chromosome damage; they do not measure cancer. Justification for their initial and continued use comes from the mechanistic relationship between mutations, chromosome damage, and cancer, as well as from the empirical correlations developed in the 1970s using model carcinogens. The high proportions of carcinogens that are not mutagenic in vitro (which became an issue in the l980s) led to the category of nongenotoxic carcinogens, which included chemicals that initiated the carcinogenic process by other than direct DNA damage—for example, hormonal, epigenetic cytotoxicity with subsequent cell proliferation. The question always arises as to why the recent test performance values (e.g., Tennant et al. 1987; Zeiger et al. 1990; Zeiger 1998; Kirkland et al. 2005) are poorer than the initial compilations in the 1970s, which showed that the Ames test could correctly identify 90% or more of carcinogens and noncarcinogens (McCann et al. 1975; Sugimura et al. 1976; Purchase et al. 1978), and other in vitro tests were similarly, or slightly less, effective (Preston et al. 1981; Clive et al. 1983).
234
CHAPTER 9 DEVELOPMENT OF GENETIC TOXICOLOGY TESTING
The known carcinogens prior to 1980 comprised primarily alkylating agents and substances that were carcinogenic following short-term administration (i.e., less than 1 year) to the test animal. Simultaneous to these genetic toxicology test performance evaluations, the U.S. National Cancer Institute began a carcinogen testing program designed to identify carcinogenic chemicals by treating mice and rats for up to two years at doses up to what was described as a maximum tolerable dose (MTD; also defined as the minimally toxic dose) and performing an extensive histopathological evaluation of the animals at the end of this time (This testing program was subsequently incorporated into the National Toxicology Program). As opposed to many of the earlier carcinogenicity tests that exposed the animals for up to 1 year, the animals in this program were exposed for up to 2 years, which led to the appearance of tumors that are not normally expressed in less than 2 years of exposure. As a consequence of this expanded testing protocol, a number of chemicals were identified as carcinogens that would not have been identified in the shorter-term studies. Unlike the original listing of carcinogenic chemicals that were primarily DNAreactive chemicals, such as direct alkylating agents, aromatic amines, polycyclic aromatic hydrocarbons, and nitrosamines, many of the chemicals identified in the longer-term studies included such non-DNA reactive chemicals as chlorinated hydrocarbons, phthalates, and hormonally active substances. These latter classes of chemicals do not adduct DNA, tend to be less toxic, and tend not to induce gene mutations. These chemicals have contributed to the shift in the performance of the Ames test from a sensitivity of ≥90% to a sensitivity of 50–60% (i.e., a false-negative rate of 40–50%). The existence of this category of nongenotoxic carcinogens not detected by the Ames or other genetic tests has been well established. These substances will only be addressed by the development and validation of tests for other precancer mechanistic endpoints. In summary, the genetic toxicity tests routinely used for identifying potential carcinogens have not performed as effectively as their original promise, partially because the initial validation studies used potent carcinogens that were known or suspected to be DNA-reactive. Subsequent to those studies, a large number of rodent carcinogens have been identified that are not DNA-reactive and therefore are not detected, or poorly detected, in genetic toxicity tests. Similarly, a relatively high proportion of substances that are positive in mammalian cell systems have been shown to be noncarcinogenic. Despite these apparent deficiencies, the tests are widely used for screening chemicals to presumptively identify carcinogens. Although the sensitivity and specificity of the tests are not as high as originally anticipated, positive results in the genetic toxicity tests are highly predictive for rodent carcinogenicity. As a consequence of these issues, extensive efforts are underway to identify tests that are more predictive than the standard tests, or as predictive but with fewer false positives, which can be used to supplement or replace the tests currently used. Similarly, protocol modifications have been proposed for mammalian cell systems— for example, reducing the toxicity levels or test chemical concentrations that must be achieved for a test to be considered valid, which are designed to reduce the number of artifact-induced positive responses.
REFERENCES
9.7.
235
SUMMARY
The initial hopes and aspirations for the short-term genetic toxicology tests was the accurate prediction of carcinogenicity and the ability to distinguish between carcinogens and noncarcinogens. The in vitro tests remain the basis of carcinogen screening tests despite the knowledge that they are far from accurate for predicting carcinogenicity and are ineffective for identifying potential noncarcinogens (i.e., the specificities of the tests currently used are about 50%). In retrospect, it was naive to expect in vitro tests that measured point mutations and chromosome breakage to accurately reflect the multiple genetic and nongenetic steps between the induction of the initial DNA damage and the development of a tumor. The predictive ability of these short-term tests for mutation and chromosome breakage needs to be placed into context. The tests are designed and used to identify chemicals that cause cancer, with the ideal being the correct identification of carcinogens (by their genetic toxicity) and noncarcinogens (by their lack of genotoxicity). This search for the ideal should be compared with the interspecies predictivity of the in vivo cancer tests. In a compilation of tests performed by the NTP on rats and mice, in parallel, typically in the same laboratory, and using a larger database that went beyond the NTP studies and where the rat and mouse studies were not always performed in the same labs, the correspondence between rat and mouse carcinogenicity was 70–75% (Haseman et al. 1987; Gold et al. 1997). This level of interspecies predictivity of carcinogenicity under highly controlled conditions puts an upper limit on the predictivity of in vitro, single-cell systems and systems that measure clastogenicity or mutagenicity in single, typically nontarget, tissues of a single rodent species.
REFERENCES Ames, B. N. (1971). The detection of chemical mutagens with enteric bacteria. In Chemical Mutagens: Principles and Methods for Their Detection, Vol. 1, Hollaender, A., ed., Plenum Press, New York, pp. 267–282. Ames, B. N., Durston, W. E., Yamasaki, E., and Lee, F. D. (1973). Carcinogens are mutagens: A simple test system combining liver homogenates for activation and bacteria for detection. Proc Natl Acad Sci USA 70, 2281–2285. Auletta, A. E., Dearfield, K. L., and Cimino, M. C. (1993). Mutagenicity test schemes and guidelines: US EPA Office of Pollution Prevention and Toxics and Office of Pesticide Programs. Environ Mol Mutagen 21, 38–45. Bridges, B. A. (1973). Some general principles of mutagenicity screening and a possible framework for testing procedures. Environ Health Perspect 6, 221–227. Bridges, B. A. (1976). Short term screening tests for carcinogens. Nature 261, 195–200. Brusick, D. (1986). Genotoxic effects in cultured mammalian cells produced by low pH treatment conditions and increased ion concentrations. Environ Mutagen 8, 879–886. Clive, D., McCuen, R., Spector, J. F. S., Piper, C., and Mavournin, K. H. (1983). Specific gene mutations in L5178Y cells in culture. A report of the US Environmental Protection Agency Gene-Tox Program. Mutat Res 115, 225–251. Crow, J. F. (1968). Chemical risk to future generations. Scientist and Citizen, June–July, 113–117.
236
CHAPTER 9 DEVELOPMENT OF GENETIC TOXICOLOGY TESTING
Demerec, M., Bertani, G., and Flint, J. (1951). A survey of chemicals for mutagenic action on E. coli. Am Naturalist 85, 119–136. DHEW (US Department of Health, Education, and Welfare). (1969). Report of the Secretary’s Commission on Pesticides and Their Relationship to Environmental Health. Parts I and II. US GPO, December 1969. Drake, J. W., Abrahamson, S., Crow, J. F., Hollaender, A., Lederberg, S., Legator, M. S., Neel, J. V., Shaw, M. W., Sutton, H. E., Von Borstel, R. C., and Zimmering, S. (1975). Environmental mutagenic hazards. Science 187, 503–514. Dunkel, V. C., Zeiger, E., Brusick, D., McCoy, E., McGregor, D., Mortelmans, K., Rosenkranz, H. S., and Simmon, V. F. (1985). Reproducibility of microbial mutagenicity assays: II. Testing of carcinogens and noncarcinogens in Salmonella typhimurium and Escherichia coli. Environ Mutagen 7(Suppl 5), 1–248. EPA (1975). Pesticide Program. Guidelines for Registering Pesticides in United States. Fed Reg (Part II) 40(123), 26802–26928 (Part VII, Subpart A, Methods for Studying Mutagenicity, pp. 26899–26900). Wednesday, June 25, 1975. EPA (1979). Environmental Protection Agency. Proposed Health Effects Test Standards for Toxic Substances Control Act Test Rules and Proposed Good Laboratory Practice Standards for Health Effects. Fed Reg (Part IV) 44(145), 44054–44093. (Subpart E, Mutagenic Effects §772.144, pp. 44080–44087). Thursday, July 26, 1979. EPA (1980). Mutagenicity Risk Assessments; Proposed Guidelines. Fed Reg 45(221), 74984–74988. Thursday, November 13, 1980. EPA (2008). OPPTS Harmonized Test Guidelines. Series 870 Health Effects Test Guidelines— Final Guidelines. http://www.epa.gov/opptsfrs/publications/OPPTS_Harmonized/870_Health_Effects_ Test_Guidelines/Series/ FDA (2000). Toxicological Principles for the Safety Assessment of Food Ingredients Redbook 2000, July 2000. IV.C.1. Short-Term Tests for Genetic Toxicity. http://vm.cfsan.fda.gov/∼redbook/red-ivc1. html. Flamm, W. G. (1974). A tier system approach to mutagen testing. Mutat Res 26, 329–333. Flamm, W. G., Valcovic, L. R., D’Aguanno, W., Fishbein, L., Green, S., Malling, H. V., Mayer, V., Prival, M., Wolff, G., and Zeiger, E. (1977). Approaches to determining the mutagenic properties of chemicals: Risk to future generations. J Environ Pathol Toxicol 1, 301–352. Galloway, S. M., Deasy, D. A., Bean, C. L., Kraynak, A. R., Armstrong, M. J., and Bradley, M. O. (1987). Effects of high osmotic strength on chromosome aberrations, sister-chromatid exchanges and DNA strand breaks, and the relation to toxicity. Mutat Res 189, 15–25. Gatehouse, D., Haworth, S., Cebula, T., Gocke, E., Kier, L., Matsushima, T., Melcion, C., Nohmi, T., Ohta, T., Venitt, S., and Zeiger, E. (1994). Mutat Res 312, 217–233. Gold, L. S., Slone, T. H., and Ames, B. N. (1997). Overview and update of analyses of the carcinogenic potency database. In Handbook of Carcinogenic Potency and Genotoxicity Databases, Gold, L. S., and Zeiger, E., eds., CRC Press, Boca Raton, FL, pp. 661–685. Haseman, J. K., Huff, J. E., Zeiger, E., and McConnell, E. E. (1987). Comparative results of 327 chemical carcinogenicity studies. Environ Health Perspect 74, 229–235. Hemmerly, J., and Demerec, M. (1955). XIII. Tests of chemicals for mutagenicity. Cancer Res Suppl 3, 69–75. Hollstein, M., McCann, J., Angelosanto, F., and Nichols, W. (1979). Short-term tests for carcinogens and mutagens. Mutat Res 65, 133–226. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (1987). Genetic and Related Effects: An updating of selected IARC Monographs from Volumes 1 to 42, Supplement 6, Lyon, France. ICH (International Conference on Harmonisation) (2008). S2(R1): Guidance on Genotoxicity Testing and Data Interpretation for Pharmaceuticals Intended for Human Use; S2A: Guidance on Specific Aspects of Regulatory Genotoxicity Tests for Pharmaceuticals; S2B: Genotoxicity: A Standard Battery for Genotoxicity Testing for Pharmaceuticals. http://www.ich.org/cache/compo/276-254-1.html. Kirkland, D., Aardema, M., Henderson, L., and Muller, L. (2005). Evaluation of the ability of a battery of three in vitro genotoxicity tests to discriminate rodent carcinogens and noncarcinogens: I. Sensitivity, specificity and relative predictivity. Mutat Res 584, 1–256.
REFERENCES
237
Kirsch-Volders, M., Sofuni, T., Aardema, M., Albertini, S., Eastmond, D., Fenech, M., Ishidate, M., Jr., Lorge, E., Norppa, H., Surralles, J., von der Hude, W., and Wakata, A. (2000). Report from the in vitro micronucleus assay working group. Environ Mol Mutagen 35, 167–172. Ku, W. W., Bigger, A., Brambilla, G., Glatt, H., Gocke, E., Guzzie, P. J., Hakura, A., Honma, M., Martus, H.-J., Obach, R. S., and Roberts, S. (2007). Strategy for genotoxicity testing—Metabolic considerations. Mutat Res 627, 59–77. Maron, D. M., and Ames, B. N. (1983). Revised methods for the Salmonella mutagenicity test. Mutat Res 113, 173–215. McCann, J., Choi, E., Yamasaki, E., and Ames, B. N. (1975). Detection of carcinogens in the Salmonella/ microsome test: Assay of 300 chemicals. Proc Natl Acad Sci USA 72, 5135–5139. Mortelmans, K., and Zeiger, E. (2000). The Ames Salmonella/microsome mutagenicity assay. Mutat Res 455, 29–60. NRC (National Research Council) (1983). Identifying and Estimating the Genetic Impact of Chemical Mutagens. Committee on Chemical Environmental Mutagens, National Academies Press, Washington D.C., 295 pages. OECD (Organization for Economic Co-operation and Development) (2001). OECD Series on Testing and Assessment, No. 33. Harmonised Integrated Hazard Classification System for Chemical Substances and Mixtures. Chapter 2.5. Harmonised System for the Classification of Chemicals which Cause Mutations in Germ Cells. OECD (Organization for Economic Co-operation and Development) (2008). OECD Guidelines for the Testing of Chemicals. Section 4: Health Effects. [specifically, Guideline nos. 471–486] http://www. oecd.org/document/55/0,3343,en_2649_34377_2349687_1_1_1_1,00.html. Owais, W. M., Kleinhofs, A., and Nilan, R. A. (1979). In vivo conversion of sodium azide to a stable mutagenic metabolite in Salmonella typhimurium. Mutat Res 68, 15–22. Parry, J. M., Parry, E. M., Bourner, R., Doherty, A., Ellard, S., O’Donovan, J., Hoebee, B., de Stoppelaar, J. M., Mohn, G. R., Onfelt, A., Renglin, A., Schultz, N., Soderpalm-Berndes, C., Jensen, K. G., KirschVolders, M., Elhajouji, A., Van Hummelen, P., Degrassi, F., Antoccia, A., Cimini, D., Izzo, M., Tanzarella, C., Adler, I.-D., Kliesch, U., Schriever-Schwemmer, G., Gasser, P., Crebelli, R., Carere, A., Andreoli, C., Benigni, R., Leopardi, P., Marcon, F., Zijno, Z., Natarajan, A. T., Boei, J. J. W. A., Kappas, A., Voutsinas, G., Zarani, F. E., Patrinelli, A., Pachierotti, F., Tiveron, C., and Hess, P. (1996). The detection and evaluation of aneugenic chemicals. Mutat Res 353, 11–46. Preston, R. J., Au, W., Bender, M. A., Brewen, J. G., Carrano, A. V., Heddle, J. A., McFee, A. F., Wolff, S., and Wassom, J. S. (1981). Mammalian in vivo and in vitro cytogenetic assays: A report of the US EPA’s Gene-Tox Program. Mutat Res 87, 143–188. Purchase, I. F. H., Longstaff, E,, Ashby, J., Styles, J. A., Anderson, D., Lefevre, P. A., and Westwood, F. R. (1978). An evaluation of 6 short-term tests for detecting organic chemical carcinogens. Br J Cancer 37, 873–959. Sugimura, T., Sato, S., Nagao, M., Yahagi, T., Matsushima, T., Seino, Y., Takeuchi, M., and Kawachi, T. (1976). Overlapping of carcinogens and mutagens. In Fundamentals of Cancer Prevention, Magee, P. N., Takayama, S., Sugimura, T., and Matsushima, T. eds., University Park Press, Baltimore, MD, pp. 191–215. Szybalski, W. (1958). Special microbiological systems. II. Observations on chemical mutagenesis in microorganisms. Ann NY Acad Sci 76, 475–489. Tennant, R. W., Margolin, B. H., Shelby, M. D., Zeiger, E., Haseman, J. K., Spalding, J., Caspary, W., Resnick, M., Stasiewicz, S., Anderson, B., and Minor, R. (1987). Prediction of chemical carcinogenicity in rodents from in vitro genetic toxicity assays. Science 236, 933–941. Zeiger, E. (1998). Identification of rodent carcinogens and noncarcinogens using genetic toxicity tests: premises, promises and performance. Regul Toxicol Pharmacol 28, 85–95. Zeiger, E. (2001). Mutagens that are not carcinogens: Faulty theory or faulty tests? Mutat Res 492, 29–38. Zeiger, E. (2004). The history and rationale of genetic toxicity testing—An impersonal, and sometimes personal, view. Environ Mol Mutagen 44, 363–371. Zeiger, E., Haseman, J. K., Shelby, M. D., Margolin, B. H., and Tennant, R. W. (1990). Evaluation of four in vitro genetic toxicity tests for predicting rodent carcinogenicity: Confirmation of earlier results with 41 additional chemicals. Environ Mol Mutagen 16(Suppl 18), 1–14.
CH A P TE R
10
GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS Lutz Müller Hans-Jörg Martus
10.1. HISTORICAL OVERVIEW OF GENOTOXICITY TESTING GUIDELINES As early as in the late 1940s, it was Auerbach who demonstrated that chemicals could be powerful mutagens (Auerbach and Robson 1946). Over the years, this created a concern that exposure to environmental chemicals could introduce deleterious alterations in the DNA of human beings in the environment. These concerns included damage to the germ line that could cause heritable disease and genetic alterations to individuals via somatic DNA damage (DHEW 1977; Meselson 1971; Wassom 1989). These concerns led to formation of the Environmental Mutagen Society in 1969 (Wassom 1989) and to the introduction of requirements for testing for mutagenic properties of chemicals in the 1970s. In this context, the U.S. Toxic Substances Control Act of 1976 specifically required the U.S. Environmental Protection Agency (EPA) to establish standards for the assessment of health and environmental effects associated with mutagenesis (TSCA 1976). During this period, the primary focus was on the potential of any chemical to induce germ-line mutations and to the development of appropriate testing methodologies for assessment of heritable mutations (Ehling et al. 1978; Meselson 1971). The thinking of the field at this key stage, when the recognition of the need for genetic toxicology testing had led to the initial formulation of testing requirements for genotoxicity, is illustrated by a key report of the department-wide working group of the U.S. Department of Health Education and Welfare (DHEW) (present name Department of Health and Human Services) issued in 1977 (DHEW 1977). This working group, formed by the DHEW Committee to Coordinate Toxicology and Related Programs, Subcommittee on Environmental Mutagenesis, was established in 1974 to develop a background document on mutagenicity test procedures and approaches to testing chemicals for mutagenic activity. The intent was “… to aid officials of regulatory agencies who have the responsibility for deciding:
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
238
10.1. HISTORICAL OVERVIEW OF GENOTOXICITY TESTING GUIDELINES
239
(1) advisability of promulgating test requirements for mutagenicity at the present time under any of their legislative authorities; (2) the appropriateness of mutagenicity tests for a wide range of product use and exposure categories; and (3) the reliability and interpretation of data from mutagenicity tests developed on substances of commerce within their regulatory purview in spite of the absence of formal testing requirements.” This report, entitled “Approaches to Determining the Mutagenic Properties of Chemicals: Risk to Future Generations,” emphasized two key points— first, that the primary concern about genotoxic damage was the potential to cause heritable genetic alterations in the human germ-line and, second, that quantitative assessment of the risk of heritable damage was necessary and that mere hazard identification was insufficient. The importance of quantitative risk assessment was emphasized: “It is not sufficient merely to identify substances which may pose a genetic hazard to the human population. Many such compounds will have a significant benefit factor and hence cannot reasonably be eliminated from use. Therefore, it is necessary to obtain quantitative data from relevant animal model systems from which extrapolation to humans can be made to predict virtually safe or tolerable levels of exposure.” Additionally, the association of mutagenesis with other toxicological endpoints such as carcinogenesis, teratogenesis, and aging was also noted. In the mid-1970s, the landmark publication of McCann et al. on the detection of carcinogens as mutagens based on an analysis of 300 chemicals, demonstrated a strong correlation of mutagenic activity in Salmonella with animal carcinogenicity (Ames et al. 1975; Maron and Ames 1983; McCann et al. 1975). This report generated great enthusiasm that inexpensive in vitro mutagenesis screening tests could be used to identify chemical carcinogens and hence control of exposure to such agents could potentially lower the human tumor burden. As regulatory guidelines were implemented during the 1970s and 1980s, there was a shift in focus from concern over germ-line mutagenesis to control of chemical carcinogens (MacGregor 1994). Though these early results in Salmonella were highly promising, it was already recognized at that time that mutations could arise by multiple mechanisms, some of which would not be detected in a nutritional reversion assay such as the Salmonella his reversion test. In particular, chromosomal interchanges, DNA strand breaks, and large chromosomal deletions, all characteristic of damage induced by ionizing radiation, which was one of the environmental mutagens of most concern during this period, are not efficiently detected in the Ames assay. Thus, an in vitro and in vivo test battery was devised that would detect the major classes of damage known to result in heritable mutations (NRC 1983). These concepts underlie the batteries currently in use (Brusick 1987; Hoffmann 1998). The types of lesions expected to be detected by the test systems most commonly used for mutagenesis screening at the present time are in line with our knowledge about the types of lesions involved in modifying the activity of oncogene products and tumor suppressor gene products. In the meantime, changes in such genes are widely accepted to be associated with cancer risk. Guidelines for testing environmental chemicals in the United States were delineated during the 1970s and 1980s (Auletta et al. 1993; Waters and Auletta 1981) and for food additives in 1982 (FDA 1982). Classically, the first batteries included (1) a bacterial test for gene mutation, (2) either an in vitro test for chromosomal
240
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
aberrations (based on the knowledge that ionizing radiation and radiomimetic chemicals produced high rates of chromosomal aberrations even when induced mutation rates were relatively low) or a mammalian cell mutagenesis test, and (3) a general test for DNA damage (FDA 1982). An in vivo test was generally encouraged, with preference for a test for bone marrow chromosomal aberrations or micronucleus induction, based on the knowledge of a few chemicals that were uniquely active in vivo (ICH 1997b; Tweats et al. 2007a). Much research effort was focused on development of appropriate mutagenicity testing methods that would detect a broad array of mutagenic chemicals. The classical series initiated by Hollaender, Chemical Mutagens: Principles and Methods for their Detection, was devoted to summarizing these methodologies (Hollaender 1971). By the time of the 1993 draft revision of the U.S. Food and Drug Administration’s (FDA) guidance on testing requirements for food and color additives, the U.S. FDA-recommended “core” testing battery consisted of the following: (1) a test for gene mutation in bacteria (S. typhimurium), (2) a test for gene mutation in mammalian cells in vitro, with the recommendation that the endpoint be based on an autosomal locus (so that events related to chromosomal interchanges could be detected), and (3) an assay for cytogenetic damage in vivo, with preference for a rodent bone marrow assay (FDA 1993). By the year 2000, these so called “Redbook” guidelines were finalized (FDA 2000). At this same time, the European, Japanese, and Canadian recommendations were similar. However, there were distinct differences in requirements both among regions and within different regulatory agencies within each region (DOH 1991; Purves et al. 1995; Shelby and Sofuni 1991). For example, the European recommendations generally included both an assay for gene mutation and an assay for chromosomal aberrations in mammalian cells (Kirkland 1993), while the Japanese relied on an in vitro mammalian cell chromosomal aberration assay and did not necessarily include the in vitro mammalian cell mutagenesis assay (Shirasu 1988). Test practices regarding potential genotoxicity of pharmaceuticals, including test quality and assessment issues, have been delineated in a series of publications communicated by members of the German regulatory authorities (Madle et al. 1987; Müller and Kasper 2000; Müller et al. 1991). The evaluation spans the period between 1982 and 1997 and addresses nearly 600 new pharmaceutical entities. These publications summarize changes in test selection, improvements in test quality and shifts in the focus of test interpretation and assessment. The initial review (Madle et al. 1987) as well as its update (Müller et al. 1991) focused on deficiencies in test quality which was at that time considered to be a major issue. By the 1990s, this was no longer considered a major issue. In addition, some genotoxicity systems which played a considerable role in the 1970s and 1980s, such as assays using yeast as indicator organisms, host-mediated assays, sister chromatid exchange (SCE) tests in vitro or in vivo, chromosomal aberration analysis in bone marrow or spermatogonia, and dominant lethal assays, were little used by the 1990s. In part this reflects changes in test philosophy including a move away from assays involving cells of the germ line (Müller and Kasper 2000). A similar evaluation addressing the experience in Germany with the review of tests for 776 new chemicals reviewed between 1982 and 1997 has been published
10.1. HISTORICAL OVERVIEW OF GENOTOXICITY TESTING GUIDELINES
241
(Broschinski et al. 1998). This evaluation focuses on the rates of positives in various standard in vitro systems. These data are correlated with chemical structure characteristics and genotoxicity effects versus cytotoxicity. A later review of pharmaceuticals on the U.S. market appeared to indicate a similar tendency including the fact that ∼20–30% of marketed pharmaceuticals seem to possess some kind of genotoxic potential especially in mammalian cells under in vitro conditions (Snyder and Green 2001). Because such data seem to indicate indirect means of genotoxicity in vitro, which may lack relevance in vivo, further evaluations have focused in a broader context on this issue. Kirkland et al. have published an updated comparison of in vitro genotoxicity assay results with The Carcinogenicity Potency Database, the most comprehensive carcinogenicity database available (CPDB 2007; Kirkland et al. 2005a). This evaluation showed in general terms that a battery of in vitro tests for genotoxicity can be pushed to high levels of sensitivity for detection of rodent carcinogens (sensitivity), but this sensitivity came at the price of inappropriately increasing the likelihood of obtaining a positive genotoxicity result for noncarcinogens (specificity) (Figure 10.1). Matthews et al. (2006) obtained confirmation of these results, which included the use of proprietary data. In the views of many scientists, there is a growing lack of confidence in the results that come out of regulatory in vitro genotoxicity tests. On the one hand, some scientists see a benefit to
100
Specificity and Sensitivity (%)
90 80 70
MN
Ames + MLA + MN MLA + MN Ames + MN MLA + Cab Ames + MLA + Cab Ames + MLA Ames + Cab
Ames MLA Cab
60
Sensitivity
Ames
Specificity
50 Cab
40 30
MLA MN
Ames + Cab Ames + MLA MLA + Cab Ames + MLA + Cab
20 10
Ames + MN
MLA + MN Ames + MLA + MN
0
Figure 10.1. Correlation data for the sensitivity and specificity for single in vitro genotoxicity assays and assay combinations. Ames, Salmonella reverse mutation test introduced by B. Ames; MLA, tk assay using the L5178Y mouse lymphoma cell line; Cab, chromosome aberration test with different mammalian cell types; MN, in vitro micronucleus assay with different mammalian cell types. [Data according to Kirkland et al. (2005a).]
242
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
approach maximum sensitivity for hazard identification followed by a weight-ofevidence approach for risk assessment (Bergman et al. 1996; Cimino 2006; Dearfield and Moore 2005; FDA 2006; Kasper et al. 2007; Müller et al. 2003). Others, however, see a need to refine the conditions for in vitro mammalian cell genotoxicity tests to optimize their predictivity or to introduce new assays (Kasper et al. 2007; Kirkland et al. 2005a,b, 2007a; MacGregor et al. 2000; Müller et al. 2003). In part, this is driven by the economic circumstances of restricted resources in the industry and the need to prioritize expenditures. In Europe, efforts in this context culminated in a publication from a workshop held under the auspices of the European Center for Validation of Alternative Methods (ECVAM) entitled “How to reduce false positive results when undertaking in vitro genotoxicity testing and thus avoid unnecessary follow-up animal tests: Report of an ECVAM Workshop” (Kirkland et al. 2007a). In addition, the International Life Sciences Institute (ILSI) has instituted a working group that tackles “Relevance and follow-up of positive results in in vitro genetic toxicity assays: An ILSI-HESI initiative” (Thybaud et al. 2007a,b). Furthermore, recent changes to regulatory genotoxicity testing have been communicated with the revised ICH Guideline “Guidance on Genotoxicity Testing and Data Interpretation for Pharmaceuticals Intended for Human Use” (ICH 2008). These changes include (a) a reduction in the top concentrations for in vitro mammalian cell tests from 10 mM to 1 mM and (b) more stringent criteria for acceptable ranges of cytotoxicity for evaluation of test compounds. The scientific reasoning for these recommendations can be found later in this chapter. With this historical overview and short elaborations on recent regulatory developments on genotoxicity testing, interpretation, and risk assessment, this chapter will now focus on major sets of internationally relevant guidelines and scientific efforts to support the concepts of regulatory testing. In the views of the authors of this chapter, there are two major sets of internationally applicable regulatory guidelines for genotoxicity testing and two major international scientific processes that dominate the regulatory landscape: 1. The Organisation for Economic Cooperation and Development (OECD) test guidelines 2. The International Conference of Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) guideline(s) for genotoxicity testing of pharmaceuticals 3. The International Workshop(s) on Genotoxicity Tests (IWGT) 4. The International Program on Chemical Safety (IPCS) under auspices of the World Health Organization (WHO) These guidelines and processes are selected because they have been driving the scientific process (IWGT and IPCS) and have set internationally acknowledged standards of testing that go beyond country or regional borders (ICH and OECD). This chapter does not focus on genotoxicity guidelines for compounds for other purposes such as pesticides, new chemicals, food additives, and so on. The reader is referred to review articles that cover these regulations in more detail (e.g., Cimino 2006; Kirkland et al. 2005b) or to the specific guidelines such as those from the
10.3. HUMAN USE ICH GUIDELINES FOR PHARMACEUTICALS
243
United Kingdom’s Committee on Mutagens (UKCOM 2000), the U.S. Environmental Protection Agency’s (EPA’s) Office of Prevention, Pesticides and Toxic Substances’ health effects test guidelines series 870 (EPA 2008), and the European Union’s (EU) guidelines for the testing of chemicals under Annex V, Part B (EC 2008). It might be worthwhile to mention here that regulatory guidance documents on genotoxicity are partly influenced by general considerations prevailing in the society such as the three R’s (replace, refine, reduce) to optimize use of animals in safety testing and to reduce their burden. For example, in the European Union this has led to a ban of animal testing for decorative cosmetics, which require a focus on in vitro genotoxicity testing only (Kirkland et al. 2005b). This approach will require work on more predictive in vitro tests for the purpose of risk assessment (Kirkland et al. 2008).
10.2. ORGANIZATION FOR ECONOMIC COOPERATION AND DEVELOPMENT (OECD) GUIDELINES FOR GENOTOXICITY The OECD has played a major role in developing recommendations for internationally harmonized testing protocols. The protocols developed by OECD are very influential, because parties to the OECD treaty, which include most of the major industrialized nations of the world, have agreed to accept testing protocols developed by the OECD consensus process. These guidelines also served as the bases from which most of the U.S. EPA’s health effects guidelines and the EU’s Annex V testing guidelines were developed. OECD testing guidelines are updated periodically and are available from the OECD in Paris, France and from the following URL: http://www.oecd.org/sourceoecd/. The OECD guidelines embrace a number of test guidelines for genetic toxicity, some of which are little used in practice nowadays (OECD 1997). A very recent addition to the collection of OECD Guidelines is a draft guideline on the in vitro micronucleus test (OECD 2006) on the basis of various validation exercises for this test (Corvi et al. 2008; Kirsch-Volders et al. 1997, 2003). There is a great interest in this guideline since this test approach is simpler in evaluation than the cytogenetic evaluation of chromosome aberrations that are normally required by guidelines.
10.3. INTERNATIONAL CONFERENCE OF HARMONIZATION OF TECHNICAL REQUIREMENTS FOR REGISTRATION OF PHARMACEUTICALS FOR HUMAN USE (ICH) GUIDELINES FOR PHARMACEUTICALS On a worldwide basis, the resources that are spent to identify new chemicals with the potential to cure diseases or help to cope with symptoms of diseases are enormous and far exceed expenses in any other area of chemistry. Enormous is also
244
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
the economic pressure of the health-care systems to give patients access to pharmaceuticals at a reasonable price while still maintaining scientific progress. In this context, any unreasonable regulatory requirements that result in redundancies in testing of pharmaceutical candidates and their registration are counterproductive. Hence, the ICH was established in the early 1990s to reduce such redundancies and to strive for international harmonization. This unique project brought together the regulatory authorities of Europe, Japan, and the United States and experts from the pharmaceutical industry in these three regions. Recommendations (ICH guidances) for the economical use of human, animal, and material resources in quality, safety, efficacy, and multidisciplinary (“Q”, “S”, “E” and “M” guidances) testing were developed by these parties and are now accepted internationally as the standards for evaluation of pharmaceuticals (http://www.ich.org/). Genotoxicity was established as an ICH guidance topic at the first ICH conference in November 1991 in Brussels. Within the subsequent years of negotiation between the parties involved, two ICH genotoxicity guidances were developed: (1) the guidance on “Specific Aspects of Regulatory Genotoxicity Tests for Pharmaceuticals” (ICH 1996) and (2) a guidance on “A Standard Battery for Genotoxicity Testing of Pharmaceuticals” (ICH 1997b). These two ICH guidances are complementary and are the principal guidances on genotoxicity studies for pharmaceuticals in the three ICH regions. Detailed information on the background of these guidances, along with their text, has been published (Müller et al. 1999). The ICH guidances address test procedures, as well as strategy and test interpretation. The ICH guidance “A Standard Battery for Genotoxicity Testing of Pharmaceuticals” recommends a core battery of tests “core” testing battery for pharmaceutical registration, which consists of the following: (1) a test for gene mutation in bacteria, (2) a test for chromosomal aberrations in mammalian cells in vitro or the L5178Y mouse lymphoma mammalian cell mutagenesis test, and (3) an in vivo test for chromosomal damage in rodent hematopoietic cells. Compounds giving negative results in this battery, performed and evaluated in accordance with current recommendations, will usually be considered to have a sufficient level of assurance of safety to allow product approval. Within 10 years of the ICH guidelines for genotoxicity testing in operation, it was realized that advances in genotoxicity testing and interpretation would require a maintenance process. This process was initiated at the end of 2006 and has resulted in a new single draft ICH “Guidance on Genotoxicity Testing and Data Interpretation for Pharmaceuticals Intended for Human Use” (ICH 2008). In the following, the main principles of this new ICH guideline are laid down and discussed. It is understood that some of these principles will bring about changes to genetic toxicology testing and its regulatory use, and we may see genetic toxicology guidelines for purposes other than testing of pharmaceuticals follow these rationales. I. The Ames test continues to be an elementary and indispensable part of regulatory testing. However, there is no continued need to repeat negative Ames tests in an independent experiment. While it is clear that there are some differences between mammalian cells and bacteria in regard to metabolism and DNA repair processes, it continues to be
10.3. HUMAN USE ICH GUIDELINES FOR PHARMACEUTICALS
245
acknowledged that there is no suitable alternative for the bacterial reverse mutation test (Ames test). The Ames test is the most widely used test for genotoxic activity with unparalleled easiness, cross-laboratory robustness, and specificity for mutagenic carcinogens (Gatehouse et al. 1994; Kirkland et al. 2006). It appears logical that, provided that the appropriate metabolic pathways are incorporated, the ability of a chemical to produce DNA damage that result in mutations, which conversely initiate cells to outgrow to tumors, is most easily measured in bacteria. II. The in vitro micronucleus test is endorsed as an alternative option to the in vitro chromosome aberration test and the mouse lymphoma tk assay. Many years of protocol evaluation and validation exercises imply that the in vitro micronucleus test has reached a status of reliability that is comparable with the mouse lymphoma tk assay in L5179Y cells or the chromosome aberration test with various cell lines of primary human lymphocytes (Corvi et al. 2008; Lorge et al. 2006, 2007; OECD 2006). Hence, it can be used interchangeably with these assays in the regulatory world. Since many industrial laboratories already screen for genotoxic activity in the in vitro micronucleus test in early stages of nonregulatory activity, this should now enable a seamless transition from early non-GLP screening activities into the regulatory GLP testing phase. III. An extensive review of exposure data to pharmaceuticals suggests that testing to a concentration of 1 mM (instead of 10 mM) for nontoxic compounds in mammalian cells in vitro is sufficient. Traditionally, in vitro tests for genotoxicity have been viewed as hazard identification tests to be followed up by in vivo tests for risk identification or risk assessment. Under such a view, a maximal sensitivity approach has often been the goal for in vitro tests, and an upper limit of 10 mM (or 5 mg/ml) for test material in the cell culture has been applied for testing of compounds that were nontoxic. It is understood that this level somehow represented worst-case assumptions that human cells might be at risk to ∼5% of the foreign test material in their in vivo environment. This upper limit was also borne out of early testing experience that some mutagenic carcinogens appeared to require such high concentrations to elicit a chromosome damaging response in mammalian cells in vitro (Scott et al. 1991). Consequently, very often, positive results in vitro were viewed as potentially relevant for a chronic low-level exposure in vivo because of the stochastic element in genetic toxicology and mutation induction where fully linear dose–response characteristics are thought to prevail. In practice, evidence for genotoxic activity in vitro has led in numerous cases to the conduct of extensive in vivo evaluations without ever reaching a conclusion that was acceptable to regulatory review. Another consequence was the discontinuation of development of potentially useful products very early before wasting resources on further activities with uneconomic delays. Over the years, however, there was growing evidence in the applied science for nonlinearity of many aspects of genotoxic activity. Thus, mistrust has been building up in the judgment of in vitro positive findings. An essential element of in vivo testing and risk assessment is the comparison of concentrations that are positive in vitro and the exposure that can be
246
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
reached under in vivo conditions. In this context, human pharmaceuticals offer the best possible judgment basis because exposure in animals and in humans is measured by default and into high dose ranges usually employed in animals studies. A limit of 1 mM maintains the element of hazard identification, being higher than clinical exposures to known pharmaceuticals, including those that concentrate in tissues (Hardman et al. 2001), and is also higher than the levels generally achievable in preclinical studies in vivo. Certain drugs are known to require quite high clinical exposures—for example, nucleoside analogs and some antibiotics. While comparison of potency with existing drugs may be of interest to sponsors, perhaps even above the 1 mM limit, it is ultimately the in vivo tests that determine relevance for human safety. IV. Concerns over growing numbers of nonrelevant positive findings in mammalian cell tests in vitro will also be counterbalanced by limiting the levels of cytotoxicity to “at most 50%” for in vitro chromosome aberration and micronucleus tests. This proposal is supported by an extensive review of results obtained with in vitro hazard identification testing and in vivo risk assessment testing in the pharmaceutical industry. Though some genotoxic carcinogens are not detectable with in vitro genotoxicity assays unless the concentrations tested induce some degree of cytotoxicity, particularly when measured by colony forming assays, DNA damaging agents are generally detectable with only moderate levels of toxicity (e.g., 30% reduction in growth measured at the time of sampling in the chromosome aberration assay) (Greenwood et al. 2004). As cytotoxicity increases, mechanisms other than direct DNA damage by a compound or its metabolites can lead to ‘positive’ results that are related to cytotoxicity and not genotoxicity. Such indirect induction of DNA damage secondary to damage to non-DNA targets is more likely to occur above a certain concentration threshold. The disruption of cellular processes is not expected to occur at lower, pharmacologically relevant concentrations. In cytogenetic assays, even weak clastogens that are known to be carcinogens are positive without exceeding a 50% reduction in cell counts. On the other hand, compounds that are not DNA damaging, mutagenic, or carcinogenic can induce chromosome breakage at toxic concentrations. For cytogenetic assays in cell lines, measurement of cell population growth over time by measuring the change in cell number during culture relative to control (e.g., by the method referred to as population doubling) has been shown to be a useful measure of cytotoxicity (Greenwood et al. 2004), because it is known that cell numbers can underestimate toxicity (Kirkland et al. 2007a). For lymphocyte cultures, an inhibition of mitotic index (MI) not exceeding about 50% is considered sufficient. For the in vitro micronucleus assay, a limit of about 50% cytotoxicity is also appropriate. Moreover, for the in vitro micronucleus assay, since micronuclei are scored in the interphase subsequent to a mitotic division, it is important to verify that cells have progressed through the cell cycle. This can be done by use of cytochalasin B to allow nuclear division but not cell division, so that micronuclei can be scored in binucleate cells (the preferred method for lymphocytes). Other methods to demonstrate cell proliferation, including cell population growth over time (PD)
10.3. HUMAN USE ICH GUIDELINES FOR PHARMACEUTICALS
247
as described above, may be used for cell lines (Kirsch-Volders et al. 2003; Lorge et al. 2006, 2007). For the mouse lymphoma tk+/− assay (MLA), appropriate sensitivity is achieved by limiting the top concentration to one with close to 20% relative total growth (RTG) both for soft agar and for microwell methods. This is based on reviews of published data using the current criteria described by Moore et al. (2006), which found very few chemicals that were positive in MLA only at concentrations with less than 20% RTG and that were rodent carcinogens, and convincing evidence of genotoxic carcinogenesis for this category is lacking. The consensus is that caution is needed in interpreting results when increases in mutation are seen only below 20% RTG, and a result would not be considered positive if the increase in mutant fraction occurred only at ≤10% RTG (Moore et al. 2006, 2007). Because of the inherent difficulties to obtain an almost exact value of 20% RTG in an MLA, it is acceptable to approach a range of 10–20 RTG for a valid assay with a compound that produces toxicity. In conclusion, caution is appropriate in interpreting positive results obtained as reduction in growth/survival approaches or exceeds 50% for cytogenetic assays or 80% for the MLA. It is acknowledged that the evaluation of cells treated at these levels of cytotoxicity/clonal survival may result in greater sensitivity, but bears an increased risk of nonrelevant positive results (Kirkland et al. 2007a). The battery approach for genotoxicity is designed to ensure appropriate sensitivity without the need to rely on single in vitro mammalian cell tests at high cytotoxicity. To obtain an appropriate toxicity range, a preliminary range-finding assay over a broad range of concentrations is useful, but in the genotoxicity assay it is often critical to use multiple concentrations that are spaced quite closely (less than twofold dilutions). Extra concentrations may be tested, but not all concentrations will need to be evaluated for genotoxicity. It is not intended that multiple experiments be carried out to reach exactly 50% reduction in growth, for example, or exactly 80% reduction in RTG. V.
Because pharmaceuticals are normally tested for toxicity in rodent repeated dose toxicity tests and because there is no longer a requirement for an acute high dose rodent toxicity test, the assessment of genotoxicity (e.g., bone marrow micronucleus test or other tissue/endpoint) should be integrated, if feasible, into the rodent repeated dose toxicity study to optimize animal usage. VI. The options for a standard battery of genotoxicity tests are expanded by the possibility to choose to conduct an in vivo test with investigation of genotoxic damage in two tissues instead of conducting an in vitro test with mammalian cells followed by an in vivo test. In conjunction with the respective ICH carcinogenicity guidances (see under http://www.ich.org), the ICH genotoxicity guidances are setting a new standard for genetic toxicology and carcinogenesis testing, assessment, and interpretation which is applicable in most parts of the world. Different criteria are proposed for so-called exploratory investigational new drug applications (exploratory IND). This is the case for clinical investigations in which volunteers will receive a low number of doses of an investigational drug at
248
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
relatively low or micro-doses, with the intent of collecting limited human information. In this scenario, it is acceptable to conduct a standard bacterial mutation assay (ICH 2008).
10.4. INTERNATIONAL WORKSHOP ON GENOTOXICITY TESTS (IWGT) Four workshops have been organised previously under the auspices of the IWGT. The International Association of Environmental Mutagen Societies (IAEMS) formalized these workshops in 2002 under the IAEMS umbrella and agreed that they would be held on a continuing basis in conjunction with the International Conferences on Environmental Mutagens (ICEM) that are held every four years (Kirkland et al. 2007b,c). In this way, an ongoing process of international scientific discussion and harmonization of testing methods and testing approaches has been established that can take advantage of the international experts who attend these meetings. These ongoing workshops have proven to be useful to ensure that different recommendations for methodology in these new assays do not arise in different parts of the world (Kirkland et al. 2007b,c) and thus avoid situations that could lead to (i) unnecessary duplication of testing to satisfy local requirements, (ii) variations in the test performance, (iii) potential differences in test outcome, and (iv) unjustified differences in the use of test data for description, assessment, and management of risk. The IWGT process is implemented through working groups of recognized international experts from industry, academia, and the regulatory sectors, with due attention to geographical, disciplinary, and sector balance. For each working group, a chairperson, deputy chair, and rapporteur are appointed. Experts in the science of each topic are invited to bring experimental data to bear on the discussions; the remit of each group is to derive recommendations based on data, and not on unsupported opinion or anecdotal information (Kirkland et al. 2007b). There are several objectives sought in bringing together representatives from around the world to share their experiences in generating and evaluating genotoxicity data from a variety of methodologic and strategic approaches. The IWGT strives to (i) attain a greater understanding of true test performance from a wide database, (ii) provide recommendations that minimise misinterpretation, (iii) recognize that no single assay can detect every genotoxicant, and (iv) achieve compromise for the sake of harmonisation or acceptance that more than one approach is both reasonable and valid. Because of the IWGT approach—in particular, development of data-driven consensus by the key global experts from academia, government, and industry—IWGT recommendations have been seen as state of the art and have high credibility. These recommendations serve as important supplements to established regulatory guidelines and provide a sound basis for updating those guidelines as the state of science advances. With OECD and ICH guidelines constituting the two major sets of internationally harmonised genotoxicity guidelines in regulatory use, the IWGT process and working group recommendations are of particular help in supplementing test design
10.5. THE IPCS UNDER THE AUSPICES OF THE WORLD HEALTH ORGANIZATION
249
and interpretation of genotoxicity test packages that are based on these guidelines. For further information on the IWGT recommendations, the reader is referred to the various special issues of the peer-reviewed journal Mutation Research that have emerged from the IWGT workshops (Kirkland et al. 2007b).
10.5. THE INTERNATIONAL PROGRAM ON CHEMICAL SAFETY (IPCS) UNDER THE AUSPICES OF THE WORLD HEALTH ORGANIZATION (WHO) The first harmonized scheme for mutagenicity testing on behalf of the IPCS was published in 1996 (Ashby et al. 1996). Similar to the maintenance process of ICH guidances (see Section 10.3), the WHO has recognized important developments in the field and has decided to update this IPCS Harmonized Scheme as part of the IPCS Harmonization of Approaches to the Assessment of Risk from Exposure to Chemicals. A draft for public and peer review was prepared by an International Drafting Group Meeting of experts held at the Fraunhofer Institute for Toxicology and Experimental Medicine in Hanover, Germany, on 11–12 April 2007 (IPCS 2007). The approach presented by this IPCS expert group focused on the identification of mutagens and genotoxic carcinogens. The approach is shortly presented below. For further details, the reader is referred to updates in the harmonized scheme emerging from this IPCS group. The term “mutation” as understood in this document comprises gene mutations, as well as structural and numerical chromosome alterations. The group is aware of other mechanisms leading to carcinogenicity and other heritable diseases, but their identification requires additional types of mechanistic studies. The group proposed to use a weight-of-evidence approach at various stages of the outlined testing strategy. However, it is also stated that a clear positive result at a single mutagenicity endpoint, even when multiple negative results in other endpoints have been reported, is generally sufficient for the classification “positive.” Most short-term tests in bacteria and mammalian cell cultures have been designed primarily for hazard identification and, thus, can represent only the starting point in the process of risk assessment. Whether or not the observed effects are relevant for human exposure depends on bioavailability, absorption, metabolism, half-lives, and other factors that require investigation in vivo. Especially when choosing in vivo assays and when proceeding into germ cell mutagenicity studies, expert judgment is required to select the appropriate test system(s) and to avoid uninformative and thus unnecessary animal experiments. Before initiating mutagenicity testing on a particular compound, the following aspects should be considered: (i) chemical structure and class of the agent (possible structure–activity relationships) and physicochemical properties, such as solubility and stability; (ii) expected routes of metabolism, chemical and biological reactivity/ activity, and relationship to known genotoxic chemicals; and (iii) routes of exposure, bioavailability, and target organ(s).
250
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
10.6.
IN VITRO TESTING
Usually two or three different tests in bacteria and mammalian cells are selected to cover the endpoints of gene mutations, clastogenicity (structural chromosome aberrations), and aneuploidy (numerical chromosome aberrations), taking into account physicochemical properties of substances under consideration (see Chapter 11).
10.6.1.
In Vitro Tests
Screening should be based on a limited number of tests that are well-validated and informative. Genetic toxicity test batteries generally include the following: (1) A test for gene mutation in bacteria (bacterial reverse mutation assay): Organisation for Economic Cooperation and Development (OECD) Guideline 471 recommends the use of at least five strains of bacteria: (i) Salmonella typhimurium TA1535, (ii) S. 12 typhimurium TA1537 or TA97 or TA97a, (iii) S. typhimurium TA98, (iv) S. typhimurium TA100, and (v) Escherichia coli WP2 or E. coli WP2uvrA or S. typhimurium TA102. The choice of additional tests depends on the chemical structure and class of the agent. (2) In vitro mammalian assays: These assays should evaluate the potential of a chemical to produce point mutations, clastogenicity, and/or aneugenicity, by using either mammalian cell lines or primary human cell cultures such as fibroblasts or lymphocytes (e.g., mouse lymphoma TK assay or cytogenetic evaluation of chromosomal damage in mammalian cells via in vitro micronucleus test).
10.6.2.
Evaluation of In Vitro Testing Results
Evaluation of results and classification into: (i) positive results, (ii) negative results, and (iii) inconsistent, conflicting, or equivocal results. Positive: Substance is positive at one or more endpoints of mutagenicity. Negative: Substance is negative in all test systems under appropriate in vitro conditions; the substance is not mutagenic (genotoxic) in vitro and is predicted not to be mutagenic in vivo [for exceptions, see Tweats et al. (2007a,b)]. Inconsistent, conflicting, or equivocal (i.e., borderline biological or statistical significance): All other substances.
10.6.3.
Follow-Up to In Vitro Testing
In case of positive results: Conduct an in vivo test with selection of an appropriate endpoint; if necessary, conduct further in vitro studies to optimize in vivo testing (e.g., kinetochore staining as an addition in the micronucleus assay of in vitro aneugens). In case of negative results: Further in vivo testing is required only in the case of “high” or “moderate and sustained” exposure, or for chemicals of high concern.
10.7. IN VIVO TESTING
251
In case of inconsistent, conflicting, or equivocal results: Further in vitro testing to clarify positive or negative results; depending on whether the situation is resolved by further in vitro testing, proceed according to “Positive” or “Negative.”
10.7.
IN VIVO TESTING
In vivo tests should be chosen carefully to avoid an uninformative outcome. Therefore, toxicokinetics, metabolism, chemical reactivity, and mode of action have to be considered carefully. Typically, a bone marrow micronucleus or clastogenicity test is conducted. However, if there are indications that point to a more appropriate assay, then this assay should be conducted instead (e.g., mutagenicity study with transgenic animals; comet assay in stomach/small intestine/colon, if there is no uptake via gastrointestinal tract; comet assay in the liver if there is metabolism to toxic species) (see Chapter 12).
10.7.1.
Follow-Up to In Vivo Testing
In case of positive results: The compound is an “in vivo somatic cell mutagen” and testing for germ cell mutagenicity may be required. In case of negative results: Further in vivo testing is required only in the case of positive in vitro studies; again, the second in vivo test is chosen on a case-by-case basis as stated above. If the test is negative, it is concluded that there is no evidence for in vivo mutagenicity. In case of equivocal results: Equivocal results may be due to low statistical power, which can be improved by increasing the number of treated animals and/or scored cells. If the situation is unresolved, a second in vivo test is required, chosen on a case-by-case basis (ordinarily on a different endpoint or in a different tissue, depending on toxicokinetics, metabolism, and mode of action); proceed according to “Positive” or “Negative.”
10.7.2.
Strategy for Germ Cell Testing
When information on the risk to the offspring of exposed individuals is important, the following germ cell testing strategy is recommended. For substances that give positive results for mutagenic effects in somatic cells in vivo, their potential to affect germ cells should be considered. If there is toxicokinetic or toxicodynamic evidence that germ cells are actually exposed to the somatic mutagen, it is reasonable to conclude that the substance may also pose a mutagenic hazard to germ cells and thus a risk to future generations. Where germ cell testing is required, judgment should be used to select the most appropriate test strategy. There are a number of tests available, which fall into two classes: (1) tests in germ cells per se (“class 1”) and (2) tests to detect effects in the offspring (or potential offspring) of exposed animals (“class 2”). Three
252
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
internationally recognized OECD guidelines are available for such studies (OECD 1997): (1) clastogenicity in rodent spermatogonial cells (class 1), (2) the dominant lethal test (class 2), and (3) the mouse heritable translocation assay (class 2). In order to minimize the use of animals in germ cell testing, it is advisable to start with tests that detect effects in germ cells per se (class 1). These methods include (but are not limited to) gene mutation tests in transgenic animals, gene mutations in the Expanded Simple Tandem Repeat (ESTR) assay (Dubrova et al. 1998), chromosomal assays including those using fluorescence in situ hybridization (FISH) (Hill et al. 2003), comet assay (Burlinson et al. 2007; Hartmann et al. 2003; Merk and Speit 1999; Pfuhler and Wolf 1996), and DNA adduct analysis (Phillips et al. 2000). Following the use of such tests, if quantification of heritable effects is required (class 2), an assay for ESTR mutations can be performed with the offspring of a low number of exposed animals. Tests used historically to investigate transmitted effects (i.e., the heritable translocation test and the specific locus test) can also be performed; however, they use large numbers of animals.
10.8. EUROPEAN UNION GUIDELINE FOR TESTING OF CHEMICALS UNDER THE REGISTRATION, EVALUATION, AUTHORIZATION AND RESTRICTION OF CHEMICAL (REACH) REACH is a new EU policy aimed at evaluating the health risks of chemicals marketed (i.e., produced in or imported) in the EU. It was proposed by the European Commission at the end of 2003, and it came into effect in June 2007 (EC 2006; EU 2003). The objective of REACH is to realize a more rapid and less expensive method to identify risks for exposed humans and at the same time minimize the use of laboratory animals (van der Jagt et al. 2004). Also, clearly the improvement of the competitiveness of the European chemicals industry is defined as another goal. In this context, it is important to understand that of the ∼30,000 chemicals which are in regular human use, and of which many are high-tonnage chemicals, only ∼3% are sufficiently tested for toxicological risk (EC 2001). Therefore, one of the aims of REACH is to provide guidance for a retrospect safety evaluation of existing chemicals, as well as for an evaluation of newly developed and marketed chemicals. The REACH regulation requires toxicological information from every chemical with a marketing volume of >1 tonnes per annum (t/a). In the context of testing for a genotoxic potential, REACH considers genotoxicity testing also a surrogate for a mechanistic link to carcinogenicity, although under some circumstances germ cell mutagenicity—that is, the risk of induction of heritable diseases—is evaluated in addition. The amount of information required is dependent on the tonnage in that a higher production volume means that more information is needed. A technical dossier is required for all registered chemicals; however, only chemicals with a production volume greater than 10 t/a require a chemical safety assessment documented in a chemical safety report. No new toxicological information is required for substances between 1 and 10 t/a, unless they are considered a priority substance (i.e., carcinogenic, mutagenic, or toxic to reproduction [CMR]) (EC 2006). Column
10.8. EUROPEAN UNION GUIDELINE FOR TESTING OF CHEMICALS UNDER THE REACH
253
1 “Standard Information Required” of the Annexes of REACH describe the standard information requirements for substances produced or imported in quantities of ≥1 t/a (Annex VII), ≥10 t/a (Annex VIII), ≥100 t/a (Annex IX), and ≥1000 t/a (Annex X); different tonnage classes have defined testing requirements (EC 2006). Also, in the context of REACH, it should be understood that existing chemicals (called “phasein” substances) are considered equivalent to newly introduced chemicals (called “non phase-in” substances) with respect to the toxicological testing data required. This means that a major part of information obtained will be on phase-in substances. In general, like for other areas of genotoxicity testing, information on gene mutations, structural chromosome aberrations (clastogenicity), and numerical chromosome aberrations (aneugenicity) is required (Aardema et al. 1998). Also, the guidance specifically defines the terms mutagenicity and genotoxicity in a way that is also acceptable for other areas of genetic toxicology (ECHA 2008a): Mutagenicity refers to the induction of permanent transmissible changes in the amount or structure of the genetic material of cells or organisms. These changes may involve a single gene or gene segment, a block of genes or chromosomes. The term clastogenicity is used for agents giving rise to structural chromosome aberrations. A clastogen can cause breaks in chromosomes that result in the loss or rearrangements of chromosome segments. Aneugenicity (aneuploidy induction) refers to the effects of agents that give rise to a change (gain or loss) in chromosome number in cells. An aneugen can cause loss or gain of chromosomes resulting in cells that have not an exact multiple of the haploid number. For example, three number 21 chromosomes or trisomy 21 (characteristic of Down syndrome) is a form of aneuploidy. Genotoxicity is a broader term and refers to processes which alter the structure, information content or segregation of DNA and are not necessarily associated with mutagenicity. Thus, tests for genotoxicity include tests which provide an indication of induced damage to DNA (but not direct evidence of mutation) via effects such as unscheduled DNA synthesis (UDS), sister chromatid exchange (SCE), DNA strand breaks, DNA adduct formation or mitotic recombination, as well as tests for mutagenicity.
Testing of chemicals under REACH is done in a tiered fashion, in that the different tonnage classes define the necessary data for which testing may be required. This is one of the fundamental differences between REACH and other regulated substances (e.g., regulation of pharmaceuticals, where the production volume is not considered at all). The second fundamental difference between chemical testing under REACH and testing of chemicals for human use (e.g., pharmaceuticals or cosmetics) lies in the fact that chemical testing is mainly conducted for the purpose of labeling and defining appropriate protection measures and exposure limitations, whereas particularly for drug ingredients, the definition and quantification of a benefit–risk ratio and a risk assessment derived therefrom is the predominant goal of a test strategy. This is mainly justified by the fact that exposure to pharmaceuticals generates a therapeutic benefit and that—in contrast to the chemicals under REACH—consideration, administration, and exposure scenarios for pharmaceuticals are well-defined. Therefore, the tonnage principle may be viewed as a surrogate for the lack of reliable exposure information, so that a higher tonnage is considered to result in the higher likelihood of human exposure in general.
254
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
For chemicals, the main product of a risk evaluation under REACH is the Chemical Safety Assessment (CSA), which is prepared for all chemicals above 10 t/a to establish the safe conditions of manufacture and use of a substance for all lifecycle stages, submitted to the European Chemicals Agency (ECHA) as part of the registration dossier (ECHA 2008b,c). Although the minimum information requirements under REACH is primarily determined by tonnage triggers, it may be adapted due to hazard, exposure, or risk considerations, as well as technical difficulties in testing the substance. In addition, the level of follow-up testing is defined by the result in one of the preceding genotoxicity assays, an approach that is reminiscent to as the regulation of other substances (e.g., pharmaceuticals). All eligible chemicals under REACH require the conductance of a bacterial reverse mutation (Ames) test (Annex VII, §8.4.1) (EC 2006). For compounds ≥10 t/a, an additional mammalian in vitro cytogenetic test is required (Annex VIII, §8.4.2) (EC 2006). For the latter, both the chromosome aberration test and the micronucleus test in vitro are considered acceptable. In the case of a negative result in the mammalian in vitro cytogenetic test and in the in vitro gene mutation test in bacteria, data from an in vitro gene mutation study in mammalian cells are required (Annex VIII, §8.4.3) (EC 2006). A positive in vitro cytogenetic result should be evaluated for the possibility that the chemical tested acts via an aneugenic mechanism, which is appropriately addressed by an in vitro assay, should the endpoint not be covered already by the cytogenetic test conducted. Appropriate in vivo mutagenicity studies should be considered in case of a positive result in any of those genotoxicity studies (Annex VIII, §8.4) (EC 2006). For compounds marketed at ≥100 t/a, an in vivo somatic cell mutagenicity test is requested (Annex IX, §8.4) (EC 2006). Dependent on the data obtained in the above tests, a second in vivo genotoxicity test may be required. If there is a positive result obtained in any in vivo somatic cell study, the potential for germ cell mutagenicity should be considered on the basis of all available data, including toxicokinetics. If no clear conclusion about germ cell mutagenicity can be made, additional investigations shall be considered (Annex IX, §8.4) (EC 2006). Obviously, the major aim of the in vivo genotoxicity tests is to confirm the positive results of in vitro genotoxicity tests in vivo and to identify those compounds where a positive in vitro result does not translate into a positive in vivo finding—that is, in vitro effects defined as false-positive. REACH suggests that this should be accomplished with only one in vivo test. A second in vivo test is considered necessary when the compound has shown both a clastogenic and a gene mutagenic potential in vitro and the in vivo test is considered inadequate to address both endpoints simultaneously. In addition, evidence for specific genotoxicity to germ cells may trigger the performance of a second in vivo study. Eventually, the assessment of the totality of data obtained after the REACH-compliant test strategy will be the classification (with corresponding labeling) into the following categories*: *In the EU, dangerous substances and preparations must be classified and labeled according to Directives 67/548/EEC and 1999/45/EC, respectively. These Directives will be repealed and replaced with the EU Regulation on classification, labeling, and packing of substances and mixtures, implementing the Globally Harmonised System (GHS) in the EU.
10.8. EUROPEAN UNION GUIDELINE FOR TESTING OF CHEMICALS UNDER THE REACH
255
Category 1 (label T R46): Substances known to be mutagenic to man. In this category, substances would fall with sufficient evidence to establish a causal association between human exposure to a substance and heritable genetic damage. Human mutation epidemiology studies are needed for chemicals to be placed in this category (EC 1967, 1999). However, so far no chemical has received this label (Appendix 3, Point 29—Mutagens: category 1) (EC 2006). Category 2 (label T R46): Substances which should be regarded as if they were mutagenic for man (EC 1967, 1999). For compounds in this category, there is sufficient evidence to provide a strong presumption that human exposure to the substance may result in the development of heritable genetic damage, generally on the basis of appropriate animal studies or other relevant information (Appendix 4, Point 29—Mutagens: category 2) (EC 2006). Category 3 (label Xn R68 or R40): Substances that cause concern to man owing to possible mutagenic effects. For compounds in this category, there is evidence from appropriate mutagenicity studies, but this is insufficient to place the substance in Category 2 (EC 1967, 1999). Obviously, those requirements will necessitate the conductance of additional animal experiments. Therefore, in order to fulfill the intention to reduce the need for animal experiments, REACH requests that all available in vitro data, in vivo human data and animal–human data, data from in silico SAR systems, and data from structurally related substances be evaluated before additional tests be carried out or that arguments be provided to waive additional testing. In addition to pharmaceuticals, for those substances regulated under REACH, an independent determination of genotoxic risk to germ cells is a requirement, whereas for pharmaceuticals, which are regulated mainly according to ICH guidances, germ cell genotoxicity tests are not foreseen. However, also under REACH it is acknowledged that heritable mutation is the consequence of a general mutagenic effect elicited in germ cells. The product of a mutagenic potential and the ability of a chemical to be distributed into the germ line is considered and therefore is not an independent genotoxic endpoint. As a consequence, all available toxicokinetic and toxicodynamic properties of the test substance are taken into consideration to estimate whether there is sufficient information to conclude that the substance poses a mutagenic hazard to germ cells. If this is the case, it can be concluded that the substance may cause heritable genetic damage and no further testing is justified. If no clear conclusion can be drawn, additional toxicokinetic experiments or tests for inheritable mutations may be conducted, although toxicokinetic investigations are preferred. The trigger to consider a compound a potential germ cell mutagen would be a positive in vivo genotoxicity test; and depending on this dataset, the appropriate germ cell genotoxicity test (i.e., germ cell clastogenicity of dominant lethal test for chromosome damage, transgenic animals for gene mutations, DNA binding, or the comet assay for direct DNA damage) would be selected. Also, like for pharmaceuticals, evidence for a relevant genotoxic potential to male or female germ cells will, without otherwise convincing
256
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
evidence, led to the assumption that a chemical is also leading to reproductive toxicity, given the mechanistic association of both endpoints. In summary, the newly introduced REACH guidance, which is regulating the safety of chemicals marketed or imported into the EU and which has been finalized recently, is, for the purpose of genetic toxicology, based on a stepwise strategy of sequential tests, each based on the results of the respective previous step and leading to a labeling of the chemical which in turn spikes measures to limit the exposure of humans (or the environment) during production, transport, or use of the chemical. The new regulation treats already existing chemicals equally to those brought newly into the market (produced or imported) in the EU. So far, it remains to be seen how the reduction in the number of animal experiments, one of the goals of REACH, will be achieved, while at the same time compiling a comprehensive database on all the existing compounds.
10.9. SPECIALTY GUIDELINES FOR GENOTOXICITY: GENOTOXIC IMPURITIES IN PHARMACEUTICALS In recent years, the testing and control of pharmaceuticals for the presence of genotoxic impurities have been under regulatory scrutiny. Because of their dedicated use—in many cases chronic use—in humans, pharmaceuticals are expected to be of high purity and little batch-to-batch variability is allowed. This focus is justified compared to chemicals of other use areas, for which exposure may be dedicated but limited (e.g., cosmetics), more of an accidental type (e.g., household chemicals), or low and chronic under workplace or use conditions (e.g., pesticides, industrial chemicals). The relevant ICH guidelines concerning the qualification of impurities in commercial manufacture are Q3A(R) and Q3B(R) that focus on impurities in drug substances and drug products, respectively, while Q3C recommends limits for residual solvents in the drug product (ICH 1997a, 2002, 2003). The guidance given in these regulatory documents is considered to be applicable at the time of registration of a new pharmaceutical entity. The first two guidelines describe threshold levels above which impurities are required to be reported, identified, and qualified either in toxicological investigations or in the clinic. The threshold levels vary according to the maximum daily dose of a drug. For drug substances, the identification thresholds are within the range of 500 and 1000 ppm (i.e., 0.05 and 0.1%). While in general very high purity of more than 98% is attained for pharmaceuticals, the presence of impurities even at low levels of 0.1% or lower may cause unwanted effects or may be of concern for chronic intake. This is of particular importance when the drug is taken at a high daily dose. Hence, the ICH Guidelines Q3A(R) and Q3B(R) take precaution for this case and state that although identification of impurities is not generally necessary at levels less than or equal to the identification threshold, “analytical procedures should be developed for those potential impurities that are expected to be unusually potent, producing toxic or pharmacological effects at a level not more than (≤) the identification threshold.” Thus in the case of impurities where a potential safety concern for genotoxicity exists, the guidelines imply that the routine identification threshold is not considered to be applicable.
10.9. SPECIALTY GUIDELINES FOR GENOTOXICITY
257
One may ask the following questions: (I) Why do genotoxic impurities exist at all in pharmaceuticals? (II) Why can’t they be totally eliminated? I: The synthesis of often complex chemical structures that bear in themselves sufficient specificity and selectivity for a pharmacological action is only manageable via reactive steps of chemistry. Such reactions often involve or result in electrophilic intermediates, which may possess genotoxic activity. It is a dream of every chemist to create reaction conditions that fully create a product from its educts. This dream can hardly be put into reality. II: Purification and control are part of the means to manage processes in a way that they can be economically reliable and result in safe products. Purification steps often result in significant loss of the product, and a balance between safety requirements and economic conditions must be found. Since genotoxic compounds are usually considered as potentially carcinogenic with a linear dose–response relationship, genotoxic impurities are considered separately from the existing ICH Guidelines and limits of acceptability have to be set. Because there was no general guidance on how to do this, considerable differences occurred between regulatory authorities, even within the same region. To reduce the differences in judgement on genotoxic impurities between EU Member States, the Committee of Human Medicinal Products (CHMP) decided to ask the Safety Working Party (SWP) to develop a guideline on genotoxic impurities. This guideline was released after much discussion in June 2006 (CHMP 2006). The central idea in this draft guideline is the concept of the Threshold of Toxicological Concern (TTC). A TTC value of 1.5 μg/day intake of a genotoxic impurity is derived from a large database of animal carcinogenicity studies. It is estimated that the intake of any genotoxic impurity, with a few exceptions of highly potent genotoxic carcinogens, below this TTC is associated with an acceptable risk of less than one excess cancer in a population of 100,000 people (i.e., <1 × 10−5). The use of this TTC approach is proposed as a pragmatic solution for the situation where a genotoxic impurity cannot be avoided and where no compound-specific information is available on the carcinogenic potential of the impurity. While the generic TTC of 1.5 μg/day is based on a risk delineation from animal data for human lifetime use, it is acknowledged that most pharmaceuticals are not given over lifetime. In addition, the investigation of pharmaceutical candidate compounds in clinical trials will generally involve fewer subjects and treatment for shorter durations than marketing authorization. Based on the knowledge that human tumor risk from exposure to genotoxic carcinogens is not only a function of dose but also a function of duration of exposure (Bos et al. 2004), a so-called “staged” TTC concept was proposed (Müller et al. 2006). According to this concept, generic TTC values can be calculated for durations of exposure that occur typically during clinical trials for pharmaceutical candidate development (Table 10.1). In addition to the back-calculation of lifetime risk data to shorter durations, two additional considerations were taken into account when Müller et al. (2006) proposed these values: (i) If clinical use of a pharmaceutical extents beyond a duration of 12 months, a
258
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
TABLE 10.1. Generic Staged TTC Values for Genotoxic Impurities During the Clinical Trial Stage for Pharmaceutical Candidate Compoundsa
Duration of Exposure (months)
Allowable Daily Intake (ADI) in μg/person/day:
≤1
>1–3
>3–6
>6–12
>12
120
40
20
10
1.5
a
For details see Müller et al. (2006).
TABLE 10.2. Generic Staged TTC Values for Genotoxic Impurities During the Clinical Trial Stage for Pharmaceutical Candidate Compounds Proposed by the CHMP in the EU
Duration of Exposurea
Allowable Daily Intake (ADI) in μg/person/day:
Single Dose
≤1
≤3
≤6
≤12
>12
120
60
20
10
5
1.5
a
Note: The U.S. FDA has proposed adoption of a similar set of values for investigational new drug applications, with the exception that the ADI of 120 μg/day be used for exposures of less than 14 days and that 60 μg/day be used for exposures from 14 days to 1 month (FDA 2008).
chronic lifetime exposure cannot be excluded; (ii) in the clinical trial stage, the benefits of a pharmaceutical candidate are not fully established. Hence, the generic TTC value of 1.5 μg/day is proposed for any intake duration of more than 12 months, and all staged TTC values for shorter duration of exposure are calculated using an acceptable risk of less than one excess cancer in a population of 1,000,000 (i.e., <1 × 10−6) instead of <10−5. In 2008, the CHMP communicated its regulatory position to the EU on staged TTC values for genotoxic impurities in pharmaceuticals (CHMP 2008). These values are given in Table 10.2. Compared to the proposal of a staged TTC according to Müller et al. (2006), these values incorporate a dose rate correction factor of 2 to account for deviations from the linear extrapolation model and hence are more conservative. In the meantime, the U.S. FDA has released an analogous draft guidance (FDA 2008). One of the main difference between the two guidances lies in partially different exposure duration intervals used by the FDA guidance. However, in general, a comparable reasoning is employed in both documents.
10.10. THE QUINTESSENCE FOR REGULATORY ASSESSMENT: IN VIVO TESTING FOR RISK ASSESSMENT In practically all cases, genotoxicity testing of a molecule is started with in vitro tests. In vitro tests are particularly suitable in that they are quick and inexpensive, and the experimental environment is well-defined and can be manipulated easily. Also, since genotoxicity is a molecular event that occurs on the single-cell level,
10.10. THE QUINTESSENCE FOR REGULATORY ASSESSMENT
259
in vitro systems are suitable for analysis of genotoxicity, in contrast to many other toxicological endpoints. According to the ICH S2B used so far (ICH 2008), the standard in vitro test battery consists of a bacterial reverse mutation (Ames) test and a mammalian cell test which can be either one for chromosomal damage or a test in L5178Y mouse lymphoma cells, which utilizes the hemizygous tk gene of this cell line. The latter test, while technically and formally a gene mutation test, has shown its sensitivity to both (a) compounds inducing point mutations and (b), those generating larger genome rearrangements and chromosome aberrations. In general, this relatively restricted test battery is considered suitable to ensure the absence of a genotoxic risk of a compound, provided that it is clearly negative, at least for early clinical development stages, which normally encompass only a limited number of involved persons. However, there are cases where additional in vivo testing is required. For pharmaceuticals, genotoxicity assessment serves the objective of supporting a benefit–risk assessment. In general, for compounds used for medicinal purposes, parameters to critically assess the in vivo kinetic behavior are relatively well understood when in vivo genetic toxicology testing is conducted. The assessment of absorption, distribution, metabolism, and excretion, generally referred to as ADME parameters, is crucial for estimating a therapeutic or toxic activity. At the stage of preclinical development, in vitro or in vivo models are established for disease endpoints that allow the estimation of a pharmacodynamic effect at the desired target, and kinetic parameters like hepatic or extrahepatic metabolism, plasma concentration profiles, tissue distribution and retention, or elimination pathways are characterized. In addition, extensive data on in vivo toxicity from single- and repeated-dose toxicity studies, generally up to several weeks, have been obtained until then. Once a compound has entered clinical development, even human experience is added to the body of knowledge about the kinetic behavior of a drug candidate. For that reason, in vivo genotoxicity testing of a pharmaceutical compound is supported by a considerable body of information of how the molecule will behave in the animal body, and data obtained can be interpreted accordingly. Internationally, the strategy of genotoxicity testing of pharmaceuticals is primarily described by ICH guidelines. Hitherto, ICH S2A and S2B guidelines were the ones dedicated to genotoxicity testing, while at the moment these two guidelines are combined into one ICH S2 guideline during a maintenance process. ICH S2B defines the test battery as a series of in vitro and in vivo tests designed to detect compounds that induce genetic damage directly or indirectly by various mechanisms. Clearly the tests should enable hazard identification with respect to damage to DNA and its fixation. Fixation of damage to DNA in the form of gene mutations, larger-scale chromosomal damage, recombination, and numerical chromosome changes is generally considered to be essential for heritable effects and in the multistep process of malignancy. Therefore, compounds which are positive in tests that detect such kinds of damage have the potential to be human carcinogens and/ or mutagens—that is, may induce cancer and/or heritable defects. Because the relationship between exposure to particular chemicals and carcinogenesis is established for humans, whereas a similar relationship has been difficult to prove for heritable diseases, genotoxicity tests have been used mainly for the prediction of
260
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
carcinogenicity. Nevertheless, because germ-line mutations are clearly associated with human heritable disease, the suspicion that a compound may induce heritable effects is considered to be just as serious as the suspicion that a compound may induce cancer. In addition, the outcome of such tests may be valuable for the interpretation of carcinogenicity studies. However, currently germ cell tests are not recommended by the current ICH guidance, but it is considered that germ cell mutagenicity can be viewed as somatic mutagenicity of a drug with toxicokinetic properties that allow the exposure of germ cells. Thus, according to ICH S2A and S2B, in vivo genotoxicity testing is triggered, up to a relatively late phase of clinical development, by positive in vitro data in mammalian cells, be it a cytogenetic or a mouse lymphoma tk test. So, in general, the strategic position of an in vivo test is therefore the confirmation or invalidation of a positive in vitro result, supported by, ideally, data that support a mechanistic hypothesis of why the observed result is due to an in vitro-specific effect. As a second objective, in vivo tests serve the purpose of covering those mechanisms that are not or are insufficiently covered by in vitro tests. As an obvious reason, in vivo metabolism can be different from the one employed by in vitro systems, which predominantly rely on the exogenous supply of metabolically active organ preparations or the recombinant expression of metabolic enzymes within the target cell. However, experience has shown that these cases are infrequent so that this second objective can be considered of inferior importance. Therefore, while in vitro systems are considered very sensitive, in vivo systems can be viewed as very specific, thus providing a very relevant readout and safety signal if positive results are obtained. In the context of the revised ICH S2 guideline, the options presented offer the possibility to abandon in vitro mammalian genotoxicity testing altogether, and replacing it by two in vivo tests, supplemented with a bacterial reverse mutation (Ames) test (ICH 2008). Thus, the in vivo test is placed into a fundamentally different strategic context. Hitherto, in vivo testing has mainly served the purpose of validating previously obtained in vitro mammalian cell genotoxicity data, which in return were considered to provide a highly sensitive readout and supply a high level of confidence that no relevant genotoxicant would go undetected. In the revised guideline, under option 2 the in vitro mammalian test can be abandoned, so that the two in vivo tests, which in many cases will be a bone marrow or peripheral blood micronucleus assay combined with an in vivo comet assay in a “to be defined” target organ, need to provide the desired level of sensitivity. Two possibilities are offered for that case. An acute test, as has been conducted under the current guidance, is considered appropriate in that it permits the application of a sufficiently high dose. Alternatively, the in vivo readout, both for the micronucleus and for the second endpoint, can be obtained from a repeated-dose treatment. In most cases, this will be a 4-week toxicity study, if GLP is required. Obviously, sensitivity is an issue under these conditions because in most cases limitations will exist as to the maximum dose and blood levels achievable under those conditions as compared to a single administration. According to the revised ICH S2R1 guidance, as a test for chromosome damage, either the analysis of chromosomal aberrations or the measurement of micronucleated polychromatic erythrocytes in bone marrow cells in vivo is appropri-
10.10. THE QUINTESSENCE FOR REGULATORY ASSESSMENT
261
ate for the detection of clastogens. Both rats and mice are appropriate for use in the bone marrow micronucleus test. Micronuclei may also be measured in immature (e.g., polychromatic) erythrocytes in peripheral blood in the mouse, or in the newly formed reticulocytes in rat blood. Likewise, immature erythrocytes can be used from any other species, which has been shown to provide adequate sensitivity for the detection of clastogens/aneuploidy inducers in bone marrow or peripheral blood; however, in general the above rodents will be used, particularly the rat, which is the standard rodent species for toxicological investigations. Chromosomal aberrations can also be analyzed in peripheral lymphocytes cultured from treated rodents. Importantly, if no in vitro mammalian cell genotoxicity assay is conducted (option 2), the micronucleus test in vivo is recommended, not the metaphase chromosome aberration assay, to include more direct capability for detection of chromosome loss (potential for aneuploidy), for which the micronucleus is technically better suited than the chromosome aberration test. Also, in situ-hybridization techniques allow the discrimination of aneugenic versus clastogenic mechanisms in that the content of micronuclei can be analyzed for whole chromosomes, identified by their centromeres, or chromosome fragments.
10.10.1.
Choice of In Vivo Test
As recommended by ICH S2R1, the bone marrow or peripheral blood micronucleus test is the first in vivo test to be employed. Developed as a rapid alternative to chromosome aberration assessment in rodent bone marrow (Heddle 1973), the rodent micronucleus assay in erythrocytes is currently the most widely utilized in vivo genotoxicity test for assessing induction of cytogenetic damage. Reasons for the popularity are the technical ease of conductance, the possibility of applying image analysis systems, the robustness of the assay, and the sensitivity toward both clastogenic and aneugenic events. Accordingly, a wealth of information has been generated to demonstrate its usefulness (CSGMT 1986, 1988, 1990, 1992, 1995; Hamada et al. 2001, 2003; Hayashi et al. 1994, 2000, 2007; MacGregor et al. 1980, 2006; OECD 1997; Romagna 1988; Suzuki et al. 2005; Wakata et al. 1998). The basis of the in vivo micronucleus assay is that actively dividing cells exposed to an agent capable of causing chromosome breakage or loss will induce the formation of micronuclei, which are readily identified in progeny cells. Micronuclei are extranuclear membrane-coated bodies that are formed after cell division around chromosome fragments or whole chromosomes and are retained in the cytoplasm of one of the daughter cells. These structures can be readily detected and counted. The fact that malsegregated chromosomes are also forming micronuclei renders this test sensitive to aneugenic events (i.e., those where aberrations in the number of chromosomes is obtained) and to clastogenicity (i.e., structural chromosome damage). The result of the test is the number of micronucleated cells in the reticulocyte (i.e., immature erythrocyte) population. Immature or newly formed erythrocytes provide an ideal cell type for micronucleus assessment as the erythroblast precursor cells are rapidly dividing in the bone marrow, with the nucleus extruded from the cell a few hours after final mitosis. Provided that there is sufficient time between exposure to the test article and recovery of erythrocytes for analyses,
262
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
micronuclei that result from induction of chromosomal breakage or malsegregation in erythroblast precursor cells can be easily detected in the newly formed anucleate erythrocyte population. The micronucleus assay was originally developed in the mouse bone marrow, and later it was demonstrated that assessment of MN induction in peripheral blood of mice was equally sensitive (MacGregor et al. 1980). Routinely, rats are used in the micronucleus test in most cases since rats are the species used for other toxicity studies so that typically a large body of historical information exists for rats, which facilitates data interpretation as well as study design and the definition of experimental parameters. Depending on the question to be asked and, most importantly, the data already obtained with a compound under investigation, other in vivo tests can be utilized and are accepted by the revised ICH S2R1 guidance, in particular if reasons are present to assume that the employed battery may have missed detection of genotoxic activity of a true genotoxic carcinogen (Brambilla and Martelli 2004). Of those, transgenic animal systems (Heddle et al. 2000; Lambert et al. 2005), alkaline elution (Speit and Hartmann 2005; Storer et al. 1996), or covalent binding assays such as the postlabeling assay (Phillips et al. 2000), all applicable to a variety of tissues, or the liver unscheduled DNA synthesis (UDS) (Madle et al. 1994; OECD 1997) test are considered acceptable according to the revised guideline. Whereas alkaline elution and UDS test can be applied, at least in principle, to various animal species or strains, the use of transgenic systems (e.g., the most popular being MutaMouse or Big Blue®) obviously requires the utilization of specific animals. All these transgenic models utilize the so-called shuttle vector principle for mutation analysis, which describes a method in which the transgenes are mutated in the animal body, whereas mutation analysis occurs after specific retrieval of the transgene and subsequent transfer of the mutational target gene into suitable bacterial hosts. Mutation analysis is carried out by extracting high-molecular-weight genomic DNA from the tissue of interest, packaging the lambda shuttle vector in vitro into lambda phage heads, and testing for mutations that arise in the transgene sequences following infection of an appropriate bacterial strain (Lambert et al. 2005). Probably the most popular second test is the single cell gel electrophoresis or comet assay (Hartmann et al. 2003; Storer et al. 1996; Tice et al. 2000). The basic principle of the test relies on the migration of DNA in an agarose minigel under electrophoretic conditions. When viewed through the microscope, a cell has the appearance of a comet, with a head (the nuclear region) and a tail containing DNA fragments or strands that have migrated in the direction of the anode. Among the various versions of the comet assay, the alkaline method, utilizing a pH value of above 13 for the unwinding step, has shown to be able to detect the broadest spectrum of DNA damage. This test detects DNA damage such as strand breaks, alkalilabile sites, and single strand breaks associated with incomplete excision repair. Under certain conditions, the assay can also detect DNA–DNA and DNA–protein crosslinking, which (in the absence of other kinds of DNA lesions) appears as a relative decrease in DNA migration compared to concurrent controls. In contrast to other DNA alterations, crosslinks may stabilize chromosomal DNA and inhibit DNA migration (Merk and Speit 1999; Pfuhler and Wolf 1996). Thus, reduced DNA migration in comparison to the negative control (which should show some degree
10.10. THE QUINTESSENCE FOR REGULATORY ASSESSMENT
263
of DNA migration) may indicate the induction of crosslinks, which are relevant lesions with regard to mutagenesis and should be investigated further. Increased DNA migration indicates the induction of DNA damage or alkali-labile DNA repair intermediates. The advantage of this test lies in the relative ease of conductance and the fact that it can be basically applied to any animal tissue (or cell culture) from which a single cell suspension can be prepared. Whatever the choice of the second in vivo test will be, it needs to be made on a case-by-case basis, taking into account all available information. Obviously, the endpoint investigated plays a pivotal role in a realistic risk assessment. Therefore, for the elucidation of a cytogenetic in vitro result, the micronucleus test can be considered the most appropriate assay. Intuitively, for in vitro-positive gene mutation assays, particularly the Ames test, transgenic animals are the systems of choice. However, because the alkaline comet assay detects DNA repair intermediates, which are also formed after DNA damage with point mutagens, this test is also able to detect gene mutations at least to some extent. Another consideration of the second in vivo test is the question whether a specific tissue needs to be investigated, which will be in most cases. The choice of this tissue will depend on factors described by ADME (e.g., in the sense that known hepatic metabolism would warrant the analysis in the liver), whereas the accumulation or retention in a specific organ, or otherwise high local concentrations, such as will be in the gastrointestinal tract with orally applied compounds, would call for an analysis there. Also, many of those systems allow the elucidation of germ cell genotoxicity without the need to conduct formal (trans-generation) germ cell genotoxicity assays, which are not recommended by the ICH S2R1 guideline since it is recognized that germ cell genotoxicity can be viewed as somatic genotoxicity in germ cells.
10.10.2.
Evaluation of In Vivo Results
Generally, in vivo results are considered to provide a very relevant readout for the assessment of a carcinogenic risk. Primarily, this is achieved by the fact that animal systems possess ADME characteristics of a functional organism, which in its full complexity cannot be replicated by in vitro systems. While in vivo systems generally are considered less sensitive but more specific than in vitro systems, situations have been described where in vivo false-positive results were obtained—that is, in vivo positive results that were considered not to indicate a relevant genotoxic potential that would be predictive of carcinogenicity. Owing to the widespread use of the in vivo micronucleus test, most of those examples have been described for this test system (Tweats et al. 2007a,b). For rodent bone marrow or peripheral blood micronucleus tests, these disturbances include changes in core body temperature (hypothermia and hyperthermia) and increases in erythropoiesis following prior toxicity to erythroblasts or by direct stimulation of cell division in these cells. Also, administration of erythropoietin or conditions such as anoxia, which stimulate endogenous erythropoietin production, have been described. As a consequence, the frequency of micronucleated cells in the population examined is increased in the absence of any genotoxic insult to the cellular DNA. Other examples include obvious mechanisms such as in vivo-relevant or in vivo-unique metabolic pathways, which
264
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
are only incompletely catalyzed in standard in vitro test systems. Examples described include urethane, where CYP2E1 and carboxylesterase isozyme hydrolase A are required (Forkert and Lee 1997), benzene, a known carcinogen where the activating metabolism has not been elucidated conclusively (Ahmad Khan 2007), or compounds where conjugation reactions are involved (Glatt 2000), which are normally not supported by in vitro systems due to the lack of the appropriate cofactors. Due to those examples, full genotoxicity evaluation of a compound requires the conduct of in vivo tests. However, overall those cases are rare. In contrast, ample evidence has accumulated to show the existence of in vitro-only positive results (Kirkland et al. 2005a, 2006), and mechanisms have been described to explain this phenomenon (Kirkland et al. 2007a). As a consequence, in vivo tests are considered to provide very specific readout on a genotoxic potential of a compound, which means that there is a high level of confidence that a positive response reflects a relevant genotoxic potential of a compound. According to the revised ICH S2R1 guideline, if there is an increase in micronuclei in vivo, all the toxicological data should be evaluated to determine whether a nongenotoxic effect may be the cause or a contributing factor. If nonspecific effects of disturbed erythropoiesis or physiology are suspected, an in vivo assay for chromosome aberrations may be more appropriate. If a “real” increase is suspected, strategies would be needed to demonstrate whether the increase is due to chromosome loss or chromosome breakage, which would allow for the definition of a threshold value in the former case, because there is evidence that aneuploidy induction (e.g., with spindle poisons) follows a nonlinear dose response. Thus, it may be possible to determine that there is a threshold exposure below which chromosome loss is not expected and to determine whether an appropriate safety margin exists compared with clinical exposure. In conclusion, the assessment of the genotoxic potential of a compound should take into account the totality of the findings and acknowledge the intrinsic values and limitations of both in vitro and in vivo tests.
10.11.
SUMMARY AND OUTLOOK
In summary, regulatory genotoxicity testing and evaluation of results has seen many changes over the past ∼30 years, with an initial focus on germ cell effects and subsequently moving mainly toward detection and risk assessment of genotoxic carcinogens. Accordingly, rodent germ cell tests have earlier dominated the field, while short-term in vivo test with somatic cells and in vitro tests using primary (human) cells or established cell lines are of widespread use nowadays. Over the last 5 years, it became more and more apparent that the currently employed in vitro genotoxicity tests do excellently fulfill their role with regard to sensitively detecting potential genotoxic carcinogens but do that at a price of an unacceptable low specificity (Kirkland et al. 2005a). Accordingly, in vitro tests need to be refined (ICH 2008; Kirkland et al. 2008; Thybaud et al. 2007a,b) and in vivo tests need to be appropriately designed for risk assessment (ICH 2008; Kirkland et al. 2008) including the adequate inclusion of metabolism pathways (Ku et al. 2007). Hence, the authors of this chapter do foresee a new era of in vitro and in vivo genotoxicity tests with a
REFERENCES
265
better focus on human relevance and mutational contribution to cancer risk assessment for humans.
REFERENCES Aardema, M. J., Albertini, S., Arni, P., Henderson, L. M., Kirsch-Volders, M., Mackay, J. M., Sarrif, A. M., Stringer, D. A., and Taalman, R. D. (1998). Aneuploidy: A report of an ECETOC task force. Mutat Res 410, 3–79. Ahmad Khan, H. (2007). Benzene’s toxicity: A consolidated short review of human and animal studies. Hum Exp Toxicol 26, 677–685. Ames, B. N., McCann, J., and Yamasaki, E. (1975). Methods for detecting carcinogens and mutagens with the Salmonella/mammalian-microsome mutagenicity test. Mutat Res 31, 347–364. Ashby, J., Waters, M. D., Preston, J., Adler, I. D., Douglas, G. R., Fielder, R., Shelby, M. D., Anderson, D., Sofuni, T., Gopalan, H. N., Becking, G., and Sonich-Mullin, C. (1996). IPCS harmonization of methods for the prediction and quantification of human carcinogenic/mutagenic hazard, and for indicating the probable mechanism of action of carcinogens. Mutat Res 352, 153–157. Auerbach, C., and Robson, J. M. (1946). Chemical production of mutations. Nature 157, 302. Auletta, A. E., Dearfield, K. L., and Cimino, M. C. (1993). Mutagenicity test schemes and guidelines: U.S. EPA Office of Pollution Prevention and Toxics and Office of Pesticide Programs. Environ Mol Mutagen 21, 38–45; discussion 46–57. Bergman, K., Muller, L., and Teigen, S. W. (1996). Series: Current issues in mutagenesis and carcinogenesis, No. 65. The genotoxicity and carcinogenicity of paracetamol: A regulatory (re)view. Mutat Res 349, 263–288. Bos, P. M., Baars, B. J., and van Raaij, M. T. (2004). Risk assessment of peak exposure to genotoxic carcinogens: A pragmatic approach. Toxicol Lett 151, 43–50. Brambilla, G., and Martelli, A. (2004). Failure of the standard battery of short-term tests in detecting some rodent and human genotoxic carcinogens. Toxicology 196, 1–19. Broschinski, L., Madle, S., and Hensel, C. (1998). Genotoxicity tests for new chemicals in Germany: routine in vitro test systems. Mutat Res 418, 121–129. Brusick, D. (1987). Screening chemicals for genotoxic properties. In Principles of Genetic Toxicology, Springer, New York, pp. 79–117. Burlinson, B., Tice, R. R., Speit, G., Agurell, E., Brendler-Schwaab, S. Y., Collins, A. R., Escobar, P., Honma, M., Kumaravel, T. S., Nakajima, M., Sasaki, Y. F., Thybaud, V., Uno, Y., Vasquez, M., and Hartmann, A. (2007). Fourth International Workgroup on Genotoxicity testing: results of the in vivo Comet assay workgroup. Mutat Res 627, 31–35. CHMP (2006). Committee for Medicinal Products for Human Use (CHMP): Guideline on the limits of genotoxic impurities. London, 28 June 2006, CPMP/SWP/5199/02, EMEA/CHMP/QWP/251344/2006, http://www.emea.europa.eu/pdfs/human/swp/519902en.pdf, 1–8. CHMP (2008). CHMP Safety Working Party (SWP): Question & Answers on the CHMP Guideline on limits of genotoxic impurities. London, 26 June 2008, Doc. Ref. EMEA/CHMP/SWP/431994/2007 Revision 1, http://www.emea.europa.eu/pdfs/human/swp/43199407en.pdf, 1–3. Cimino, M. C. (2006). Comparative overview of current international strategies and guidelines for genetic toxicology testing for regulatory purposes. Environ Mol Mutagen 47, 362–390. Corvi, R., Albertini, S., Hartung, T., Hoffmann, S., Maurici, D., Pfuhler, S., van Benthem, J., and Vanparys, P. (2008). ECVAM retrospective validation of in vitro micronucleus test (MNT). Mutagenesis 23, 271–283. CPDB (2007). The Carcinogenicity Potency Database (CPDB). http://potency.berkeley.edu/cpdb.html. CSGMT (1986). Sex difference in the micronucleus test. The Collaborative Study Group for the Micronucleus Test. Mutat Res 172, 151–163. CSGMT (1988). Strain difference in the micronucleus test. The Collaborative Study Group for the Micronucleus Test. Mutat Res 204, 307–316. CSGMT (1990). Single versus multiple dosing in the micronucleus test: the summary of the fourth collaborative study by CSGMT/JEMS.MMS. Collaborative Study Group for the Micronucleus Test,
266
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
the Mammalian Mutagenesis Study Group of the Environmental Mutagen Society, Japan (CSGMT/ JEMS.MMS). Mutat Res 234, 205–222. CSGMT (1992). Micronucleus test with mouse peripheral blood erythrocytes by acridine orange supravital staining: the summary report of the 5th collaborative study by CSGMT/JEMS.MMS. The Collaborative Study Group for the Micronucleus Test. Mutat Res 278, 83–98. CSGMT (1995). Protocol recommended by the CSGMT/JEMS.MMS for the short-term mouse peripheral blood micronucleus test. The Collaborative Study Group for the Micronucleus Test (CSGMT) (CSGMT/JEMS.MMS, The Mammalian Mutagenesis Study Group of the Environmental Mutagen Society of Japan). Mutagenesis 10, 153–159. Dearfield, K. L., and Moore, M. M. (2005). Use of genetic toxicology information for risk assessment. Environ Mol Mutagen 46, 236–245. DHEW (1977). Approaches to determining the mutagenic properties of chemicals: Risk to future generations. J Environ Pathol Toxicol 1, 301–352. DOH (1991). Genotoxicity assays recommended in the Health Protection Branch Genotoxicity Committee Report. In Canada Department of National Health and Welfare, Health Protection Branch, Ottawa. Dubrova, Y. E., Plumb, M., Brown, J., Fennelly, J., Bois, P., Goodhead, D., and Jeffreys, A. J. (1998). Stage specificity, dose response, and doubling dose for mouse minisatellite germ-line mutation induced by acute radiation. Proc Natl Acad Sci USA 95, 6251–6255. EC (1967). Annex VI General classification and labelling requirements for dangerous substances and preparations. Council Directive 67/548/EEC, 1–47. EC (1999). Directive 1999/45/EC of the European Parliament and the Council of 31 May 1999 concerning the approximation of the laws, regulations and administrative provisions of the Member States relating to the classification, packaging and labelling of dangerous preparations. Official J Eur Commun L200, 1–68. EC (2001). Commission of the European Communities, White paper—Strategy for a future chemicals policy. Brussels, 27.2.2001, COM(2001) 88 final, http://eurlex.europa.eu/LexUriServ/site/en/com/2001/ com2001_0088en01.pdf, 1–32. EC (2006). Regulation (EC) No 1907/2006 of the European Parliament and of the Council of 18 December 2006 concerning the registration, Evaluation, Authorisation and Restriction of Chemicals (REACH), establishing a European Chemicals Agency, amending Directive 1999/45/EC and repealing Council Regulation (EEC) No. 793/93 and Commission Regulation (EC) No. 1488/94 as well as Council Directive 76/769/EEC and Commission Directives 91/155/EEC, 93/67/EEC, 93/105/EC, and 2000/21/EC. Official J Eur Commun L136, 3. EC (2008). Annex V. Part B: Methods for the determination of toxicity. Directive 67/548/EEC, http:// ecb.jrc.ec.europa.eu/testing-methods/annex5/. ECHA (2008a). Guidance on information requirements and chemical safety assessment—Chapter R.7a: Endpoint specific guidance. Guidance for the implementation of REACH, http://guidance.echa.europa. eu/docs/guidance_document/information_requirements_r7a_en.pdf?vers=20_08_08, 1–428. ECHA (2008b). Guidance on information requirements and chemical safety assessment. Part A: Introduction to the Guidance Document. Guidance for the implementation of REACH, http://reach.jrc. it/docs/guidance_document/information_requirements_part_a_en.pdf?vers=20_08_08, 1–44. ECHA (2008c). Guidance on information requirements and chemical safety assessment. Part F: Chemical Safety Report. Guidance for the implementation of REACH, http://reach.jrc.it/docs/guidance_document/information_requirements_part_f_en.pdf?vers=20_08_08, 1–10. Ehling, U. H., Machemer, L., Buselmaier, W., Dycka, J., Frohberg, H., Kratochvilova, J., Lang, R., Lorke, D., Muller, D., Peh, J., Rothrborn, G., Roll, R., Schulze-Schencking, M., and Wiemann, H. (1978). Standard protocol for the dominant lethal test on male mice set up by the work group “Dominant Lethal Mutations of the ad hoc Committe Chemogenetics”. Arch Toxicol 39, 173–185. EPA (2008). OPPTS Harmonized Test Guidelines Series 870 Health Effects Test Guidelines. http://www. epa.gov/opptsfrs/publications/OPPTS_Harmonized/870_Health_Effects_Test_Guidelines/Series/. EU (2003). Proposal for a Regulation of the European Parliament and of the Council concerning the Registration, Evaluation, Authorization and Restriction of Chemicals (REACH), establishing a European Chemicals Agency and Amending Directive 1999/45/EC and Regulation (EC) {on Persistent Organic Pollutants} Proposal for a DIRECTIVE OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL amending Council Directive 67/548/EEC in order to adapt it to Regulation (EC) of the
REFERENCES
267
European Parliament and of the Council concerning the registration, evaluation, authorization and restriction of chemicals. Brussels, 29.10.2003 COM(2003) 644 final 1–6, http://europa.eu/eur-lex/en/ com/pdf/2003/com2003_0644en.html. FDA (1982). Toxicological Principles for the Safety Assessment of Direct Food Additives and Color Additives Used in Food, Bureau of Foods, U.S. Food and Drug Adminstration (FDA), Washington, D.C. FDA (1993). Toxicological Principles for the Safety Assessment of Direct Food Additives and Color Additives Used in Food, Redbook II Draft Guidance, http://www.cfsan.fda.gov/∼redbook/redtoc93. html, US Food and Drug Administration, Rockville, MD. FDA (2000). IV.C.1. Short-term tests for genetic toxicity. In Toxicological Principles for the Safety Assessment of Food Ingredients Redbook 2000, http://www.cfsan.fda.gov/∼redbook/red-ivc1.html. US Food and Drug Administration, Rockville, MD. FDA (2006). Guidance for industry and review staff recommended approaches to integration of genetic toxicology study results, http://www.fda.gov/cder/guidance/6848fnl.pdf. FDA (2008). Draft guidance for industry on genotoxic and carcinogenic impurities in drug substances and products: Recommended approaches. Fed Reg 73, 76361–76362. Forkert, P. G., and Lee, R. P. (1997). Metabolism of ethyl carbamate by pulmonary cytochrome P450 and carboxylesterase isozymes: Involvement of CYP2E1 and hydrolase A. Toxicol Appl Pharmacol 146, 245–254. Gatehouse, D., Haworth, S., Cebula, T., Gocke, E., Kier, L., Matsushima, T., Melcion, C., Nohmi, T., Ohta, T., Venitt, S., et al. (1994). Recommendations for the performance of bacterial mutation assays. Mutat Res 312, 217–233. Glatt, H. (2000). Sulfotransferases in the bioactivation of xenobiotics. Chem Biol Interact 129, 141–170. Greenwood, S. K., Hill, R. B., Sun, J. T., Armstrong, M. J., Johnson, T. E., Gara, J. P., and Galloway, S. M. (2004). Population doubling: A simple and more accurate estimation of cell growth suppression in the in vitro assay for chromosomal aberrations that reduces irrelevant positive results. Environ Mol Mutagen 43, 36–44. Hamada, S., Nakajima, K., Namiki, C., Serikawa, T., and Hayashi, M. (2003). Sex differences in the chemical induction of micronuclei in the rat. Environ Mutagen Res 25, 33–37. Hamada, S., Sutou, S., Morita, T., Wakata, A., Asanami, S., Hosoya, S., Ozawa, S., Kondo, K., Nakajima, M., Shimada, H., Osawa, K., Kondo, Y., Asano, N., Sato, S., Tamura, H., Yajima, N., Marshall, R., Moore, C., Blakey, D. H., Schechtman, L. M., Weaver, J. L., Torous, D. K., Proudlock, R., Ito, S., Namiki, C., and Hayashi, M. (2001). Evaluation of the rodent micronucleus assay by a 28-day treatment protocol: Summary of the 13th Collaborative Study by the Collaborative Study Group for the Micronucleus Test (CSGMT)/Environmental Mutagen Society of Japan (JEMS)–Mammalian Mutagenicity Study Group (MMS). Environ Mol Mutagen 37, 93–110. Hardman, J. G., Limbird, L. E., and Gilman, A. G., eds. (2001). Goodman & Gilman’s—The Pharmacological Basis of Therapeutics, 10th edition. McGraw-Hill Professional, New York. Hartmann, A., Agurell, E., Beevers, C., Brendler-Schwaab, S., Burlinson, B., Clay, P., Collins, A., Smith, A., Speit, G., Thybaud, V., and Tice, R. R. (2003). Recommendations for conducting the in vivo alkaline Comet assay. 4th International Comet Assay Workshop. Mutagenesis 18, 45–51. Hayashi, M., MacGregor, J. T., Gatehouse, D. G., Adler, I. D., Blakey, D. H., Dertinger, S. D., Krishna, G., Morita, T., Russo, A., and Sutou, S. (2000). In vivo rodent erythrocyte micronucleus assay. II. Some aspects of protocol design including repeated treatments, integration with toxicity testing, and automated scoring. Environ Mol Mutagen 35, 234–252. Hayashi, M., MacGregor, J. T., Gatehouse, D. G., Blakey, D. H., Dertinger, S. D., Abramsson-Zetterberg, L., Krishna, G., Morita, T., Russo, A., Asano, N., Suzuki, H., Ohyama, W., and Gibson, D. (2007). In vivo erythrocyte micronucleus assay III. Validation and regulatory acceptance of automated scoring and the use of rat peripheral blood reticulocytes, with discussion of non-hematopoietic target cells and a single dose-level limit test. Mutat Res 627, 10–30. Hayashi, M., Tice, R. R., MacGregor, J. T., Anderson, D., Blakey, D. H., Kirsh-Volders, M., Oleson, F. B., Jr., Pacchierotti, F., Romagna, F., Shimada, H., et al. (1994). In vivo rodent erythrocyte micronucleus assay. Mutat Res 312, 293–304. Heddle, J. A. (1973). A rapid in vivo test for chromosomal damage. Mutat Res 18, 187–190.
268
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
Heddle, J. A., Dean, S., Nohmi, T., Boerrigter, M., Casciano, D., Douglas, G. R., Glickman, B. W., Gorelick, N. J., Mirsalis, J. C., Martus, H. J., Skopek, T. R., Thybaud, V., Tindall, K. R., and Yajima, N. (2000). In vivo transgenic mutation assays. Environ Mol Mutagen 35, 253–259. Hill, F. S., Marchetti, F., Liechty, M., Bishop, J., Hozier, J., and Wyrobek, A. J. (2003). A new FISH assay to simultaneously detect structural and numerical chromosomal abnormalities in mouse sperm. Mol Reprod Dev 66, 172–180. Hoffmann, G. R. (1998). Genetic toxicology. In Casarett & Doull’s Toxicology—The Basic Science of Poisons, 5th edition, Klaassen, C. D., ed., McGraw-Hill, New York, pp. 269–300. Hollaender, A. (1971). Chemical Mutagens: Principles and Methods for Their Detection, Plenum Press, New York. ICH (1996). Specific aspects of regulatory genotoxicity tests for pharmaceuticals. Guideline for Industry ICH S2A, http://www.ich.org. ICH (1997a). ICH Guideline Q3C: Impurities: Guidelines for residual solvents. ICH Harmonized Tripartite Guideline Q3C, http://www.ich.org. ICH (1997b). S2B Genotoxicity: A standard battery for genotoxicity testing of pharmaceuticals. Guidance for Industry S2B, http://www.ich.org. ICH (2002). Guideline Q3A(R): Impurities in new drug substances. ICH Harmonized Tripartite Guideline Q3A(R), http://www.ich.org. ICH (2003). ICH Guideline Q3B(R): Impurities in new drug products. ICH Harmonized Tripartite Guideline Q3B(R), http://www.ich.org. ICH (2008). Guidance on genotoxicity testing and data interpretation for pharmaceuticals intended for human use. ICH Harmonized Tripartite Guideline S2(R1), http://www.ich.org. IPCS (2007). Draft guidance on mutagenicity testing for chemical risk assessment open for comment. The International Programme on Chemical Safety (IPCS), http://www.who.int/ipcs/methods/ harmonization/areas/mutagenicity/en/index.html. Kasper, P., Uno, Y., Mauthe, R., Asano, N., Douglas, G., Matthews, E., Moore, M., Mueller, L., Nakajima, M., Singer, T., and Speit, G. (2007). Follow-up testing of rodent carcinogens not positive in the standard genotoxicity testing battery: IWGT workgroup report. Mutat Res 627, 106–116. Kirkland, D., Aardema, M., Henderson, L., and Muller, L. (2005a). Evaluation of the ability of a battery of three in vitro genotoxicity tests to discriminate rodent carcinogens and non-carcinogens I. Sensitivity, specificity and relative predictivity. Mutat Res 584, 1–256. Kirkland, D., Aardema, M., Muller, L., and Makoto, H. (2006). Evaluation of the ability of a battery of three in vitro genotoxicity tests to discriminate rodent carcinogens and non-carcinogens II. Further analysis of mammalian cell results, relative predictivity and tumour profiles. Mutat Res 608, 29–42. Kirkland, D., Kasper, P., Muller, L., Corvi, R., and Speit, G. (2008). Recommended lists of genotoxic and non-genotoxic chemicals for assessment of the performance of new or improved genotoxicity tests: A follow-up to an ECVAM workshop. Mutat Res 653, 99–108. Kirkland, D., Pfuhler, S., Tweats, D., Aardema, M., Corvi, R., Darroudi, F., Elhajouji, A., Glatt, H., Hastwell, P., Hayashi, M., Kasper, P., Kirchner, S., Lynch, A., Marzin, D., Maurici, D., Meunier, J. R., Muller, L., Nohynek, G., Parry, J., Parry, E., Thybaud, V., Tice, R., van Benthem, J., Vanparys, P., and White, P. (2007a). How to reduce false positive results when undertaking in vitro genotoxicity testing and thus avoid unnecessary follow-up animal tests: Report of an ECVAM Workshop. Mutat Res 628, 31–55. Kirkland, D. J. (1993). Genetic toxicology testing requirements: Official and unofficial views from Europe. Environ Mol Mutagen 21, 8–14; discussion 46–57. Kirkland, D. J., Hayashi, M., Jacobson-Kram, D., Kasper, P., MacGregor, J. T., Muller, L., and Uno, Y. (2007b). The International Workshops on Genotoxicity Testing (IWGT): History and achievements. Mutat Res 627, 1–4. Kirkland, D. J., Hayashi, M., Jacobson-Kram, D., Kasper, P., MacGregor, J. T., Muller, L., and Uno, Y. (2007c). Summary of major conclusions from the 4th IWGT, San Francisco, 9–10 September, 2005. Mutat Res 627, 5–9. Kirkland, D. J., Henderson, L., Marzin, D., Muller, L., Parry, J. M., Speit, G., Tweats, D. J., and Williams, G. M. (2005b). Testing strategies in mutagenicity and genetic toxicology: An appraisal of the guidelines of the European Scientific Committee for Cosmetics and Non-Food Products for the evaluation of hair dyes. Mutat Res 588, 88–105.
REFERENCES
269
Kirsch-Volders, M., Elhajouji, A., Cundari, E., and Van Hummelen, P. (1997). The in vitro micronucleus test: A multi-endpoint assay to detect simultaneously mitotic delay, apoptosis, chromosome breakage, chromosome loss and non-disjunction. Mutat Res 392, 19–30. Kirsch-Volders, M., Sofuni, T., Aardema, M., Albertini, S., Eastmond, D., Fenech, M., Ishidate, M., Jr., Kirchner, S., Lorge, E., Morita, T., Norppa, H., Surralles, J., Vanhauwaert, A., and Wakata, A. (2003). Report from the in vitro micronucleus assay working group. Mutat Res 540, 153–163. Ku, W. W., Bigger, A., Brambilla, G., Glatt, H., Gocke, E., Guzzie, P. J., Hakura, A., Honma, M., Martus, H. J., Obach, R. S., and Roberts, S. (2007). Strategy for genotoxicity testing—Metabolic considerations. Mutat Res 627, 59–77. Lambert, I. B., Singer, T. M., Boucher, S. E., and Douglas, G. R. (2005). Detailed review of transgenic rodent mutation assays. Mutat Res 590, 1–280. Lorge, E., Lambert, C., Gervais, V., Becourt-Lhote, N., Delongeas, J. L., and Claude, N. (2007). Genetic toxicity assessment: Employing the best science for human safety evaluation. Part II: Performances of the in vitro micronucleus test compared to the mouse lymphoma assay and the in vitro chromosome aberration assay. Toxicol Sci 96, 214–217. Lorge, E., Thybaud, V., Aardema, M. J., Oliver, J., Wakata, A., Lorenzon, G., and Marzin, D. (2006). SFTG international collaborative study on in vitro micronucleus test I. General conditions and overall conclusions of the study. Mutat Res 607, 13–36. MacGregor, J. T. (1994). Environmental mutagenesis: past and future directions. Environ Mol Mutagen 23(Suppl 24), 73–77. MacGregor, J. T., Bishop, M. E., McNamee, J. P., Hayashi, M., Asano, N., Wakata, A., Nakajima, M., Saito, J., Aidoo, A., Moore, M. M., and Dertinger, S. D. (2006). Flow cytometric analysis of micronuclei in peripheral blood reticulocytes: II. An efficient method of monitoring chromosomal damage in the rat. Toxicol Sci 94, 92–107. MacGregor, J. T., Casciano, D., and Muller, L. (2000). Strategies and testing methods for identifying mutagenic risks. Mutat Res 455, 3–20. MacGregor, J. T., Wehr, C. M., and Gould, D. H. (1980). Clastogen-induced micronuclei in peripheral blood erythrocytes: The basis of an improved micronucleus test. Environ Mutagen 2, 509–514. Madle, S., Dean, S. W., Andrae, U., Brambilla, G., Burlinson, B., Doolittle, D. J., Furihata, C., Hertner, T., McQueen, C. A., and Mori, H. (1994). Recommendations for the performance of UDS tests in vitro and in vivo. Mutat Res 312, 263–285. Madle, S., Korte, A., and Bass, R. (1987). Experience with mutagenicity testing of new drugs: Viewpoint of a regulatory agency. Mutat Res 182, 187–192. Maron, D. M., and Ames, B. N. (1983). Revised methods for the Salmonella mutagenicity test. Mutat Res 113, 173–215. Matthews, E. J., Kruhlak, N. L., Cimino, M. C., Benz, R. D., and Contrera, J. F. (2006). An analysis of genetic toxicity, reproductive and developmental toxicity, and carcinogenicity data: I. Identification of carcinogens using surrogate endpoints. Regul Toxicol Pharmacol 44, 83–96. McCann, J., Choi, E., Yamasaki, E., and Ames, B. N. (1975). Detection of carcinogens as mutagens in the Salmonella/microsome test: Assay of 300 chemicals. Proc Natl Acad Sci USA 72, 5135– 5139. Merk, O., and Speit, G. (1999). Detection of crosslinks with the comet assay in relationship to genotoxicity and cytotoxicity. Environ Mol Mutagen 33, 167–172. Meselson, M. (1971). Preface. In Chemical Mutagens: Principles and Methods for Their Detection, Hollaender, A. ed., Plenum Press, New York, pp. 9–12. Moore, M. M., Honma, M., Clements, J., Bolcsfoldi, G., Burlinson, B., Cifone, M., Clarke, J., Clay, P., Doppalapudi, R., Fellows, M., Gollapudi, B., Hou, S., Jenkinson, P., Muster, W., Pant, K., Kidd, D. A., Lorge, E., Lloyd, M., Myhr, B., O’Donovan, M., Riach, C., Stankowski, L. F., Jr., Thakur, A. K., and Van Goethem, F. (2007). Mouse lymphoma thymidine kinase gene mutation assay: Meeting of the International Workshop on Genotoxicity Testing, San Francisco, 2005, recommendations for 24-h treatment. Mutat Res 627, 36–40. Moore, M. M., Honma, M., Clements, J., Bolcsfoldi, G., Burlinson, B., Cifone, M., Clarke, J., Delongchamp, R., Durward, R., Fellows, M., Gollapudi, B., Hou, S., Jenkinson, P., Lloyd, M., Majeska, J., Myhr, B., O’Donovan, M., Omori, T., Riach, C., San, R., Stankowski, L. F., Jr., Thakur, A. K., Van Goethem, F., Wakuri, S., and Yoshimura, I. (2006). Mouse lymphoma thymidine
270
CHAPTER 10 GENETIC TOXICOLOGY TESTING GUIDELINES AND REGULATIONS
kinase gene mutation assay: Follow-up meeting of the International Workshop on Genotoxicity Testing, Aberdeen, Scotland, 2003—Assay acceptance criteria, positive controls, and data evaluation. Environ Mol Mutagen 47, 1–5. Müller, L., Blakey, D., Dearfield, K. L., Galloway, S., Guzzie, P., Hayashi, M., Kasper, P., Kirkland, D., MacGregor, J. T., Parry, J. M., Schechtman, L., Smith, A., Tanaka, N., Tweats, D., and Yamasaki, H. (2003). Strategy for genotoxicity testing and stratification of genotoxicity test results—Report on initial activities of the IWGT Expert Group. Mutat Res 540, 177–181. Müller, L., and Kasper, P. (2000). Human biological relevance and the use of threshold-arguments in regulatory genotoxicity assessment: Experience with pharmaceuticals. Mutat Res 464, 19–34. Müller, L., Kasper, P., and Madle, S. (1991). The quality of genotoxicity testing of drugs. Experiences of a regulatory agency with new and old compounds. Mutagenesis 6, 143–149. Müller, L., Kikuchi, Y., Probst, G., Schechtman, L., Shimada, H., Sofuni, T., and Tweats, D. (1999). ICH-harmonised guidances on genotoxicity testing of pharmaceuticals: Evolution, reasoning and impact. Mutat Res 436, 195–225. Müller, L., Mauthe, R. J., Riley, C. M., Andino, M. M., Antonis, D. D., Beels, C., DeGeorge, J., De Knaep, A. G., Ellison, D., Fagerland, J. A., Frank, R., Fritschel, B., Galloway, S., Harpur, E., Humfrey, C. D., Jacks, A. S., Jagota, N., Mackinnon, J., Mohan, G., Ness, D. K., O’Donovan, M. R., Smith, M. D., Vudathala, G., and Yotti, L. (2006). A rationale for determining, testing, and controlling specific impurities in pharmaceuticals that possess potential for genotoxicity. Regul Toxicol Pharmacol 44, 198–211. NRC (1983). Identifying and Estimating the Genetic Impact of Chemical Mutagens, National Academy Press, Washington, D.C. OECD (1997). Guidelines for genetic toxicology testing (TGs No. 471-486). OECD Guideline for the Testing of Chemicals, http://www.oecd.org/dataoecd/9/11/33663321.pdf. OECD (2006). Draft proposal for a new guideline 487: In vitro micronucleus test (2nd version, 21 December 2006). OECD Guideline for the Testing of Chemicals 487, 1–16. Pfuhler, S., and Wolf, H. U. (1996). Detection of DNA-crosslinking agents with the alkaline comet assay. Environ Mol Mutagen 27, 196–201. Phillips, D. H., Farmer, P. B., Beland, F. A., Nath, R. G., Poirier, M. C., Reddy, M. V., and Turteltaub, K. W. (2000). Methods of DNA adduct determination and their application to testing compounds for genotoxicity. Environ Mol Mutagen 35, 222–233. Purves, D., Harvey, C., Tweats, D., and Lumley, C. E. (1995). Genotoxicity testing: Current practices and strategies used by the pharmaceutical industry. Mutagenesis 10, 297–312. Romagna, F. (1988). Series: “Current issues in mutagenesis and carcinogenesis.” 12. Improved method of preparing bone marrow micronucleus assay slides. Mutat Res 206, 307–309. Scott, D., Galloway, S. M., Marshall, R. R., Ishidate, M., Jr., Brusick, D., Ashby, J., and Myhr, B. C. (1991). International Commission for Protection Against Environmental Mutagens and Carcinogens. Genotoxicity under extreme culture conditions. A report from ICPEMC Task Group 9. Mutat Res 257, 147–205. Shelby, M. D., and Sofuni, T. (1991). Toxicology testing requirements and the U.S.–Japan collaborative study on in vitro tests for chromosomal aberrations. Environ Health Perspect 94, 255–259. Shirasu, Y. (1988). The Japanese mutagenicity studies guidelines for pesticide registration. Mutat Res 205, 393–395. Snyder, R. D., and Green, J. W. (2001). A review of the genotoxicity of marketed pharmaceuticals. Mutat Res 488, 151–169. Speit, G., and Hartmann, A. (2005). The comet assay: A sensitive genotoxicity test for the detection of DNA damage. Methods Mol Biol 291, 85–95. Storer, R. D., McKelvey, T. W., Kraynak, A. R., Elia, M. C., Barnum, J. E., Harmon, L. S., Nichols, W. W., and DeLuca, J. G. (1996). Revalidation of the in vitro alkaline elution/rat hepatocyte assay for DNA damage: Improved criteria for assessment of cytotoxicity and genotoxicity and results for 81 compounds. Mutat Res 368, 59–101. Suzuki, H., Ikeda, N., Kobayashi, K., Terashima, Y., Shimada, Y., Suzuki, T., Hagiwara, T., Hatakeyama, S., Nagaoka, K., Yoshida, J., Saito, Y., Tanaka, J., and Hayashi, M. (2005). Evaluation of liver and peripheral blood micronucleus assays with 9 chemicals using young rats. A study by the Collaborative
REFERENCES
271
Study Group for the Micronucleus Test (CSGMT)/Japanese Environmental Mutagen Society (JEMS)— Mammalian Mutagenicity Study Group (MMS). Mutat Res 583, 133–145. Thybaud, V., Aardema, M., Casciano, D., Dellarco, V., Embry, M. R., Gollapudi, B. B., Hayashi, M., Holsapple, M. P., Jacobson-Kram, D., Kasper, P., MacGregor, J. T., and Rees, R. (2007a). Relevance and follow-up of positive results in in vitro genetic toxicity assays: An ILSI-HESI initiative. Mutat Res 633, 67–79. Thybaud, V., Aardema, M., Clements, J., Dearfield, K., Galloway, S., Hayashi, M., Jacobson-Kram, D., Kirkland, D., MacGregor, J. T., Marzin, D., Ohyama, W., Schuler, M., Suzuki, H., and Zeiger, E. (2007b). Strategy for genotoxicity testing: Hazard identification and risk assessment in relation to in vitro testing. Mutat Res 627, 41–58. Tice, R. R., Agurell, E., Anderson, D., Burlinson, B., Hartmann, A., Kobayashi, H., Miyamae, Y., Rojas, E., Ryu, J. C., and Sasaki, Y. F. (2000). Single cell gel/comet assay: Guidelines for in vitro and in vivo genetic toxicology testing. Environ Mol Mutagen 35, 206–221. TSCA (1976). Toxic Substances Control Act. 15 U.S.C. §2601 et seq. Tweats, D. J., Blakey, D., Heflich, R. H., Jacobs, A., Jacobsen, S. D., Morita, T., Nohmi, T., O’Donovan, M. R., Sasaki, Y. F., Sofuni, T., and Tice, R. (2007a). Report of the IWGT working group on strategies and interpretation of regulatory in vivo tests I. Increases in micronucleated bone marrow cells in rodents that do not indicate genotoxic hazards. Mutat Res 627, 78–91. Tweats, D. J., Blakey, D., Heflich, R. H., Jacobs, A., Jacobsen, S. D., Morita, T., Nohmi, T., O’Donovan, M. R., Sasaki, Y. F., Sofuni, T., and Tice, R. (2007b). Report of the IWGT working group on strategy/ interpretation for regulatory in vivo tests II. Identification of in vivo-only positive compounds in the bone marrow micronucleus test. Mutat Res 627, 92–105. UKCOM (2000). Guidance on a strategy for testing of chemicals for mutagenicity. Committee on Mutagenicity of Chemicals in Food, Consumer Products and the Environment (COM), http://www. dh.gov.uk/assetRoot/04/07/71/96/04077196.pdf, 1–36. van der Jagt, K., Munn, S., Torslov, J., and de Bruijn, J. (2004). Alternative approaches can reduce the use of test animals under REACH—Addendum to the report “Assessment of additional testing needs under REACH. Effects of (Q)SARS, risk based testing and voluntary industry initiatives.” Institute for Health and Consumer Protection (IHCP), European Commission, http://ecb.jrc.ec.europa.eu/ documents/REACH/PUBLICATIONS/Reducing_the_use_of_test_animals_under_REACH_IHCP_ report.pdf. Wakata, A., Miyamae, Y., Sato, S., Suzuki, T., Morita, T., Asano, N., Awogi, T., Kondo, K., and Hayashi, M. (1998). Evaluation of the rat micronucleus test with bone marrow and peripheral blood: summary of the 9th collaborative study by CSGMT/JEMS. MMS. Collaborative Study Group for the Micronucleus Test. Environmental Mutagen Society of Japan. Mammalian Mutagenicity Study Group. Environ Mol Mutagen 32, 84–100. Wassom, J. S. (1989). Origins of genetic toxicology and the Environmental Mutagen Society. Environ Mol Mutagen 14(Suppl 16), 1–6. Waters, M. D., and Auletta, A. (1981). The GENE-TOX program: Genetic activity evaluation. J Chem Inf Comput Sci 21, 35–38.
CH A P TE R
11
IN VITRO GENOTOX ASSAYS David Kirkland David Gatehouse
11.1.
INTRODUCTION
As discussed elsewhere in this volume, exposure to particular chemicals or complex mixtures can cause damage to DNA, and the resulting mutations can lead to cancer in later life. It is therefore important to determine whether new or existing chemicals possess the ability to damage DNA. This is determined by testing for genotoxic potential, and data concerning the genotoxicity of chemicals are an integral part of the basic toxicological information package used for decision-making and risk assessment (see Chapter 10). Genotoxicity testing was first described in the 1960s when several seminal conferences were held focusing on chemical mutagens, their effects on germ cells, and the risk to future generations. The area of concern was broadened in the 1970s when evidence relating genotoxicity and carcinogenicity began to accumulate. This was further supported by the use of in vitro metabolic activation systems capable of producing electrophilic metabolites, along with the fact that early analysis of rodent carcinogens and noncarcinogens suggested that almost all carcinogens were also genotoxic (Ames et al. 1973). Over the ensuing decades, it has become clear that nongenotoxic carcinogens also exist. Nonetheless, genotoxicity testing is conducted to provide evidence of both carcinogenic and germ cell mutagenic risk. Genotoxic potential can be manifest in a variety of ways, but the most common endpoints studied are: • • • •
Gene mutations Structural and numerical chromosomal aberrations DNA damage (strand breaks) DNA repair (unscheduled DNA synthesis, or UDS)
These endpoints can be studied both in vitro and in vivo, but no single test system can detect changes to all these endpoints. It has therefore become necessary to use a battery of complementary tests in order to effectively detect genotoxic potential. Various national expert committees have developed numerous guidelines
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
272
11.3. IN VITRO TESTS FOR GENE MUTATION IN BACTERIA
273
describing the tests that should be used to investigate the genotoxicity of chemicals. Over the decades some tests have been included in guidelines, but have subsequently lost favor or been superseded by improved methods. It is not the intention of this chapter to discuss all the in vitro genotoxicity tests that have been included in various guidelines, but instead to focus on those in vitro tests most commonly used at this time and to comment on their strengths and limitations. It is not intended to describe in detail the performance of these tests, but references that describe the recommended methods are given in Chapters 10 and 12.
11.2.
IN VITRO METABOLIC ACTIVATION
With the exception of primary hepatocytes, most of the indicator cells (bacteria and mammalian cells) used for in vitro genotoxicity testing possess a very limited capacity for endogenous metabolism of xenobiotics. Many carcinogens and mutagens are unable to interact with DNA unless they have undergone some degree of metabolism. To improve the ability of the test systems to detect as many authentic in vivo carcinogens and mutagens as possible, extracts of mammalian liver (usually rat) are incorporated. The liver is a rich source of mixed-function oxygenases capable of converting carcinogens and mutagens to reactive electrophiles. Crude homogenate such as the 9000g supernatant (S9 fraction) is used, which is composed of free endoplasmic reticulum, microsomes, soluble enzymes, and some cofactors. Normal uninduced S9 preparations are of limited value for screening because they are deficient in particular enzyme activities. In addition, species and tissue differences are most divergent in such preparations. These problems are reduced when enzyme inducers are used, and most commonly preparations are made from rat livers after enzyme induction with Aroclor 1254, which is a mixture of polychlorinated biphenyls, or with a combination of phenobarbitone and β-naphthoflavone which induce a similar range of mono-oxygenases and have been recommended as a safer alternative to Aroclor (Elliott et al. 1992). It should be noted that this system is only a first approximation to the complex metabolic processes that occur in vivo, and in particular there is little account taken of the phase II detoxification reactions. Such factors should be considered when interpreting positive in vitro results that are only seen in the presence of S9 mix. It should be noted that some cytochrome P450s in induced S9 are so greatly elevated above the levels in normal liver (see Table 11.1) that reactive metabolites may be produced in significant quantities in vitro that would be negligible in normal livers in vivo.
11.3. IN VITRO TESTS FOR GENE MUTATION IN BACTERIA The most widely used assays for detecting chemically induced gene mutations are those employing bacteria. These assays are utilized in all test batteries for genotoxicity because it is relatively straightforward to use them as a sensitive indirect indicator
274
CHAPTER 11
IN VITRO GENOTOX ASSAYS
TABLE 11.1. The Impact of Aroclor on the Induction of Various CYPs in Comparison to Normal Rat and Human Livera
CYP (nmol/mg microsomal protein)
Enzyme CYP1A1 CYP1A2 CYP2B1 CYP2B2 CYP2C6 CYP2C11 CYP2D1 CYP3A
Untreated Rat
Aroclor-Treated Rat
Induction Factor (Rat)
Level in Human Liver
0.04 <0.03 0.03 0.07 0.36 1.20 0.15 0.39
1.45 1.23 1.29 1.46 0.36 0.27 0.15 0.77
36 >41 43 21 1 0.23 1 2
0 0/+b + ++ 0/+c +++
a
From (Guengerich et al. 1982).
b
Depending on induction state.
c
Depending on genotype.
of DNA damage. Bacteria can be grown in large numbers overnight, permitting the detection of rare mutational events. The most commonly used bacteria are the S. typhimurium strains, which contain defined mutations in the histidine operon. These were developed by Bruce Ames, and they form the basis of the “reverse” mutation assays (Ames 1971). In these assays, bacteria that are already mutant at the histidine locus are treated with a range of concentrations of test chemical to determine whether the compound can induce a second mutation that directly reverses or suppresses the original mutations. This simple concept underlines the great strength of these assays because it provides enormous selective power that can identify a small number of the chosen mutants from a population of millions of unmutated cells and cells mutated in other genes. Each of the S. typhimurium strains contains one of a number of possible mutations in the histidine operon, and each can be reverted by either base-change or frameshift mutations. In order to make the bacteria more sensitive to mutation by chemical agents, an rfa mutation has been introduced into the Salmonella strains, resulting in defective lipopolysaccharide and increased permeability to large molecules. In addition, the test strains were constructed with a deletion removing the uvrB gene, which codes for the first enzyme in the error-free excision repair pathway. This renders the strains excision repair deficient, thus increasing their sensitivity to many genotoxins by several orders of magnitude. Lastly, some of the bacterial strains do not appear to possess classical error-prone repair as found in other members of the Enterobacteria such as E. coli. This results from a deficiency in umuD activity. This deficiency is overcome by insertion of a plasmid containing umuDC genes. Plasmid pKM101 is the most useful conferring on the bacteria sensitivity to mutation without a concomitant increase in sensitivity to the lethal effects of test compounds (Mortelmans and
11.3. IN VITRO TESTS FOR GENE MUTATION IN BACTERIA
275
Dousman 1986). The incorporation of strain TA102 into the test battery has been recommended, because the target mutation has an AT base pair at the critical site. This allows the detection of genotoxins not detected by the usual battery of S. typhimurium strains that possess mutations exclusively at GC base pairs. As an alternative, many guidelines recommend the use of the E. coli WP2 trpE strains, which contain a terminating ochre mutation in the trpE gene. The ochre mutation involves an AT base pair, and so reverse mutation can take place at the original site of mutation or in the relevant tRNA loci. A combination of E. coli WP2 trpE (pKM101) and E. coli WP2 trpE uvrA (pKM101) can be used as alternatives to S. typhimurium TA102 for the detection of point mutations at AT sites. In most regulatory test guidelines, the following base set of bacterial test strains is recommended: S. typhimurium: TA98, TA100, TA1535 S. typhimurium: TA1537 or TA97 or TA97a S. typhimurium: TA102 or E. coli WP2 uvrA or E. coli WP2 uvrA (pKM101) The use of the repair-proficient E. coli strain WP2 (pKM101) allows the detection of crosslinking agents that require an intact excision repair pathway to generate mutations, and this strain may also be selected. Various authors have described the ways to conduct the assay (Gatehouse et al. 1994; Tweats and Gatehouse 1999); however, a schematic representation of the conduct of the assay is given in Figure 11.1 with typical appearance of Ames plates in Figure 11.2. Bacterial mutation tests have been subjected to several large-scale
Bacterial Culture
Overlay Onto Minimal Agar Mix 37oC
Test Article Solution
Molten Soft Agar (+ his or tryp) Number of colonies = 7
Incubate for 2-3 Days
S9 Mix or Buffer
Score Colonies Using Automated Counter Figure 11.1. Schematic representation of the conduct of the Ames test (plate incorporation method). See insert for color representation of this figure.
276
CHAPTER 11 Control
Figure 11.2. this figure.
IN VITRO GENOTOX ASSAYS Mutagen-treated
Example of appearance of Ames plates. See insert for color representation of
trials over the years [e.g., Tennant et al. (1987)]. These studies were primarily concerned with assessing the correlation between results obtained in the assays and the carcinogenic activity of chemicals. Most of the studies suggest that there is a good qualitative relationship between genotoxicity in the Salmonella assay and carcinogenicity for many, although not all, chemical classes. This figure varies between a sensitivity of 60% and 90%, dependent upon chemical class. The bacterial assays seem to be particularly efficient in detecting trans-species, multiorgan animal carcinogens (Ashby and Tennant 1988). More recently, a comprehensive evaluation by Kirkland et al. (2005) using a database for the Ames test of over 540 chemicals confirmed that the sensitivity of the test was around 60% and the specificity (i.e., giving negative results with noncarcinogens) was higher at 73.9% (Kirkland et al. 2005).
11.4. IN VITRO TESTS FOR GENE MUTATION IN MAMMALIAN CELLS Although the prokaryotic systems described above are extremely versatile, rapid and mostly accurate in detecting genotoxins, the intrinsic differences between prokaryotic and eukaryotic cells in terms of genome structure and organization necessitate the use of mammalian test systems within any screening battery designed to detect the widest spectrum of genotoxins. A variety of in vitro mutation systems have been described in the literature, but only a few have been defined adequately for quantitative studies (Cole et al. 1990). Unlike the bacterial reverse mutation systems, these tests are based upon the detection of forward mutations. A defined large number of cells are treated with the test agent and then, after a set interval (termed the expression period), the cells are exposed to a selective toxic agent, so only those cells that have mutated can survive.
11.4. IN VITRO TESTS FOR GENE MUTATION IN MAMMALIAN CELLS
277
Usually, mutation is measured either in genes located on the X chromosome in male cells (XY), where only one copy of the target gene is present, or in cells (called heterozygotes) where two copies of the gene are present but only one copy is active, because the other has been inactivated through mutation or deletion. The most common systems make use of genes that are not essential for cell survival but allow the cell to salvage nucleic acid breakdown products (nucleotides) from the culture medium for reuse in metabolism. Toxic nucleotides placed in the culture medium will be transported into normal (nonmutated) cells that consequently die. However, loss of the salvage enzyme through genotoxic damage (mutation, chromosome deletions or rearrangements) of the corresponding gene will render the cell resistant to the toxic nucleotide and so it will survive. The surviving mutant cells can be detected by colony formation in tissue culture plates. The two most popular genes for measuring mutation in vitro are those coding for hypoxanthine-guanine phosphoribosyl transferase (hprt) and thymidine kinase (tk). The former is located on the X chromosome in both human and Chinese hamster cells, and loss of activity in this gene can be measured by resistance to the antimetabolite 6-thioguanine. The TK gene is located on chromosome 11 in mouse cells and on chromosome 17 in humans. Loss of activity in this gene can be measured by resistance to the toxic chemical trifluorothymidine (Clive et al. 1987). Three cell lines have been used most extensively for routine in vitro mutation assays. Two are Chinese hamster cell lines (V79 and CHO), and one is a heterozygous mouse line, mouse lymphoma (L5178Y). The Chinese hamster cells have been used extensively over the past 15–25 years; but more recently, TK mutation in the heterozygous mouse lymphoma L5178Y tk+/− cell line has become most popular because of the variety of genetic events detected in this system. In particular, the mouse lymphoma TK assay is able to detect large deletion mutations that tend to be lethal in the HPRT assay and can therefore be used as an alternative to assays detecting chromosomal aberrations (see below). The theoretical basis for the mouse lymphoma assay is shown in Figure 11.3. Two main protocols have been devised for performing the assay, plating the cells in soft agar or a microwell fluctuation test approach. In each case, treatments are performed for short periods (3–6 hours) in the absence and presence of S9 mix, but a longer treatment in the absence of S9 is also usually included to detect those mutagens that need to be present for at least one full cell cycle (e.g., nucleoside analogues). The most recent recommendations on the correct performance of this assay are contained within a number of publications by Moore et al. (2000, 2002, 2003). At least two types of colony are obtained when mutations at the tk locus are selected, large colonies, which grow at the normal rate, and slow-growing small colonies (examples are shown in Figure 11.4). Initial molecular analysis of these colonies indicated that a high percentage of small colony mutants have a wide variety of visible chromosome 11b aberrations, whereas large colonies do not. However, recent chromosome painting analysis of colonies indicates that this initial generalized premise on colony size may be oversimplistic. What is clear is that a wide range of mutations and genetic events can be detected by the L5178Y system, including both point mutations and chromosomal damage.
278
CHAPTER 11
IN VITRO GENOTOX ASSAYS
Figure 11.3. The theoretical basis of the L5178Y TK+/− mouse lymphoma assay.
Large colony mutant
Small colony mutant
Figure 11.4. Examples of small and large mutant colonies in the microwell version of the mouse lymphoma assay. See insert for color representation of this figure.
11.5. IN VITRO TESTS FOR CHROMOSOME DAMAGE IN MAMMALIAN CELLS
279
The p53 status of a number of L5178Y cell lines has been investigated, and it has been found that the cells contain two mutant p53 alleles resulting in a dysfunctional p53 protein (Storer et al. 1997). Because this tumor-suppressor protein is so important in regulating cellular responses to DNA damage, this may account for the sensitivity of these cells to genotoxins. It has been suggested that the p53 status of the cells makes them even more appropriate for use in screens for genotoxic carcinogens, as the development of cancer is often associated with mutant p53 protein. A system that contains a component evaluating a chemical’s ability to induce additional mutations in p53-deficient cells may provide a more appropriate model for the human situation.
11.5. IN VITRO TESTS FOR CHROMOSOME DAMAGE IN MAMMALIAN CELLS An alternative method to measuring mutation induction within mammalian genes involves the examination of mammalian chromosomes microscopically for the presence of visible damage. Cultures of established cells (e.g., Chinese hamster fibroblasts) or primary cells (e.g., human peripheral lymphocytes) can be used. In simplest terms, these tests generally involve exposure of cultured cells to the test material for short periods (3–6 hours) in the presence and absence of a metabolic activation system (e.g., S9 mix), and they also involve a prolonged treatment in the absence of S9. The cells are then harvested at an appropriate time (around 1.5 normal cell cycles after the start of treatment), metaphase spreads are prepared, and gross damage to the chromosomes such as terminal deletions, breaks, and exchanges are recorded. An example of a damaged human lymphocyte metaphase is shown in Figure 11.5. Much of this damage is lethal to the cell during the cell
1
2
Figure 11.5. Example of a human lymphocyte metaphase showing structural chromosome damage (chromatid exchanges arrowed as 1 and 2).
280
CHAPTER 11
IN VITRO GENOTOX ASSAYS
cycle following induction of the damage. However, such changes are used as indicators of the presence of nonlethal more subtle changes (e.g., reciprocal translocations and small deletions) which are difficult to observe microscopically, but which may have important consequences in both somatic and germ cells. A detailed discussion of the test procedures employed is contained within Galloway et al. (1994). Detectable levels of chromosome aberrations are often found only at doses that induce some evidence of cytotoxicity, and consequently most current recommended protocols require that the maximum test concentration should induce >50% reduction in cell number or culture confluency for cell lines, or an inhibition of mitotic index by >50% for lymphocytes. However, there are growing concerns as to the relevance of genotoxic effects that are found only at highly cytotoxic concentrations (Galloway 2000), and this may be reflected by the poor specificity of mammalian cells tests (Kirkland et al. 2005).
11.6.
THE IN VITRO MICRONUCLEUS TEST
Analysis of structural chromosomal aberrations requires considerable training. Normal chromosomes in metaphase preparations can display a variety of appearances, and a full understanding of these different manifestations of normality is necessary before abnormal chromosomes can be satisfactorily identified. The demands of training and the diligent approach needed for thorough analysis means that only 100 cells per replicate culture (usually 2 replicates per test concentration) are scored for aberrations. A more recent technique that will also detect structural chromosomal damage, but requires much less training, is more rapid and allows analysis of larger numbers of cells is the in vitro micronucleus test. Another reason for the increased interest in this test is its ability to detect aneugens. The need for a specific assay to detect genome mutation (i.e., chromosome loss/gain) has been considered by a number of regulatory authorities. Aneuploidy is considered to be a condition in which the chromosome number of a cell or individual deviates from a multiple of the haploid set. The maintenance of karyotype during cell division depends upon the fidelity of chromosome replication and the accurate segregation of chromosomes to daughter cells. In turn, these events depend upon different cell organelles functioning correctly and a number of metabolic activities related specifically to cell division (e.g., synthesis of nuclear spindle, proteins, etc.). Aneuploidy can occur through errors of many types; hence there are numerous cellular targets that can lead to chromosome gain or loss. Briefly, the mechanisms by which aneuploidy can occur fall into several categories, including damage to the mitotic spindle and associated elements, damage to chromosomal substructures, chromosome rearrangements, alterations to cellular physiology, and mechanical disruption. The importance of aneuploidy to adverse human health is well accepted, and the effects of aneuploidy include birth defects, spontaneous abortions, and infertility. Tumor cells frequently have alterations in chromosome number, and several specific aneuploidies have been associated with tumor development, although whether this is the cause or the effect of tumorigenesis is not clear.
11.6. THE IN VITRO MICRONUCLEUS TEST
281
When chromosomes fail to segregate correctly, this process of nondisjunction can lead to the production of both monosomic and trisomic progeny cells. If chromosomes are lost from the dividing nucleus, they produce monosomic progeny without the reciprocal trisomic cell and the expelled chromosomes become membrane-bound and are detected as micronuclei outside the main progeny nuclei. Consequently, chromosome loss can be measured by monitoring micronucleus formation, and the in vitro micronucleus assay using mammalian cells provides such a technique (Fenech and Morley 1985). As mentioned above, this methodology has also been developed and validated as a simpler method for detecting structural chromosome damage because micronuclei may also arise from acentric chromosome fragments (lacking a centromere), which are unable to migrate with the rest of the chromosomes during the anaphase of cell division. Because micronuclei in interphase cells can be assessed much more objectively than chromosomal aberrations in metaphase cells, there is not such rigorous a requirement for training personnel and slides can be scored more quickly. This makes it practical to score thousands instead of hundreds of cells per treatment and thus imparts greater statistical power to the assay. Recently, methodology has been published that allows micronucleus analysis to be conducted using flow cytometry, thus further enhancing the utility of this assay (Avlasevich et al. 2006). The in vitro micronucleus assay may employ cultures of established cell lines, cell strains, or primary cultures, including human and Chinese hamster fibroblasts, mouse lymphoma cells, and human lymphocytes. Guidelines have been published recommending suitable protocols (Kirsch-Volders et al. 2000, 2003). To analyze the induction of micronuclei, it is essential that nuclear division has occurred in both treated and untreated cultures. It is therefore important to provide evidence that cell proliferation has occurred after test chemical exposure. Analysis of the induction of micronuclei in human lymphocytes has indicated that the most convenient stage to score micronuclei in this cell system is the binucleate interphase stage. Such cells have completed one mitotic division after chemical treatment and are thus capable of expressing micronuclei. Treatment of the cells with the inhibitor of actin polymerization cytochalasin B inhibits microfilament assembly and cytokinesis, thus preventing the separation of daughter cells after mitosis and trapping them at the binucleate stage. A schematic for this method is shown in Figure 11.6, and an example of a binucleate cell with a micronucleus is shown in Figure 11.7. The principle of the method is to expose cell cultures to the test substance in both the presence and absence of an in vitro metabolizing system. After exposure, the cultures are grown for a period sufficient to allow chromosome damage or chromosome loss to lead to the formation of micronuclei in interphase cells (usually 1.5–2 normal cell cycles after the start of treatment). Harvested and stained interphase cells are then analyzed microscopically for the presence of micronuclei. If the cytokinesis-block technique is used, micronucleus analysis is restricted to binucleate cells, and at least 1000 lymphocytes per duplicate culture should additionally be classified as mononucleates, binucleates, or multinucleates to estimate the cytokinesis-block proliferation index, which is a measure of cell cycle delay. Micronuclei formed by aneuploidy induction can be distinguished from those produced by clastogenic activity by the presence of centromeric DNA or kinetochore
282
CHAPTER 11
IN VITRO GENOTOX ASSAYS
Figure 11.6. The basis for the in vitro micronucleus assay using cytochalasin B.
Figure 11.7. Example of a binucleate cell with a micronucleus. See insert for color representation of this figure.
proteins in the micronuclei. Fluorescent in situ hybridization (FISH) with pancentromeric DNA probes can be used to detect the former, whereas specific antibodies can be used to detect the presence of kinetochores (Migliore et al. 1996; Schuler et al. 1997). If aneuploidy is the suspected cause of micronucleus induction, further analysis of binucleate cells can be performed with chromosome-specific probes and the mal-segregation of chromosomes between daughter nuclei can be studied. An
11.7. IN VITRO TEST FOR UNSCHEDULED DNA SYNTHESIS IN RAT HEPATOCYTES
283
Figure 11.8. Example of nondisjunction of 2 chromosomes induced by an aneugen in a binucleate cell using fluorescence in situ hybridization (FISH). See insert for color representation of this figure.
example of a cell showing nondisjunction of 2 chromosomes as a result of aneugen treatment is shown in Figure 11.8. Until recently, most regulatory guidelines have focused mainly on tests for gene mutations and structural chromosome damage. However, the validation of the in vitro micronucleus test and the development of an OECD guideline indicates that its use will become much more widespread (OECD 2007). It will continue to be of considerable importance to establish a specific role for chromosome loss in tumor development. The analysis of aneuploidy in interphase cells of solid tumors using FISH will be greatly advantageous in this respect. For cancer risk assessment purposes, results from aneuploidy assays can be considered particularly useful when the mode of action of a chemical is known to result in chromosome loss or nondisjunction.
11.7. IN VITRO TEST FOR UNSCHEDULED DNA SYNTHESIS IN RAT HEPATOCYTES Because the liver is the major organ of xenobiotic metabolism, hepatocytes are an appropriate cell in which to conduct genotoxicity tests. Because they do not divide readily in culture, they are not so useful in tests for chromosomal aberrations or mutations, but they can be used to detect DNA damage or repair where dividing cells are not needed. In studies with fresh hepatocytes, it is not necessary to include S9 mix.
284
CHAPTER 11
IN VITRO GENOTOX ASSAYS
UDS assays quantify the resultant excision repair of DNA following a permanent change such as covalent binding of an activated mutagen or a reactive chemical species generated intracellularly. Cells undergoing such repair synthesize DNA at stages of the cell cycle other than S-phase, where normal replicative (scheduled) DNA synthesis takes place, hence the term “unscheduled DNA synthesis.” This technique is potentially highly sensitive because the whole genome is theoretically a target for chemical reaction. A detailed description of the methodology was given by Madle et al. (1994). Briefly, viable hepatocyte populations are prepared by perfusing the livers of rats with collagenase (Madle et al. 1994). The hepatocytes are treated in vitro for 16–20 hours in the presence of the radiolabeled nucleotide [3H]thymidine. Uptake of the tritium into nuclear or cytoplasmic DNA is identified by autoradiography. Cells undergoing repair are identified by increases in the number of silver grains overlying the nuclei compared to those overlying the cytoplasm. S-phase cells exhibit extremely high numbers of nuclear silver grains and are excluded from analysis.
11.8.
IN VITRO COMET ASSAY
A useful way to measure direct damage to DNA is the single-cell gel electrophoresis assay or “comet” assay. This is a rapid and simple system for measuring alkali labile sites and overt strand breaks in the DNA of mammalian cells (Fairbairn et al. 1995). During electrophoresis, damaged (fragmented) DNA penetrates further than undamaged DNA into the agar gel in which the cells are embedded. The basis for this assay is represented in Figure 11.9. The technique can be applied to virtually any cell culture from which a single-cell suspension can be prepared (McKelvey-Martin et al. 1993). After treatment, the cells are suspended in agar and exposed to strong alkali, which denatures the proteins and permits DNA unfolding. Electrophoresis is then performed, during which time the supercoiled DNA relaxes and fragmented DNA is pulled toward the anode. After electrophoresis, the slides are neutralized and stained with a DNA-specific stain such as propidium iodide or ethidium bromide, when the cell ghosts with damaged DNA are visible as comets of various sizes
Figure 11.9. The theoretical basis for the formation of DNA comets.
11.9. STRENGTHS AND LIMITATIONS
Damaged cells
285
Control cells
Figure 11.10. Visualization of comets using ethidium bromide. See insert for color representation of this figure.
(hence the name), whereas those with undamaged DNA are visible as round images (examples are shown in Figure 11.10). DNA can be determined visually by the categorization of comets into different “classes” of migration or by using an eyepiece micrometer to estimate image or tail length. However, image analysis is recommended with the measurement of parameters such as the percentage of DNA in the tail (percent migrated DNA), tail length, and tail moment (fraction of migrated DNA multiplied by some measure of tail length). Of these, tail moment and/or tail length measurements are the most commonly reported, but there is much to recommend the use of percent DNA in tail, because this gives a clear indication of the appearance of the comets and is linearly related to the DNA break frequency over a wide range of levels of damage (Collins 2004). Cell death is associated with increased levels of DNA strand breaks, and in the comet assay the microscopic image resulting from necrotic or apoptotic cells can be comets with small or nonexistent heads and large diffuse tails, commonly called “clouds” or “hedgehogs.” However, such cells can also be seen after treatment with high concentrations of strong mutagens indicating that these images are not uniquely diagnostic for apoptosis/necrosis (Collins 2004). For this reason, it is recommended that relatively high levels of viability are achieved (e.g., no more than 30% cytotoxicity) at the end of treatment. Further validation and development of this methodology is continuing and recommended guidance on the correct conduct of the comet assay have been published (Hartmann et al. 2003; Tice et al. 2000).
11.9.
STRENGTHS AND LIMITATIONS
Analysis of the ability of in vitro genotoxicity tests to predict carcinogenicity in rodents has shown them to have a high sensitivity (i.e., giving positive results with carcinogens), particularly when combined in batteries of complementary tests
286
CHAPTER 11
IN VITRO GENOTOX ASSAYS
(Kirkland et al. 2005; Tennant et al. 1987). Predictivity rises above 90% for batteries of 2 or 3 tests, and only carcinogens with a known nongenotoxic mode of cancer induction, or those chemicals that are extremely weak carcinogens, are missed. However, apart from the Ames test, the specificity (i.e., ability to give negative results with noncarcinogens) of the in vitro tests is poor. For the mammalian tests (e.g., mouse lymphoma, chromosomal aberration, micronucleus) used alone, the specificity was less than 50%, and when combined in batteries of two or three tests the specificity fell to 33% or 25%, respectively (i.e., the chances of a wrong prediction of carcinogenicity rose to 2/3 or 3/4). Snyder and Green (2001) also reported that from a total of 467 pharmaceuticals examined in the 1999 Physicians Desk Reference and from the open literature, 75% were positive in at least one in vitro assay (Snyder and Green 2001). Experimental culture conditions such as changes in pH and high osmolality are known to cause false-positive results in in vitro mammalian assays. However, other biochemical and physiological stresses such as inhibition of protein synthesis, inhibition of DNA synthesis or repair, inhibition of topoisomerases, overload of metabolism, and so on, can also lead indirectly to DNA damage and genotoxic responses, particularly in mammalian cells. Consequently a positive result in any one in vitro assay does not necessarily mean that the chemical poses a genotoxic/carcinogenic hazard to humans. It is believed that some of these irrelevant positives result from the lack of functional p53 in many cell lines, from artifacts resulting from high levels of cytotoxicity or high concentrations that overload normal metabolism and defense mechanisms. These have been discussed at length in Kirkland et al. (2007). There have therefore been calls for in vitro tests in mammalian cells in particular to be more robust using p53 and DNA repair-proficient cells, with a stable karyotype, avoiding excessive cytotoxicity and test chemical concentrations. Further investigation in relevant in vivo assays is usually required to put any positive in vitro results into perspective. The in vivo tests have advantages in terms of relevant metabolism, and so on, and also allow the influence of detoxification mechanisms to be assessed. However, some understanding of the mode of action leading to the genotoxic response, and whether this is relevant for humans or may have a threshold, is important. Nonetheless, positive results from in vitro tests that are wrong or irrelevant predictors of in vivo mutagenic or carcinogenic hazard will lead to either abandoning the development of certain products or the unnecessary use of animals in follow-up studies. Thus while in vitro genotoxicity tests are very useful, we need new or modified tests for the future that demonstrate improved specificity without compromising sensitivity.
REFERENCES Ames, B. N. (1971). The detection of chemical mutagens with enteric bacteria. In Chemical Mutagens, Principles and Methods for Their Detection, Vol. 1, ed. H. Hollaender. Plenum Press, New York, pp. 267–282. Ames, B. N., Durston, W. E., Yamasaki, E., and Lee, F. D. (1973). Carcinogens are mutagens: A simple test system combining liver homogenates for activation and bacteria for detection. Proc Natl Acad Sci USA 70, 2281–2285.
REFERENCES
287
Ashby, J., and Tennant, R. W. (1988). Chemical structure, Salmonella mutagenicity and extent of carcinogenicity as indicators of genotoxic carcinogenesis among 222 chemicals tested in rodents by the U.S. NCI/NTP. Mutat Res 204, 17–115. Avlasevich, S. L., Bryce, S. M., Cairns, S. E., and Dertinger, S. D. (2006). In vitro micronucleus scoring by flow cytometry: Differential staining of micronuclei versus apoptotic and necrotic chromatin enhances assay reliability. Environ Mol Mutagen 47, 56–66. Clive, D., Caspary, W., Kirby, P. E., Krehl, R., Moore, M., Mayo, J., and Oberly, T. J. (1987). Guide for performing the mouse lymphoma assay for mammalian cell mutagenicity. Mutat Res 189, 143–156. Cole, J., Fox, M., Garner, R. C., McGregor, D. B., and Thacker, J. (1990). Gene mutation assays in cultured mammalian cells. In Basic Mutagenicity Tests: UKEMS Recommended Procedures Kirkland, D. J., ed., Cambridge University Press, Cambridge, pp. 87–114. Collins, A. R. (2004). The comet assay for DNA damage and repair: Principles, applications, and limitations. Mol Biotechnol 26, 249–261. Elliott, B. M., Combes, R. D., Elcombe, C. R., Gatehouse, D. G., Gibson, G. G., Mackay, J. M., and Wolf, R. C. (1992). Alternatives to Aroclor 1254-induced S9 in in vitro genotoxicity assays. Mutagenesis 7, 175–177. Fairbairn, D. W., Olive, P. L., and O’Neill, K. L. (1995). The comet assay: A comprehensive review. Mutat Res 339, 37–59. Fenech, M., and Morley, A. A. (1985). Measurement of micronuclei in lymphocytes. Mutat Res 147, 29–36. Galloway, S. M. (2000). Cytotoxicity and chromosome aberrations in vitro: Experience in industry and the case for an upper limit on toxicity in the aberration assay. Environ Mol Mutagen 35, 191–201. Galloway, S. M., Aardema, M. J., Ishidate, M., Jr., Ivett, J. L., Kirkland, D. J., Morita, T., Mosesso, P., and Sofuni, T. (1994). Report from working group on in vitro tests for chromosomal aberrations. Mutat Res 312, 241–261. Gatehouse, D., Haworth, S., Cebula, T., Gocke, E., Kier, L., Matsushima, T., Melcion, C., Nohmi, T., Ohta, T., Venitt, S., et al. (1994). Recommendations for the performance of bacterial mutation assays. Mutat Res 312, 217–233. Guengerich, F. P., Dannan, G. A., Wright, S. T., Martin, M. V., and Kaminsky, L. S. (1982). Purification and characterization of liver microsomal cytochromes p-450: Electrophoretic, spectral, catalytic, and immunochemical properties and inducibility of eight isozymes isolated from rats treated with phenobarbital or beta-naphthoflavone. Biochemistry 21, 6019–6030. Hartmann, A., Agurell, E., Beevers, C., Brendler-Schwaab, S., Burlinson, B., Clay, P., Collins, A., Smith, A., Speit, G., Thybaud, V., and Tice, R. R. (2003). Recommendations for conducting the in vivo alkaline comet assay. 4th International Comet Assay Workshop. Mutagenesis 18, 45–51. Kirkland, D., Aardema, M., Henderson, L., and Muller, L. (2005). Evaluation of the ability of a battery of three in vitro genotoxicity tests to discriminate rodent carcinogens and noncarcinogens I. Sensitivity, specificity and relative predictivity. Mutat Res 584, 1–256. Kirkland, D., Pfuhler, S., Tweats, D., Aardema, M., Corvi, R., Darroudi, F., Elhajouji, A., Glatt, H., Hastwell, P., Hayashi, M., Kasper, P., Kirchner, S., Lynch, A., Marzin, D., Maurici, D., Meunier, J. R., Muller, L., Nohynek, G., Parry, J., Parry, E., Thybaud, V., Tice, R., van Benthem, J., Vanparys, P., and White, P. (2007). How to reduce false positive results when undertaking in vitro genotoxicity testing and thus avoid unnecessary follow-up animal tests: Report of an ECVAM Workshop. Mutat Res 628, 31–55. Kirsch-Volders, M., Sofuni, T., Aardema, M., Albertini, S., Eastmond, D., Fenech, M., Ishidate, M., Jr., Kirchner, S., Lorge, E., Morita, T., Norppa, H., Surralles, J., Vanhauwaert, A., and Wakata, A. (2003). Report from the in vitro micronucleus assay working group. Mutat Res 540, 153–163. Kirsch-Volders, M., Sofuni, T., Aardema, M., Albertini, S., Eastmond, D., Fenech, M., Ishidate, M., Jr., Lorge, E., Norppa, H., Surralles, J., von der Hude, W., and Wakata, A. (2000). Report from the In Vitro Micronucleus Assay Working Group. Environ Mol Mutagen 35, 167–172. Madle, S., Dean, S. W., Andrae, U., Brambilla, G., Burlinson, B., Doolittle, D. J., Furihata, C., Hertner, T., McQueen, C. A., and Mori, H. (1994). Recommendations for the performance of UDS tests in vitro and in vivo. Mutat Res 312, 263–285.
288
CHAPTER 11
IN VITRO GENOTOX ASSAYS
McKelvey-Martin, V. J., Green, M. H., Schmezer, P., Pool-Zobel, B. L., De Meo, M. P., and Collins, A. (1993). The single cell gel electrophoresis assay (comet assay): A European review. Mutat Res 288, 47–63. Migliore, L., Cocchi, L., and Scarpato, R. (1996). Detection of the centromere in micronuclei by fluorescence in situ hybridization: Its application to the human lymphocyte micronucleus assay after treatment with four suspected aneugens. Mutagenesis 11, 285–290. Moore, M. M., Honma, M., Clements, J., Awogi, T., Bolcsfoldi, G., Cole, J., Gollapudi, B., HarringtonBrock, K., Mitchell, A., Muster, W., Myhr, B., O’Donovan, M., Ouldelhkim, M. C., San, R., Shimada, H., and Stankowski, L. F., Jr. (2000). Mouse lymphoma thymidine kinase locus gene mutation assay: International Workshop on Genotoxicity Test Procedures Workgroup Report. Environ Mol Mutagen 35, 185–190. Moore, M. M., Honma, M., Clements, J., Bolcsfoldi, G., Cifone, M., Delongchamp, R., Fellows, M., Gollapudi, B., Jenkinson, P., Kirby, P., Kirchner, S., Muster, W., Myhr, B., O’Donovan, M., Oliver, J., Omori, T., Ouldelhkim, M. C., Pant, K., Preston, R., Riach, C., San, R., Stankowski, L. F., Jr., Thakur, A., Wakuri, S., and Yoshimura, I. (2003). Mouse lymphoma thymidine kinase gene mutation assay: International Workshop on Genotoxicity Tests Workgroup report—Plymouth, UK 2002. Mutat Res 540, 127–140. Moore, M. M., Honma, M., Clements, J., Harrington-Brock, K., Awogi, T., Bolcsfoldi, G., Cifone, M., Collard, D., Fellows, M., Flanders, K., Gollapudi, B., Jenkinson, P., Kirby, P., Kirchner, S., Kraycer, J., McEnaney, S., Muster, W., Myhr, B., O’Donovan, M., Oliver, J., Ouldelhkim, M. C., Pant, K., Preston, R., Riach, C., San, R., Shimada, H., and Stankowski, L. F., Jr. (2002). Mouse lymphoma thymidine kinase gene mutation assay: Follow-up International Workshop on Genotoxicity Test Procedures, New Orleans, Louisiana, April 2000. Environ Mol Mutagen 40, 292–299. Mortelmans, K. E., and Dousman, L. (1986). Mutagenesis and plasmids. In Chemical Mutagens, Principles and Methods for Their Detection, Vol. 10, de Serres, F. J., ed., Plenum Press, New York, pp. 469–508. OECD (2007). Draft proposal for a new guideline 487: In vitro mammalian cell micronucleus test (MNvit). OECD Guideline for the Testing of Chemicals 487, 1–21. Schuler, M., Rupa, D. S., and Eastmond, D. A. (1997). A critical evaluation of centromeric labeling to distinguish micronuclei induced by chromosomal loss and breakage in vitro. Mutat Res 392, 81–95. Snyder, R. D., and Green, J. W. (2001). A review of the genotoxicity of marketed pharmaceuticals. Mutat Res 488, 151–169. Storer, R. D., Kraynak, A. R., McKelvey, T. W., Elia, M. C., Goodrow, T. L., and DeLuca, J. G. (1997). The mouse lymphoma L5178Y Tk+/− cell line is heterozygous for a codon 170 mutation in the p53 tumor suppressor gene. Mutat Res 373, 157–165. Tennant, R. W., Margolin, B. H., Shelby, M. D., Zeiger, E., Haseman, J. K., Spalding, J., Caspary, W., Resnick, M., Stasiewicz, S., Anderson, B., et al. (1987). Prediction of chemical carcinogenicity in rodents from in vitro genetic toxicity assays. Science 236, 933–941. Tice, R. R., Agurell, E., Anderson, D., Burlinson, B., Hartmann, A., Kobayashi, H., Miyamae, Y., Rojas, E., Ryu, J. C., and Sasaki, Y. F. (2000). Single cell gel/comet assay: Guidelines for in vitro and in vivo genetic toxicology testing. Environ Mol Mutagen 35, 206–221. Tweats, D., and Gatehouse, D. (1999). Mutagenicity. In General and Applied Toxicology, Vol. 2, Ballantyne, B., ed., Macmillan Press, London, pp. 1017–1078.
CH A P TE R
12
IN VIVO GENOTOXICITY ASSAYS Véronique Thybaud
12.1.
INTRODUCTION
12.1.1. Endpoints Used for In Vivo Genetic Toxicology Assays Genetic changes affecting cell-cycle control or genome integrity, such as oncogene activation and inactivation of DNA repair or tumor-suppressor genes, are key events in the multistep process of carcinogenesis and are also associated with other human diseases and with aging (Hanahan and Weinberg 2000). In addition, germ-line mutations can lead to inheritable diseases. A variety of in vivo genotoxicity assays (Figure 12.1) have been developed over the last 35 years to detect genotoxicity, and no single assay is currently able to detect all genotoxic agents (MacGregor et al. 2000; Müller et al. 2003). In vivo genotoxicity assays can use either (a) somatic cells for the prediction of cancer and aging or (b) germ cells for inheritable diseases. When a compound is found to be genotoxic in somatic cells and is able to reach germ cells, it is reasonable to conclude that it may also pose a mutagenic hazard to germ cells and thus a risk to future generations (Waters et al. 1993, 1994; Shelby 1996). Genotoxic evaluation on germ cells (OECD 1984, 1986a, 1997c; Bishop and Kodell 1980; Russell and Matter 1980; Russo 2000) can use either germ cells themselves (e.g., spermatogonial cells) or the offspring (or potential offspring) of exposed animals (dominant lethal test and mouse heritable translocation assay). This chapter will mainly focus on (a) the evaluation of genetic toxicity in somatic cells and (b) its use in cancer risk assessment. The ability of chemicals to damage DNA, either directly or after metabolic activation, can be evaluated by detecting covalent DNA binding (for example, by measuring DNA adduct formation) or by detecting single or double DNA strandbreaks, crosslinking, and apurinic sites with methods such as the comet assay. Primary DNA damage (DNA adducts and strand-breaks) is considered a biomarker of exposure because it provides an integrated measurement of compound absorption, metabolic activation, and delivery to target organ macromolecules (Swenberg et al. 2008). Primary DNA damage can be repaired and is not therefore systematically
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
289
290
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
Direct Carcinogen
Indirect Carcinogen Metabolic activation
Electrophilic compound/metabolite interactions with DNA Markers of exposure
DNA PRIMARY DAMAGE
Apoptosis Cell death
No DNA repair, errors in repair or proliferation before DNA repair
DNA repair
DNA primary damage assays: • DNA strand breaks (Comet Assay) • DNA adducts Indicator assays: • Unscheduled DNA synthesis • Sister Chromatid Exchanges
Markers of exposure Gene mutation assays: • Endogenous genes • Transgenes Chromosome damage assays: • Chromosome aberrations • Micronuclei
Figure 12.1.
MUTATIONS
Back to normal
TUMOR
End-points detected with the available in vivo genotoxicity assays.
transmitted to daughter cells (Norbury and Hickson 2001; Baute and Depicker 2008; Fousteri and Mullenders 2008; Hegde et al. 2008). Thus, primary DNA damage can also be assessed in indicator assays, such as those detecting increased DNA excision repair or DNA recombination activity [e.g., unscheduled DNA synthesis (UDS) and sister chromatid exchange (SCE)]. If DNA damage is not repaired and/or the cell bearing DNA damage does not undergo apoptosis or death through other mechanisms, then the damage may lead to replication errors and to irreversible changes in DNA structure. These changes are then transmitted to the progeny of the mutated cell and can potentially lead to inheritable genetic changes. Gene mutations and chromosome damage, because they are stable and transmissible genetic changes, are considered as biomarkers of effect (Swenberg et al. 2008). Gene mutation assays measure mutagenicity—that is, the ability of a product to induce point mutations in single genes or blocks of genes resulting from basepair substitutions, frameshifts, and small deletions or insertions. These methods generally rely on reporter genes, which may be endogeneous (e.g., hprt, tk, aprt, Dlb-1, and Pig-a) or transgenes (e.g., lacZ, lacI, and gpt), based on the assumption that a product able to induce mutations in a reporter gene can also provoke mutations in genes involved in the initiation and progression of cancer, such as oncogenes (e.g., ras) and tumor suppressor genes (e.g., p53) (Nestmann et al. 1996; Hemminki et al. 2000). Clastogenicity—the ability to induce structural chromosome damage— is measured by detecting gross chromosome abnormalities (e.g., chromatid breaks, and chromosome rearrangements) with standard cytogenetic methods. Such changes are rarely compatible with cell viability or with transmission to daughter cells. It is assumed that if such abnormalities are detected, then more discrete stable
12.1. INTRODUCTION
291
rearrangements (translocations or small deletions) are also generated (Obe et al. 2002). Changes in chromosome numbers (aneuploidy) can also be associated with cancer and other human diseases (Chi and Jeang 2007). The capacity of the different assays to detect relevant genetic changes, and especially those associated with a risk of cancer, is controversial. Gene mutations and chromosome damage are considered more relevant than primary DNA damage, and it is generally accepted that only gene mutation and chromosome damage assays can establish a genotoxic mode of action (Kasper et al. 2007).
12.1.2. Contribution of In Vivo Genetic Toxicology Assays to Risk Assessment As initially recommended in the 1970s and 1980s (MacGregor et al. 2000), in vivo genotoxicity assays are generally combined with two in vitro genotoxicity tests (a bacterial gene mutation assay and a chromosome damage assay using mammalian cells)(Cimino 2006)(for discussion, see Chapter 11). This is because some genotoxic carcinogens provoke genetic damage in animals, of a nature that cannot readily be detected in vitro (Tweats et al. 2007). In vivo genetic toxicology assays are part of the standard battery of regulatory tests required for the development and registration of products like pharmaceuticals, food additives, and pesticides [for review see Cimino (2006) and Chapter 10]. For other products, such as cosmetics and chemicals, in vivo genetic toxicology assays do not necessarily belong to the minimal battery of regulatory tests, at least in Europe. For ethical reasons, animal testing of the latter products is mainly recommended (1) if in vitro genetic toxicity tests are positive, or (2) in case of “high” or “moderate and sustained” human exposure, or (3) when large amounts of the product are to be released on the market and, potentially, in the environment. In this regulatory context, in vivo genetic toxicity assays contribute to genotoxic hazard identification and to human risk assessment, by predicting carcinogenic activity and inheritable genetic changes. They are also crucial for mechanistic interpretation of 2-year bioassay findings (Kasper et al. 2007). One or more in vivo genetic toxicology assays may be necessary when positive results are obtained in vitro, in order to increase the weight of evidence and to better evaluate the human risk. A major advantage of in vivo genetic toxicology assays over in vitro tests is that they take into account not only intrinsic genotoxic potential, but also toxicokinetic parameters such as bioavailability, absorption, tissue distribution, metabolism (activation, detoxification, and excretion), and other factors that only exist in vivo. Individually, they are generally considered less sensitive (more false-negatives) but more specific (fewer false-positives) than in vitro assays. They are also thought to be more relevant to human exposure, being less prone to experimental artifacts, confounding factors, and irrelevant results. Consequently, a positive result in an in vivo genetic toxicology assay represents strong evidence for genotoxic carcinogenicity. Moreover, in vivo genetic toxicology assays are helpful for understanding and interpreting the results of carcinogenicity studies. In the case of genotoxic carcinogens, gene mutations and chromosome damage are generally necessary but not sufficient for carcinogenesis, because additional key events such
292
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
as cell proliferation are required for tumor formation. In other words, most genotoxic compounds are carcinogens, especially those showing genotoxic activity in vivo, but not all carcinogens exhibit genotoxic activity: Some have other mechanisms of action (e.g., hormone imbalance, enzymatic induction, cell proliferation, and inhibition of apoptosis). Importantly, genetic changes can also secondarily occur after exposure to nongenotoxic carcinogens as a result of genetic instability (Moore et al. 2008). It is generally agreed that a no-effect dose level can be determined for nongenotoxic carcinogens that act via well-understood thresholded mechanisms (KirschVolders et al. 2003; Müller and Kasper 2000; Pratt and Barron 2003). This notion is not yet fully accepted for genotoxic carcinogens, despite some evidence of thresholds in metabolic activation, repair, and other key mechanisms involved in the formation of gene mutations and chromosome damage, along with data recently reported for ethyl methanesulfonate (Gocke et al. 2009; Gocke and Muller 2009). Therefore, no dose below which no tumorigenic effect would occur can usually be determined for genotoxic carcinogens, unless mechanistic studies demonstrate that (a) the primary target is not DNA itself but other cell components such as proteins and (b) DNA damage occurs secondarily. For example, effects on components of the mitotic apparatus are responsible for chromosome gain or loss (aneuploidy) as a result of mechanisms such as improper attachment of chomosomes to the mitotic spindle, failed cytokinesis, and an abnormal number of mitotic spindle poles (Chi and Jeang 2007). In this case the primary target is generally not DNA but instead multiprotein complexes involved in chromosome segregation (Bharadwaj and Yu 2004; Chi and Jeang 2007). Thus, a no-effect level can be determined and a safety margin can be estimated compared with actual human exposure. Consequently, understanding the mode of action is essential for risk assessment (Dearfield and Moore 2005; Kasper et al. 2007). Numerous in vivo genotoxicity assays have been developed for research purposes. The present chapter does not aim to provide an exhaustive list of all models described in the literature. Rather, it focuses on the main assays using somatic cells, which may be part of the standard battery of genetic toxicology tests or be requested as part of a weight-of-evidence approach, mechanistic investigations, or more accurate human risk assessment. General considerations applicable to all tests are presented first. Then, for each test system, the principles, the protocol design, the advantages and limitations, data interpretation, and regulatory status are discussed.
12.2. PARAMETERS AND CRITERIA FOR VALID IN VIVO GENOTOXICITY ASSAYS AND IMPLICATIONS FOR EXPERIMENTAL DESIGN The general criteria for the selection of parameters used in in vivo genotoxicity assays are listed below and are indicated for each type of assay in Table 12.1. Selection of the Top Dose. In the absence of toxicity, the highest dose tested is generally 2000 mg/kg/day for treatments up to 2 weeks and 1000 mg/kg/day for
12.2. PARAMETERS AND CRITERIA
293
treatments longer than 2 weeks. In the absence of toxicity and genotoxicity, effective tissue exposure should generally be documented. For the most commonly used tissues (e.g., bone marrow and liver), plasma levels are used. No specific measurements are therefore needed to demonstrate tissue exposure. For other tissues, such as site-ofcontact tissues, specific toxicokinetic data may be required. If toxicity is noted, the highest dose should be the maximal tolerated dose, usually defined as “the dose producing signs of toxicity such as higher dose levels, based on the same dosing regimen, would be expected to produce lethality” (Mackay 1995). Observation of cytotoxicity in a given tissue (e.g., impact on erythopoeisis in bone marrow, hepatotoxicity, etc.) can contribute to limiting the top dose. For example, at least a 50% reduction in the mitotic index in the bone marrow chromosome aberration test is required. When used to interpret results obtained in carcinogenicity studies, the doses selected for in vivo genotoxicity assays can be those used in two-year bioassays. Number of Doses. A maximum of three doses are generally required for cytogenetic assays in bone marrow and peripheral blood. In the case of complex and labor-intensive assays (e.g., gene mutation assays in transgenic animals, liver UDS test), two doses might be considered sufficient. In the absence of toxicity, a “limit test” using a single dose of 1000 or 2000 mg/kg/day is considered acceptable for some assays (see OECD guidelines). Duration of Exposure. In most cases, acute treatment with one or two administrations of high doses is recommended. For chronic administration, there is a balance between the duration of treatment and the dose levels, because the doses evaluated during long-term treatment are frequently lower because of more pronounced toxicity. When DNA lesions, mutations, and damaged cells do not accumulate over time (because of efficient damage repair or elimination of damaged cells by apoptosis or cell turnover), longer treatment might result in lower sensitivity. When DNA lesions accumulate over time and damaged cells are not eliminated (e.g., transgenic gene mutation assays in slowly proliferating tissues), multiple administrations may improve assay sensitivity. In vivo genetic toxicity endpoints can be evaluated after multiple administrations in other toxicology studies—for example, 14- or 28-day general toxicity studies. After more than 28 days (e.g., 3-month studies), confounding effects such as oxidative stress, enzymatic induction, and preneoplasic lesions might be expected, leading to the measurement of secondary rather than primary effects on DNA. Genotoxicity findings obtained after long-term exposure should therefore be interpreted with caution. Sampling Time. The optimal sampling time for a given assay in a given tissue should correspond to the likely maximal effect. The time needed for the compound to reach the tissue of interest, for metabolic activation, for the formation and accumulation of DNA lesions and their fixation into mutations, as well as for the possible elimination of damaged cells and cellular turnover, should be taken into account. Sampling times vary from one assay to another: 3–24 hours after treatment in the UDS and comet assays, 24–48 hours for the detection of chromosome damage, and up to several weeks for the measurement of gene mutations.
294
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
TABLE 12.1. Summary Table of Principle, Study Design, Parameters and Criteria Used in the In Vivo Genotoxicity Assays
In Vivo Assays
Micronucleus Assay in Bone Marrow and Peripheral Blood
Bone Marrow Chromosome Aberration Assay
Comet Assay
Endpoints
Structural and numerical chromosome damage. Request cell proliferation.
Chromosome damage (structural and some indication of potential numerical damage, e.g., polyploidy). Request cell proliferation.
DNA primary damage: DNA single and double strand-breaks, alkali-labile sites, incomplete excision repair sites, DNA-DNA and DNA-protein crosslinks. Does not request cell proliferation.
Parameters measured
Chromosome damage: Incidence of micronucleated polychromatic erythrocytes (PCE) or normochromatic erythrocytes (NCE).
Chromosome damage: Incidence of cells with structural chromosome aberrations including and excluding gaps (including number and type of aberrations), as well as polyploid cells and cells with endoreduplicated chromosomes.
DNA migration: In case of image analysis: tail length, % tail DNA, tail moment. In case of manual scoring: tail length, or incidence of cells with and without DNA migration, from undamaged to highly damaged (∼4 categories).
Regulatory acceptance
Protocol described in OECD 474 (OECD 1997c).
Protocol described in OECD 475 (OECD 1997b).
No OECD guideline available. International validation ongoing.
Part of standard battery and/or of follow-up testing
Standard battery, or first follow-up when in vivo test are not part of the minimal battery of test.
Standard battery. Equally acceptable alternative to the in vivo bone marrow micronucleus test.
Follow-up testing.
12.2. PARAMETERS AND CRITERIA
DNA Adducts
Unscheduled DNA Synthesis Assay in Liver Cells
Sister-Chromatid Exchange Assay
Repair of DNA lesions and unscheduled DNA synthesis in response to DNA primary damage. Does not request cell proliferation.
Repair of DNA lesions by homologous recombination (interchange between sister chromatids). Request cell proliferation.
Net nuclear grain count (nuclear grain count minus cytoplasm grain count). Proportion of cells “in repair.”
Incidence of sister chromatid exchange (SCE) per cell.
No OECD guideline available.
Protocol described in OECD 486 (OECD 1997d).
Follow-up testing.
Follow-up testing.
No OECD guideline available for in vivo SCE, OECD guideline only available for in vitro SCE (OECD 1997a). Follow-up testing.
DNA primary damage: DNA adducts (alkyl and bulky adducts) are nucleotide bases covalently modified by reactive electrophilic chemical intermediates or free radicals. Does not request cell proliferation. Number of adducts per normal nucleotides. Chemical structure of adducts, only for some methods. For more details see Table 12.2 for the in vivo measurement of DNA adducts.
295
Gene Mutation Assays Gene mutations: Point mutations such as base pair substitutions, frame shifts, small deletions or insertions. Request DNA replication and cell proliferation for fixation of DNA primary damage into stable gene mutations. Mutant frequency (e.g., number of mutants per million cells). Mutation spectrum for some assays (e.g., transgenic models, hprt). For more details see Table 12.3 for In vivo gene mutation assays in transgenic models, and Table 12.4 for in vivo gene mutation assays in endogenous genes of somatic cells. No OECD guideline available. OECD detailed review paper recently released. Follow-up testing.
(Continued)
296
CHAPTER 12
TABLE 12.1.
IN VIVO GENOTOXICITY ASSAYS
(Continued)
In Vivo Assays
Micronucleus Assay in Bone Marrow and Peripheral Blood
Bone Marrow Chromosome Aberration Assay
Comet Assay
Practicability
Widely used. Easy cell sampling and preparation. Easy to score and possible automation of micronucleated cell scoring (image analysis or flow cytometry).
Less widely used than the bone marrow micronucleus test. Easy cell sampling and preparation. Time-consuming and tedious scoring of chromosome aberrations.
Relatively widely used. Still under validation. Easy cell/nuclei sampling and preparation. Easy to score and possible automation of comet scoring.
Integration in general toxicity
Possible if criteria for top dose selection acceptable.
Possible if criteria for top dose selection acceptable. Administration of mitotic inhibitor shortly before tissue sampling might impact the measurement of other parameters.
Sampling times not compatible with general toxicity studies. Would require extra treatment 2–6 hours before sampling.
Selection of top dose
Maximal tolerated dose, or a dose inducing bone marrow cytotoxicity, or in the absence of toxicity 1000 mg/kg if more than 14 days of treatment; and 2000 mg/kg/day if less than 14 days of treatment.
Maximal tolerated dose, or a dose inducing bone marrow cytotoxicity (at least 50% reduction in mitotic index), or in the absence of toxicity 1000 mg/kg if more than 14 days of treatment; and 2000 mg/kg/day if less than 14 days of treatment.
Maximal tolerated dose, or a dose inducing cytotoxicity in the selected tissue, or in the absence of toxicity 1000 mg/ kg if more than 14 days of treatment; and 2000 mg/kg/ day if less than 14 days of treatment.
12.2. PARAMETERS AND CRITERIA
DNA Adducts Rarely used. Different methods available, some being more straightforward than the others (for more details see Table 12.2).
Possible for methods that do not request animal treatment with radio-labeled compounds.
Maximal tolerated dose, or maximal dose used in 2-year bioassay, if study conducted as a follow-up of carcinogenicity studies.
Unscheduled DNA Synthesis Assay in Liver Cells Rarely used. Labor-intensive. Treatment and sampling times not compatible with normal working day. Possible automation of grain scoring. Require handling of radio-labeled thymidine, and appropriate authorization. Sampling times not compatible with general toxicity studies. Extra treatment 2–6 hours before sampling, and liver perfusion required. Maximal tolerated dose, or a dose inducing liver cytotoxicity (e.g., pyknotic nuclei) or 2000 mg/kg/day.
Sister-Chromatid Exchange Assay
297
Gene Mutation Assays
Rarely used. Easy cell sampling and preparation, as for chromosome aberrations. Relatively easy to score.
Rarely used. Generally labour intensive. Different methods available, some being more straightforward than the others (for more details see Table 12.3 and 12.4).
Possible if criteria for top dose selection acceptable. Administration of mitotic inhibitor and thymidine analogue shortly before tissue sampling might impact the measurement of other parameters. Maximal tolerated dose, or a dose inducing cytotoxicity in the selected tissue.
Possible for a few endogenous genes. Most often specific strain and/or study design requested.
Maximal tolerated dose, or maximal dose used in 2-year bioassay, if study conducted as a follow-up of carcinogenicity studies.
(Continued)
298
CHAPTER 12
TABLE 12.1.
IN VIVO GENOTOXICITY ASSAYS
(Continued)
In Vivo Assays
Micronucleus Assay in Bone Marrow and Peripheral Blood
Bone Marrow Chromosome Aberration Assay
Comet Assay
Number of doses
3 (only 1 acceptable, in absence of toxicity, i.e., limit test using 1000 or 2000 mg/kg).
3 (only 1 acceptable, in absence of toxicity, i.e., limit test using 1000 or 2000 mg/kg).
2 (limit test not acceptable).
Duration of exposure
Single or multiple treatments.
Single or multiple treatments.
One or two administration. Multiple administrations under evaluation.
Sampling time
In bone marrow, generally two samplings after a single administration (24 and 48 hours after treatment) or one sampling, 24 hours after the last treatment in case of multiple administrations. Sampling should not be before 24 hours and after 48 hours. In peripheral blood, generally 48-hour sampling time Sampling time should not be before 36 and after 72 hours.
One sampling time, 12–18 hours (i.e., 1.5 cell cycle) after a single or last administration. One additional sampling 24 hours later is optional.
Two sampling times 2–6 and 16–26 hours after treatment. Multiple administrations: 2–6 hours after the last treatment, under evaluation.
12.2. PARAMETERS AND CRITERIA
DNA Adducts No clear recommendations. As dose–response is generally linear at low doses, an evidence of dose response is an important confirmation of a positive response. Single or multiple administrations.
No clear recommendations. Should take place before repair and adduct removal after single administration and at steady state after repeated administrations (e.g., 10 days to 1–2 months).
Unscheduled DNA Synthesis Assay in Liver Cells
Sister-Chromatid Exchange Assay
299
Gene Mutation Assays
2 (only 1 acceptable, in absence of toxicity, i.e., limit test using 1000 or 2000 mg/kg).
Generally three doses.
No clear recommendations. Two to three for transgenic models.
Single administration.
Single or multiple administrations.
Two sampling times: 12–16, and 2–4 hours after single administration.
Depends on the cell turn over in the evaluated tissue, generally second cell cycle after the last treatment.
Depends on the model and cell turn over in the evaluated tissue, e.g., 28 days recommended for any tissue in transgenic models. Depends on the model and cell turn over in the evaluated tissue, e.g., for transgenic models, 3 and 28 days after the last administration are recommended.
(Continued)
300
CHAPTER 12
TABLE 12.1.
IN VIVO GENOTOXICITY ASSAYS
(Continued)
In Vivo Assays
Micronucleus Assay in Bone Marrow and Peripheral Blood
Bone Marrow Chromosome Aberration Assay
Comet Assay
Species
Mainly rodents, potentially applicable to nonrodent species, including human.
Mainly rodents, potentially applicable to nonrodent species, including human.
Applicable to rodent and nonrodent species, including human.
Genders
Males are sufficient if no gender difference anticipated.
Males are sufficient if no gender difference anticipated.
One gender is sufficient if no gender difference anticipated.
Tissue(s)
Mainly bone marrow and peripheral blood. Other tissues described in the literature. 5
Mainly bone marrow. Other tissues described in the literature.
Any tissue/organ from which cell/ nuclei can be properly isolated.
5
4–5
2000 for chromosome damage. 200 and 1000 for cytotoxicity, in bone marrow and peripheral blodd, respectively.
100 for chromosome damage, 1000 for cytotoxicity
100–150 cells, preferably 150 (depends on the number of animals per group).
Number of animal per group
Number of cells per animal
12.2. PARAMETERS AND CRITERIA
DNA Adducts
Unscheduled DNA Synthesis Assay in Liver Cells
Sister-Chromatid Exchange Assay
Applicable to rodent and nonrodent species, including human.
Mainly rodents, potentially applicable to nonrodent species.
Mainly rodents, potentially applicable to nonrodent species, including human.
No clear recommendations. One gender should be sufficient if no gender difference anticipated. Any tissue/organ from which DNA can be properly isolated.
Males are sufficient if no gender difference anticipated.
No clear recommendations. One gender should be sufficient if no gender difference anticipated.
Mainly liver cells, but other tissues described in the literature.
Any dividing tissue/ organ from which cell suspensions can be properly isolated.
No clear recommendations, depends on the statistical power of the method used for the detection.
3
No clear recommendations, depends on the statistical power of the method used for the detection. Generally at least three.
Not appropriate.
100 liver cells.
Generally 25 to 50.
301
Gene Mutation Assays Mainly rodents, especially for transgenic models. Endogenous genes potentially applicable to nonrodent species, including human. No clear recommendations. One gender should be sufficient if no gender difference anticipated. Single or limited number of tissues (most of endogenous genes). Any tissue (transgenic models). No clear recommendations, depends on the statistical power of the method used for the detection. From about 5 for most models to ∼50 dams and 100’s offsprings for spot tests. Depends on the models. Not appropriate when DNA is extracted.
302
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
Species. In vivo genotoxicity assays mainly use rodents (rats and mice), because they are small (standard animal facilities and low amount of test article needed) and because abundant toxicologic and metabolic data are available on these species. Most assays are also theoretically applicable to nonrodent animals, but limited data have been reported in the literature. Most of the assays are applicable to wild-type and readily available animal strains. It should nevertheless be noted that a few models require specific strains (e.g., some gene mutation assays). For follow-up testing, the choice of species should take previous data into consideration (e.g., organ-specific generation of DNA-reactive metabolites, tumor findings, etc.). Selection of Gender. Unless clear sex differences (in e.g., metabolism, toxicity, pharmacological activity, two-year bioassay findings, gender specificity) are anticipated, only one gender needs to be tested. Males are generally considered most sensitive for genotoxicity studies. Selection of Tissues. The in vivo genotoxicity assay required for the standard battery is generally a chromosome damage test performed with erythrocytes from bone marrow or peripheral blood. Other in vivo models are mostly used as complementary or follow-up tests. They can also be used as first-line tests if considered appropriate for the compound, in terms of its known properties, metabolism, exposure, target organ, and endpoints. For example, when the compound is known to be chemically unstable, contact tissues might be preferred (e.g., skin for dermal application, lung for inhalation, and gastrointestinal tract for ingested compounds). Similarly, if the compound is metabolized into toxic species, the liver might be preferred. Depending on the assay limitations, genotoxic endpoints can be evaluated either in a restricted number of tissues (e.g., micronucleus and UDS assays) or in almost any tissue (e.g., comet assay, transgenic gene mutation models). Moreover, some assays are dependent on cell proliferation status: The UDS and comet assays can be conducted with nonproliferating cells such as liver cells soon after exposure, while cell proliferation is needed for chromosome damage and gene mutation tests. Depending on the tissues of interest, in vivo assays can be used to detect primary DNA damage, gene mutations, and/or chromosome damage. When in vivo assays are conducted as a follow-up to positive 2-year bioassays, the selected tissues are preferably those in which the tumors arose. In other cases, genotoxicity is evaluated in surrogate tissues. Number of Animals. For statistical reasons, about five to ten animals per dose group are generally used, except in the case of labor-intensive tests where a minimum of three (e.g., for the UDS test in liver cells) or four animals is recommended. A much larger number of animals must be used in some assays (e.g., up to 50 dams treated and several hundred F1 animals examined in mouse spot tests). For ethical reasons, such assays are seldom used. Animals should be evaluated individually,
12.3. IN VIVO GENOTOXICITY ASSAYS REQUIRED IN THE STANDARD BATTERY OF TESTS
303
and pooled samples should be avoided, if possible. Both individual and mean group values are usually reported. Negative and Positive Control Animals. Negative control animals should only receive the vehicle. If an unknown “exotic” vehicle is used, an absolute negative control group of animals receiving no treatment might be necessary. A positive control group of animals treated with a well-known genotoxic carcinogen, preferably requiring metabolic activation to express its genotoxic activity, is generally used. While positive control groups are mandatory for new assays and for laboratories with limited experience, it is increasingly recommended, for ethical reasons, to generate positive control samples every 6–12 months (for example) and to use them for different studies (e.g., DNA samples for DNA adducts, slides for chromosome damage, comet, and UDS assays). A study is considered valid if the results obtained with positive and negative controls are consistent with the laboratory’s historical data and with the literature. Statistical analysis is usually applied to compare treated and negative control groups. Both pairwise and linear trend tests can be used. Because of the low background and Poisson distribution, data transformation (e.g., log transformation) is sometimes needed before using tests applicable to normally distributed data. Otherwise, nonparametric analyses should be preferred. The 3Rs. The three Rs—Reduce, Refine, and possibly Replace the use of animals—are increasingly being taken into consideration before initiating in vivo studies, including genotoxicity assays. It is recommended (1) to evaluate only one gender unless gender differences are anticipated, (2) not to include positive control groups in all studies (see above), (3) to avoid assays requiring large numbers of animals when alternatives exist, (4) to evaluate multiple genotoxicity endpoints in a single animal whenever possible, and (5) to integrate genotoxicity studies with other toxicology studies (organ toxicity, etc.). In specific cases, it may also be advisable to only rely on data obtained from in vitro assays.
12.3. IN VIVO GENOTOXICITY ASSAYS REQUIRED IN THE STANDARD BATTERY OF TESTS Given that the genotoxicity of some compounds can only be detected in vivo, in vivo genotoxicity assays—generally those able to detect chromosome damage in bone marrow or peripheral blood—are often recommended in the standard battery as a complement to in vitro genotoxicity tests (Brambilla and Martelli 2004; Cimino 2006). They either directly quantify and analyze different types of chromosome aberrations in metaphase cells (chromatid and chromosome deletions and exchanges) or indirectly measure the induction of chromosome damage by scoring micronuclei resulting from chromosome breaks, chromosome rearragements, and chromosome lagging (Mateuca et al. 2006). The two methods are considered equally acceptable and interchangeable (Shelby and Witt 1995).
304
CHAPTER 12
12.3.1.
IN VIVO GENOTOXICITY ASSAYS
Mammalian Erythrocyte Micronucleus Test
Purpose. The purpose of the in vivo micronucleus test is to evaluate the potential of the test substance to cause chromosomal damage (clastogenicity) or damage to the mitotic apparatus (aneugenicity) by the analysis of micronuclei in erythrocytes sampled in bone marrow and/or in peripheral blood of experimental animals (usually rodents) (Schmid 1975; Heddle 1973; Heddle and Salamone 1981). Regulatory Acceptance. The rodent erythrocyte micronucleus assay is relatively easy to conduct and is considered able to detect the vast majority of clastogens and rodent carcinogens (when combined with the Ames test) and all human genotoxic carcinogens (Shelby and Zeiger 1990; Rosenkranz and Cunningham 2000). This test is well-validated and widely accepted by regulatory agencies as part of the standard battery of genetic toxicity assays, in addition to the two in vitro assays (Brambilla and Martelli 2004; Cimino 2006). Because of the mechanism of micronuclei formation, the micronucleus test does not in principle detect gene mutations. It should therefore be considered complementary to in vitro gene mutation assays, and not as a follow-up assay to confirm in vitro gene mutations. The experimental conditions and data interpretation are described in OECD guideline 474. It is the most commonly used in vivo assay. Many laboratories are highly experienced, and a large database has been generated for comparison with carcinogenicity and other effects. Principle. During the anaphase of mitosis, acentric chromosome fragments and/ or unseparated chromosomes lag and fail to become incorporated into daughter cell nuclei. After telophase, most of these fragments and/or lagging chromosomes are not included in the nuclei of the daughter cells, but condense to form one or several micronuclei (smooth-boundaried bodies that stain strongly and specifically for chromatin, one-fifth to one-twentieth the size of the main nucleus, also named Howell– Jolly bodies in hematology). During erythropoiesis, when the erythroblast develops into an immature or polychromatic erythrocyte (PCE), the main nucleus is expelled while micronuclei are retained in the cells, facilitating their detection (Heddle et al. 1991). Typically, micronucleus induction is measured in bone marrow PCEs after acute treatment. After repeated administration, it is also advisable to evaluate the induction of micronuclei in mature or normochromatic erythrocytes (NCEs). As erythrocytes persist for about one month in peripheral blood (named reticulocytes), the measurement of micronucleated reticulocytes in peripheral blood is considered equally acceptable in any species, provided that the spleen does not remove the micronucleated erythrocytes from blood and that both aneugens and clastogens are efficiently detected (see more in “assays limitations and confounding factors” section). Micronuclei are generally analyzed in the youngest (i.e., immature) reticulocytes in peripheral blood. Because bone marrow is a relatively well-perfused tissue, its exposure to systemically distributed compounds is generally adequate and can be extrapolated from the plasma concentration. Moreover, the high rate of cell proliferation during erythropoeisis in bone marrow facilitates the formation and detection of micronuclei.
12.3. IN VIVO GENOTOXICITY ASSAYS REQUIRED IN THE STANDARD BATTERY OF TESTS
305
Study Design. The sampling time should take into account the interval between the last mitosis in erythroblasts and the formation of polychromatic erythrocytes (i.e., about 6–8 hours) and the lifespan of polychromatic erythrocytes (i.e., 18–24 hours) (Mavournin et al. 1990). Following a single administration, the incidence of micronucleated PCEs is generally scored at 24 and 48 hours after the administration (not earlier than 24 hours or later than 48 hours) in bone marrow and at 36 and 72 hours in peripheral blood (not earlier than 36 hours or later than 72 hours). After multiple administrations (2 or more), the bone marrow should be sampled at least once between 18 and 24 hours, and peripheral blood should be sampled between 36 and 48 hours after the last administration. In the acute version of the bone marrow micronucleus assay, the bone marrow is generally sampled 24 hours after two administrations 24 hours apart in order to save animals (only one group of animals per dose) without impacting the sensitivity (Salamone et al. 1980; Ashby et al. 1985; CSGMT 1990). The incidence of micronucleated PCEs in bone marrow and in peripheral blood (also termed reticulocytes) should be evaluated among at least 2000 erythrocytes per animal, in order to take into account the low spontaneous rate of micronucleated erythrocytes (0–3/4 micronucleated PCEs per thousand PCEs) and to ensure adequate statistical power (Adler 1984; Hayashi et al. 1994). After continuous treatment for 4 weeks or more, micronuclei should also be analyzed in bone marrow NCEs among at least 2000 erythrocytes per animal. Data suggest that the in vivo micronucleus assay can be integrated into a 28-day toxicological study. For the detection of a majority genotoxic compounds, the incidence of micronucleated erythrocytes should be determined both 4 and 28 days after the beginning of treatment in the peripheral blood and at 28-day necropsy time for bone marrow (Hamada et al. 2001). In addition, a reduction in the ratio of PCEs to NCEs is usually considered to indicate inhibition of erythroblast proliferation or maturation, or destruction of nucleated cells. The cytotoxic effect on bone marrow is therefore measured as PCE/ (PCE + NCE) by counting a total of at least 200 erythrocytes in bone marrow and 1000 erythrocytes in peripheral blood per animal. The distinction between polychromatic and normochromatic erythrocytes is based on the presence of RNA in PCEs, and it is visualized under the microscope after differential labeling with a nonfluorescent stain such as Giemsa (blue for PCEs and pink for NCEs), or a fluorescent stain such as acridine orange (e.g., orange fluorescence in PCEs and no staining in NCEs) (Krishna and Hayashi 2000). For reticulocytes, acridine orange supravital staining is used to distinguish the youngest reticulocytes (i.e., type I and II reticulocytes) based on RNA content (RNA disappears from reticulocytes older than 3 days) (MacGregor et al. 1980, 1987; Hayashi et al. 1990). Immediately after sacrifice, bone marrow cells are collected from the femurs or tibias, spread on slides and fixed. Similarly, slides can be prepared from blood samples. Slides are then stained (as briefly described above) and visually scored under the microscope. Because manual scoring is time-consuming, the scoring can also be done automatically with validated methods (Hayashi et al. 2007). Bone marrow micronucleated cells can be analyzed by image analysis following cellulose column separation of rodent bone marrow samples to remove nucleated cells (Romagna and Staniforth 1989; Frieauff and Romagna 1994) and to make the
306
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
automatic scoring more efficient. Automatic scoring of micronuclei in bone marrow erythrocytes and peripheral blood reticulocytes can also be done by flow cytometry using fluorescein-conjugated monoclonal antibodies against the transferrin receptor (anti-CD71-FITC) to stain polychromatic erythocytes and young (immature) reticulocytes, and propidium iodide with RNase treatment is used to identify micronuclei (Dertinger et al. 1997; Hayashi et al. 2000; MacGregor et al. 2006; Weaver and Torous 2000). A three-color labeling method, using antiplatelet-PE antibody in addition to anti-CD71-FITC and propodium iodide, was recently developed to improve the flow cytometry method, especially for human blood samples (Dertinger et al. 2004). Both automatic methods result in more objective measurements, allow more cells to be analyzed than with microscopic scoring (up to 10-fold more in the case of flow cytometry), and improve the statistical power (Torous et al. 2000, 2003, 2005; MacGregor et al. 2006; Kissling et al. 2007). The flow cytometric procedure was first developed for the analysis of micronuclei in mouse peripheral blood reticulocytes (Dertinger et al. 1996, 1997) before being applied to rat, dog, monkey, and human reticulocytes, and rodent bone marrow (Torous et al. 2000; Dertinger et al. 2002; MacGregor et al. 2006; Harper et al. 2007; Hotchkiss et al. 2008). The main advantage of the micronucleus test in peripheral blood is the possibility of using small volumes of blood (a few microliters) obtained during other studies such as general toxicity studies, in any species. Interpretation. A significant increase in the number of micronucleated polychromatic erythrocytes or young reticulocytes is usually considered indicative of structural and/or numerical chromosome damage caused by exposure to a clastogenic and/or aneugenic substance. In order to distinguish between clastogenic and aneugenic effects and to improve the risk assessment, it can be useful to conduct additional mechanistic investigations. Because the formation of micronuclei containing whole chromosomes results from an impact on the mitotic apparatus and not from a direct effect on DNA, this effect is considered to exhibit a threshold dose–response (Aardema et al. 1998), while this is generally not considered to be the case of micronuclei containing acentric fragments resulting from direct interaction with DNA. Whole chromosomes can be detected in micronuclei by using specific kinetochore antibodies and immunofluorescent CREST staining, or pancentromeric DNA probes and fluorescence in situ hybridization (FISH) (Iarmarcovai et al. 2006). A few authors have recommended the use of flow cytometry because, as noted previsouly, it increases the statistical power and allows better assessment of low doses (Grawé et al. 1998; Asano et al. 2006). Assay Limitations and Confounding Factors. In mice, the spleen does not destroy circulating micronucleated erythocytes, and the results obtained in bone marrow and peripheral blood are similar. In other species, and in particularly rats, the spleen efficiently removes circulating micronucleated reticulocytes, especially those induced by aneugens, owing to the large size of the micronuclei (Cammerer et al. 2007a,b). The sensitivity of the micronucleus test on rat peripheral blood has been reported to be lower than on rat bone marrow, from both a qualitative and
12.3. IN VIVO GENOTOXICITY ASSAYS REQUIRED IN THE STANDARD BATTERY OF TESTS
307
quantitative point of view (Wakata et al. 1998). Moreover, published data on the effect of aneugens on rat reticulocytes are conflicting. Recent data show that aneugens such as colchicine, vincristine, and vinblastin are barely detected after a single administration (Cammerer et al. 2007a), while colchicine was properly detected after five consecutive administrations (Cammerer et al. 2007b; MacGregor et al. 2006). Numerous studies are being conducted to further evaluate this problematic and controversial issue (Hayashi et al. 2007). Recently, the scoring of large numbers of reticulocytes with flow cytometry was described as being able to compensate for the low rate of micronucleated cells in peripheral blood resulting from spleen removal of micronucleated erythrocytes in some species, including rats (Torous et al. 2000; Witt et al. 2008). Provided that the ability of the flow cytometric approach to solve the sensitivity issue is confirmed, the micronucleus test on peripheral blood reticulocytes represents a promising noninvasive in vivo assay for the detection of chromosome damage in any species, including rat (MacGregor et al. 2006), dog (Harper et al. 2007), monkey (Hotchkiss et al. 2008), and human (Dertinger et al. 2002). Some toxic compounds (e.g., mitomycin C and dimethylhydrazine-2HCl), while they clearly increase the incidence of micronucleated erythrocytes after acute treatment, were not readily detected after multiple administrations, when integrated in general toxicity studies because the doses reached are much lower doses (Hamada et al. 2001). Even if the bone marrow is a well-perfused tissue, chemically unstable compounds and/or metabolites may not reach it in sufficient quantities to induce detectable effects (Brambilla and Martelli 2004; Morita et al. 1997). In a large collaborative study of IARC carcinogens (groups 1, 2A, and 2B), the in vivo bone marrow micronucleus test easily detected compounds able to induce tumors in hematopoietic tissues and lung. In contrast, it predicted only 40% of liver carcinogens (Morita et al. 1997). To solve this issue, chromosome damage, especially micronuclei, can be measured in tissues other than bone marrow and peripheral blood erythrocytes (Hayashi et al. 2007). The measurement of micronuclei on lymphocytes (from spleen and peripheral blood) consists of animal exposure, cell isolation, in vitro lymphocyte stimulation and micronucleus evaluation (Ren et al. 1991; Benning et al. 1992, 1994). The in vivo micronucleus test has also been developed in skin (Nishikawa et al. 1999, 2002) and gastrointestinal tract (e.g., colon) (Vanhauwaert et al. 2001; Ohyama et al. 2002). When evaluated in liver, cell proliferation is provoked by partial hepatectomy (Tates et al. 1980; Cliet et al. 1989; Igarashi and Shimada 1997) or by treatment with products able to induce cell proliferation such as 4-acetlyaminofluorene (Braithwaite and Ashby 1988) and carbon tetrachloride. Recently, a hepatocyte micronucleus assay in young rats was reported. Because hepatocytes are still able to proliferate in young animals, no mitogen stimulus is required. Moreover, young animals have a metabolic capacity similar to that of adults (Suzuki et al. 2004; Hayashi et al. 2007). Despite promising results, these models present technical difficulties (e.g., the need to induce liver cell proliferation) and are only used in specific cases (Hayashi et al. 2007). Confounding factors can lead to irrelevant findings in the in vivo micronucleus test on bone marrow [for review, see Tweats et al. (2007)]. They include changes
308
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
in core body temperature, because hypothermia and hyperthermia can disrupt chromosome binding to mitotic spindles and cause chromosome loss. Similarly, an increase in erythropoeisis, as a result of toxicity for erythroblasts or through direct stimulation of cell division (e.g., after bleeding, hemolysis, or erythropoietin production), can enhance the incidence of micronucleated erythrocytes. It is suggested that acceleration of erythroblast maturation and proliferation can lead to errors in erythrocyte enucleation (expulsion of the main nucleus from erythroblasts) or differentiation, as well as errors in genetic repair processes, resulting in higher rates of micronucleated cells (Tweats et al. 2007). Induction of apoptosis can also be a confounding factor, but it is generally easy to recognize because the micronuclei are much more numerous or pyknotic as compared to those induced by clastogens and aneugens.
12.3.2.
Bone Marrow Chromosome Aberration Test
Purpose. The purpose of the in vivo chromosome aberration test is to evaluate the potential of the test substance to cause chromosomal aberrations in bone marrow cells of experimental animals (usually rodents) (Tice et al. 1994; Preston et al. 1987). Regulatory Acceptance. The rodent erythrocyte chromosome aberration assay is well-validated and widely accepted by regulatory agencies as part of the standard battery of genetic toxicity assays, in addition to the two in vitro assays. It is considered as an equally acceptable alternative to the in vivo micronucleus test on rodent erythrocytes (Shelby and Witt 1995). The experimental conditions and data interpretation have been published in OECD guideline 475. Nevertheless, it is less widely and commonly used than the in vivo micronucleus test, because it is less simple. Chromosome aberration scoring is more tedious and time-consuming than micronucleus scoring, and requires skilled and experienced personnel. Principle. Chromosome aberrations are scored in bone marrow cells in first metaphase after compound administration—that is, 1.5 normal cell cycles. In order to take into account possible delays in absorption, metabolism, and cell cycling, bone marrow cells can also be collected 24 hours after the first sample. The structural aberrations are classified into two types (i.e., chromatid and chromosome aberrations) and three different categories (i.e., gaps, deletions, and rearrangements or exchanges) (Preston et al. 1987; Tice et al. 1994). Even if this test is not specifically designed to detect aneuploidy, an increased incidence of polyploid cells and of cells with endoreduplicated chromosomes suggests a potential to induce numerical chromosome aberrations. As in the in vivo micronucleus test, bone marrow is preferred because it is a well-perfused and rapidly dividing tissue. Study Design. Bone marrow cells are sampled 1.5 normal cell cycles (12–18 hours) after compound administration, and they are sampled optionally 24 hours later in additional groups of animals. In case of multiple administrations, cells are sampled 1.5 normal cell cycles (12–18 hours) after the last treatment. In order to accumulate cells in metaphase and make the scoring easier, the animals receive a
12.3. IN VIVO GENOTOXICITY ASSAYS REQUIRED IN THE STANDARD BATTERY OF TESTS
309
metaphase arresting agent (e.g., colchicine) by the intraperitoneal route 3–5 hours before sampling, depending on the species (OECD 1997b). Immediately after sacrifice, bone marrow cells are collected from femurs or tibias, exposed to hypotonic treatment, fixed, spread on slides, and stained. The slides are then scored visually under the microscope. Structural chromosome aberrations should be analyzed in at least 100 metaphase cells per animal. To this end, for each cell, the number and type of aberrations (chromatid or chromosome breaks and gaps, as well as the different types of chromatid and chromosome rearrangements) should be recorded. The incidence of polyploid cells and of cells with endoreduplicated chromosomes should also be reported because they potentially reflect the induction of numerical chromosome aberrations and/or inhibition of cell cycle progression. In addition, the mitotic index, used as a cytotoxicity parameter, is calculated for 1000 cells per animal. With cytotoxic compounds, the highest dose level should induce at least a 50% reduction in the mitotic index (OECD 1997b; Richold et al. 1990; Tice et al. 1994). Interpretation. A significant increase in the number of cells with structural chromosome aberrations, excluding gaps (gaps are reported separately), is usually considered as indicative of structural chromosome damage caused by exposure to a clastogenic agent. Moreover, an increased incidence of polyploid cells or cells with endoreduplicated chromosomes suggests that the compound is potentially an aneugen (Kirsch-Volders et al. 2002). The biological significance of polyploidy and endoreduplication (i.e., DNA replication without cell division) is controversial (Storchova and Pellman 2004). Formation of polyploid cells in normal tissues is far from negligible in nature. Moreover, polyploid cells that possess more than two sets of homologous chromosomes are generated by various mechanisms (endoreduplication, cell fusion, abortive cell cycle, mitotic slippage, and cytokinesis failure) in case of cellular stress, aging, and diseases, because they are thought to confer a metabolic advantage. While polyploid cells are normally blocked in G1 cell cycle arrest or eliminated by apoptosis in the case of cells bearing functional ploidy-sensing (e.g., p53, and Rb proteins) checkpoints, aneuploidy can arise from genetically instable tetraploid intermediates (Ganem et al. 2007). Finally, polyploidy cells are formed during stress conditions and therefore do not always reflect genotoxicity (Storchova and Kuffer 2008). In order to confirm the compound ability to provoke aneuploidy, as well as to improve the risk assessment, it can be useful to conduct an additional mechanistic evaluation and/or an in vivo micronucleus test including the presence of kinetochores in micronuclei (Aardema et al. 1998; Kirsch-Volders et al. 2002). Assay Limitations and Confounding Factors. Owing to the relatively small number of cells evaluated (100 cells per animal) and the low background (0–5% of cells with chromosome aberrations), the statistical power is lower than with the in vivo micronucleus test (Adler 1984). Moreover, chromosomes can be lost during metaphase spread. Therefore, aneuploidy cannot be directly assessed by counting the number of chromosomes per cell, but only by looking at polyploid cells and cells with endoreduplicated chromosomes (Aardema et al. 1998; Kirsch-Volders et al. 2002). As described above, not all compounds with such effects are aneugens. Some
310
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
might only impact cell division, without inducing chromosome loss or nondisjunction. Therefore, while the in vivo chromosome aberration test is considered equally acceptable for the detection of structural aberrations, it is less sensitive than the in vivo micronucleus test for the detection of numerical changes. This issue is nevertheless offset by the ability of this test to reveal the type of structural aberrations and to provide mechanistic information. As in the bone marrow micronucleus test, chemically unstable compounds and/or metabolites that cannot reach the bone marrow in sufficient quantities to induce detectable effects cannot induce chromosome aberrations in the bone marrow and are not appropriately detected in this test. Chromosome aberration can also be measured in peripheral blood or spleen lymphocytes after acute or chronic administrations (e.g., as part of general toxicity studies). Because lymphocytes remain quiescent in G0 cell-cycle phase for about one month in rodent blood and spleen, they have the opportunity to accumulate DNA lesions. Chromosome aberrations are analyzed after an in vitro 48-hour stimulation of lymphocyte proliferation by mitogenic compounds [e.g., phytohemagglutinin or concanavalin A (Kligerman et al. 1984)] and addition of a metaphase arresting agent (e.g., colcemid) for the last 2–3 hours. S-dependent genotoxins would not be detected with this method, as the compound is not in contact with the cells during the S-phase.
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS More than one in vivo genotoxicity assay is generally required when positive results are obtained in in vitro assays and 2-year bioassays (Cimino 2006; Kasper et al. 2007). In case of positive results in in vitro assays, the second assay is generally conducted in a different tissue (e.g., the most exposed tissue, a target organ for toxicity, or a tissue with a high capacity for metabolic activation). The endpoint can be primary DNA damage (in order to further evaluate DNA reactivity) and/or the endpoint found to be impacted in in vitro assays and not yet evaluated in vivo (i.e., gene mutations), in order to better evaluate the relevance of the in vitro findings. The most promising approach to determining whether tumors found in specific rodent tissues are attributable to genotoxic events is the assessment of genotoxicity in cancer target tissues (Kasper et al. 2007; Kirkland and Speit 2008; Lambert et al. 2005). Genotoxicity assays applicable to any tissue comprise rodent transgenic mutation assays, the in vivo comet assay, and determination of DNA adducts; other models such as the liver UDS and micronucleus assay in tissues other than bone marrow and peripheral blood are also valuable, but they are restricted to one or a limited number of tissues. The following section presents (1) primary DNA damage assays that detect the ability of compounds to interact with DNA and to cause primary DNA damage such as DNA strand-breaks (comet assay) and DNA adducts, (2) indicator assays (UDS and SCE assays) that reveal DNA repair or recombination in response to DNA lesions, and (3) gene mutation assays.
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
12.4.1.
311
The Comet Assay
Purpose. The purpose of the comet assay is to evaluate the potential of the test substance to induce primary DNA damage—that is, DNA strand-breaks in treated animals, usually rodents (Ostling and Johanson 1984; Singh et al. 1988; Olive et al. 1990a,b; 1991; Olive 2002; Collins 2002). It is applicable to any tissue from which a sufficient amount of cells or nuclei can be isolated without damaging DNA or triggering DNA repair processes. Regulatory Acceptance. The comet assay is widely accepted by regulatory agencies, even though it was only relatively recently developed and is still being validated. No OECD guidelines are yet available, but several publications provide internationally agreed protocol recommendations (Tice et al. 2000; Hartmann et al. 2003; Burlinson et al. 2007). The standard protocol for regulatory purposes is currently being refined. These validation exercises and collaborative efforts are currently coordinated by the Japanese Center for the Validation of Alternative Methods or JaCVAM. The comet assay is mentioned or clearly recommended as a complementary assay in the vast majority of regulatory documents: (1) in case of positive in vitro results not confirmed in the in vivo bone marrow micronucleus test, in order to improve the weight of evidence and risk assessment, (2) in case of negative results in the standard battery of genotoxicity tests and tumor findings in 2-year bioassays, in order to better understand the mechanisms responsible for carcinogenesis, to confirm the absence of a genotoxic impact in a target organ of carcinogenicity, and to allow the compound to be classified as a nongenotoxic carcinogen provided that the nongenotoxic mechanism is elucidated (Kasper et al. 2007), and (3) to evaluate the genotoxic impact in the first contact tissues in the case of (a) poorly absorbed compounds giving little or no systemic exposure, and (b) chemically unstable short-lived compounds and metabolites. Examples of application are given in Brendler-Schwaab et al. (2005) and Hartmann et al. (2004). Principle. The comet assay (or single-cell gel electrophoresis assay) is a rapid and simple method for the detection of DNA breakage in mammalian cells. The recommended and most commonly used method, the alkaline version (pH > 13) of the comet assay (Singh et al. 1988, 1994; Burlinson et al. 2007), detects in individual cells double strand-breaks, single strand-breaks (including those resulting from alkali-labile sites and incomplete excision repair), and both DNA–DNA and DNA– protein crosslinks. Thus, the comet assay detects most types of primary DNA damage that could later become fixed as either gene mutations (e.g., resulting from alkali-labile or abasic sites and damage not eliminated by excision repair), or structural chromosome damage (e.g., owing to DNA strand-breaks). Study Design. The compound is administered by the most appropriate route to at least four to five animals per group and at a minimum of two doses (MTD and 25–50% MTD). After a single administration, the tissues are sampled respectively 2–6 hours (preferably 3) and 16–26 hours (preferably 21) after treatment, in order
312
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
to detect the effects of (a) rapidly absorbed, unstable and directly acting compounds and (b) compounds requiring more time for absorption, distribution, and metabolic activation. After multiple treatments (2 or more) at 24-hour intervals, the tissues/ organs should be collected once, 2–6 hours (preferably 3) after the last administration (Hartmann et al. 2003). The cells are isolated from solid tissues by using digestive enzymes (trypsin or collagenase), by brief mincing with scissors, or by pushing the tissue through a mesh (Brendler-Schwaab et al. 1994). The nuclei are obtained by tissue mincing and homogenization (Miyamae et al. 1998; Sasaki et al. 1997a–d). During cell or nuclei isolation, EDTA and radical scavengers (e.g., dimethylsulfoxide) can be added to prevent degradation by endonucleases and oxidative DNA damage. The isolated cells or nuclei are embedded in agarose gel and layered on microscopic slides. The slides are incubated first in lysis buffer containing detergents and high salt concentrations in order to release the DNA. The next steps are DNA unwinding in the presence of alkaline buffer in order to produce single-stranded DNA and to express the DNA alkali-labile sites as single strand-breaks, followed by electrophoresis in alkaline conditions. At the end of the electrophoresis step, the slides are neutralized, dried, and stained with fluorescent dyes (propidium iodide, ethidium bromide, SYGR green, or yoyo-1) in order to visualize the DNA. When examined under the microscope, the cells look like comets, with (a) a head corresponding to undamaged DNA in the nuclear region and (b) a tail containing DNA strands and fragments or loops (Shaposhnikov et al. 2008) resulting from DNA breakages, which have migrated in the direction of the anode. DNA migration should be measured with 100–150 cells per animal (Lovell et al. 1999; Wiklund and Agurell 2003) on two to three different slides by manual scoring or by using an interactive or fully automated image analysis systems (Böcker et al. 1999; Frieauff et al. 2001; Schunck et al. 2004; Dehon et al. 2008). The experimental conditions should allow some DNA migration in vehicle controls in order to detect a delay in DNA migration after treatment with crosslinking agents (Hartmann et al. 2003). Because cell death, necrosis, and apoptosis can lead to DNA fragmentation and to possibly irrelevant findings, cytotoxicity should be evaluated, especially in case of a positive result. First, comets with small or nonexistent heads and large diffuse tails (Fairbairn et al. 1996; Olive et al. 1993), named “hedgehogs,” “ghost cells,” “clouds,” or “nondetectable nuclei cells,” are either excluded from the analysis or scored separately as potentially apoptotic/necrotic cells. The presence of low-molecular-weight DNA fragments in apoptotic/necrotic cells can be evaluated by omitting the electrophoresis step after alkaline unwinding, in a neutral diffusion assay (Tice et al. 2000). Finally, it is strongly recommended to collect samples of tissues/organs for histopathological examination (Burlinson et al. 2007). Interpretation. Manual scoring generally quantifies the distance of DNA migration, the percentage of cells with and without DNA migration, and/or the proportion of cells in four to five different categories, from undamaged to heavily damaged cells (Miyamae et al. 1998). When the scoring is done by image analysis, in addition to the tail length, the percentage of DNA in the tail (% tail DNA) is also measured. The product of the tail length and the % tail DNA, named the tail moment, is also calculated. The values of the three parameters (tail length, % tail DNA, and tail
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
313
moment) should be reported for each individual cell, along with the mean and/or median values for each animal and treatment group. The % tail DNA is the preferred parameter, because it is less dependent than tail length on the technology used, is linear with respect to the dose, and is more easily reproduced between laboratories (Dehon et al. 2008). Moreover, the distribution of migration among cells from each animal and group are useful for data interpretation (Olive and Durand 2005; Burlinson et al. 2007). An increase in DNA migration parameters indicates that the test substance has induced DNA strand-breaks, while a decrease suggests DNA–DNA and DNA– protein crosslinks. Advantages. The main advantage of the comet assay is that it allows simple, rapid (hours to days after sampling), and cost-effective DNA damage evaluation. It has the added advantage of detecting low levels of DNA damage in single cells of any organ (Burlinson et al. 2007; Anderson et al. 1998). Its ability to visualize and quantify DNA strand-breaks in individual cells is clearly seen as an advantage as compared to alkaline elution method measuring breaks in DNA from a cell pool (Kohn and Grimek-Ewig 1973; Kohn et al. 1976). It does not require cell division, contrary to the in vivo chromosome aberration and micronucleus tests, or manipulation of radiolabeled compounds as required by the UDS test and some DNA adduct assays. Moreover, in contrast to the micronucleus, chromosome aberration, and UDS tests, mainly conducted with bone marrow and liver, the comet assay is applicable to any tissue and can be performed on a limited number of cells or nuclei. The ability to measure DNA strand-breaks in site-of-contact tissues is particularly important in case of low systemic exposure and chemically unstable compounds. The tissue selection should take into consideration all available information on structural analogues, absorption, distribution, metabolism, excretion, and/or toxicology. When no information is available, one or preferably two tissues should be examined: the liver for orally absorbed compounds and a site of first contact tissue—that is, the gastrointestinal tract, respiratory tract, and skin for the oral route, inhalation, and dermal application, respectively (Hartmann et al. 2003). Tremendous amounts of data have already been obtained with the comet assay in vivo (McKelvey-Martin et al. 1993; Fairbairn et al. 1995; Burlinson et al. 2007; Rojas et al. 1999; Kirkland and Speit 2008). Some publications focus on its ability to detect organ specificity (Burlinson et al. 2007). In general, these publications confirm the ability of the comet assay to detect DNA damage in carcinogenicity target tissues, when the appropriate route of administration is used—for example, when first site-of-contact tissues (stomach, skin, and basal mucosa) are examined. Other authors assessed a large number of compounds on a selected set of tissues (stomach, colon, liver, kidney, urinary bladder, lung, brain, and bone marrow, using intraperitoneal injection or oral gavage) in order to assess its sensitivity and specificity for the detection of carcinogens. These data, summarized by Sasaki et al. (2000), show that among 208 compounds selected from IARC monographs and the US NTP Carcinogenicity Database, 94% of rodent genotoxic carcinogens were positive in the mouse comet assay and 80% of rodent noncarcinogens were negative (Sasaki et al. 2000). Moreover, 91% of the carcinogens that did not induce
314
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
micronuclei in hematopoietic cells in bone marrow and/or peripheral blood were positive in the comet assay. The authors also pointed out that DNA damage was detected not only in almost all carcinogenicity target organs, but also in nontarget organs. Therefore, the comet assay can be used to confirm the induction of DNA damage in target organs, but not to predict where tumors will occur. When DNA migration is evaluated at the two recommended sampling times, 62% of genotoxic carcinogens are detected in liver while 69% are detected in colon and stomach. The detection rate is vastly improved when liver and stomach (86%) or liver and colon are combined (87%). Limitations and Confounding Factors. The standard alkaline method does not differentiate between the different types of damage (single and double strand-breaks, alkali-labile and incomplete excision repair sites). Additional evaluations and/or modifications are needed to obtain mechanistic information. These include (1) comparing strand-breaks obtained with the alkaline and neutral versions of the protocol to potentially differentiate double from single strand-break induction, (2) using DNA repair enzymes such as UV-specific endonucleases, endonuclease III for oxidized pyrimidines, or formamido pyrimidine glycosylase (FPG) for 8-hydroxyguanines (Collins et al. 1993, 2008) to provoke specific breaks, and (3) combining the comet assay with DNA-damage-specific antibodies revealed by immunofluorescence or with chromosome painting probes measured by fluorescence in situ hybridization (COMET-FISH) to visualize DNA damage and specific genomic regions in the comet (Santos et al. 1997; Sauvaigo et al. 1998; Rapp et al. 2005). It should also be emphasized that rapidly repaired damage can be missed unless early sampling times and high doses, at which the repair capacity is generally overwhelmed, are appropriately selected. One of the main potential confounding factors is the formation of DNA strandbreaks through cell death, necrosis, or apoptosis. It is therefore important to confirm the absence of toxicity in the case of positive results (Tice et al. 2000; Burlinson et al. 2007), using the neutral diffusion assay and histological tissue evaluation. Another confounding effect is the induction of DNA strand-breaks through indirect mechanisms, such as production of free radicals in the case of enzymatic induction, inflammation, and preneoplastic lesions. Therefore, positive results obtained after long-term administration, during which tissue remodeling may have taken place, should be considered with caution.
12.4.2.
DNA Adducts
Purpose. DNA adducts are nucleotide bases (i.e., purines and pyrimidines) that have been covalently modified by reactive electrophilic chemical intermediates or free radicals. The chemical structures of DNA adducts are diverse and vary from simple alkyl adducts induced by alkylating agents to complex bulky adducts such as those resulting from metabolic activation of polycyclic aromatic hydrocarbons, aromatic amines, and aflatoxins (Dipple 1995; Chiarelli and Jackson 1992; Rundle 2006; Xue and Warshawsky 2005). The purpose of measuring DNA adducts is to determine whether a DNA-reactive compound or a metabolically activated
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
315
intermediate can covalently bind to nucleophilic sites of DNA nucleotide bases. If replication takes place before DNA adducts are removed—for example, through base and nucleotide excision repair or dealkylation mechanisms—or if DNA adducts are misrepaired, they can be fixed into point mutations (Hemminki et al. 2000) or lead to chromosome damage. For many years, DNA adducts have been used as biomarkers of exposure: If measured shortly after treatment, they provide an integrated marker of compound intake (absorption and distribution), metabolic activation, delivery to target macromolecules in target tissues, and interaction with DNA (Farmer 2004a; Koc and Swenberg 2002; Swenberg et al. 2008; Farmer and Singh 2008). The level of adducts measured at a given time point also depends on the formation and stability of electrophilic entities, adduct stability and DNA repair, and tissue turnover (i.e., cell proliferation and cell death, including apoptosis). DNA damage such as DNA adducts is considered to be necessary but not sufficient for tumorigenesis (Poirier et al. 2000). Mutagenesis and cell proliferation must also take place. However, DNA adducts are considered to be early key events in carcinogenesis induced by genotoxic carcinogens. Their measurement contributes to understanding the metabolism and action of carcinogens through molecular dosimetry across species, including humans [e.g., aflatoxins, tamoxifen (Gamboa da Costa et al. 2003), benzo[a]pyrene (Beland et al. 2005), and 2-amino-3,8-dimethylimidazo [4,5-f]quinoxaline (MeIQx) (Mauthe et al. 1999)], and provides useful information for cancer risk assessment (Poirier and Beland 1992; Weston 1993). Regulatory Acceptance. The detection of DNA adducts is not part of the standard battery of tests, but can be recommended as a follow-up investigation, in the case of positive results in in vitro genetic toxicity assays and negative results in the bone marrow chromosome damage test, as well as in the case of negative results in the standard battery of genotoxicity tests and tumor findings in 2-year bioassay (Reddy 2000; Phillips et al. 2000). In the latter case, DNA adducts are sought in the target organ of carcinogenicity. No guidelines are available, but several publications provide protocol recommendations [for general review, see Farmer (2004b), Farmer and Singh (2008), Hemminki et al. (2000), Phillips et al. (2000), Poirier et al. (2000), Reddy (2000), Singh and Farmer (2006), Garner (1998)]. Principle. After animal treatment with single or multiple administrations, depending on the method used for DNA adduct detection and the purpose of the study, DNA is isolated from the tissue(s) of interest. DNA is preferably analyzed in tissue from individual animals and is pooled only if necessary to make up the required quantity (Phillips et al. 2000). The measurement of DNA adducts generally consists of four main steps: (1) DNA isolation from the tissue or organ of interest, (2) DNA hydrolysis or digestion, (3) DNA adduct enrichment and isolation, and (4) DNA adduct analysis with or without the addition of a standard. Qualitative and quantitative analyses of DNA adducts have gradually improved over the last 40 years. In the past two decades, significant efforts have been made to elucidate the chemical structures of DNA adducts by using chemically specific techniques such as mass, fluorescence, and nuclear magnetic resonance spectrometry (Poirier 2004). The main methods for the detection and analysis of DNA adducts are, from the oldest to the
316
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
most recent: (1) administration of radiolabeled compounds and measurement of radioactive decay by scintillation counting, (2) immunoassays using antibodies to carcinogen-modified DNA, also used to locate DNA adducts in tissues by immunohistochemistry, (3) 32P-postlabeling (PPL), (4) high-performance liquid chromatography (HPLC) combined with physicochemical detection methods such as fluorescence or electrochemical detection, (5) various chromatographic techniques with mass spectrometric (MS) detection, and (6) administration of radiolabeled compounds and measurement of isotope ratios with accelerator mass spectrometry (AMS). A brief description of the principle, study design, sensitivity, specificity, limitations, and strengths of the different techniques is given in Table 12.2. Interpretation. The biological significance of DNA adducts is still controversial (Nestmann et al. 1996; Phillips et al. 2000). The formation of DNA adducts is considered to be a key event in human cancer formation. However, it has been clearly demonstrated only for a limited number of compounds—for example, aflatoxin and aromatic hydrocarbons (see examples in Poirier et al., 2000). Comparison of DNA adduct formation and tumorigenesis after chronic administration of genotoxic carcinogens to animals showed that steady-state DNA adduct levels are generally reached after 1–2 months of chronic administration (Poirier et al. 2000). No tumors were observed in the absence of DNA adducts, but, on the other hand, the presence of DNA adducts was not always synonymous with tumor formation, suggesting either that a threshold level of DNA adducts is needed or that other key events such as cell proliferation are necessary for tumour formation. Ottender and Lutz (1999)
TABLE 12.2.
In Vivo Measurement of DNA Adducts
Methods Used for the Measurement of DNA Adducts References
Radiolabeling Method Coupled with Liquid Scintillation Counting Baird (1979), Buss et al. (1990), Lutz (1979, 1986), Martin et al. (1993), Phillips et al. (2000), Reddy (2000), Swenberg et al. (2008).
Immunoassays Müller et al. (1982), Müller and Rajewsky (1980, 1981), Den Engelse et al. (1990), Farmer (2004a,b), Hsu et al. (1981), Kriek et al. (1984), Phillips etal. (2000), Poirier et al. (2000), Poirier and Beland (1992), Poirier (1981, 1993, 2004), Reddy (2000), Santella (1999), Strickland and Boyle (1984), Wild (1990).
32
P-Postlabeling Assay
Farmer (2004a,b), Gupta et al. (1982), Phillips (1997), Phillips and Arlt (2007), Phillips and Castegnaro (1999), Phillips et al. (2000, 2005), Poirier et al. (2000), Randerath et al. (1981), Randerath and Randerath (1994), Reddy (2000), Reddy and Randerath (1986), Reddy et al. (1984), Shibutani et al. (2006), Terashima et al. (2002), Whong et al. (1992).
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
317
concluded from data obtained with 27 carcinogens that a 50% tumor incidence rate was associated with 53–5543 adducts per 108 nucleotides, depending on the compound. Evaluation of the risk associated with DNA adducts depends on the chemical nature, quantity, and stability of the adducts, the rate of cell proliferation and adduct fixation into stable mutations, adduct mutagenic and repair efficiencies, and the extent of changes to critical genes (Nestmann et al. 1996). Some DNA adducts, including minor adducts, are associated with mutagenesis and tumorigenesis, while others are not (Yuspa and Poirier 1988; Hemminki et al. 2000). Current data, obtained with highly sensitive methods able to detect one adduct among 1010–1012 nucleotides, suggest that DNA adduct formation is linear at low doses. There may thus be adduct levels (e.g., 1 adduct among 1010 normal nucleotides) at which the risk of mutations and tumors is indistinguishable from the background risk (Nestmann et al. 1996). Therefore, adducts are mainly used as a marker of exposure, given that the DNA binding of many compounds is linear over the dose range. Because DNA binding efficiency does not strictly correlate with the incidence of mutations and tumors, the detection of DNA adducts does not necessarily predict (a) tumorigenicity for a given tissue or (b) the human cancer risk (Hemminki et al. 2000). Thus, DNA adducts should be interpreted in view of other in vivo endpoints—that is, stable mutations and carcinogenicity (Nestmann et al. 1996). Assay Limitations and Advantages. Each method mentioned above has its strengths and weaknesses (see Table 12.2) that impact its specificity, sensitivity, cost, and practicality. The radiolabeling method coupled with liquid scintillation
High-Performance Liquid Chromatography with UV, Fluorescence, or Electrochemical Detection
Mass Spectrometry Coupled with Liquid Chromatography, Gas Chromatography or Capillary Electrophoresis
Kriek et al. (1984), Farmer (2004a,b), Poirier et al. (2000), Poirier (2004), Reddy (2000), Weston et al. (1989), Weston (1993).
Beland et al. (2005), Chiarelli and Jackson (1992), Farmer (2004a,b), Farmer et al. (2005), Gamboa da Costa et al. (2003), Koc and Swenberg (2002), Phillips et al. (2000), Poirier (2004), Poirier et al. (2000), Reddy (2000), Singh and Farmer (2006).
Radiolabeling Method Associated with Accelerator Mass Spectrometry Dingley et al. (1998, 2005), Farmer (2004a,b), Farmer et al. (2005), Garner (1998), Goldman et al. (2000), Mauthe et al. (1999), Phillips et al. (2000), Poirier (2004), Poirier et al. (2000), Reddy (2000), Tompkins et al. (2006), Turteltaub and Dingley (1998).
(Continued)
318 TABLE 12.2.
CHAPTER 12
(Continued)
Methods Used for the Measurement of DNA Adducts Principle
IN VIVO GENOTOXICITY ASSAYS
Radiolabeling Method Coupled with Liquid Scintillation Counting Measurement of radioactive decay after single administration of [14C]labeled or [3H]-labeled compounds. A few hours to days after exposure, DNA is extracted from tissue of interest and purified. Increase in radioactivity, measured by liquid scintillation counting, as compared to control DNA is considered to reflect DNA adduct formation. Further characterization can be done on hydrolyzed DNA after isolation of modified nucleotides by highperformance liquid chromatography (HPLC).
Immunoassays DNA extraction from the tissue of interest after compound administration. Competitive or direct immunoassays are conducted and inhibition of antibody binding is measured using antibodies against DNA adducts or modified DNA obtained from immunized rabbit. Changes in antibody binding are indicative of the presence of DNA adducts.
32
P-Postlabeling Assay
DNA extraction from the tissue of interest after compound administration followed by (1) enzymatic digestion of DNA in 3′-monophosphates of normal and adducted nucleotides, (2) optional enrichment step to select or isolate adducted nucleotides (typically butanol extraction or nuclease P1 treatment), (3) radio-labeling of the adducts by incorporation of 32P-orthophosphate at nucleotides 5′-end using a polynucleotide kinase and [γ32P]-ATP, (4) separation of DNA adducts using multidirectional thin-layer chromatography (TLC) on polyethyleneimine (PEI) cellulose, HPLC or electrophoretic separation, (5) radioactivity quantification by measurement of radioactive decay using autoradiography or electronic imaging. Intensity changes in background spots and observation of additional spots are indicative of a positive response.
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
319
High-Performance Liquid Chromatography with UV, Fluorescence, or Electrochemical Detection
Mass Spectrometry Coupled with Liquid Chromatography, Gas Chromatography or Capillary Electrophoresis
Radiolabeling Method Associated with Accelerator Mass Spectrometry
DNA extraction from the tissue of interest after compound administration. High-performance liquid chromatography (HPLC) separation of DNA adducts, followed by fluorescence or electrochemical detection and quantification. Observation of additional peaks/signals is indicative of the presence of adducts.
DNA isolation from any tissue of interest after compound administration followed by (1) addition of stable isotopelabeled internal standard for quantification, (2) hydrolysis/ digestion of DNA, (3) enrichment of DNA adducts of interest (e.g., solid-phase extraction, immunoaffinity chromatography, or DNA repair enzymes), (4) quantification by mass spectrometry (MS) using gas chromatography–MS (after derivization) with electron impact ionisation, or liquid chromatography–MS using electrospray or other ionspray capillary electrophoresis interfaces. Tandem MS–MS is often used to increase the specificity of the assay and is the favored technique. A signal at the correct chromatographic retention time with intensity greater than a specified signal-to-noise ratio is indicative of the presence of DNA adducts.
Measurement of radioactive decay (accelerated mass spectrometry) after single administration of [14C]-labeled or [3H]-labeled compounds (other isotopes also applicable). After exposure (hours to days), DNA is extracted from tissue of interest and purified. Incorporated radioactivity, is measured by separation of isotope based on mass followed by quantification in gas ionization detector to measure isotope ratio. Another alternative is the use of postlabeling, i.e., digestion of DNA to nucleotides, depurination, HPLC separation to isolate adducts, [14C]-acetylation reaction, HPLC to collect [14C]-labeled adducts, and AMS analysis. Increase in radioactivity, measured by AMS, as compared to control DNA is considered to reflect the presence of adduct.
(Continued)
320 TABLE 12.2.
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
(Continued)
Methods Used for the Measurement of DNA Adducts
Radiolabeling Method Coupled with Liquid Scintillation Counting
Immunoassays
Sensitivity
Typically 1 adduct per 106 nucleotides to as low as 1 adduct per 109 nucleotides.
Typically 1 adduct out 106 nucleotides. Depends on the method used for the detection of antiserum bound in microtiter plates, i.e., colorimetry, fluorescence, or chemiluminescence. Could be as low as 1 adduct per 108–109 nucleotides with chemiluminescence.
Specificity
Limited specificity as no information on chemical structure of the adducts.
Limitations
Difficult and expensive synthesis of quite large amount of radio-labeled compounds with high and stable specific activity, i.e., millicurie amounts per animal needed. Multidose treatment is difficult, if not impossible, because of the cost and difficulty to use large volume of radiolabeled compounds. Need to ensure that the location of labeling is resistant to loss during metabolism and adduct formation. Need to verify that radioactivity is not due to contamination or metabolism, i.e., to avoid possible artifacts, HPLC of DNA hydrolysates and digestion with proteases and ribonucleases can be used.
Potentially highly specific. Depends on antibody preparation, and specificity. Preparation of large quantities of specific antibodies is needed. Relatively large amounts of DNA (up to 100 μg) required. Information on DNA adduct structure is needed for the preparation of specific antibodies, when highly specific antibodies are prepared. Therefore not applicable for unknown compounds, unless whole damaged DNA is used for antiserum preparation.
32
P-Postlabeling Assay
Highly sensitive: Generally as low as 1 adduct in 1010 nucleotides, when butanol extraction or nuclease P1 treatment are applied. Very efficient for bulky adducts, and N7alkylguanines and other positively charged adducts, less suitable for other nonaromatic, small or depurinating adducts (1 adduct out of 105–106 nucleotides). Sensitivity can be improved by about one order of magnitude using HPLC combined with radioisotope detector after TLC or PAGE. Moderate specificity, as it does not provide information on chemical structure of the adducts. Use of high quantities of 32 P radioactivity (25–50 μCi per sample) with very high specific activity. Relatively labor intensive and low throughput (completed in 3 days), except with PAGE method that allows the concomitant analysis of a large number of samples in parallel within a few hours. The efficiency of the enrichment and labelling steps varies according to the adduct structures and can result in adduct loss or low sensitivity for some types of adducts.
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
High-Performance Liquid Chromatography with UV, Fluorescence, or Electrochemical Detection
Mass Spectrometry Coupled with Liquid Chromatography, Gas Chromatography or Capillary Electrophoresis
321
Radiolabeling Method Associated with Accelerator Mass Spectrometry
As low as 1 adduct per 107–108 nucleotides.
As low as 1 adduct per 108–109 nucleotides. Sensitivity is increasing with new equipments.
Extremely sensitive: As low as 1 adduct per 1011–1012 nucleotides following administration of [14C]-labeled compounds, i.e., less than one adduct per cell. Lower sensitivity generally obtained when [3H]-labeled compounds are used.
Highly specific.
Very high specificity and accurate information on chemical structure of adducts.
Limited as it provides no information on the chemical structure of the adducts.
Only applicable to adducts chemically characterized possessing fluorophore (e.g., polycyclic aromatic hydrocarbons) or electrochemically active groups (e.g., some oxidative DNA lesions). Relatively large amounts of DNA (up to 100 μg) needed.
Inability to screen unknown mixtures. High purity of DNA is required to avoid artefacts due to protein and RNA. Relatively large amounts of DNA (up to 100 μg) needed.
Very expensive instrumentation and limited number of laboratories equipped. Requests administration of radio-labeled compounds, unless post-labeling is applied. Requires synthesis of radio-labeled compounds ([14C]-labeled or [3H]-labeled compounds). Artifacts can result from metabolism and DNA contamination by proteins and RNA, and crosscontamination between samples. Further work is needed to verify that isotopes are integrated in DNA adducts.
(Continued)
322
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
TABLE 12.2. (Continued) Methods Used for the Measurement of DNA Adducts Strengths
Radiolabeling Method Coupled with Liquid Scintillation Counting Simple and straightforward method as compared to the other techniques.
Immunoassays No need for radiolabeled compounds. Useful for highthroughput analysis (e.g., ELISA) of specific adducts, because cost effective and relatively easy to conduct. Antibodies also used for immunohistochemistry in order to get information on morphological localization of adducts, and for immunoaffinity chromatography used for preliminary steps of DNA adduct enrichment before any other techniques.
32
P-Postlabeling Assay
Radio-labeled compounds or specific antibodies are not requested. Limited amount of DNA needed (less than 10 μg DNA). Ability to detect adducts from chemicals with various structures, e.g., polycyclic aromatic hydrocarbons, aromatic amines, heterocyclic amines, small aromatic compounds, alkylating agents, unsaturated aldehydes from lipid peroxidation, reactive oxygen species and UV radiation. Attempts to define a standardized protocol. Applicable after multiple administrations and for complex mixtures.
counting and AMS has the drawback of involving radioactive manipulations. AMS is the only method able to detect DNA adducts after exposure to very low doses (Dingley et al. 2005; Brown et al. 2005). It also requires much lower isotope doses than liquid scintillation counting. This is a clear advantage for risk assessment, because it eliminates the need to extrapolate from high dose levels (Turteltaub and Dingley 1998). Moreover, the very low isotope dose levels required for AMS make the method applicable in humans (Turteltaub and Dingley 1998; Mauthe et al. 1999; Farmer et al. 2005). Immunoassays (in the case of specific antibodies) and methods associated with physicochemical detection methods such as fluorescence or electrochemical detection can only be used if the structure of the DNA adducts is known. Similarly, depending on the chemical nature of the adducts, different enrichment methods are used in PPL such as further enzymatic digestion with nuclease P1 for aromatic hydrocarbon-like bulky adducts, immunoaffinity chromatography with specific antibodies, and anion-exchange column chromatography for methyl- and ethylsubstituted compounds bearing a positive charge. There is no PPL method available for other simple alkylated adducts. Moreover, false-negative results or underestimation of adduct levels can result from the use of an inappropriate method. Indeed, the efficiency of the enrichment and labeling steps varies according to the adduct
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
High-Performance Liquid Chromatography with UV, Fluorescence, or Electrochemical Detection
Mass Spectrometry Coupled with Liquid Chromatography, Gas Chromatography or Capillary Electrophoresis
Radio-labeled compounds or specific antibodies are not requested.
The most promising method for the detection of DNA adducts, especially as the new equipments are less expensive and more easy to use, and more the method readily automated. No need for radio-labeled compounds or specific antibodies. Use of chemical-specific stable isotope internal standards ensures very accurate quantification, and confirmation with certainty of the chemical nature of adducts. Possible association with 32 P-postlabeling and immunochemical methodologies. Detection of very different chemicals, from bulky adducts to modified DNA bases.
323
Radiolabeling Method Associated with Accelerator Mass Spectrometry Quantification of DNA adducts in samples at very low exposure levels, and after administration of very low quantities of isotopes/ radiolabeled compounds (as low as 1 μCi/40 kBq). Measurement independent of radioactive decay. Being improved by postlabeling techniques, i.e., incorporating 14 C into specific DNA adducts after formation.
structure, and this can result in adduct loss or in poor sensitivity for some types of adduct (e.g., adducts formed at N7 positions of purines). Because PPL protocols may underestimate adduct levels, several methods (i.e., two enrichment methods) should be used in the case of unknown adducts, and preliminary studies with standards are therefore very important (Whong et al. 1992). AMS, immunoassays, HPLC combined with fluorescence, and mass spectrometry using stable isotope-labeled internal standards can accurately quantify DNA adducts (Koc and Swenberg 2002). PPL and AMS are extremely sensitive (detecting 1 adduct per 1012 nucleotides) but lack specificity, and they provide no information on the chemical nature of the adducts. MS is less sensitive than PPL or AMS but is highly specific and can provide structural information (Farmer and Singh 2008). In addition, mass spectrometers are becoming cheaper and simpler to use. MS methods are also improving with the development of high-resolution mass spectrometry and tandem MS, for example, and are expected to become the most specific and sensitive approaches in the near future. All this means that MS is playing an increasingly important role in DNA adduct analysis (Farmer and Singh 2008; Farmer et al. 2005; Koc and Swenberg 2002; Singh and Farmer 2006). The available methods have rarely been compared but appear to give similar qualitative results (Weston et al. 1989; Eide et al. 1999).
324
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
The main artifacts and possible confounding effects result from DNA contamination, lack of purity, and a lack of specificity of certain steps of DNA adduct preparation or analysis (Phillips et al. 2000). Recommendations for DNA isolation and storage have been made (Phillips and Castegnaro 1999). Endogenous DNA adducts and adducts formed by nongenotoxic carcinogens (e.g., peroxisome proliferators and estrogens) can sometimes interfere with the detection of DNA adducts, especially when highly sensitive methods (e.g., PPL) are used.
12.4.3.
Unscheduled DNA Synthesis Test in Liver Cells
Purpose. The purpose of the ex vivo unscheduled DNA synthesis (UDS) test on mammalian liver cells is to evaluate the potential of the test substance to induce DNA excision repair in liver cells of treated animals (usually rodents and preferably rats). An increase in UDS activity is indicative of primary DNA damage and subsequent excision repair (Butterworth et al. 1987; Mirsalis and Butterworth 1980; Mirsalis et al. 1982). Regulatory Acceptance. The UDS test on mammalian liver cells is wellvalidated and widely accepted by regulatory agencies. The optimal experimental conditions and rules for data interpretation have been published in OECD guideline 486. The UDS test has never been widely used because it is relatively timeconsuming (in vivo and in vitro experimental steps, and treatment and sampling times that do not fit with usual working day). In addition, it requires the use of radiolabeled compounds with relatively high specific activity and, therefore, special authorization. For many years, this assay has been recommended as an in vivo follow-up assay, in the case of positive results in the in vitro genetic toxicity assays and negative results in the bone marrow chromosome damage test, or findings in 2-year bioassays, especially in liver. Its potential lack of sensitivity has been highlighted, and other, more recent in vivo assays (e.g., the comet assay) are now preferred (Kirkland and Speit 2008). Principle. During nucleotide excision repair (NER), a stretch of 20–30 nucleotides (up to 100 nucleotides) containing the DNA damage (e.g., bulky adducts) is removed and replaced; but during base excision repair (BER), only 1 to about 10 nucleotides, including the modified nucleotide, are excised (Shuck et al. 2008; Fousteri and Mullenders 2008; Baute and Depicker 2008; Hegde et al. 2008). The UDS test detects the incorporation of radiolabeled thymidine during DNA resynthesis in the excised region (Butterworth et al. 1987; Mirsalis and Butterworth 1980; Mirsalis et al. 1982). Liver is the preferred tissue for UDS measurement because it is well-perfused and the main metabolic site for absorbed compounds (during first-pass of compounds administered by the oral and intraperitoneal routes). Moreover, it is a slow-dividing tissue, and only a small proportion of cells undergo replicative DNA synthesis. Therefore DNA synthesis in most liver cells is limited to DNA repair (Butterworth et al. 1987; Mirsalis and Butterworth 1980).
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
325
Study Design. This is an ex vivo assay. The compound is administered once, to at least three animals per group, then the liver cells are isolated and UDS activity is measured in primary culture. The amount of tritiated thymidine incorporated into liver cell DNA is measured in individual cells by autoradiography; liquid scintillation is not recommended. Respectively 2–4 and 12–16 hours after treatment, the liver cells are isolated by liver perfusion with collagenase, in order to detect the effects of (a) rapidly absorbed and/or metabolized compounds (e.g., methyl methanesulfonate, and dimethyl nitrosamine) and (b) compounds requiring more complex metabolic activation (e.g., 2-acetylamidofluorene). Amphlett et al. (1996) showed that in order to save animals and simplify the assay the animals can be treated twice, 2–4 and 12–16 hours, before sampling. After liver cell attachment to culture dishes, the cells are incubated for 3–8 hours with tritiated thymidine. At the end of the incubation period, which can be followed by a cold chase (incubation with unlabeled thymidine), the cells are fixed, dipped into radiographic emulsion, and kept in the dark for 7–14 days. Grains are counted over the nuclei (nuclear grain count) and cytoplasm (cytoplasm grain count), in 100 cells per animal. The average net nuclear grain (NNG) count (cytoplasmic grain count subtracted from nuclear grain count) and the percentage of cells undergoing repair (increase in the NNG value over spontaneous background) are then calculated. The grains are counted with an interactive or fully automated image analysis system. Cells undergoing DNA replication are easily identified (completely covered by grains) and are excluded from the analysis, because they contain extremely large numbers of grains over the nucleus. Interpretation. Statistical analysis is not generally used, and each laboratory should consider the distribution of its negative control values to define an NNG cutoff value for positive results. In general, it is considered that negative controls always have NNG values lower than zero. An increase in the mean NNG value over the threshold value (usually zero) obtained for a given animal and for a given dose group is indicative of enhanced DNA repair activity (Hamilton and Mirsalis 1987). The proportion of cells undergoing repair (i.e., with NNG values over the threshold) is also taken into consideration. Therefore, a compound that increases the average NNG value above zero, as well as the number of cells undergoing repair, would be considered positive in this test. Assay Limitations. The UDS test is only an indirect measurement of primary DNA damage. It does not provide information on the nature of the primary damage or on the fidelity of repair. Moreover, its sensitivity depends on the number of nucleotides removed and replaced and, thus, the amount of tritiated thymidine incorporated into DNA. Compounds inducing bulky adducts and long-patch repair via NER are generally more efficiently detected by the UDS test than those removed by short-patch repair or BER. Moreover, the UDS assay can only detect DNA damage repaired via the excision repair process: Single strand-breaks and oxidative base damage, for example, are not detected. Like tests done with bone marrow, the liver UDS test has limited value for the detection of labile direct-acting compounds, which cannot readily reach the liver.
326
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
Site-of-contact tissues might be preferable for such compounds (Burlinson 1989; Furihata et al. 1984; Furihata and Matsushima 1987; Mori et al. 1999; Sawyer et al. 1988), if sufficiently validated. The main technical limitation of this assay with respect to other tissues is the need to isolate the cells after in vivo treatment and to get them to incorporate tritiated thymidine in vitro. Because UDS measurement does not require cell division, it can potentially be applied to many different tissues, provided that the cells can be isolated and maintained in primary culture for the few hours required for tritiated thymidine incorporation. The literature contains reports of UDS-based studies of stomach, colon, kidney, pancreas, tracheal epithelium, nasal epithelium, epidermis, keratinocytes, and spermatocytes (Burlinson 1989; Furihata et al. 1984; Furihata and Matsushima 1987; Sawyer et al. 1988; Loury et al. 1987; Mori et al. 1999; Latt et al. 1981; Helleday 2003).
12.4.4.
Sister-Chromatid Exchange Assay
Purpose. The purpose of the sister chromatid exchange (SCE) assay is to evaluate the potential of the test substance to induce repair of DNA lesions by homologous recombination in cells of treated animals (potentially all species, usually rodents) (Latt et al. 1981; Helleday 2003). It can easily be applied to any dividing tissue, such as bone marrow and peripheral blood, from which cell suspensions can be isolated and analyzed. Regulatory Acceptance. A regulatory OECD guideline (476) describes the protocol of the in vitro SCE assay conducted with mammalian cells, but no OECD guideline exists for the in vivo assay. Though popular (especially for biomonitoring) and widely used to detect exposure to mutagens and carcinogens (Perry and Evans 1975; Latt 1981; Tucker et al. 1993), the use of SCE data for risk assessment is more controversial because the mechanism of SCE formation and the biological significance of the increased incidence of SCE are not fully elucidated (Latt 1981; Tucker et al. 1993; Wilson and Thompson 2007). Principle. SCE consist of an interchange of DNA replication products and parental DNA strands between two sister chromatids at homologous sites. They require DNA breakage and rejoining steps (Wilson and Thompson 2007; Latt 1981; Latt and Schreck 1980). They generally do not alter chromosome morphology. SCE are conservative and error-free end-products of homologous recombination associated with the repair of persisting single strand breaks. They do not occur during double strand-break repair by homologous recombination, which mostly results in gene conversion, deletions, and tandem duplications (Helleday 2003; Wilson and Thompson 2007). SCE take place during S-phase of the cell cycle and can therefore only be evaluated in actively dividing cells. Any agent or mechanism that stalls the replication progression fork is able to induce SCE. In contrast, inhibition of replication initiation (by X rays and bleomycin for example) rarely and only mildly affect the incidence of SCE (Latt 1981; Painter 1980). Study Design. Chromatids can be visualized in the late prophase and early metaphase of mitosis, before chromatid segregation in daughter cells. In order to increase
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
327
the number of cells in metaphase and thus facilitate SCE scoring, a mitotic inhibitor such as colchicine is administered a few hours before tissue sampling (e.g., about 3 hours for bone marrow), and cells are then prepared in the same way as for chromosome aberration analysis. Techniques for SCE visualization take advantage of the semiconservative nature of DNA replication (Wilson and Thompson 2007). In the 1950s, the first method developed for the detection of SCE was based on the incorporation of tritiated thymidine in newly synthesized DNA. Subsequently, the most commonly used techniques consist of injecting a thymidine base analogue (e.g., 5-bromo-2-deoxyuridine or BrdU) to animals, shortly after test compound administration (e.g., 2–3 hours after administration in the case of bone marrow). Alternatively, small tablets of BrdU can be implanted subcutaneously (e.g., 8 hours before treatment) for sustained and continuous base analogue release (Allen et al. 1977; Latt 1981; King et al. 1982; Madrigal-Bujaidar and Sanchez-Sanchez 1991). Tissues— for example, rodent bone marrow—are generally sampled 21 hours after BrdU infusion, or 24 hours after tablet implantation. BrdU is only incorporated into the nascent daughter strand of each DNA duplex, and after the second division the two sister chromatids bear different amounts of BrdU. One of the two sister chromatids has the original template strand that contains no BrdU and the second strand that has incorporated BrdU. The other sister chromatid has BrdU incorporated on both strands. SCE are visualized as asymmetrically stained chromosomes, or “harlequin” chromosomes, after differential chromatid staining with nonfluorescent dyes (e.g., Giemsa), or fluorescent dyes (e.g., Hoechst dyes), the staining being generally less pronounced in the chromatid or fragments of chromatid containing BrdU on both strands (Korenberg and Freedlender 1974; Latt 1974; Latt and Schreck 1980). One of the standard techniques, named the fluorescent plus Giemsa method, combines fluorescent (e.g., Hoescht) and nonfluorescent (e.g., Giemsa) dyes (Perry and Wolff 1974; Speit 1984; Speit and Haupter 1985; Spencer and Butler 1987). The sister chromatid with BrdU incorporation shows a less pronounced staining with Giemsa. SCE can also be visualized with fluorescent BudR antibodies that specifically label BrdU-substituted DNA (Natarajan et al. 1986). DNA is generally counterstained with 4′,6-diamidino-2-phenylindole (DAPI) or propidium iodide (Pinkel et al. 1985). The parameter measured in this assay is the incidence of SCE per cell. Assay Limitations and Interpretation. Incorporation of BrdU itself can contribute to SCE, because it results in single strand-breaks and alkali-labile sites. Immunofluorescence methods that only necessitate small doses of BrdU for SCE visualization yield a lower SCE background than other techniques (Wilson and Thompson 2007). A method using Biotin-dUTP instead of BrdU was recently reported to overcome this technical issue (Wojcik et al. 2004). The SCE indicator assay measures error-free homologous recombination occurring in case of fork collapse and replication-blocking lesions. An increased incidence of SCE is considered as a potential marker of exposure (Wilson and Thompson 2007). It has also been linked to the induction of single strand breaks. Numerous studies have described the induction of SCE after exposure to DNAdamaging compounds such as alkylating agents, crossslinking agents, heavy metals, agents that form bulky adducts, and UV (Helleday 2003).
328
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
Most clastogenic and mutagenic compounds are SCE inducers, although some qualitative and quantitative differences have been reported. However, it has been suggested that SCE can also be provoked by nongenotoxic mechanisms able to disturb DNA synthesis. Moreover, recent epidemiological studies suggest that studies of SCEs cannot replace the analysis of structural chromosome damage (Gebhart 1981) and that SCEs might not be indicative of cancer risk (Liou et al. 1999; Hagmar et al. 2001; Norppa et al. 2006). Consequently, even if a large number of genotoxic products can enhance the incidence of SCE, an increased incidence of SCE does not always reflect a genotoxic mechanism (Bradley et al. 1979) and is therefore difficult to interpret.
12.4.5.
Gene Mutation Assays
Purpose. The purpose of in vivo gene mutation assays is to detect point mutations, such as base-pair substitutions, frameshifts, small deletions, and insertions. As mentioned in Tables 12.3 and 12.4, a few assays are also able to detect large deletions and reciprocal recombination. TABLE 12.3. In Vivo Gene Mutation Assays
In Vivo Gene Mutation Assays in Endogenous Genes of Somatic Cells
Mouse (Coat) Spot Assay
Mouse Retinoblast (Eye Spot) Assay
References
OECD Test Guideline 484 (OECD 1986b), Fahrig (1975, 1995), Fahrig and Neuhauser-Klaus (1985), Russell (1977), Russell and Major (1957).
Bishop et al. (2000), Gondo et al. (1993), Searle (1977), Schiestl et al. (1997).
Casciano et al. (1999), Deubel et al. (1996), Dobrovolsky et al. (1999b, 2005), Jones et al. (1985), Aidoo et al. (1991), Tates et al. (1994), van Dam et al. (1992), Walker et al. (1999).
Endpoints
Gene mutations (point mutations, small deletions/insertions), reciprocal recombination and chromosome aberrations.
Gene deletions resulting from intrachromosomal recombination (intrachromosomal crossing-over, single-strand annealing, unequal sister-chromatid exchange and sister-chromatid exchange).
Gene mutations: point mutations, small deletions/ insertions.
Hprt Assay
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
329
Regulatory Acceptance. Detection of gene mutations is not generally part of the standard battery of genotoxicity tests. The mouse spot test (Fahrig 1975, 1995; Fahrig and Neuhauser-Klaus 1985; Russell 1977) is the only gene mutation test on somatic cells described in an OECD guideline (OECD 1986b), but it is rarely used because it is restricted to a single tissue and requires a very large number of animals. No official guidelines are yet available for assays applicable to any tissue, such as transgenic gene mutation assays, but several publications provide protocol recommendations (Heddle et al. 2000; Nohmi et al. 2000; Thybaud et al. 2003). Use of transgenic gene mutation assays is frequently recommended by international guidelines as follow-up tests, in the case of positive results in in vitro genetic toxicity assays and negative results in the bone marrow chromosome damage test, as well as in the case of negative results in the standard battery of genotoxicity tests and tumor findings in 2-year bioassays. A general review has been published by Lambert and colleagues (Lambert et al. 2005; OECD 2009). Principle, Interpretation, Limitations, and Strengths. Tables 12.3 and 12.4 summarize the main characteristics of the different gene mutation assays that are further compared below.
Aprt Assay Gupta et al. (1994), Van Sloun et al. (1998).
Gene mutations (point mutations, small deletions/insertions) and events conducting the loss of heterozygosity (large deletions, mitotic nondisjunctions, mitotic recombination and gene conversions).
Dbl-1 Assay Cosentino and Heddle (1995), Tao et al. (1993a,b), Tao and Heddle (1994), Uiterdijk et al. (1986), Winton et al. (1990), Winton and Ponder (1990). Gene mutations (point mutations, small deletions/ insertion) and events conducting the loss of heterozygosity (large deletions, mitotic nondisjunctions, mitotic recombination and gene conversions).
Tk Assay
Pig-a Assay
Dobrovolsky et al. (1999a,b, 2005), Tischfield et al. (1994).
Bryce et al. (2008).
Gene mutations (point mutations, small deletions/ insertion) and events conducting the loss of heterozygosity (large deletions, mitotic nondisjunctions, mitotic recombination and gene conversions).
Gene mutations: point mutations, small deletions/insertions.
(Continued)
330
CHAPTER 12
TABLE 12.3. In Vivo Gene Mutation Assays Principle
Reporter gene
Species, strain
IN VIVO GENOTOXICITY ASSAYS
(Continued) Mouse (Coat) Spot Assay
Mouse Retinoblast (Eye Spot) Assay
Detection in F1 animals of spots on fur/coat resulting from point mutations at c locus or recombination between in c and p genes, induced in melanoblasts in utero. Dam mice (∼50 per group) are treated on day 8, 9, and 10 of gestation. 3–4 weeks after birth, ∼300 offsprings are examined for coat spots, resulting from somatic mutations considered genetically relevant (i.e., gene mutations at c locus in melanocytes). c locus
Detection in F1 animals of spots on eyes resulting from deletion mutations at p locus, induced in utero in precursor cells of retinal pigment epithelial cells. C57Bl/6J pun/pun dam mice are treated at about day 10 of gestation. F1 animals are euthanized 20 days after birth. A large number of dams per group needs to be exposed, and numerous offsprings should be examined for eye spots at single-cell level.
Detection of hprt mutant in peripheral blood T lymphocytes or splenocytes, using 6-thioguanine (6TG)resistant phenotype.
pun locus (pink-eye unstable mutation) in the tandem duplication. Mouse C57Bl/6J pun/pun carrying an autosomal recessive mutation that produces pink eyes.
hprt locus. Located on X chromosome. Only one active copy per cell.
Mouse F1 animals that are heterozygous for different recessive coat color genes, as a result of mating of T-strain mice with HT or C57/Bl mice.
Hprt Assay
Wild-type animals, mainly rodents. Also applicable in other species, including human.
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
331
Aprt Assay
Dbl-1 Assay
Tk Assay
Detection of aprt mutant in T lymphocytes from peripheral blood or splenocytes, and fibroblasts, using 8-azaadenine (AZA) or 2,6-diaminopurine (DAP) resistant phenotype.
Detection of Dlb-1b mutant cells in vertical stripes of small intestine and colon villi in heterozygous mice (Dlb-1a/ Dlb-1b) at Dlb-1 locus, using differential staining.
Detection of tk mutant in peripheral blood T lymphocytes or splenocytes, using 5-bromo-2′deoxyuridine (BrdU)-resistant phenotype.
Detection of Pig-a mutant cells in red blood cells, showing CD59− phenotype.
aprt locus. Located in chromosome 8 in mouse.
Dlb-1 locus. Located in chromosome 11 in mouse. Mouse heterozygous at Dlb-1 locus (Dlb-1a/Dlb-1b). Dlb-1b is an autosomal dominant gene that determines the expression of the binding site for the lectin Dolichos biflorus agglutinin in small intestine and colon epithelium. (Dlb-1a leads to its expression in vascular epithelium.)
Tk locus. Located in chromosome 11 in mouse.
Pig-a locus. Located on X chromosome. Only one active copy per cell. Wild-type animals, mainly rodents. Also applicable in other species, including human.
Aprt+/− C57BL/6 heterozygous mouse.
Tk+/− C57BL/6 heterozygous mouse.
Pig-a Assay
(Continued)
332
CHAPTER 12
TABLE 12.3. In Vivo Gene Mutation Assays
IN VIVO GENOTOXICITY ASSAYS
(Continued) Mouse (Coat) Spot Assay
Mouse Retinoblast (Eye Spot) Assay
Tissues
Target cells are melanoblasts in utero, and scoring is done in melanocytes in F1 animals.
Method for mutant selection
Changes in fur color: brown, gray, black, and nearly white spots randomly distributed over the whole coat. Different types of mutations can be identified. For example, gene mutations in c (albino) and c (chinchilla) alleles lead to brown spots, while reciprocal recombination by crossing over involving p (pink-eyed dilution) and c (albino) loci result in black and near white black spots.
Target cells are precursor cells of retinal pigment epithelium cells in utero, and scoring is done in retinal pigment epithelium cells in F1 animals. Changes in pigmentation of retinal pigment epithelium, appearance of pigmented mutant cells. Retinas are examined under microscope and spots of pigmented cells are counted. Deletion of one of two pun copies in the tandem duplication causes reversion from pun to wild-type p. The wild-type cells are easily identified as black pigmented cells/spots.
Hprt Assay Any tissue that can be subcultured in vitro. Mainly T-lymphocytes from spleen or peripheral blood.
Isolation of T-lymphocytes, followed by an in vitro mitogen stimulation and selection of mutant cells in the presence of a selective agent, such as 6TG, cytotoxic for nonmutant cells. The wild-type hprt gene encodes for hypoxanthineguanine phosphoribosyl transferase that phosphorylates 6TG into a cytotoxic compound. In the presence of 6TG only cells bearing a mutated hprt gene survive. The mutant frequency is the number of cell clones in the presence of 6TG versus the number cell clones in the absence of 6TG.
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
Aprt Assay
333
Dbl-1 Assay
Tk Assay
Pig-a Assay
Any tissue that can be subcultured in vitro. Mainly T-lymphocytes from spleen, or skin fibroblasts.
Small intestine and colon epithelial cells.
Any tissue that can be subcultured in vitro. Mainly splenic lymphocytes.
Peripheral blood erythrocytes.
Isolation of T-lymphocytes or fibroblasts, followed by in vitro mitogen stimulation (for lymphocytes) and selection of mutant cells in the presence of a selective agent, such as AZA or DAP, cytotoxic for nonmutant cells. The wild-type aprt gene encodes for adenine phosphoribosyltransferase that phosphorylates the selective agent into a cytotoxic compound. In the presence of the selective agent only cells bearing a mutated aprt gene survive. The mutant frequency is the number of cell clones in the presence of selective agent versus the number cell clones in the absence of selective agent.
Differential staining of Dlb-1b mutant cells. Dlb-1b mutant cells do not express the lectin binding site, and can easily be identified as they remain unstained in the presence of peroxidase conjugate of Dolichos biflorus agglutinin, while nonmutant cells are dark-brown stained.
Isolation of splenocytes, followed by an in vitro mitogen stimulation and selection of mutant cells in the presence of a selective agent, such as BrdU, cytotoxic for nonmutant cells. The wild type tk gene encodes for thymidine kinase that phosphorylates the selective agent into a cytotoxic compound. In the presence of the selective agent only cells bearing a mutated tk gene survive. The mutant frequency is the number of cell clones in the presence of selective agent versus the number cell clones in the absence of selective agent.
Blood samples are analyzed by flow cytometry to numerate the CD59-negative erythrocytes. The Pig-A gene product is involved in the first step in glycosylphosphatidylinositol (GPI) anchor biosynthesis. Mutations in Pig-a gene impact the cell surface expression of all GPI-anchored proteins, including CD59. The incidence of GPI-anchor deficient cells, i.e., Pig-a mutant cells, is quantified by flow cytometry, using anti-CD59-Pe and other fluorescent reagents to differrentiate the blood cell populations.
TABLE 12.4.
In Vivo Gene Mutation Assays in Transgenic Models
Transgenic Gene Mutation Assays
Muta™Mouse
Big Blue® Models
Plasmid lacZ
References
Gossen and Vijg (1993), Gossen et al. (1989, 1991, 1992), Douglas et al. (1994), Dean and Myhr (1994), Blakey et al. (1995), Heddle et al. (2000), Lambert et al. (2005), Mientjes et al. (1994), Nohmi et al. (2000), Piegorsch et al. (1997), Vijg and Douglas (1996), Thybaud et al. (2003).
Dolle et al. (1996), Gossen et al. (1995), Gossen and Vijg (1993), Heddle et al. (2000), Lambert et al. (2005), Nohmi et al. (2000), Thybaud et al. (2003), Vijg and Douglas (1996).
Endpoints
Gene mutations: point mutations, small deletions/ insertions.
Principle
Detection of gene mutations in the lacZ bacterial reporter gene. Main steps are: (1) administration of the compound to mouse, (2) isolation of highmolecular-weight genomic DNA from the tissues of interest, (3) rescue of bacteriophage DNA bearing the bacterial reporter gene using packaging kits, (4) infection of host bacteria and (5) quantification of lacZ gene mutations in host bacteria in medium containing either X-Gal and P-Gal.
Dycaico et al. (1994), Heddle et al. (2000), Kohler et al. (1990, 1991a,b), Lambert et al. (2005), Nohmi et al. (2000), Piegorsch et al. (1995), Stiegler and Stillwell (1993), Thybaud et al. (2003), Wyborski et al. (1995). Gene mutations: point mutations, small deletions/ insertions. Detection of gene mutations in the lacI bacterial reporter gene. Main steps are: (1) administration of the compound to mouse or rat, (2) isolation of high molecular weight genomic DNA from the tissues of interest, (3) rescue of bacteriophage DNA bearing the bacterial reporter gene using packaging kits, (4) infection of host bacteria and (5) quantification of lacI gene mutations in host bacteria in medium containing X-Gal.
Gene mutations (point mutations, small deletions/insertions) and large deletions. Detection of gene mutations in the lacZ bacterial reporter gene. Main steps are: (1) administration of the compound to mouse, (2) isolation of high-molecularweight genomic DNA from the tissues of interest, (3) rescue of plasmid DNA bearing the bacterial reporter gene using HindIII digestion, absorption on lac represssor coated magnetic beads and plasmid recircularization, (4) electroporation of plasmids into the host bacteria and (5) quantification of lacZ gene mutations in host bacteria in medium containing P-Gal.
cII lambda phage Gene
Gpt delta (6-thioguanine)
Gpt Delta (spi)
Heddle et al. (2000), Jakubczak et al. (1996), Lambert et al. (2005), Nohmi et al. (2000), Swiger et al. (1999, 2001), Thybaud et al. (2003).
Hayashi et al. (2003), Heddle et al. (2000), Lambert et al. (2005), Masumura et al. (1999), Nohmi et al. (1996, 2000), Swiger et al. (2001), Thybaud et al. (2003).
Gunther et al. (1993), Hayashi et al. (2003), Heddle et al. (2000), Lambert et al. (2005), Nohmi and Masumura (2004), Nohmi et al. (1996, 1999, 2000), Thybaud et al. (2003).
Gene mutations: point mutations, small deletions/ insertions.
Gene mutations: point mutations, small deletions/insertions.
Small and large deletions.
Detection of gene mutations in the cII lambda phage gene. Main steps are: (1) administration of the compound to mouse or rat, (2) isolation of high-molecular-weight genomic DNA from the tissues of interest, (3) rescue of bacteriophage DNA bearing the bacterial reporter gene lacZ or lacI using packaging kits, (4) infection of host bacteria and (5) quantification of cII gene mutations in host bacteria maintained at 25 °C.
Detection of gene mutations in the gpt bacterial reporter gene. Main steps are: (1) administration of the compound to mouse or rat, (2) isolation of high-molecular-weight genomic DNA from the tissues of interest, (3) rescue of bacteriophage DNA bearing the bacterial reporter gene using packaging kits, (4) infection of host bacteria and (5) quantification of gpt point mutations in host bacteria in medium containing 6-thioguanine.
Detection of deletion leading to inactivation of both redBA and gam genes. Main steps are: (1) administration of the compound to mouse or rat, (2) isolation of highmolecular-weight genomic DNA from the tissues of interest, (3) rescue of bacteriophage DNA bearing the bacterial reporter gene using packaging kits, (4) infection of host bacteria and (5) quantification of deletions with spi selection (spi stands for sensitive to P2 interference).
(Continued)
336
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
TABLE 12.4. (Continued)
Transgenic Gene Mutation Assays
Muta™Mouse
Big Blue® Models
Plasmid lacZ
Reporter gene
lacZ bacterial gene in λgtı0 bacteriophage vector. 40 copies in head-to-tail manner in chromosome 3 of all mouse cells.
LacZ bacterial gene in pUR288 plasmid. ∼20 copies per genome in multiple chromosomes, e.g., chromosomes 3 and 4 in mouse line 60.
Species, strains
Muta™Mouse: CD2F1 (BALB/CxDBA2) mouse strain 40.6
lacI bacterial gene in λLIZα bacteriophage vector, also containing α-lacZ gene. The commercially available Big Blue® mouse (mouse lineage A1) and rat contains 40 and 15–20 copies in a head-to-tail manner in chromosome 4, i.e., 80 and 30–40 copies per genome in homozygous strain, respectively. Big Blue® mouse and rat: B6C3F1 and C57BL/6 background for mice, and Fischer 344 background for rat.
Tissues
Any tissue from which DNA can be properly extracted.
Any tissues from which DNA can be properly extracted.
Any tissues from which DNA can be properly extracted.
LacZ plasmid mouse: C57BL/6 background mouse strain 60.
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
337
cII lambda phage Gene
Gpt delta (6-thioguanine)
Gpt Delta (spi)
λ phage cII gene. CII encodes for a protein that regulates the λ phage lysogenic/lytic cycle.
gpt E. coli bacterial gene in λEG10 bacteriophage vector. 80 copies in head-to-tail manner in mouse chromosome 17, i.e., 160 copies per genome, the mouse strain being maintained as homozygous. About 10 copies in rat chromosome 4q24-q31, the strain being only available as hemizygote.
Red and gam genes in λEG10 bacteriophage vector. 80 copies in head-to-tail manner in mouse chromosome 17, i.e., 160 copies per genome, the mouse strain being maintained as homozygous. About 10 copies in rat chromosome 4q24-q31, the strain being only available as hemizygote.
Applicable to all assays that use λ phage as shuttle vector (e.g., Muta™Mouse, and Big Blue® mouse and rat), except gpt delta models. The λEG10 phage used as vector in the later model bears a mutation in the chiC gene, involved in positive mutant selection. Any tissues from which DNA can be properly extracted.
gpt delta mouse (C57BL/6J background), and gpt delta rat (Sprague– Dawley background).
gpt delta mouse (C57BL/6J background), and gpt delta rat (Sprague–Dawley background).
Any tissues from which DNA can be properly extracted.
Any tissues from which DNA can be properly extracted.
(Continued)
338
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
TABLE 12.4. (Continued)
Transgenic Gene Mutation Assays Method for mutant selection
Muta™Mouse
Big Blue® Models
Plasmid lacZ
Two methods: (1) Colorimetric method: Infection of E. coli C (ΔlacZ) lacZ deficient host bacteria with the λgt10 bacteriophage vectors containing mutated or nonmutated lacZ genes rescued from the mouse genomic DNA. Mutant bacteria are detected on medium supplemented with X-Gal (a substrate of β-galactosidase that produces a blue product). Blue and white plaques contain wild-type lacZ, and mutant lacZ− genes, respectively. The mutant frequency is the frequency of white plaques out of the total number of plaques. (2) Positive selection method: Infection of E. coli C (ΔgalE ΔlacZ) host bacteria, i.e., deficient in galE and lacZ, with the λgt10 bacteriophage vectors containing mutated or nonmutated lacZ genes rescued from the mouse genomic DNA. Mutant bacteria are detected on medium containing P-Gal toxic for lacZ+ galE− bacteria. Only lacZ− phages lead to plaque formation after bacteria infection. The mutant frequency is expressed as the number of plaques observed in the presence of P-Gal out of the total number of plaques formed in the absence of P-Gal.
Colorimetric method: Infection of E. coli SCS-8 (lacZΔM15) host bacteria, i.e., lacI-deficient, with the λLIZα bacteriophage vectors containing mutated or nonmutated lacI genes rescued from the mouse genomic DNA. Mutant bacteria are detected on medium supplemented with X-Gal (a substrate of β-galactosidase that produces a blue product). No β-galactosidase is synthetized when lacZ gene is repressed by lacI product. White and blue plaques contain wild-type lacI, and mutant lacI− genes, respectively. The mutant frequency corresponds to frequency of blue plaques out of the total number of plaques.
Positive selection method: Electroporation of plasmids containing mutated or nonmutated lacZ genes in E. coli C (ΔgalE ΔlacZ) host bacteria, i.e., deficient in galE and lacZ. Mutant bacteria are detected on medium containing P-Gal toxic for lacZ+ galE− bacteria. Only bacteria that integrate lacZ− plasmid are able to survive. The mutant frequency is expressed as the number of colonies observed in the presence of P-Gal out of the total number of colonies formed in the absence of P-Gal.
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
339
cII lambda phage Gene
Gpt delta (6-thioguanine)
Gpt Delta (spi)
Positive selection method: Infection of E. coli G1225 (hfl−) host bacteria with λgt10 bacteriophage vectors containing mutated or nonmutated cII genes rescued from the mouse genomic DNA. When E. coli hfl− host bacteria are infected by phages bearing nonmutated CII genes, the CII protein is not degraded by Hfl protease and induces cI and int gene expression that leads to phage lysogeny: No plaques are formed. In contrast, phages bearing cII mutated gene do not produce CII protein and enter in lytic instead of lysogenic cycle: plaques are formed. The mutant frequency is expressed as the number of plaques observed in hfl− host bacteria versus the number of plaques obtained in hfl+ host bacteria after incubation at 25 °C.
6-thioguanine (6TG) selection: Infection of E. coli YG6020 (cre+) host bacteria with λEG10 bacteriophage vectors containing mutated or nonmutated gpt genes rescued from the mouse genomic DNA. In the bacteria expressing Cre recombinase λEG10 DNA is converted in multi-copy-number of circularized plasmids carrying gpt and gene for chloramphenical acetyltransferase (CAT). The wild-type gpt gene encodes for guanine phosphoribosyltransferase that phosphorrylates 6TG into a cytotoxic compound for the bacteria. In the presence of 6TG and chloramphenicol, only bacteria bearing mutated gpt gene survive. Total number of infected bacteria is evaluated in medium containing chloramphenicol only. The mutant frequency is the number of colonies in medium with 6TG and chloramphenicol versus the number colonies observed in medium containing chloramphenicol only.
Spi selection: Infection of E. coli XL1 Blue host bacteria, carrying P2 phage DNA (i.e., P2 lysogen) with λEG10 bacteriophage vectors containing or not deletions in red/gam region. Only mutant phages deficient in both red and gam genes, as a result of a deletion in this region, are able to grow in the P2 lysogen host bacteria and display spi− phenotype (as long as they carry chi site and the host bacteria is recA+). Mutant frequency is the number of spi− plaques out of the total number of rescued plaques measured in E. coli XL1 Blue host bacteria not carrying the P2 phage DNA.
340
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
Some publications report the measurement of mutations in cancer-related genes such as p53, but these models are currently only used for research purposes [e.g., McKinzie et al. (2001), McKinzie and Parsons (2002), and Parsons and Heflich (1998)]. The most frequently used in vivo gene mutation assays evaluate mutations not in genes specifically involved in carcinogenesis but in surrogate reporter genes. Only a few assays detect mutations in endogenous genes (see Table 12.3), and most are restricted to a limited number of tissues, species, and developmental stages, such as melanocytes in the mouse (coat) spot test, mouse retinal epithelial cells for the retinoblast (eye spot) assay, mouse small intestine and colon for mutations in Dlb-1 gene, mostly splenocytes for mutations in Hprt gene, and lymphocytes for mutations in Tk gene. During the past 20 years, transgenic animal models have been developed to detect gene mutations in any organ or tissue (see Table 12.4), provided that high-molecular-weight genomic DNA can be properly extracted (e.g., lacZ in Muta™mouse, lacI in Big Blue® mouse and rat, and gpt delta mouse and rat). These models can be used to measure gene mutations in the most appropriate tissues, according to the mode of administration, absorption, distribution, and tissue-specific metabolism. They facilitate the evaluation of gene mutations at site-of-contact tissues after inhalation, topical application, or oral administration. They are also claimed to be able to specifically detect site-of-contact mutagens (Dean et al. 1999). Most endogenous genes are transcriptionally active. DNA lesions are therefore actively removed by transcription-coupled repair and cannot accumulate over time. By contrast, the bacterial reporter genes in transgenic animals, as well as the endogenous gene Pig-A, are neutral and nontranscribed. In neutral genes, DNA lesions accumulate over time in reporter genes (Tao and Heddle 1994), and especially in nondividing or slow-dividing tissues. The study design (i.e., treatment and sampling times) should take this information into account in order to optimize the detection of gene mutations (Thybaud et al. 2003). First, the duration and number of treatments, (i.e., the administration time) should be sufficient to permit the accumulation of primary DNA damage. Then, a second period after compound administration, generally called the fixation time, is needed to allow the distribution and metabolization of the compound, the formation of primary DNA damage, and its fixation into stable mutations during DNA replication or repair. A short fixation time is sufficient for highly proliferative tissues (e.g., 3–10 days for bone marrow and the gastrointestinal tract), while a longer period is required for slowly proliferating tissues (e.g., 28 days or more for liver or mammary gland). Finally, in the case of actively transcribed endogenous genes, the expression time is the time required for the turnover and disappearance of preexisting nonmutated protein in the tissues. This is especially important when selective agents are used to visualize the mutants (e.g., 6–8 days of in vitro culture before adding the selective agent in the case of Aprt assay in splenocytes). These three periods—the administration, fixation and expression times—determine the optimal sampling time for each organ in a given model, also named the manifestation time (Heddle 1999). For endogenous genes (e.g., hprt, aprt and tk), up to 5 weeks may be required between animal exposure and analysis of mutant cells in order to ensure optimal evaluation of the mutant frequency. Preliminary data suggest that Pig-a mutations could be detected more rapidly after
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
341
the last administration, a plateau in mutant frequency being reached after 1–2 weeks (Bryce et al. 2008). In transgenic models, and most probably in all models using neutral genes, multiple administrations increase mutant frequency in an approximately additive manner, and the time needed to reach a mutant frequency plateau varies from one organ to another—from a few days with highly proliferative tissues to several weeks with slowly proliferative tissues (Tao and Heddle 1994; Thybaud et al. 2003). An increased mutant frequency obtained only after chronic administration (3 months or more) should be interpreted with caution, because it can result from secondary nongenotoxic effects, such as clonal expansion, genomic instability inherent to preneoplastic foci and tumors, and oxidative damage resulting from chronic induction of cytochrome P-450 monooxygenases (Heddle et al. 2000). A 28-day administration period with sampling 3 and/or 28 days after the last administration is considered appropriate for most tissues (except maybe for germ cells, for which the timing of cell development should also be taken into consideration; more than 50 days may be necessary to detect gene mutations in sperm) and is, by default, considered the optimal design for transgenic models (Thybaud et al. 2003). Another issue with gene mutation assays is how they allow mutant cells to be identified, visualized, and counted, in order to calculate the mutant frequency. Available methods for endogenous genes consist of using (1) a selective agent to select mutant cells in vitro after cell isolation and transfer to culture medium (e.g., 6-thiogunaine for hprt, 2,6-diaminopurine for aprt and bromodeoxyuridine for tk), (2) histopathological colorimetric methods (Dlb-1), and (3) pigmentation differences to visualize mutant cells in the whole animal (e.g., spot tests). Transgenic models use transgenes bearing a bacterial reporter gene (e.g., lacZ, lacI, or gpt) integrated in a shuttle vector (e.g., a phage or plasmid) in order to allow DNA exposure and the formation of gene mutations in animals, the rescue of reporter gene vectors from animal genomic DNA by in vitro packaging of shuttle vectors or excision/relegation of integrated plasmids, and the measurement of mutations in reporter genes (lacI, lacZ, or gpt) in host bacteria after phage infection or plasmid electroporation (see Table 12.4 for more details). The selection of mutants in the presence of X-Gal, the first method developed for the Muta™Mouse and Big Blue® models (Dycaico et al. 1994), is quite labor-intensive and time-consuming, because at least 100,000 plaques per animal have to be scored. A more straightforward positive selection system using P-Gal and E. coli galE- is now available for the Muta™Mouse model (Gossen et al. 1992; Gossen and Vijg 1993; Mientjes et al. 1994; Dean and Myhr 1994; Vijg and Douglas 1996), but is not applicable to Big Blue®. Alternatively, mutations can easily be detected in the cII gene of the lambda phage used as shuttle vector in the Muta™ Mouse and Big Blue® models (Jakubczak et al. 1996; Swiger et al. 2001): results obtained with cII are similar to those obtained with lacZ and lacI (Zimmer et al. 1999). Gene mutation assays measure mutant frequency, which is generally expressed as the incidence of mutant cells per million cells. A higher mutant frequency in treated animals than in controls indicates that the compound has the potential to induce gene mutations. The background mutant frequency obtained in transgenic models (∼5 × 10−5 range) is generally 5- to 10-fold higher than that of endogenous genes (as low as 10−6) (Cosentino and Heddle 2000; Lambert et al. 2005).
342
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
This relatively high mutation background might have an impact on the sensitivity of these assays, as more induced mutations would be needed to detect an effect. This disadvantage is thought to be at least partially overcome by lesion accumulation after multiple administrations, as reporter genes are neutral, not transcribed, and therefore inefficiently repaired (Swiger et al. 2001). With some models, in addition to measuring the incidence of mutant cells, DNA sequencing is used to determine the mutation frequency. Mutations in endogeneous genes can only be measured if the mutants can be isolated and if the gene sequence is known (e.g., hprt and tk). The ability to isolate the mutants and to sequence them relatively easily is one advantage of transgenic models. This is particularly true for lacI (1080 bp) (Stiegler and Stillwell 1993), gpt (456 bp) (Masumura et al. 1999) and cII (294 bp) (Kohler et al. 1991a,b). Because of its length (3021 bp), molecular analysis of the lacZ gene is more complex and is generally done after genetic complementation analysis to determine in which of the three complementation regions (α, β, Ω) the mutation occurred (Douglas et al. 1994). Molecular analysis of mutations is not considered essential, but is useful for understanding the mechanism of mutation formation and for further evaluation of interindividual differences owing to potential jackspot mutations or clonal expansion (Heddle et al. 2000; Lambert et al. 2005). In this case, it can be necessary to sequence 10–25 mutants. Moreover, whenever possible, the mutational spectra can be analyzed in different genes in the same tissues (e.g., both lacZ and cII in the Muta™Mouse, and lacI and cII Big Blue® models). Assays of autosomal endogenous genes generally reveal both gene mutations and chromosome damage (e.g., aprt, tk, Dlb-1), while those using nonautosomal genes (e.g., hprt on the X chromosome) are limited to the detection of point mutations and small deletions (Tao et al. 1993b). In somatic cells, only one copy of the X chromosome gene is active. Males have only one copy of the X chromosome, and in females the second copy is inactivated. As a result, the loss of essential genes adjacent to the reporter gene cannot be compensated for by the homologous region present on the second copy of the gene. Thus, large deletions and chromosome rearrangements that impact adjacent essential genes are generally lethal for cells. The mouse (coat) spot test uses F1 animals to detect point mutations at the c locus or recombination between the c and p genes induced in melanoblasts in utero. The mouse (eye) spot test detects eye spots resulting from induction of deletions at the p locus in utero, in precursor cells of retinal pigmented epithelial cells. This assay is considered useful for specifically detecting deletions in vivo, as well as for mechanistic studies and research purposes. In transgenic models the large deletions that inactivate essential phage sequences (e.g., cos sites at both ends of the bacteriophage lambda required for phage DNA packaging) and the large deletions or insertions that strongly affect phage size prevent phage packaging and reporter gene recovery. Thus, transgenic models using a lambda phage as shuttle vector (e.g., Muta™Mouse and Big Blue®) are only able to reveal point mutations and small deletions (Heddle et al. 2000). The lacZ plasmid and gpt delta (spi) models do not have this disadvantage, and were designed to detect large deletions (Hayashi et al. 2003). The mutants in the lacZ plasmid mouse model result from both point mutations and large deletions (Gossen
12.4. IN VIVO GENOTOXICITY ASSAYS USED MAINLY AS COMPLEMENTARY OR FOLLOW-UP TESTS
343
and Vijg 1993; Gossen et al. 1995; Dolle et al. 1996, 1999). In the gpt delta mouse and rat models (Hayashi et al. 2003; Nohmi et al. 2000; Nohmi and Masumura 2004), two different selection methods are used in parallel to detect (a) point mutations by means of 6-thioguanine selection and (b) large deletions by means of spi selection (Gunther et al. 1993). Another important aspect of in vivo gene mutation assays is the number of animals used per group. Spot tests are seldom used, partly because they require (a) large numbers of treated animals in each group (up to 50 dams) and (b) the observation of up to 300 offspring. The use of such a large number of animals, when other alternatives exist, raises clear ethical issues. The other gene mutation assays, including transgenic models, generally require no more than 5–10 animals per group (Heddle et al. 2000; Lambert et al. 2005). A collaborative study has shown the reproducibility of data obtained with the Muta™Mouse and BigBlue® models across laboratories. The same study showed that, despite the multiple technical steps (DNA isolation, phage packaging, bacterial infection and mutant selection), the principal source of variability in these assays is inter-animal variability, but that 5–10 animals per group are sufficient (Piegorsch et al. 1995, 1997). It is nevertheless strongly recommended to use a block design protocol and to make sure that all samples are handled in parallel at the different steps of the analysis (Piegorsch et al. 1995, 1997; Heddle et al. 2000). Only a few assays (e.g., hprt and pigA) can be done with readily available wild-type animals. Recently published data identify the endogenous gene pigA as a potentially useful model gene for the detection of mutations in peripheral blood erythrocytes of any wild-type species, the gene being conserved across species (Bryce et al. 2008). Moreover, the analysis requires only a small volume of blood and does not require in vitro mitogen stimulation for mutant selection. Therefore, once fully validated, this assay could easily be integrated in all toxicology studies. The other models (e.g., both spot tests, aprt, tk, and Dlb-1 assays, and transgenic models) can only be performed with a specific strain of mouse or rat. Some animal strains are commercially available but quite expensive (Muta™Mouse and BigBlue®), whereas others are more difficult to purchase and are mainly used for research purpose. Other transgenic models have been developed for research purposes and are not described in this chapter. The majority of in vivo gene mutation data have been obtained with transgenic models, and especially with Muta™Mouse and Big Blue® (about 80%). In a review issued in 2005, Lambert and colleagues indicated that 165 agents have been evaluated in transgenic models (Lambert et al. 2005). In a recent update (OECD 2009), the same authors analyzed data for 228 different types of exposure (chemical, radiation, diet, infectious agents and mixtures). Among the 165 agents described in the first paper, 105 have already been evaluated in carcinogenicity studies (92 carcinogens and 13 noncarcinogens). In the updated report, 141 agents have been evaluated in 2-year bioassays (118 carcinogens and 23 noncarcinogens). It should be noted that the vast majority of the gene mutation results were obtained after short-term treatment and that the recommended optimal protocol was not applied (i.e., 28 days of treatment and sampling 3 and/or 28 days later). Nevertheless, for the 105 compounds analyzed in 2005, the concordance with carcinogenicity data is
344
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
77% (same value in the recent update), with 78% sensitivity and 69% specificity. Transgenic in vivo gene mutation assays show a slightly stronger correlation with the in vivo UDS and comet assays (about 85%), than with in vivo chromosome damage tests (about 70–80%). They are therefore a useful complement to in vivo chromosome damage tests (i.e., bone marrow micronucleus and chromosome aberration tests), in case of conflicting results. At least 70% of carcinogen compounds considered to induce tumors through a genotoxic mechanism, and previously found to be positive in in vitro genotoxicity assays and negative in in vivo chromosome damage tests conducted with bone marrow, are properly detected in gene mutation assays (Lambert et al. 2005; Kirkland and Speit 2008). Moreover, these models are particularly appropriate for site-of-contact effects (Dean et al. 1999). It was recently recommended to use gene mutation assays to evaluate the effect of low doses (Moore et al. 2008) and to better assess the shape of the dose–response curve (i.e., the existence of a threshold).
12.5.
CONCLUSION AND PERSPECTIVES
In vivo genotoxicity assays provide an integrated and pertinent approach for evaluating genetic changes. Numerous in vivo assays have been developed and continuously improved during the past 30–40 years in order to detect different genotoxicity endpoints in different tissues. However, no currently available in vivo assays are able to detect all genotoxic carcinogens. Some assays are used as markers of exposure— to detect the ability of the compound to interact with DNA in tissues—while other assays represent markers of effect, able to reveal stable genetic changes. The latter are considered more relevant to risk assessment. In the weight-of-evidence approach, more weight is generally given to positive results obtained in in vivo assays than in in vitro assays. In this context it is important to select in vivo assays that accurately identify genotoxic carcinogens, with a low rate of false-negative results. More than one in vivo endpoint and/or tissue might be necessary to reach an appropriate level of sensitivity (e.g., combination of the in vivo bone marrow micronucleus test and the liver comet assay). In order to optimize the use of animals and to respect the 3R’s engagement, the choice of assays and the data interpretation should take into account all available information (e.g., physicochemical properties, metabolism, potential pharmacological and toxicological activities, in vitro data and specificities related to the tissue, gender and/or species). They can also be integrated in studies conducted for other purposes, such as organ toxicity. Moreover, models that can be applied both in vitro and in vivo (e.g., micronucleus test and comet assay, and more recently gene mutations in the lacZ reporter gene), as well as endpoints/tissues that can be evaluated in both animals and humans (e.g., comet assay, DNA adducts, micronucleus in peripheral blood reticulocytes and chromosome aberrations in blood lymphocytes, and, maybe soon, Pig-a gene mutations), contribute to a better understanding of (a) conflicting results and (b) extrapolation to humans. A mode-of-action approach is being developed that uses all available information to understand carcinogenicity mechanisms and to identify for a given compound the key events involved in this multistage process, in order to better evaluate the risk for humans. Even if
REFERENCES
345
genetic changes are necessary for tumor induction by genotoxic carcinogens, they are often not sufficient; other mechanisms such as cell proliferation could be the key events and limiting factors for tumor development. In this context, the development of “omic” techniques (e.g., proteomics, toxicogenomics, etc.) might contribute to a more global analysis of mechanisms involved in early and late steps of carcinogenicity, as well as to the identification of new markers. In vivo genotoxicity studies are also being conducted to better define the impact of low doses and to determine if a threshold dose may also exist for genotoxic carcinogens. In vivo genotoxicity assays, despite their limitations and the need to consider the results together with all other available information, remain crucial for genotoxicity and carcinogenicity risk assessment.
REFERENCES Aardema, M. J., Albertini, S., Arni, P., Henderson, L. M., Kirsch-Volders, M., Mackay, J. M., Sarrif, A. M., Stringer, D. A., and Taalman, R. D. (1998). Aneuploidy: A report of an ECETOC task force. Mutat Res 410, 3–79. Adler, I. D. (1984). Cytogenetic tests in mammals. In Mutagenicity Testing: A Practical Approach, Venitt, S., and Parry, J. M., eds., IRL Press, Oxford, pp. 275–306. Aidoo, A., Lyn-Cook, L. E., Mittelstaedt, R. A., Heflich, R. H., and Casciano, D. A. (1991). Induction of 6-thioguanine-resistant lymphocytes in Fischer 344 rats following in vivo exposure to N-ethyl-Nnitrosourea and cyclophosphamide. Environ Mol Mutagen 17, 141–151. Allen, J. W., Shuler, C. F., Mendes, R. W., and Latt, S. A. (1977). A simplified technique for in vivo analysis of sister-chromatid exchanges using 5-bromodeoxyuridine tablets. Cytogenet Cell Genet 18, 231–237. Amphlett, N. W., Mitchell, I. D., Rees, R. W., and Haynes, G. A. (1996). A proposal for a two-dose/ single-sample in vivo/in vitro rat hepatocyte unscheduled DNA synthesis assay. Mutagenesis 11, 19–26. Anderson, D., Yu, T. W., and McGregor, D. B. (1998). Comet assay responses as indicators of carcinogen exposure. Mutagenesis 13, 539–555. Asano, N., Torous, D. K., Tometsko, C. R., Dertinger, S. D., Morita, T., and Hayashi, M. (2006). Practical threshold for micronucleated reticulocyte induction observed for low doses of mitomycin C, Ara-C and colchicine. Mutagenesis 21, 15–20. Ashby, J., Lefevre, P. A., Burlinson, B., and Penman, M. G. (1985). An assessment of the in vivo rat hepatocyte DNA-repair assay. Mutat Res 156, 1–18. Baird, W. M. (1979). The use of radioactive carcinogens to detect DNA modification. In Chemical Carcinogens and DNA, Grover, P. L., ed., CRC Press, Boca Raton, FL, pp. 59–83. Baute, J., and Depicker, A. (2008). Base excision repair and its role in maintaining genome stability. Crit Rev Biochem Mol Biol 43, 239–276. Beland, F. A., Churchwell, M. I., Von Tungeln, L. S., Chen, S., Fu, P. P., Culp, S. J., Schoket, B., Gyorffy, E., Minarovits, J., Poirier, M. C., Bowman, E. D., Weston, A., and Doerge, D. R. (2005). Highperformance liquid chromatography electrospray ionization tandem mass spectrometry for the detection and quantitation of benzo[a]pyrene–DNA adducts. Chem Res Toxicol 18, 1306–1315. Benning, V., Brault, D., Duvinage, C., Thybaud, V., and Melcion, C. (1994). Validation of the in vivo CD1 mouse splenocyte micronucleus test. Mutagenesis 9, 199–204. Benning, V., Depasse, F., Melcion, C., and Cordier, A. (1992). Detection of micronuclei after exposure to mitomycin C, cyclophosphamide and diethylnitrosamine by the in vivo micronucleus test in mouse splenocytes. Mutat Res 280, 137–142. Bharadwaj, R., and Yu, H. (2004). The spindle checkpoint, aneuploidy, and cancer. Oncogene 23, 2016–2027.
346
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
Bishop, A. J., Kosaras, B., Sidman, R. L., and Schiestl, R. H. (2000). Benzo(a)pyrene and X-rays induce reversions of the pink-eyed unstable mutation in the retinal pigment epithelium of mice. Mutat Res 457, 31–40. Bishop, J. B., and Kodell, R. L. (1980). The heritable translocation assay: Its relationship to assessment of genetic risk for future generations. Teratog Carcinog Mutagen 1, 305–332. Blakey, D. H., Douglas, G. R., Huang, K. C., and Winter, H. J. (1995). Cytogenetic mapping of lambda gt10 lacZ sequences in the transgenic mouse strain 40.6 (Muta Mouse). Mutagenesis 10, 145–148. Böcker, W., Rolf, W., Bauch, T., Müller, W.-U., and Streffer, C. (1999). Automated comet assay analysis. Cytometry 35, 134–144. Bradley, M. O., Hsu, I. C., and Harris, C. C. (1979). Relationship between sister chromatid exchange and mutagenicity, toxicity and DNA damage. Nature 282, 318–320. Braithwaite, I., and Ashby, J. (1988). A non-invasive micronucleus assay in the rat liver. Mutat Res 203, 23–32. Brambilla, G., and Martelli, A. (2004). Failure of the standard battery of short-term tests in detecting some rodent and human genotoxic carcinogens. Toxicology 196, 1–19. Brendler-Schwaab, S., Hartmann, A., Pfuhler, S., and Speit, G. (2005). The in vivo comet assay: use and status in genotoxicity testing. Mutagenesis 20, 245–254. Brendler-Schwaab, S. Y., Schmezer, P., Liegibel, U., Weber, S., Michalek, K., Tompa, A., and PoolZobel, B. L. (1994). Cells of different tissues for in vitro and in vivo studies in toxicology: Compilation of isolation methods. Toxicology in Vitro 8, 1285–1302. Brown, K., Dingley, K. H., and Turteltaub, K. W. (2005). Accelerator mass spectrometry for biomedical research. Methods Enzymol 402, 423–443. Bryce, S. M., Bemis, J. C., and Dertinger, S. D. (2008). In vivo mutation assay based on the endogenous Pig-a locus. Environ Mol Mutagen 49, 256–264. Burlinson, B. (1989). An in vivo unscheduled DNA synthesis (UDS) assay in the rat gastric mucosa: Preliminary development. Carcinogenesis 10, 1425–1428. Burlinson, B., Tice, R. R., Speit, G., Agurell, E., Brendler-Schwaab, S. Y., Collins, A. R., Escobar, P., Honma, M., Kumaravel, T. S., Nakajima, M., Sasaki, Y. F., Thybaud, V., Uno, Y., Vasquez, M., and Hartmann, A. (2007). Fourth International Workgroup on Genotoxicity testing: Results of the in vivo Comet assay workgroup. Mutat Res 627, 31–35. Buss, P., Caviezel, M., and Lutz, W. K. (1990). Linear dose–response relationship for DNA adducts in rat liver from chronic exposure to aflatoxin B1. Carcinogenesis 11, 2133–2135. Butterworth, B. E., Ashby, J., Bermudez, E., Casciano, D., Mirsalis, J., Probst, G., and Williams, G. (1987). A protocol and guide for the in vivo rat hepatocyte DNA-repair assay. Mutat Res 189, 123–133. Cammerer, Z., Elhajouji, A., Kirsch-Volders, M., and Suter, W. (2007a). Comparison of the peripheral blood micronucleus test using flow cytometry in rat and mouse exposed to aneugens after single-dose applications. Mutagenesis 22, 129–134. Cammerer, Z., Elhajouji, A., and Suter, W. (2007b). In vivo micronucleus test with flow cytometry after acute and chronic exposures of rats to chemicals. Mutat Res 626, 26–33. Casciano, D. A., Aidoo, A., Chen, T., Mittelstaedt, R. A., Manjanatha, M. G., and Heflich, R. H. (1999). Hprt mutant frequency and molecular analysis of Hprt mutations in rats treated with mutagenic carcinogens. Mutat Res 431, 389–395. Chi, Y. H., and Jeang, K. T. (2007). Aneuploidy and cancer. J Cell Biochem 102, 531–538. Chiarelli, M. P., and Jackson, O. L., Jr. (1992). Mass spectrometry for the analysis of carcinogen–DNA adducts. Mass Spectrom Rev 11, 447–493. Cimino, M. C. (2006). Comparative overview of current international strategies and guidelines for genetic toxicology testing for regulatory purposes. Environ Mol Mutagen 47, 362–390. Cliet, I., Fournier, E., Melcion, C., and Cordier, A. (1989). In vivo micronucleus test using mouse hepatocytes. Mutat Res 216, 321–326. Collins, A. R. (2002). The comet assay. Principles, applications, and limitations. Methods Mol Biol 203, 163–177. Collins, A. R., Duthie, S. J., and Dobson, V. L. (1993). Direct enzymic detection of endogenous oxidative base damage in human lymphocyte DNA. Carcinogenesis 14, 1733–1735.
REFERENCES
347
Collins, A. R., Oscoz, A. A., Brunborg, G., Gaivao, I., Giovannelli, L., Kruszewski, M., Smith, C. C., and Stetina, R. (2008). The comet assay: Topical issues. Mutagenesis 23, 143–151. Cosentino, L., and Heddle, J. A. (1995). The induction of dominant somatic mutations at the Dlb-1 locus. Mutat Res 346, 115–119. Cosentino, L., and Heddle, J. A. (2000). Differential mutation of transgenic and endogenous loci in vivo. Mutat Res 454, 1–10. CSGMT (1990). Single versus multiple dosing in the micronucleus test: the summary of the fourth collaborative study by CSGMT/JEMS.MMS. Collaborative Study Group for the Micronucleus Test, the Mammalian Mutagenesis Study Group of the Environmental Mutagen Society, Japan (CSGMT/JEMS. MMS). Mutat Res 234, 205–222. Dean, S. W., Brooks, T. M., Burlinson, B., Mirsalis, J., Myhr, B., Recio, L., and Thybaud, V. (1999). Transgenic mouse mutation assay systems can play an important role in regulatory mutagenicity testing in vivo for the detection of site-of-contact mutagens. Mutagenesis 14, 141–151. Dean, S. W., and Myhr, B. (1994). Measurement of gene mutation in vivo using Muta Mouse and positive selection for lacZ- phage. Mutagenesis 9, 183–185. Dearfield, K. L., and Moore, M. M. (2005). Use of genetic toxicology information for risk assessment. Environ Mol Mutagen 46, 236–445. Dehon, G., Catoire, L., Duez, P., Bogaerts, P., and Dubois, J. (2008). Validation of an automatic comet assay analysis system integrating the curve fitting of combined comet intensity profiles. Mutat Res 650, 87–95. Den Engelse, L., van Benthem, J., and Scherer, E. (1990). Immunocytochemical analysis of in vivo DNA modification. Mutat Res 233, 265–287. Dertinger, S. D., Camphausen, K., Macgregor, J. T., Bishop, M. E., Torous, D. K., Avlasevich, S., Cairns, S., Tometsko, C. R., Menard, C., Muanza, T., Chen, Y., Miller, R. K., Cederbrant, K., Sandelin, K., Ponten, I., and Bolcsfoldi, G. (2004). Three-color labeling method for flow cytometric measurement of cytogenetic damage in rodent and human blood. Environ Mol Mutagen 44, 427–435. Dertinger, S. D., Torous, D. K., Hall, N. E., Murante, F. G., Gleason, S. E., Miller, R. K., and Tometsko, C. R. (2002). Enumeration of micronucleated CD71-positive human reticulocytes with a single-laser flow cytometer. Mutat Res 515, 3–14. Dertinger, S. D., Torous, D. K., and Tometsko, K. R. (1996). Simple and reliable enumeration of micronucleated reticulocytes with a single-laser flow cytometer. Mutat Res 371, 283–292. Dertinger, S. D., Torous, D. K., and Tometsko, K. R. (1997). Flow cytometric analysis of micronucleated reticulocytes in mouse bone marrow. Mutat Res 390, 257–262. Deubel, W., Bassukas, I. D., Schlereth, W., Lorenz, R., and Hempel, K. (1996). Age dependent selection against HPRT deficient T lymphocytes in the HPRT+/− heterozygous mouse. Mutat Res 351, 67–77. Dingley, K. H., Roberts, M. L., Velsko, C. A., and Turteltaub, K. W. (1998). Attomole detection of 3H in biological samples using accelerator mass spectrometry: Application in low-dose, dual-isotope tracer studies in conjunction with 14C accelerator mass spectrometry. Chem Res Toxicol 11, 1217–1222. Dingley, K. H., Ubick, E. A., Vogel, J. S., and Haack, K. W. (2005). DNA isolation and sample preparation for quantification of adduct levels by accelerator mass spectrometry. Methods Mol Biol 291, 21–27. Dipple, A. (1995). DNA adducts of chemical carcinogens. Carcinogenesis 16, 437–441. Dobrovolsky, V. N., Casciano, D. A., and Heflich, R. H. (1999a). Tk+/− mouse model for detecting in vivo mutation in an endogenous, autosomal gene. Mutat Res 423, 125–136. Dobrovolsky, V. N., Chen, T., and Heflich, R. H. (1999b). Molecular analysis of in vivo mutations induced by N-ethyl-N-nitrosourea in the autosomal Tk and the X-linked Hprt genes of mouse lymphocytes. Environ Mol Mutagen 34, 30–38. Dobrovolsky, V. N., Shaddock, J. G., and Heflich, R. H. (2005). Analysis of in vivo mutation in the hprt and tk genes of mouse lymphocytes. Methods Mol Biol 291, 133–144. Dolle, M. E., Martus, H. J., Gossen, J. A., Boerrigter, M. E., and Vijg, J. (1996). Evaluation of a plasmidbased transgenic mouse model for detecting in vivo mutations. Mutagenesis 11, 111–118. Dolle, M. E., Snyder, W. K., van Orsouw, N. J., and Vijg, J. (1999). Background mutations and polymorphisms in lacZ-plasmid transgenic mice. Environ Mol Mutagen 34, 112–120.
348
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
Douglas, G. R., Gingerich, J. D., Gossen, J. A., and Bartlett, S. A. (1994). Sequence spectra of spontaneous lacZ gene mutations in transgenic mouse somatic and germline tissues. Mutagenesis 9, 451–458. Dycaico, M. J., Provost, G. S., Kretz, P. L., Ransom, S. L., Moores, J. C., and Short, J. M. (1994). The use of shuttle vectors for mutation analysis in transgenic mice and rats. Mutat Res 307, 461–478. Eide, I., Zhao, C., Kumar, R., Hemminki, K., Wu, K., and Swenberg, J. A. (1999). Comparison of (32) P-postlabeling and high-resolution GC/MS in quantifying N7-(2-hydroxyethyl)guanine adducts. Chem Res Toxicol 12, 979–984. Fahrig, R. (1975). A mammalian spot test: induction of genetic alterations in pigment cells of mouse embryos with x-rays and chemical mutagens. Mol Gen Genet 138, 309–314. Fahrig, R. (1995). The mouse spot test. In Environmental Mutagenesis, Phillips, D. H., and Venitt, S., eds., Bios Scientific Publishers, Oxford, pp. 180–199. Fahrig, R., and Neuhauser-Klaus, A. (1985). Similar pigmentation characteristics in the specific-locus and the mammalian spot test. A way to distinguish between induced mutation and recombination. J Hered 76, 421–426. Fairbairn, D. W., Olive, P. L., and O’Neill, K. L. (1995). The comet assay: A comprehensive review. Mutat Res 339, 37–59. Fairbairn, D. W., Walburger, D. K., Fairbairn, J. J., and O’Neill, K. L. (1996). Key morphologic changes and DNA strand breaks in human lymphoid cells: Discriminating apoptosis from necrosis. Scanning 18, 407–416. Farmer, P. B. (2004a). DNA and protein adducts as markers of genotoxicity. Toxicol Lett 149, 3–9. Farmer, P. B. (2004b). Exposure biomarkers for the study of toxicological impact on carcinogenic processes. IARC Sci Publ, 71–90. Farmer, P. B., Brown, K., Tompkins, E., Emms, V. L., Jones, D. J., Singh, R., and Phillips, D. H. (2005). DNA adducts: mass spectrometry methods and future prospects. Toxicol Appl Pharmacol 207, 293–301. Farmer, P. B., and Singh, R. (2008). Use of DNA adducts to identify human health risk from exposure to hazardous environmental pollutants: The increasing role of mass spectrometry in assessing biologically effective doses of genotoxic carcinogens. Mutat Res 659, 68–76. Fousteri, M., and Mullenders, L. H. (2008). Transcription-coupled nucleotide excision repair in mammalian cells: Molecular mechanisms and biological effects. Cell Res 18, 73–84. Frieauff, W., Hartmann, A., and Suter, W. (2001). Automatic analysis of slides processed in the Comet assay. Mutagenesis 16, 133–137. Frieauff, W., and Romagna, F. (1994). Technical aspects of automatic micronucleus analysis in rodent bone marrow assays. Cell Biol Toxicol 10, 283–289. Furihata, C., and Matsushima, T. (1987). Use of in vivo/in vitro unscheduled DNA synthesis for identification of organ-specific carcinogens. Crit Rev Toxicol 17, 245–277. Furihata, C., Yamawaki, Y., Jin, S. S., Moriya, H., Kodama, K., Matsushima, T., Ishikawa, T., Takayama, S., and Nakadate, M. (1984). Induction of unscheduled DNA synthesis in rat stomach mucosa by glandular stomach carcinogens. J Natl Cancer Inst 72, 1327–1334. Gamboa da Costa, G., Marques, M. M., Beland, F. A., Freeman, J. P., Churchwell, M. I., and Doerge, D. R. (2003). Quantification of tamoxifen DNA adducts using on-line sample preparation and HPLCelectrospray ionization tandem mass spectrometry. Chem Res Toxicol 16, 357–366. Ganem, N. J., Storchova, Z., and Pellman, D. (2007). Tetraploidy, aneuploidy and cancer. Curr Opin Genet Dev 17, 157–162. Garner, R. C. (1998). The role of DNA adducts in chemical carcinogenesis. Mutat Res 402, 67–75. Gebhart, E. (1981). Sister chromatid exchange (SCE) and structural chromosome aberration in mutagenicity testing. Hum Genet 58, 235–254. Gocke, E., Ballantyne, M., Whitwell, J., and Müller, L. (2009). MNT and MutaMouse studies to define the in vivo dose response relations of the genotoxicity of EMS and ENU. Toxicol Lett 190, 286–297. Gocke, E., and Müller, L. (2009). In vivo studies in the mouse to define a threshold for the genotoxicity of EMS and ENU. Mutat Res 678, 101–107. Goldman, R., Day, B. W., Carver, T. A., Mauthe, R. J., Turteltaub, K. W., and Shields, P. G. (2000). Quantitation of benzo[a]pyrene–DNA adducts by postlabeling with 14C-acetic anhydride and accelerator mass spectrometry. Chem Biol Interact 126, 171–183.
REFERENCES
349
Gondo, Y., Gardner, J. M., Nakatsu, Y., Durham-Pierre, D., Deveau, S. A., Kuper, C., and Brilliant, M. H. (1993). High-frequency genetic reversion mediated by a DNA duplication: The mouse pink-eyed unstable mutation. Proc Natl Acad Sci USA 90, 297–301. Gossen, J. A., de Leeuw, W. J., Tan, C. H., Zwarthoff, E. C., Berends, F., Lohman, P. H., Knook, D. L., and Vijg, J. (1989). Efficient rescue of integrated shuttle vectors from transgenic mice: A model for studying mutations in vivo. Proc Natl Acad Sci USA 86, 7971–7975. Gossen, J. A., de Leeuw, W. J., Verwest, A., Lohman, P. H., and Vijg, J. (1991). High somatic mutation frequencies in a LacZ transgene integrated on the mouse X-chromosome. Mutat Res 250, 423–429. Gossen, J. A., Martus, H. J., Wei, J. Y., and Vijg, J. (1995). Spontaneous and X-ray-induced deletion mutations in a LacZ plasmid-based transgenic mouse model. Mutat Res 331, 89–97. Gossen, J. A., Molijn, A. C., Douglas, G. R., and Vijg, J. (1992). Application of galactose-sensitive E. coli strains as selective hosts for LacZ-plasmids. Nucleic Acids Res 20, 3254. Gossen, J. A., and Vijg, J. (1993). A selective system for lacZ-phage using a galactose-sensitive E. coli host. Biotechniques 14, 326, 330. Grawé, J., Abramsson-Zetterberg, L., and Zetterberg, G. (1998). Low dose effects of chemicals as assessed by the flow cytometric in vivo micronucleus assay. Mutat Res 405, 199–208. Gunther, E. J., Murray, N. E., and Glazer, P. M. (1993). High efficiency, restriction-deficient in vitro packaging extracts for bacteriophage lambda DNA using a new E. coli lysogen. Nucleic Acids Res 21, 3903–3904. Gupta, P. K., Sahota, A., Boyadjiev, S. A., Bye, S., O’Neill, J. P., Hunter, T. C., Albertini, R. J., and Tischfield, J. A. (1994). Analysis of in vivo somatic mutations at the APRT locus. Adv Exp Med Biol 370, 653–656. Gupta, R. C., Reddy, M. V., and Randerath, K. (1982). 32P-postlabeling analysis of nonradioactive aromatic carcinogen–DNA adducts. Carcinogenesis 3, 1081–1092. Hagmar, L., Stromberg, U., Tinnerberg, H., and Mikoczy, Z. (2001). The usefulness of cytogenetic biomarkers as intermediate endpoints in carcinogenesis. Int J Hyg Environ Health 204, 43–47. Hamada, S., Sutou, S., Morita, T., Wakata, A., Asanami, S., Hosoya, S., Ozawa, S., Kondo, K., Nakajima, M., Shimada, H., Osawa, K., Kondo, Y., Asano, N., Sato, S., Tamura, H., Yajima, N., Marshall, R., Moore, C., Blakey, D. H., Schechtman, L. M., Weaver, J. L., Torous, D. K., Proudlock, R., Ito, S., Namiki, C., and Hayashi, M. (2001). Evaluation of the rodent micronucleus assay by a 28-day treatment protocol: Summary of the 13th Collaborative Study by the Collaborative Study Group for the Micronucleus Test (CSGMT)/Environmental Mutagen Society of Japan (JEMS)-Mammalian Mutagenicity Study Group (MMS). Environ Mol Mutagen 37, 93–110. Hamilton, C. M., and Mirsalis, J. C. (1987). Factors that affect the sensitivity of the in vivo–in vitro hepatocyte DNA repair assay in the male rat. Mutat Res 189, 341–347. Hanahan, D., and Weinberg, R. A. (2000). The hallmarks of cancer. Cell 100, 57–70. Harper, S. B., Dertinger, S. D., Bishop, M. E., Lynch, A. M., Lorenzo, M., Saylor, M., and MacGregor, J. T. (2007). Flow cytometric analysis of micronuclei in peripheral blood reticulocytes III. An efficient method of monitoring chromosomal damage in the beagle dog. Toxicol Sci 100, 406–414. Hartmann, A., Agurell, E., Beevers, C., Brendler-Schwaab, S., Burlinson, B., Clay, P., Collins, A., Smith, A., Speit, G., Thybaud, V., and Tice, R. R. (2003). Recommendations for conducting the in vivo alkaline Comet assay. 4th International Comet Assay Workshop. Mutagenesis 18, 45–51. Hartmann, A., Schumacher, M., Plappert-Helbig, U., Lowe, P., Suter, W., and Mueller, L. (2004). Use of the alkaline in vivo comet assay for mechanistic genotoxicity investigations. Mutagenesis 19, 51–59. Hayashi, H., Kondo, H., Masumura, K., Shindo, Y., and Nohmi, T. (2003). Novel transgenic rat for in vivo genotoxicity assays using 6-thioguanine and Spi− selection. Environ Mol Mutagen 41, 253–259. Hayashi, M., MacGregor, J. T., Gatehouse, D. G., Adler, I. D., Blakey, D. H., Dertinger, S. D., Krishna, G., Morita, T., Russo, A., and Sutou, S. (2000). In vivo rodent erythrocyte micronucleus assay. II. Some aspects of protocol design including repeated treatments, integration with toxicity testing, and automated scoring. Environ Mol Mutagen 35, 234–252. Hayashi, M., MacGregor, J. T., Gatehouse, D. G., Blakey, D. H., Dertinger, S. D., Abramsson-Zetterberg, L., Krishna, G., Morita, T., Russo, A., Asano, N., Suzuki, H., Ohyama, W., and Gibson, D. (2007). In vivo erythrocyte micronucleus assay III. Validation and regulatory acceptance of automated scoring
350
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
and the use of rat peripheral blood reticulocytes, with discussion of non-hematopoietic target cells and a single dose-level limit test. Mutat Res 627, 10–30. Hayashi, M., Morita, T., Kodama, Y., Sofuni, T., and Ishidate, M., Jr. (1990). The micronucleus assay with mouse peripheral blood reticulocytes using acridine orange-coated slides. Mutat Res 245, 245–249. Hayashi, M., Tice, R. R., MacGregor, J. T., Anderson, D., Blakey, D. H., Kirsh-Volders, M., Oleson, F. B., Jr., Pacchierotti, F., Romagna, F., Shimada, H., et al. (1994). In vivo rodent erythrocyte micronucleus assay. Mutat Res 312, 293–304. Heddle, J. A. (1973). A rapid in vivo test for chromosomal damage. Mutat Res 18, 187–190. Heddle, J. A. (1999). Mutant manifestation: The time factor in somatic mutagenesis. Mutagenesis 14, 1–3. Heddle, J. A., Cimino, M. C., Hayashi, M., Romagna, F., Shelby, M. D., Tucker, J. D., Vanparys, P., and MacGregor, J. T. (1991). Micronuclei as an index of cytogenetic damage: Past, present, and future. Environ Mol Mutagen 18, 277–291. Heddle, J. A., Dean, S., Nohmi, T., Boerrigter, M., Casciano, D., Douglas, G. R., Glickman, B. W., Gorelick, N. J., Mirsalis, J. C., Martus, H. J., Skopek, T. R., Thybaud, V., Tindall, K. R., and Yajima, N. (2000). In vivo transgenic mutation assays. Environ Mol Mutagen 35, 253–259. Heddle, J. A., and Salamone, M. F. (1981). The micronucleus assay: In vivo. In Proceedings of the International Workshop on short term tests for chemical carcinogens, Stich, H., and San, R. H. C., eds., Springer-Verlag, New York, pp. 243–249. Hegde, M. L., Hazra, T. K., and Mitra, S. (2008). Early steps in the DNA base excision/single-strand interruption repair pathway in mammalian cells. Cell Res 18, 27–47. Helleday, T. (2003). Pathways for mitotic homologous recombination in mammalian cells. Mutat Res 532, 103–115. Hemminki, K., Koskinen, M., Rajaniemi, H., and Zhao, C. (2000). Dna adducts, mutations, and cancer 2000. Regul Toxicol Pharmacol 32, 264–275. Hotchkiss, C. E., Bishop, M. E., Dertinger, S. D., Slikker, W., Jr., Moore, M. M., and Macgregor, J. T. (2008). Flow cytometric analysis of micronuclei in peripheral blood reticulocytes IV: An index of chromosomal damage in the rhesus monkey (Macaca mulatta). Toxicol Sci 102, 352–358. Hsu, I. C., Poirier, M. C., Yuspa, S. H., Grunberger, D., Weinstein, I. B., Yolken, R. H., and Harris, C. C. (1981). Measurement of benzo(a)pyrene–DNA adducts by enzyme immunoassays and radioimmunoassay. Cancer Res 41, 1091–1095. Iarmarcovai, G., Botta, A., and Orsiere, T. (2006). Number of centromeric signals in micronuclei and mechanisms of aneuploidy. Toxicol Lett 166, 1–10. Igarashi, M., and Shimada, H. (1997). An improved method for the mouse liver micronucleus test. Mutat Res 391, 49–55. Jakubczak, J. L., Merlino, G., French, J. E., Müller, W. J., Paul, B., Adhya, S., and Garges, S. (1996). Analysis of genetic instability during mammary tumor progression using a novel selection-based assay for in vivo mutations in a bacteriophage lambda transgene target. Proc Natl Acad Sci USA 93, 9073–9078. Jones, I. M., Burkhart-Schultz, K., and Carrano, A. V. (1985). A method to quantify spontaneous and in vivo induced thioguanine-resistant mouse lymphocytes. Mutat Res 147, 97–105. Kasper, P., Uno, Y., Mauthe, R., Asano, N., Douglas, G., Matthews, E., Moore, M., Mueller, L., Nakajima, M., Singer, T., and Speit, G. (2007). Follow-up testing of rodent carcinogens not positive in the standard genotoxicity testing battery: IWGT workgroup report. Mutat Res 627, 106–116. King, M. T., Wild, D., Gocke, E., and Eckhardt, K. (1982). 5-Bromodeoxyuridine tablets with improved depot effect for analysis in vivo of sister-chromatid exchanges in bone-marrow and spermatogonial cells. Mutat Res 97, 117–129. Kirkland, D., and Speit, G. (2008). Evaluation of the ability of a battery of three in vitro genotoxicity tests to discriminate rodent carcinogens and non-carcinogens III. Appropriate follow-up testing in vivo. Mutat Res 654, 114–132. Kirsch-Volders, M., Vanhauwaert, A., De Boeck, M., and Decordier, I. (2002). Importance of detecting numerical versus structural chromosome aberrations. Mutat Res 504, 137–148. Kirsch-Volders, M., Vanhauwaert, A., Eichenlaub-Ritter, U., and Decordier, I. (2003). Indirect mechanisms of genotoxicity. Toxicol Lett 140–141, 63–74.
REFERENCES
351
Kissling, G. E., Dertinger, S. D., Hayashi, M., and MacGregor, J. T. (2007). Sensitivity of the erythrocyte micronucleus assay: Dependence on number of cells scored and inter-animal variability. Mutat Res 634, 235–240. Kligerman, A. D., Erexson, G. L., and Wilmer, J. L. (1984). Development of rodent peripheral blood lymphocyte culture systems to detect cytogenetic damage in vivo. Basic Life Sci 29(Pt B), 569–584. Koc, H., and Swenberg, J. A. (2002). Applications of mass spectrometry for quantitation of DNA adducts. J Chromatogr B Analyt Technol Biomed Life Sci 778, 323–343. Kohler, S. W., Provost, G. S., Fieck, A., Kretz, P. L., Bullock, W. O., Putman, D. L., Sorge, J. A., and Short, J. M. (1991a). Analysis of spontaneous and induced mutations in transgenic mice using a lambda ZAP/lacI shuttle vector. Environ Mol Mutagen 18, 316–321. Kohler, S. W., Provost, G. S., Fieck, A., Kretz, P. L., Bullock, W. O., Sorge, J. A., Putman, D. L., and Short, J. M. (1991b). Spectra of spontaneous and mutagen-induced mutations in the lacI gene in transgenic mice. Proc Natl Acad Sci USA 88, 7958–7962. Kohler, S. W., Provost, G. S., Kretz, P. L., Fieck, A., Sorge, J. A., and Short, J. M. (1990). The use of transgenic mice for short-term, in vivo mutagenicity testing. Genet Anal Tech Appl 7, 212–218. Kohn, K. W., Erickson, L. C., Ewig, R. A., and Friedman, C. A. (1976). Fractionation of DNA from mammalian cells by alkaline elution. Biochemistry 15, 4629–4637. Kohn, K. W., and Grimek-Ewig, R. A. (1973). Alkaline elution analysis, a new approach to the study of DNA single-strand interruptions in cells. Cancer Res 33, 1849–1853. Korenberg, J. R., and Freedlender, E. F. (1974). Giemsa technique for the detection of sister chromatid exchanges. Chromosoma 48, 355–360. Kriek, E., Den Engelse, L., Scherer, E., and Westra, J. G. (1984). Formation of DNA modifications by chemical carcinogens. Identification, localization and quantification. Biochim Biophys Acta 738, 181–201. Krishna, G., and Hayashi, M. (2000). In vivo rodent micronucleus assay: Protocol, conduct and data interpretation. Mutat Res 455, 155–166. Lambert, I. B., Singer, T. M., Boucher, S. E., and Douglas, G. R. (2005). Detailed review of transgenic rodent mutation assays. Mutat Res 590, 1–280. Latt, S. A. (1974). Sister chromatid exchanges, indices of human chromosome damage and repair: Detection by fluorescence and induction by mitomycin C. Proc Natl Acad Sci USA 71, 3162–3166. Latt, S. A. (1981). Sister chromatid exchange formation. Annu Rev Genet 15, 11–55. Latt, S. A., Allen, J., Bloom, S. E., Carrano, A., Falke, E., Kram, D., Schneider, E., Schreck, R., Tice, R., Whitfield, B., and Wolff, S. (1981). Sister-chromatid exchanges: A report of the GENE-TOX program. Mutat Res 87, 17–62. Latt, S. A., and Schreck, R. R. (1980). Sister chromatid exchange analysis. Am J Hum Genet 32, 297–313. Liou, S. H., Lung, J. C., Chen, Y. H., Yang, T., Hsieh, L. L., Chen, C. J., and Wu, T. N. (1999). Increased chromosome-type chromosome aberration frequencies as biomarkers of cancer risk in a blackfoot endemic area. Cancer Res 59, 1481–1484. Loury, D. J., Smith-Oliver, T., and Butterworth, B. E. (1987). Assessment of unscheduled and replicative DNA synthesis in rat kidney cells exposed in vitro or in vivo to unleaded gasoline. Toxicol Appl Pharmacol 87, 127–140. Lovell, D. P., Thomas, G., and Dubow, R. (1999). Issues related to the experimental design and subsequent statistical analysis of in vivo and in vitro comet studies. Teratog Carcinog Mutagen 19, 109–119. Lutz, W. K. (1979). In vivo covalent binding of organic chemicals to DNA as a quantitative indicator in the process of chemical carcinogenesis. Mutat Res 65, 289–356. Lutz, W. K. (1986). Quantitative evaluation of DNA binding data for risk estimation and for classification of direct and indirect carcinogens. J Cancer Res Clin Oncol 112, 85–91. MacGregor, J. T., Bishop, M. E., McNamee, J. P., Hayashi, M., Asano, N., Wakata, A., Nakajima, M., Saito, J., Aidoo, A., Moore, M. M., and Dertinger, S. D. (2006). Flow cytometric analysis of micronuclei in peripheral blood reticulocytes: II. An efficient method of monitoring chromosomal damage in the rat. Toxicol Sci 94, 92–107. MacGregor, J. T., Casciano, D., and Müller, L. (2000). Strategies and testing methods for identifying mutagenic risks. Mutat Res 455, 3–20.
352
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
MacGregor, J. T., Heddle, J. A., Hite, M., Margolin, B. H., Ramel, C., Salamone, M. F., Tice, R. R., and Wild, D. (1987). Guidelines for the conduct of micronucleus assays in mammalian bone marrow erythrocytes. Mutat Res 189, 103–112. MacGregor, J. T., Wehr, C. M., and Gould, D. H. (1980). Clastogen-induced micronuclei in peripheral blood erythrocytes: The basis of an improved micronucleus test. Environ Mutagen 2, 509–514. Mackay, J. M. (1995). Dose selection in in vivo genetic toxicology assays. Environ Mol Mutagen 25, 323–327. Madrigal-Bujaidar, E., and Sanchez-Sanchez, M. A. (1991). Sister-chromatid exchange analysis in vivo using different 5-bromo-2’-deoxyuridine-labeling systems. Mutat Res 262, 15–19. Martin, C. N., Lawley, P. D., Phillips, D. H., Venitt, S., and Waters, R. (1993). Measurement for covalent binding to DNA in vivo. In Supplementary Mutagenicity Tests: UKEMS Recommended Procedures, Kirkland, D. J., and Fox, M., eds., Cambridge University Press, Cambridge, pp. 78–104. Masumura, K., Matsui, M., Katoh, M., Horiya, N., Ueda, O., Tanabe, H., Yamada, M., Suzuki, H., Sofuni, T., and Nohmi, T. (1999). Spectra of gpt mutations in ethylnitrosourea-treated and untreated transgenic mice. Environ Mol Mutagen 34, 1–8. Mateuca, R., Lombaert, N., Aka, P. V., Decordier, I., and Kirsch-Volders, M. (2006). Chromosomal changes: Induction, detection methods and applicability in human biomonitoring. Biochimie 88, 1515–1531. Mauthe, R. J., Dingley, K. H., Leveson, S. H., Freeman, S. P., Turesky, R. J., Garner, R. C., and Turteltaub, K. W. (1999). Comparison of DNA–adduct and tissue-available dose levels of MeIQx in human and rodent colon following administration of a very low dose. Int J Cancer 80, 539–545. Mavournin, K. H., Blakey, D. H., Cimino, M. C., Salamone, M. F., and Heddle, J. A. (1990). The in vivo micronucleus assay in mammalian bone marrow and peripheral blood. A report of the U.S. Environmental Protection Agency Gene-Tox Program. Mutat Res 239, 29–80. McKelvey-Martin, V. J., Green, M. H., Schmezer, P., Pool-Zobel, B. L., De Meo, M. P., and Collins, A. (1993). The single cell gel electrophoresis assay (comet assay): A European review. Mutat Res 288, 47–63. McKinzie, P. B., Delongchamp, R. R., Heflich, R. H., and Parsons, B. L. (2001). Prospects for applying genotypic selection of somatic oncomutation to chemical risk assessment. Mutat Res 489, 47–78. McKinzie, P. B., and Parsons, B. L. (2002). Detection of rare K-ras codon 12 mutations using allelespecific competitive blocker PCR. Mutat Res 517, 209–220. Mientjes, E. J., van Delft, J. H. M., Op’t Hof, B. M., Gossen, J. A., Vijg, J., Lohman, P. H., and Baan, R. A. (1994). An improved selection method for lambda lacZ-phages based on galactose sensitivity. Transgenic Res 3, 67–69. Mirsalis, J. C., and Butterworth, B. E. (1980). Detection of unscheduled DNA synthesis in hepatocytes isolated from rats treated with genotoxic agents: An in vivo–in vitro assay for potential carcinogens and mutagens. Carcinogenesis 1, 621–625. Mirsalis, J. C., Tyson, C. K., and Butterworth, B. E. (1982). Detection of genotoxic carcinogens in the in vivo–in vitro hepatocyte DNA repair assay. Environ Mutagen 4, 553–562. Miyamae, Y., Yamamoto, M., Sasaki, Y. F., Kobayashi, H., Igarashi-Soga, M., Shimoi, K., and Hayashi, M. (1998). Evaluation of a tissue homogenization technique that isolates nuclei for the in vivo single cell gel electrophoresis (comet) assay: A collaborative study by five laboratories. Mutat Res 418, 131–140. Moore, M. M., Heflich, R. H., Haber, L. T., Allen, B. C., Shipp, A. M., and Kodell, R. L. (2008). Analysis of in vivo mutation data can inform cancer risk assessment. Regul Toxicol Pharmacol 51, 151–161. Mori, M., Kobayashi, H., Sugiyama, C., Katsumura, Y., and Furihata, C. (1999). Induction of unscheduled DNA synthesis in hairless mouse epidermis by skin carcinogens. J Toxicol Sci 24, 217–226. Morita, T., Asano, N., Awogi, T., Sasaki, Y. F., Sato, S., Shimada, H., Sutou, S., Suzuki, T., Wakata, A., Sofuni, T., and Hayashi, M. (1997). Evaluation of the rodent micronucleus assay in the screening of IARC carcinogens (groups 1, 2A and 2B) the summary report of the 6th collaborative study by CSGMT/JEMS MMS. Collaborative Study of the Micronucleus Group Test. Mammalian Mutagenicity Study Group. Mutat Res 389, 3–122. Müller, L., Blakey, D., Dearfield, K. L., Galloway, S., Guzzie, P., Hayashi, M., Kasper, P., Kirkland, D., MacGregor, J. T., Parry, J. M., Schechtman, L., Smith, A., Tanaka, N., Tweats, D., and Yamasaki, H.
REFERENCES
353
(2003). Strategy for genotoxicity testing and stratification of genotoxicity test results—Report on initial activities of the IWGT Expert Group. Mutat Res 540, 177–181. Müller, L., and Kasper, P. (2000). Human biological relevance and the use of threshold-arguments in regulatory genotoxicity assessment: Experience with pharmaceuticals. Mutat Res 464, 19–34. Müller, R., Adamkiewicz, J., and Rajewsky, M. F. (1982). Immunological detection and quantification of carcinogen-modified DNA components. IARC Sci Publ, 463–479. Müller, R., and Rajewsky, M. F. (1980). Immunological quantification by high-affinity antibodies of O6-ethyldeoxyguanosine in DNA exposed to N-ethyl-N-nitrosourea. Cancer Res 40, 887–896. Müller, R., and Rajewsky, M. F. (1981). Antibodies specific for DNA components structurally modified by chemical carcinogens. J Cancer Res Clin Oncol 102, 99–113. Natarajan, A. T., Rotteveel, A. H., van Pieterson, J., and Schliermann, M. G. (1986). Influence of incorporated 5-bromodeoxyuridine on the frequencies of spontaneous and induced sister-chromatid exchanges, detected by immunological methods. Mutat Res 163, 51–55. Nestmann, E. R., Bryant, D. W., and Carr, C. J. (1996). Toxicological significance of DNA adducts: Summary of discussions with an expert panel. Regul Toxicol Pharmacol 24, 9–18. Nishikawa, T., Haresaku, M., Adachi, K., Masuda, M., and Hayashi, M. (1999). Study of a rat skin in vivo micronucleus test: Data generated by mitomycin C and methyl methanesulfonate. Mutat Res 444, 159–166. Nishikawa, T., Haresaku, M., Fukushima, A., Nakamura, T., Adachi, K., Masuda, M., and Hayashi, M. (2002). Further evaluation of an in vivo micronucleus test on rat and mouse skin: Results with five skin carcinogens. Mutat Res 513, 93–102. Nohmi, T., Katoh, M., Suzuki, H., Matsui, M., Yamada, M., Watanabe, M., Suzuki, M., Horiya, N., Ueda, O., Shibuya, T., Ikeda, H., and Sofuni, T. (1996). A new transgenic mouse mutagenesis test system using Spi− and 6-thioguanine selections. Environ Mol Mutagen 28, 465–470. Nohmi, T., and Masumura, K. (2004). Gpt delta transgenic mouse: A novel approach for molecular dissection of deletion mutations in vivo. Adv Biophys 38, 97–121. Nohmi, T., Suzuki, M., Masumura, K., Yamada, M., Matsui, K., Ueda, O., Suzuki, H., Katoh, M., Ikeda, H., and Sofuni, T. (1999). Spi(−) selection: An efficient method to detect gamma-ray-induced deletions in transgenic mice. Environ Mol Mutagen 34, 9–15. Nohmi, T., Suzuki, T., and Masumura, K. (2000). Recent advances in the protocols of transgenic mouse mutation assays. Mutat Res 455, 191–215. Norbury, C. J., and Hickson, I. D. (2001). Cellular responses to DNA damage. Annu Rev Pharmacol Toxicol 41, 367–401. Norppa, H., Bonassi, S., Hansteen, I. L., Hagmar, L., Stromberg, U., Rossner, P., Boffetta, P., Lindholm, C., Gundy, S., Lazutka, J., Cebulska-Wasilewska, A., Fabianova, E., Sram, R. J., Knudsen, L. E., Barale, R., and Fucic, A. (2006). Chromosomal aberrations and SCEs as biomarkers of cancer risk. Mutat Res 600, 37–45. Obe, G., Pfeiffer, P., Savage, J. R., Johannes, C., Goedecke, W., Jeppesen, P., Natarajan, A. T., MartinezLopez, W., Folle, G. A., and Drets, M. E. (2002). Chromosomal aberrations: Formation, identification and distribution. Mutat Res 504, 17–36. OECD (1984). Genetic toxicology: Rodent dominant lethal test. OECD Guideline for the Testing of Chemicals 478, 1–6. OECD (1986a). Genetic toxicology: Mouse heritable translocation assay. OECD Guideline for Testing of Chemicals 485, 1–6. OECD (1986b). Genetic toxicology: Mouse spot test. OECD Guideline for Testing of Chemicals 484, 1–4. OECD (1997a). In vitro mammalian cell gene mutation test. OECD Guideline for Testing of Chemicals 476, 1–10. OECD (1997b). Mammalian bone marrow chromosome aberration test. OECD Guideline for Testing of Chemicals 475, 1–8. OECD (1997c). Mammalian erythrocyte micronucleus test. OECD Guideline for Testing of Chemicals 474, 1–10. OECD (1997d). Unscheduled DNA synthesis (UDS) test with mammalian liver cells in vivo. OECD Guideline for Testing of Chemicals 486, 1–8.
354
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
OECD (2009). Detailed review paper on transgenic rodent mutation assays. ENV/JM/MONO(2009)7 103, 1–281, http://www.olis.oecd.org/olis/2009doc.nsf/linkto/ENV-JM-MONO(2009)7. Ohyama, W., Gonda, M., Miyajima, H., Kondo, K., Noguchi, T., Yoshida, J., Hatakeyama, S., Watabe, E., Ueno, Y., Hayashi, M., and Tokumitsu, T. (2002). Collaborative validation study of the in vivo micronucleus test using mouse colonic epithelial cells. Mutat Res 518, 39–45. Olive, P. L. (2002). The comet assay. An overview of techniques. Methods Mol Biol 203, 179–194. Olive, P. L., Banath, J. P., and Durand, R. E. (1990a). Detection of etoposide resistance by measuring DNA damage in individual Chinese hamster cells. J Natl Cancer Inst 82, 779–783. Olive, P. L., Banath, J. P., and Durand, R. E. (1990b). Heterogeneity in radiation-induced DNA damage and repair in tumor and normal cells measured using the “comet” assay. Radiat Res 122, 86–94. Olive, P. L., and Durand, R. E. (2005). Heterogeneity in DNA damage using the comet assay. Cytometry A 66, 1–8. Olive, P. L., Frazer, G., and Banath, J. P. (1993). Radiation-induced apoptosis measured in TK6 human B lymphoblast cells using the comet assay. Radiat Res 136, 130–136. Olive, P. L., Wlodek, D., and Banath, J. P. (1991). DNA double-strand breaks measured in individual cells subjected to gel electrophoresis. Cancer Res 51, 4671–4676. Ostling, O., and Johanson, K. J. (1984). Microelectrophoretic study of radiation-induced DNA damages in individual mammalian cells. Biochem Biophys Res Commun 123, 291–298. Ottender, M., and Lutz, W. K. (1999). Correlation of DNA adduct levels with tumor incidence: carcinogenic potency of DNA adducts. Mutat Res 424, 237–247. Painter, R. B. (1980). A replication model for sister-chromatid exchange. Mutat Res 70, 337–341. Parsons, B. L., and Heflich, R. H. (1998). Detection of a mouse H-ras codon 61 mutation using a modified allele-specific competitive blocker PCR genotypic selection method. Mutagenesis 13, 581–588. Perry, P., and Evans, H. J. (1975). Cytological detection of mutagen–carcinogen exposure by sister chromatid exchange. Nature 258, 121–125. Perry, P., and Wolff, S. (1974). New Giemsa method for the differential staining of sister chromatids. Nature 251, 156–158. Phillips, D. H. (1997). Detection of DNA modifications by the 32P-postlabelling assay. Mutat Res 378, 1–12. Phillips, D. H., and Arlt, V. M. (2007). The 32P-postlabeling assay for DNA adducts. Nat Protoc 2, 2772–2781. Phillips, D. H., and Castegnaro, M. (1999). Standardization and validation of DNA adduct postlabelling methods: Report of interlaboratory trials and production of recommended protocols. Mutagenesis 14, 301–315. Phillips, D. H., Farmer, P. B., Beland, F. A., Nath, R. G., Poirier, M. C., Reddy, M. V., and Turteltaub, K. W. (2000). Methods of DNA adduct determination and their application to testing compounds for genotoxicity. Environ Mol Mutagen 35, 222–233. Phillips, D. H., Hewer, A., and Arlt, V. M. (2005). 32P-postlabeling analysis of DNA adducts. Methods Mol Biol 291, 3–12. Piegorsch, W. W., Lockhart, A. C., Carr, G. J., Margolin, B. H., Brooks, T., Douglas, G. R., Liegibel, U. M., Suzuki, T., Thybaud, V., van Delft, J. H., and Gorelick, N. J. (1997). Sources of variability in data from a positive selection lacZ transgenic mouse mutation assay: An interlaboratory study. Mutat Res 388, 249–289. Piegorsch, W. W., Margolin, B. H., Shelby, M. D., Johnson, A., French, J. E., Tennant, R. W., and Tindall, K. R. (1995). Study design and sample sizes for a lacI transgenic mouse mutation assay. Environ Mol Mutagen 25, 231–245. Pinkel, D., Thompson, L. H., Gray, J. W., and Vanderlaan, M. (1985). Measurement of sister chromatid exchanges at very low bromodeoxyuridine substitution levels using a monoclonal antibody in Chinese hamster ovary cells. Cancer Res 45, 5795–5798. Poirier, M. C. (1981). Antibodies to carcinogen–DNA adducts. J Natl Cancer Inst 67, 515–519. Poirier, M. C. (1993). Antisera specific for carcinogen–DNA adducts and carcinogen-modified DNA: Applications for detection of xenobiotics in biological samples. Mutat Res 288, 31–38. Poirier, M. C. (2004). Chemical-induced DNA damage and human cancer risk. Nat Rev Cancer 4, 630–637.
REFERENCES
355
Poirier, M. C., and Beland, F. A. (1992). DNA adduct measurements and tumor incidence during chronic carcinogen exposure in animal models: Implications for DNA adduct-based human cancer risk assessment. Chem Res Toxicol 5, 749–755. Poirier, M. C., Santella, R. M., and Weston, A. (2000). Carcinogen macromolecular adducts and their measurement. Carcinogenesis 21, 353–359. Pratt, I. S., and Barron, T. (2003). Regulatory recognition of indirect genotoxicity mechanisms in the European Union. Toxicol Lett 140–141, 53–62. Preston, R. J., Dean, B. J., Galloway, S., Holden, H., McFee, A. F., and Shelby, M. (1987). Mammalian in vivo cytogenetic assays. Analysis of chromosome aberrations in bone marrow cells. Mutat Res 189, 157–165. Randerath, K., and Randerath, E. (1994). 32P-postlabeling methods for DNA adduct detection: Overview and critical evaluation. Drug Metab Rev 26, 67–85. Randerath, K., Reddy, M. V., and Gupta, R. C. (1981). 32P-labeling test for DNA damage. Proc Natl Acad Sci USA 78, 6126–6129. Rapp, A., Hausmann, M., and Greulich, K. O. (2005). The Comet-FISH technique: A tool for detection of specific DNA damage and repair. Methods Mol Biol 291, 107–119. Reddy, M. V. (2000). Methods for testing compounds for DNA adduct formation. Regul Toxicol Pharmacol 32, 256–263. Reddy, M. V., Gupta, R. C., Randerath, E., and Randerath, K. (1984). 32P-postlabeling test for covalent DNA binding of chemicals in vivo: Application to a variety of aromatic carcinogens and methylating agents. Carcinogenesis 5, 231–243. Reddy, M. V., and Randerath, K. (1986). Nuclease P1-mediated enhancement of sensitivity of 32P-postlabeling test for structurally diverse DNA adducts. Carcinogenesis 7, 1543–1551. Ren, L., Yang, J. P., and Zhang, H. X. (1991). Use of the cytokinesis-block micronucleus method in mouse splenocytes. Mutat Res 262, 119–124. Richold, M., Chandley, A., Ashby, J., Gatehouse, D. G., Bootman, J., and Henderson, L. (1990). In vivo cytogenetic assays. In Basic Mutagenicity Tests, UKEMS Recommended Procedures, UKEMS Subcommittee on Guidelines for Mutagenicity Testing, Report part I revised, Kirkland, D. J., ed., Cambridge University Press, Cambridge, pp. 115–141. Rojas, E., Lopez, M. C., and Valverde, M. (1999). Single cell gel electrophoresis assay: Methodology and applications. J Chromatogr B Biomed Sci Appl 722, 225–254. Romagna, F., and Staniforth, C. D. (1989). The automated bone marrow micronucleus test. Mutat Res 213, 91–104. Rosenkranz, H. S., and Cunningham, A. R. (2000). The high production volume chemical challenge program: The relevance of the in vivo micronucleus assay. Regul Toxicol Pharmacol 31, 182–189. Rundle, A. (2006). Carcinogen–DNA adducts as a biomarker for cancer risk. Mutat Res 600, 23–36. Russell, L. B. (1977). Validation of the in vivo somatic mutation method in the mouse as a prescreen for germinal point mutations. Arch Toxicol 38, 75–85. Russell, L. B., and Major, M. H. (1957). Radiation-induced presumed somatic mutations in the house mouse. Genetics 42, 161–175. Russell, L. B., and Matter, B. E. (1980). Whole-mammal mutagenicity tests: Evaluation of five methods. Mutat Res 75, 279–302. Russo, A. (2000). In vivo cytogenetics: Mammalian germ cells. Mutat Res 455, 167–189. Salamone, M., Heddle, J., Stuart, E., and Katz, M. (1980). Towards an improved micronucleus test: studies on 3 model agents, mitomycin C, cyclophosphamide and dimethylbenzanthracene. Mutat Res 74, 347–356. Santella, R. M. (1999). Immunological methods for detection of carcinogen–DNA damage in humans. Cancer Epidemiol Biomarkers Prev 8, 733–739. Santos, S. J., Singh, N. P., and Natarajan, A. T. (1997). Fluorescence in situ hybridization with comets. Exp Cell Res 232, 407–411. Sasaki, Y. F., Izumiyama, F., Nishidate, E., Matsusaka, N., and Tsuda, S. (1997a). Detection of rodent liver carcinogen genotoxicity by the alkaline single-cell gel electrophoresis (comet) assay in multiple mouse organs (liver, lung, spleen, kidney, and bone marrow). Mutat Res 391, 201–214.
356
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
Sasaki, Y. F., Izumiyama, F., Nishidate, E., Ohta, T., Ono, T., Matsusaka, N., and Tsuda, S. (1997b). Simple detection of in vivo genotoxicity of pyrimethamine in rodents by the modified alkaline singlecell gel electrophoresis assay. Mutat Res 392, 251–259. Sasaki, Y. F., Nishidate, E., Izumiyama, F., Matsusaka, N., and Tsuda, S. (1997c). Simple detection of chemical mutagens by the alkaline single-cell gel electrophoresis (Comet) assay in multiple mouse organs (liver, lung, spleen, kidney, and bone marrow). Mutat Res 391, 215–231. Sasaki, Y. F., Sekihashi, K., Izumiyama, F., Nishidate, E., Saga, A., Ishida, K., and Tsuda, S. (2000). The comet assay with multiple mouse organs: Comparison of comet assay results and carcinogenicity with 208 chemicals selected from the IARC monographs and U.S. NTP Carcinogenicity Database. Crit Rev Toxicol 30, 629–799. Sasaki, Y. F., Tsuda, S., Izumiyama, F., and Nishidate, E. (1997d). Detection of chemically induced DNA lesions in multiple mouse organs (liver, lung, spleen, kidney, and bone marrow) using the alkaline single cell gel electrophoresis (comet) assay. Mutat Res 388, 33–44. Sauvaigo, S., Serres, C., Signorini, N., Emonet, N., Richard, M. J., and Cadet, J. (1998). Use of the singlecell gel electrophoresis assay for the immunofluorescent detection of specific DNA damage. Anal Biochem 259, 1–7. Sawyer, T. W., Gill, R. D., Smith-Oliver, T., Butterworth, B. E., and DiGiovanni, J. (1988). Measurement of unscheduled DNA synthesis in primary cultures of adult mouse epidermal keratinocytes. Carcinogenesis 9, 1197–1202. Schiestl, R. H., Aubrecht, J., Khogali, F., and Carls, N. (1997). Carcinogens induce reversion of the mouse pink-eyed unstable mutation. Proc Natl Acad Sci USA 94, 4576–4581. Schmid, W. (1975). The micronucleus test. Mutat Res 31, 9–15. Schunck, C., Johannes, T., Varga, D., Lorch, T., and Plesch, A. (2004). New developments in automated cytogenetic imaging: Unattended scoring of dicentric chromosomes, micronuclei, single cell gel electrophoresis, and fluorescence signals. Cytogenet Genome Res 104, 383–389. Searle, A. G. (1977). The use of pigment loci for detecting reverse mutations in somatic cells of mice. Arch Toxicol 38, 105–108. Shaposhnikov, S. A., Salenko, V. B., Brunborg, G., Nygren, J., and Collins, A. R. (2008). Single-cell gel electrophoresis (the comet assay): Loops or fragments? Electrophoresis 29, 3005–3012. Shelby, M. D. (1996). Selecting chemicals and assays for assessing mammalian germ cell mutagenicity. Mutat Res 352, 159–167. Shelby, M. D., and Witt, K. L. (1995). Comparison of results from mouse bone marrow chromosome aberration and micronucleus tests. Environ Mol Mutagen 25, 302–313. Shelby, M. D., and Zeiger, E. (1990). Activity of human carcinogens in the Salmonella and rodent bonemarrow cytogenetics tests. Mutat Res 234, 257–261. Shibutani, S., Kim, S. Y., and Suzuki, N. (2006). 32P-postlabeling DNA damage assays: PAGE, TLC, and HPLC. Methods Mol Biol 314, 307–321. Shuck, S. C., Short, E. A., and Turchi, J. J. (2008). Eukaryotic nucleotide excision repair: From understanding mechanisms to influencing biology. Cell Res 18, 64–72. Singh, N. P., McCoy, M. T., Tice, R. R., and Schneider, E. L. (1988). A simple technique for quantitation of low levels of DNA damage in individual cells. Exp Cell Res 175, 184–191. Singh, N. P., Stephens, R. E., and Schneider, E. L. (1994). Modifications of alkaline microgel electrophoresis for sensitive detection of DNA damage. Int J Radiat Biol 66, 23–28. Singh, R., and Farmer, P. B. (2006). Liquid chromatography–electrospray ionization–mass spectrometry: The future of DNA adduct detection. Carcinogenesis 27, 178–196. Speit, G. (1984). Considerations on the mechanism of differential Giemsa staining of BrdU-substituted chromosomes. Hum Genet 67, 264–269. Speit, G., and Haupter, S. (1985). On the mechanism of differential Giemsa staining of bromodeoxyuridine-substituted chromosomes. II. Differences between the demonstration of sister chromatid differentiation and replication patterns. Hum Genet 70, 126–129. Spencer, T., and Butler, L. J. (1987). A rapid modified fluorescence-plus-Giemsa technique for sister chromatid exchanges and DNA replication patterns. Cytobios 50, 145–150. Stiegler, G. L., and Stillwell, L. C. (1993). Big Blue transgenic mouse lacI mutation analysis. Environ Mol Mutagen 22, 127–129. Storchova, Z., and Kuffer, C. (2008). The consequences of tetraploidy and aneuploidy. J Cell Sci 121, 3859–3866.
REFERENCES
357
Storchova, Z., and Pellman, D. (2004). From polyploidy to aneuploidy, genome instability and cancer. Nat Rev Mol Cell Biol 5, 45–54. Strickland, P. T., and Boyle, J. M. (1984). Immunoassay of carcinogen-modified DNA. Prog Nucleic Acid Res Mol Biol 31, 1–58. Suzuki, H., Shirotori, T., and Hayashi, M. (2004). A liver micronucleus assay using young rats exposed to diethylnitrosamine: Methodological establishment and evaluation. Cytogenet Genome Res 104, 299–303. Swenberg, J. A., Fryar-Tita, E., Jeong, Y. C., Boysen, G., Starr, T., Walker, V. E., and Albertini, R. J. (2008). Biomarkers in toxicology and risk assessment: informing critical dose–response relationships. Chem Res Toxicol 21, 253–265. Swiger, R. R., Cosentino, L., Masumura, K. I., Nohmi, T., and Heddle, J. A. (2001). Further characterization and validation of gpt delta transgenic mice for quantifying somatic mutations in vivo. Environ Mol Mutagen 37, 297–303. Swiger, R. R., Cosentino, L., Shima, N., Bielas, J. H., Cruz-Munoz, W., and Heddle, J. A. (1999). The cII locus in the MutaMouse system. Environ Mol Mutagen 34, 201–207. Tao, K. S., and Heddle, J. A. (1994). The accumulation and persistence of somatic mutations in vivo. Mutagenesis 9, 187–191. Tao, K. S., Urlando, C., and Heddle, J. A. (1993a). Comparison of somatic mutation in a transgenic versus host locus. Proc Natl Acad Sci USA 90, 10681–10685. Tao, K. S., Urlando, C., and Heddle, J. A. (1993b). Mutagenicity of methyl methanesulfonate (MMS) in vivo at the Dlb-1 native locus and a lacI transgene. Environ Mol Mutagen 22, 293–296. Tates, A. D., Neuteboom, I., and Hofker, M. (1980). A micronucleus technique for detecting clastogenic effects of mutagens/carcinogens (DEN, DMN) in hepatocytes of rat liver in vivo. Mutat Res 74, 11–20. Tates, A. D., van Dam, F. J., de Zwart, F. A., van Teylingen, C. M., and Natarajan, A. T. (1994). Development of a cloning assay with high cloning efficiency to detect induction of 6-thioguanineresistant lymphocytes in spleen of adult mice following in vivo inhalation exposure to 1,3-butadiene. Mutat Res 309, 299–306. Terashima, I., Suzuki, N., and Shibutani, S. (2002). 32P-Postlabeling/polyacrylamide gel electrophoresis analysis: Application to the detection of DNA adducts. Chem Res Toxicol 15, 305–311. Thybaud, V., Dean, S., Nohmi, T., de Boer, J., Douglas, G. R., Glickman, B. W., Gorelick, N. J., Heddle, J. A., Heflich, R. H., Lambert, I., Martus, H. J., Mirsalis, J. C., Suzuki, T., and Yajima, N. (2003). In vivo transgenic mutation assays. Mutat Res 540, 141–151. Tice, R. R., Agurell, E., Anderson, D., Burlinson, B., Hartmann, A., Kobayashi, H., Miyamae, Y., Rojas, E., Ryu, J. C., and Sasaki, Y. F. (2000). Single cell gel/comet assay: Guidelines for in vitro and in vivo genetic toxicology testing. Environ Mol Mutagen 35, 206–221. Tice, R. R., Hayashi, M., MacGregor, J. T., Anderson, D., Blakey, D. H., Holden, H. E., Kirsch-Volders, M., Oleson, F. B., Jr., Pacchierotti, F., Preston, R. J., et al. (1994). Report from the working group on the in vivo mammalian bone marrow chromosomal aberration test. Mutat Res 312, 305–312. Tischfield, J. A., Engle, S. J., Gupta, P. K., Bye, S., Boyadjiev, S., Shao, C., O’Neill, P., Albertini, R. J., Stambrook, P. J., and Sahota, A. S. (1994). Germline and somatic mutation at the APRT locus of mice and man. Adv Exp Med Biol 370, 661–664. Tompkins, E. M., Farmer, P. B., Lamb, J. H., Jukes, R., Dingley, K., Ubick, E., Turteltaub, K. W., Martin, E. A., and Brown, K. (2006). A novel 14C-postlabeling assay using accelerator mass spectrometry for the detection of O6-methyldeoxy-guanosine adducts. Rapid Commun Mass Spectrom 20, 883–891. Torous, D. K., Dertinger, S. D., Hall, N. E., and Tometsko, C. R. (2000). Enumeration of micronucleated reticulocytes in rat peripheral blood: A flow cytometric study. Mutat Res 465, 91–99. Torous, D. K., Hall, N. E., Illi-Love, A. H., Diehl, M. S., Cederbrant, K., Sandelin, K., Ponten, I., Bolcsfoldi, G., Ferguson, L. R., Pearson, A., Majeska, J. B., Tarca, J. P., Hynes, G. M., Lynch, A. M., McNamee, J. P., Bellier, P. V., Parenteau, M., Blakey, D., Bayley, J., van der Leede, B. J., Vanparys, P., Harbach, P. R., Zhao, S., Filipunas, A. L., Johnson, C. W., Tometsko, C. R., and Dertinger, S. D. (2005). Interlaboratory validation of a CD71-based flow cytometric method (Microflow) for the scoring of micronucleated reticulocytes in mouse peripheral blood. Environ Mol Mutagen 45, 44–55. Torous, D. K., Hall, N. E., Murante, F. G., Gleason, S. E., Tometsko, C. R., and Dertinger, S. D. (2003). Comparative scoring of micronucleated reticulocytes in rat peripheral blood by flow cytometry and microscopy. Toxicol Sci 74, 309–314.
358
CHAPTER 12
IN VIVO GENOTOXICITY ASSAYS
Tucker, J. D., Auletta, A., Cimino, M. C., Dearfield, K. L., Jacobson-Kram, D., Tice, R. R., and Carrano, A. V. (1993). Sister-chromatid exchange: Second report of the Gene-Tox Program. Mutat Res 297, 101–180. Turteltaub, K. W., and Dingley, K. H. (1998). Application of accelerated mass spectrometry (AMS) in DNA adduct quantification and identification. Toxicol Lett 102–103, 435–439. Tweats, D. J., Blakey, D., Heflich, R. H., Jacobs, A., Jacobsen, S. D., Morita, T., Nohmi, T., O’Donovan, M. R., Sasaki, Y. F., Sofuni, T., and Tice, R. (2007). Report of the IWGT working group on strategies and interpretation of regulatory in vivo tests I. Increases in micronucleated bone marrow cells in rodents that do not indicate genotoxic hazards. Mutat Res 627, 78–91. Uiterdijk, H. G., Ponder, B. A., Festing, M. F., Hilgers, J., Skow, L., and Van Nie, R. (1986). The gene controlling the binding sites of Dolichos biflorus agglutinin, Dlb-1, is on chromosome 11 of the mouse. Genet Res 47, 125–129. van Dam, F. J., Natarajan, A. T., and Tates, A. D. (1992). Use of a T-lymphocyte clonal assay for determining HPRT mutant frequencies in individual rats. Mutat Res 271, 231–242. Van Sloun, P. P., Wijnhoven, S. W., Kool, H. J., Slater, R., Weeda, G., van Zeeland, A. A., Lohman, P. H., and Vrieling, H. (1998). Determination of spontaneous loss of heterozygosity mutations in Aprt heterozygous mice. Nucleic Acids Res 26, 4888–4894. Vanhauwaert, A., Vanparys, P., and Kirsch-Volders, M. (2001). The in vivo gut micronucleus test detects clastogens and aneugens given by gavage. Mutagenesis 16, 39–50. Vijg, J., and Douglas, G. R. (1996). Bacteriophage lambda and plasmid lacZ transgenic mice for studying mutations in vivo. In Technologies for Detection of DNA Damage and Mutations, Pfeifer, G. P., ed., Plenum Press, New York, pp. 391–410. Wakata, A., Miyamae, Y., Sato, S., Suzuki, T., Morita, T., Asano, N., Awogi, T., Kondo, K., and Hayashi, M. (1998). Evaluation of the rat micronucleus test with bone marrow and peripheral blood: Summary of the 9th collaborative study by CSGMT/JEMS. MMS. Collaborative Study Group for the Micronucleus Test. Environmental Mutagen Society of Japan. Mammalian Mutagenicity Study Group. Environ Mol Mutagen 32, 84–100. Walker, V. E., Jones, I. M., Crippen, T. L., Meng, Q., Walker, D. M., Bauer, M. J., Reilly, A. A., Tates, A. D., Nakamura, J., Upton, P. B., and Skopek, T. R. (1999). Relationships between exposure, cell loss and proliferation, and manifestation of Hprt mutant T cells following treatment of preweanling, weanling, and adult male mice with N-ethyl-N-nitrosourea. Mutat Res 431, 371–388. Waters, M. D., Stack, H. F., Jackson, M. A., and Bridges, B. A. (1993). Hazard identification: Efficiency of short-term tests in identifying germ cell mutagens and putative nongenotoxic carcinogens. Environ Health Perspect 101(Suppl 3), 61–72. Waters, M. D., Stack, H. F., Jackson, M. A., Bridges, B. A., and Adler, I. D. (1994). The performance of short-term tests in identifying potential germ cell mutagens: A qualitative and quantitative analysis. Mutat Res 341, 109–131. Weaver, J. L., and Torous, D. (2000). Flow cytometry assay for counting micronucleated erythrocytes: development process. Methods 21, 281–287. Weston, A. (1993). Physical methods for the detection of carcinogen-DNA adducts in humans. Mutat Res 288, 19–29. Weston, A., Rowe, M. L., Manchester, D. K., Farmer, P. B., Mann, D. L., and Harris, C. C. (1989). Fluorescence and mass spectral evidence for the formation of benzo[a]pyrene anti-diol-epoxide–DNA and –hemoglobin adducts in humans. Carcinogenesis 10, 251–257. Whong, W. Z., Stewart, J. D., and Ong, T. (1992). Comparison of DNA adduct detection between two enhancement methods of the 32P-postlabelling assay in rat lung cells. Mutat Res 283, 1–6. Wiklund, S. J., and Agurell, E. (2003). Aspects of design and statistical analysis in the Comet assay. Mutagenesis 18, 167–175. Wild, C. P. (1990). Antibodies to DNA alkylation adducts as analytical tools in chemical carcinogenesis. Mutat Res 233, 219–233. Wilson, D. M., 3rd, and Thompson, L. H. (2007). Molecular mechanisms of sister-chromatid exchange. Mutat Res 616, 11–23. Winton, D. J., Gooderham, N. J., Boobis, A. R., Davies, D. S., and Ponder, B. A. (1990). Mutagenesis of mouse intestine in vivo using the Dlb-1 specific locus test: Studies with 1,2-dimethylhydrazine,
REFERENCES
359
dimethylnitrosamine, and the dietary mutagen 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline. Cancer Res 50, 7992–7996. Winton, D. J., and Ponder, B. A. (1990). Stem-cell organization in mouse small intestine. Proc Biol Sci 241, 13–18. Witt, K. L., Livanos, E., Kissling, G. E., Torous, D. K., Caspary, W., Tice, R. R., and Recio, L. (2008). Comparison of flow cytometry- and microscopy-based methods for measuring micronucleated reticulocyte frequencies in rodents treated with nongenotoxic and genotoxic chemicals. Mutat Res 649, 101–113. Wojcik, A., Bruckmann, E., and Obe, G. (2004). Insights into the mechanisms of sister chromatid exchange formation. Cytogenet Genome Res 104, 304–309. Wyborski, D. L., Malkhosyan, S., Moores, J., Perucho, M., and Short, J. M. (1995). Development of a rat cell line containing stably integrated copies of a lambda/lacI shuttle vector. Mutat Res 334, 161–165. Xue, W., and Warshawsky, D. (2005). Metabolic activation of polycyclic and heterocyclic aromatic hydrocarbons and DNA damage: A review. Toxicol Appl Pharmacol 206, 73–93. Yuspa, S. H., and Poirier, M. C. (1988). Chemical carcinogenesis: from animal models to molecular models in one decade. Adv Cancer Res 50, 25–70. Zimmer, D. M., Harbach, P. R., Mattes, W. B., and Aaron, C. S. (1999). Comparison of mutant frequencies at the transgenic lambda LacI and cII/cI loci in control and ENU-treated Big Blue mice. Environ Mol Mutagen 33, 249–256.
PART
IV
ASSESSING THE HUMAN RELEVANCE OF CHEMICALINDUCED TUMORS
CH A P TE R
13
FRAMEWORK ANALYSIS FOR DETERMINING MODE OF ACTION AND HUMAN RELEVANCE R. Julian Preston
13.1.
INTRODUCTION
The overall aim of a cancer risk assessment is to characterize the risk to humans from environmental exposures.* This risk characterization includes both qualitative and quantitative components and relies on the development of separate hazard, dose–response and exposure assessments. The specific approach currently used by the U.S. Environmental Protection Agency (EPA) can be found in its Guidelines for Carcinogen Risk Assessment (EPA 2005). A similar approach is applied by other national and international organizations. In general terms, the risk characterization summarizes, in a narrative form, the analyses of hazard, dose–response, and exposure assessment. These three assessments are summarized in light of “the extent and weight of evidence, major points of interpretation and rationale for their selection, strengths and weaknesses of the evidence and the analysis, and [a discussion] of alterative conclusions and uncertainties that deserve serious consideration” (EPA 2000). This summary serves as the starting materials for the overall risk characterization process that completes the risk assessment. This chapter will concentrate on a specific feature of this risk characterization process, namely the importance of developing approaches for incorporating mechanistic data into the hazard, dose– response, and exposure assessments to reduce uncertainties in the process and thereby reduce the reliance on default factors that are used in the absence of reliable data. Given that the risk characterization is for the estimation of risks to humans from low, environmental exposures, then the issues that cover the necessary defaults are as follows (see Part I, this volume): *This chapter has been reviewed in accordance with the policy of the U.S. Environmental Protection Agency and approved for publication, although it does not necessarily reflect Agency policy.
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
363
364
CHAPTER 13 FRAMEWORK ANALYSIS FOR DETERMINING MODE OF ACTION
• Is the presence or absence of effects observed in a human population predictive of effects in another exposed human population? • Is the presence or absence of effects observed in an animal population predictive of effects in exposed humans? • How do metabolic pathways relate across species and among different age groups and between sexes in humans? • How do toxicokinetic processes relate across species and among different age groups and between sexes in humans? • What is the relationship between the observed dose–response relationship to the relationship at lower doses? These issues are ones of extrapolation and as noted by Preston (2005) such extrapolations are “the Achilles heel of risk assessment” (Preston 2005). The U.S. EPA, The International Program on Chemical Safety (IPCS), and The International Life Sciences Institute (ILSI), for example, have proposed a framework based on the mode of action of a chemical, the key events that define a particular mode of action, and a human relevance framework for assessing the plausibility of an animal mode of action to humans. It is this approach that will be described and discussed in this chapter.
13.2. FRAMEWORK ANALYSIS: MODE OF ACTION AND KEY EVENTS 13.2.1.
Definitions
The following definitions are taken from the U.S. EPA Guidelines for Carcinogen Risk Assessment (EPA 2005). Mode of action (MOA) is defined as “a sequence of key events and processes, starting with interaction of an agent with a cell, proceeding through operational and anatomical changes, and resulting in cancer formation.” MOA is contrasted with Mechanism of Action, which implies a more detailed understanding and description of key events, often at the molecular level, than for MOA. Examples of MOA are DNA-reactivity, mitogenicity, inhibition of cell death, cytotoxicity with regenerative cell proliferation, immune suppression, and epigenetic effects, such as changes in gene expression and DNA methylation patterns. A key event is an “empirically observable precursor step that is itself a necessary element of the mode of action or is a biologically based marker for such an element.” In this regard and for this chapter, a biomarker is considered to be a surrogate marker of exposure or an early biological marker of effect (e.g., mutations in reporter genes, total chromosome alterations). In contrast, a biological marker of effect that is itself a key event along the pathway from a normal cell to a transformed one is described as a bioindicator (e.g., mutation in a critical gene for cancer, cancerspecific chromosome translocation). This distinction is useful for considering those cellular events that can be used only in a qualitative way for predicting tumor responses (biomarkers) and those that can be both qualitative and quantitative endpoints in a tumor dose–response assessment (bioindicators).
13.2. FRAMEWORK ANALYSIS: MODE OF ACTION AND KEY EVENTS
365
A point of departure (POD) is a point on a dose–response curve at which the range of data is extended from the observable range to lower dose ranges by extrapolation. Such extrapolation can be by default linear, predicted linear, or predicted nonlinear (to also include a threshold).
13.2.2. An Overview of the Framework for Analyzing Mode of Action The EPA Cancer Guidelines (EPA 2005) provide an analytical framework for assessing if available evidence supports a predicted MOA for a particular agent. As noted by Wiltse and Dellarco (2000), this “framework is based on considerations for causality in epidemiological investigations originally articulated by Hill (1965) but later modified by others and extended to experimental studies (DHHS, 1982; Faustman et al., 1997)” (Wiltse and Dellarco 2000). In outline, this framework first requires a description of a postulated MOA (or MOAs). This postulated MOA is then queried by addressing the pertinent available empirical data and experimental observations. These specific queries are: • • • • • •
What are the key events that lead to the postulated MOA? What is the strength, consistency, and specificity of association? What are the dose–response relationships? What are the temporal relationships? What is the biological plausibility and coherence? Are there other MOAs that are supportable by the available data?
Each of these queries will be discussed in more detail in Section 13.2.4.1.
13.2.3. Framework for Assessing Human Relevance of Animal MOA The majority of the tumor data available for conducting cancer risk assessments for exposure to environmental chemicals come from 2-year cancer bioassays using rats and mice. Thus, a MOA based on key events is inevitably developed for laboratory animals and not humans. There are, of course, a few exceptions for which human tumor data are available (NTP 2005). These human data are generally used together with rodent tumor data as part of dose–response characterization. Thus, the need in all cases is to demonstrate that the animal MOA is plausible in humans. This can be accomplished by use of a human relevance framework (described below in this section and in Table 13.1 and in Figure 13.1). It is of note that this considerable reliance on laboratory animal data for risk assessment purposes for environmental chemicals is in sharp contrast to the situation with ionizing radiation. The cancer risk estimates for ionizing radiation (X rays and γ rays) are based to a very great extent on human tumor data obtained from the Life Stage Study (LSS) of the atomic bomb survivors in Hiroshima and Nagasaki, Japan
366
CHAPTER 13 FRAMEWORK ANALYSIS FOR DETERMINING MODE OF ACTION
TABLE 13.1.
Framework for Evaluation of an Animal MOA
1. Postulated MOA. Brief description of the sequence of measured effects, starting with chemical administration, to cancer formation at a given site. 2. Key events. Clear description of each of the key events (measurable parameters) that are thought to underlie the MOA. 3. Dose–response relationships. Dose-response relationships identified for each key event, and comparisons presented of dose–response relationships among key events and with cancer. 4. Temporal association. Sequence of key events over time that lead to tumor formation. 5. Strength, consistency, and specificity of association of key events and tumor response. Complete assessment and presentation of the relationships among the key events, precursor lesions, and tumors. Portrayal of the consistency of observations across studies of different designs. 6. Biological plausibility and coherence. Determination of whether key events and the sequence of events are consistent with current biological thinking, both regarding carcinogenesis in general and for the specific chemical under review. 7. Other MOAs. Alternative MOAs that may be applicable for the chemical under review. Comparison of their likelihood vis-à-vis the proposed MOA. 8. Conclusion about the MOA. Overall indication of the level of confidence in the postulated MOA. 9. Uncertainties, inconsistencies, and data gaps. Identification of information deficiencies in the case; description of inconsistent findings in the data at large; evaluation of uncertainties; proposal of pointed research that could significantly inform the case. Source: Meek et al. (2003); adapted from EPA (1999) and Sonich-Mullin et al. (2001).
Animal MOA (and related endpoints) not relevant to humans
Animal MOA relevant or potentially relevant to humans
Is the weight of evidence sufficient to establish the MOA in animals? No Yes
•MOA: Data insufficient to characterize animal MOA
Are key events in the animal MOA plausible in humans? •MOA: Species-specific protein •MOA: Species-specific hormone suppression
No
Yes
•MOA: Species-specific enhanced hormone clearance rate
No need to continue risk assessment for this endpoint
No
Taking into account kinetic and dynamic factors, are key events in the animal MOA plausible in humans?
Yes
•MOA: Comparable cytotoxicity and cell proliferation response •MOA: Comparable tissue response (different animalhuman exposure potential) Continue risk assessment, including dose-response human exposure analysis, and risk characterization
Figure 13.1. General schematic illustrating how the Human Relevance Framework can be used to assess whether or not an animal MOA has a human counterpart, thereby indicating if a quantitative risk assessment is required. [Adapted from Meek et al., (2003).]
13.2. FRAMEWORK ANALYSIS: MODE OF ACTION AND KEY EVENTS
367
(Preston et al. 2007, 2008). This study has been ongoing for over 50 years and utilizes a large study cohort. An extremely thorough dose reconstruction program was used to estimate individual doses (reviewed in Cullings et al. (2006)). Data on cancer mortality and cancer incidence have been obtained for a very wide range of age groups at the time of the bombing, including in utero. Additional epidemiological data for medically and occupationally exposed groups are used for enhancing the lower dose tumor database. Animal tumor data and mechanistic studies are not used directly in the risk assessment process, but really only to provide general support to the conclusions from the human tumor data (ICRP 2008; NRC 2006). Risk estimates at low dose and low dose rates are currently obtained by extrapolation of the dose–response curve for observed tumors over the medium to high dose ranges. The extrapolation used is a linear no threshold (LNT) one. Thus, for radiation an MOA and key events could be developed for human tumors without the need for a human relevance framework. The Risk Science Institute (RSI) of the International Life Sciences Institute (ILSI) developed a logical framework (Human Relevance Framework, HRF) for deciding if a mode of action by which a particular chemical induces tumors in an animal model could plausibly be acting in humans based upon the available human data and considerations of kinetic and dynamic factors (Meek et al. 2003). The HRF analysis has an initial focus on key events in the animal MOA (Table 13.1) and then considers the weight of evidence for the relevance to humans of the animal tumors being studied (Figure 13.1). This approach relies upon addressing three overriding questions: (1) Is the weight of evidence sufficient to establish the MOA in animals? (2) Are key events in the animal MOA plausible in humans? (3) Taking into account kinetic and dynamic factors, is the animal MOA plausible in humans? The outcome is presented in a Yes/No format. If the answer to the third question is No, the decision is “No need to continue risk assessment for this endpoint”; if the answer is Yes, the decision is “Continue risk assessment including dose–response, human exposure analysis, and risk characterization.” Further discussion of the components is presented in the Section 13.3.
13.2.4. Establishing and Applying Key Events in Support of MOA This section will build upon the overview information presented in the previous sections to lay out for known rodent carcinogens, what specific information is required, and how it is used to establish an MOA in animals and its relevance to humans. Specific examples are provided also for illustrating the application of the framework approach leading to risk characterization. 13.2.4.1. Is the Weight of Evidence Sufficient to Establish the MOA in Animals? The following information is used as a guide for evaluating each hypothesized carcinogenic MOA (EPA 2005; Meek et al. 2003) and is, in a sense, a corollary to the queries that were presented in the framework overview above.
368
CHAPTER 13 FRAMEWORK ANALYSIS FOR DETERMINING MODE OF ACTION
There are a number of different MOAs (or combinations of MOAs) whereby a chemical exerts its carcinogenicity. In general, these can be characterized as DNAreactivity and nonDNA-reactivity [to include oxidative stress, cytotoxicity and associated regenerative proliferation, mitogenicity, receptor mediation, and epigenetic effects (e.g., changes in gene expression, DNA methylation, chromatin organization)]. This broad classification is pertinent to the subsequent discussion because a chemical with a so-called mutagenic MOA (used synonymously with DNA-reactive) is considered to follow a default linear dose–response extrapolation for tumors whereas chemicals with non-DNA-reactive MOAs are not considered to be a default linear in this regard (EPA 2005). 13.2.4.2. Key Events for Characterizing MOA. A critical component of MOA characterization is the establishment of a set of key events that define how a normal cell can be converted into a malignant one and ultimately to a metastatic tumor as a consequence of exposure to a DNA-reactive or non-DNA-reactive chemical carcinogen. The enormous enhancement of our knowledge of the cancer process over the past decade or so has greatly enhanced the process of identifying key events (Weinberg 2006). In general terms, carcinogenesis is a multistep process that requires an integrated set of genetic and epigenetic alterations to produce the cancer phenotype. This allows for key events to be described in terms of these steps, with any particular event being a specific genetic or other cellular change that characterizes the step. An example has been developed and applied by Preston and Williams (2005). The particular approach presented by Preston and Williams (2005) was for DNA-reactive carcinogens, although as shown here it is readily adaptable to nonDNA-reactive chemicals. The framework is the description of key events for tumor development and is shown in Table 13.2. There is an essential temporal sequence
TABLE 13.2.
Key Events for Tumor Development: DNA-Reactive MOAs
1. Exposure of target cells (e.g., stem cells) to ultimate DNA-reactive and mutagenic species; in some cases this requires metabolism. 2. Reaction with DNA in target cells to produce DNA damage. 3. Misreplication on damaged DNA template or misrepair of DNA damage. 4. Mutations in critical genes in replicating target cell. 5. These mutations result in initiation of new DNA/cell replication. 6. New cell replication leads to clonal expansion of mutant cells. 7. DNA replication can lead to further mutations in critical genes. 8. Imbalanced and uncontrolled clonal growth of mutant cells may lead to preneoplastic lesions. 9. Progression of preneoplastic cells results in emergence of overt neoplasms, solid tumors (which require neoangiogenesis), or leukemia. 10. Additional mutations in critical genes as a result of uncontrolled cell division results in malignant behavior. Note: Key events along the pathway to tumor development for DNA-reactive carcinogens can be assessed both qualitatively and quantitatively by experimental and human studies. For each of the chemicals selected for the case studies, the available data were matched to one or more of these key events to help establish a MOA for human cancer.
13.2. FRAMEWORK ANALYSIS: MODE OF ACTION AND KEY EVENTS
369
to these key events since each one (except the first) is dependent on the occurrence of the previous ones. This becomes of particular importance when considering quantitative risk assessments. For non-DNA-reactive chemicals, the only differences would be that there would be no need for binding of the parent chemical or a metabolite to DNA—all other steps could be the same. There might be additional differences in cases where a chemical does not interact with DNA directly or indirectly (for example, via reactive oxygen species). The DNA replication errors as a key event would still hold for non-DNA-reactive chemicals; it is just that the chemically produced mutations arise from an enhanced probability of replication errors as a result of increased cell proliferation in response to cytotoxicity as opposed to being from a more damaged DNA template, which is the case for DNA-reactive chemicals. The initial step in using key events in an MOA framework is a qualitative one, as described in Table 13.2, namely to match the key events for a particular chemical with those for a particular MOA (DNA-reactivity in Table 13.2). However, it is also possible to utilize key events to develop informative biomarkers of exposure and effect as well as bioindicators of disease outcome that can be utilized in a quantitative risk assessment process. A distinction is made between biomarkers and bioindicators because they can be used in quite different ways and for different purposes in a risk assessment context. As mentioned above, a biomarker is considered to be a surrogate marker of exposure or an early biological marker of effect (e.g., mutations in reporter genes, total chromosome alterations). In contrast, a biological marker of effect that is itself a key event along the pathway from a normal cell to a transformed one is described as a bioindicator (e.g., mutation in critical gene for cancer, cancer-specific chromosome translocation). Biomarkers can be used to inform the dose–response for tumors in a qualitative manner. Bioindicators can be used in a qualitative and quantitative way to inform tumor dose–response curves. Use of these biomarkers and bioindicators can make it feasible to characterize a dose–response curve at exposure levels below those at which increases in tumor frequency can be assessed. Recent advances in knowledge of the underlying mechanisms of carcinogenesis and the ever-increasing portfolio of whole genome molecular assay techniques have made it much more feasible to identify and select informative bioindicators of disease processes, especially cancer (Block et al. 2008; Conrad et al. 2008; Preston 2005). The emphasis for the key events in tumor development is on the essential ingredients for driving the process, namely alterations in critical genes (e.g., oncogenes and tumor suppressor genes) and enhanced cell proliferation. Carcinogenesis can be viewed as an evolutionary process that requires an accumulation of genomic alterations that then provide a substrate for selection of an advantageous phenotype that is usually related to a growth advantage (Cahill et al. 1999; Gatenby and Vincent 2003; Maley et al. 2004; Vincent and Gatenby 2008). However, it is not necessary to characterize the precise mechanism of tumor formation in response to an exposure to an environmental agent but rather to establish that certain key genetic and phenotypic changes take place. This difference between precise mechanism and key changes can be exemplified by considering two major cancer models. Fearon and
370
CHAPTER 13 FRAMEWORK ANALYSIS FOR DETERMINING MODE OF ACTION
Vogelstein (1990) described a model for colorectal tumorigenesis in which each step along the multistep process from a normal cell to a malignant tumor was characterized at the phenotypic level as specific preneoplastic lesions and at the genome level by specific genetic alterations (gene mutations and chromosomal changes). In contrast, Hanahan and Weinberg in their “Hallmarks of Cancer” (Hanahan and Weinberg 2000) described tumors, irrespective of site or species, as acquiring six characteristics (self-sufficiency in growth signals, insensitivity to anti-growth signals, evading apoptosis, limitless replicative potential, sustained angiogenesis, and tissue invasion/metastasis) without the requirement to know precisely which genetic or epigenetic alteration resulted in the development of the characteristic. It is the latter approach which best exemplifies the application of the framework for MOA and key events. 13.2.4.3. Dose–Response Characteristics. This is the area of the risk assessment process that engenders the most debate. This is to a large extent because the need is to estimate the human risks at environmental exposure levels and yet there are few data at these exposure levels that can be used directly in a qualitative, let alone quantitative, way to support a particular form of dose–response curve (see Part VI, this volume). The approach has to rely upon extrapolation from tumor data (usually for rodents) to predict responses outside the range where tumor data themselves are available. This has in the past been done in a relatively pragmatic fashion, relying to a great extent on default positions for dose–response curve shape. With the advent of the incorporation of mechanistic data into the process, it is possible to utilize data on key events to enhance the process for interspecies, high to low dose, acute to chronic exposure extrapolations. This enhancement can be at a qualitative level (shape of dose–response curve) or a quantitative one, depending on the predictive value of the particular biological marker of the key event for tumor outcomes. The onus is on research investigators to develop linkages between the types of data they collect and their characteristics for use in a risk assessment framework. Such an approach will clearly help to reduce uncertainty in the risk assessment process and will lead to a reduction in the use of default approaches that have to be used in the absence of appropriate datasets. These considerations will be enhanced by the use of systems biology approaches, namely to treat normal tissues and altered ones as systems for comparison and selection of key event bioindicators (Edwards and Preston 2008; NRC 2007). Since the key events that define an animal MOA are in a temporal sequence, the dose–response curve for each successive key event can be viewed as the probability of converting one key event into the next one—for example, for converting a DNA adduct into a mutation by an error of replication. In the absence of any rate-limiting step, the overall probability of inducing a tumor from a chemical exposure is the integration of the probabilities for all key events. A ratelimiting step such as absence of exposure to the target cells (Table 13.2) would lead to a threshold dose–response curve because no other key events could be manifest. This is a simple example, but others are clearly feasible based upon, for example, fidelity of DNA repair processes or interacting (offsetting) gene expression changes.
13.2. FRAMEWORK ANALYSIS: MODE OF ACTION AND KEY EVENTS
371
13.2.4.4. Temporal Sequence of Key Events. As noted above, the key events represent a progression from a normal cell to a metastatic tumor, with each key event being dependent on the previous ones in order. Thus, there is a requirement for a temporal sequence to the induction of key events. Such a temporal sequence is most likely to be met when the exposures are chronic, given the propensity for continued induction of key events. 13.2.4.5. Additional MOAs. Even though a specific MOA has been established, based on the use of key events, it is possible that a particular chemical can operate through additional MOAs. This feature could impact the shape of the dose– response curve and the quantitative assessment of risk, depending on the nature of the multiple MOAs. There is, in fact, an expectation that chemicals will function through more than one MOA; for example, a mutagenic chemical is very likely also to be cytotoxic as a consequence of the induction of DNA damage—both features can enhance mutation rate above background and thus act additively or synergistically. 13.2.4.6. Examples of the Use of Key Events in an MOA Framework. There are a number of available examples for which the MOA framework has been used; published examples co-authored by the author of this chapter have been selected. (a) Dichloromethane (from Preston and Williams, 2005). The details for the conduct of the MOA analysis can be found in the publication. For this example, and indeed for many cases, it is possible to identify data for key events 1–3 (Table 13.2) and for key event 9 (carcinogenicity) and this is sufficient for describing a MOA as DNA-reactivity. In summary, Table 13.3 identifies the key events in the animal MOA for dichloromethane. This analysis leads to the conclusion that dichloromethane acts in rodents by a DNA-reactive MOA. The human relevance of this conclusion is considered in Section 13.3, paragraph (a).
TABLE 13.3.
Key Events in the Animal MOA for Dichloromethane
Key Events Metabolism by GSTTI-I (Key Event I) DNA damage induced (Key Event 2) Genotoxicity and mutagenicity (Key Event 3) Carcinogenicity (Key Event 9)
Evidence in Animals In mice metabolism by GSTTI-I in target tissues for tumor formation. In rats metabolism by GSTTI-I much lower in target tissues than for mice DNA–protein crosslink and single strand-breaks induced in mouse cells in vivo and in vitro but either not in rat cells or a lower frequency Reduced DNA damage in presence of GST-depleting agent Mutagenic in bacteria Genotoxic/mutagenic in mouse cells in vitro and in vivo Less genotoxic or nongenotoxic in rat cells in vitro and in vivo Carcinogenic in mice by inhalation (lung and liver); carcinogenic, at a reduced level, in rats by inhalation (mammary gland)
372
CHAPTER 13 FRAMEWORK ANALYSIS FOR DETERMINING MODE OF ACTION
TABLE 13.4.
Key Events in the Carcinogenicity of 4-Aminobiphenyl in Animals
Key Event 1:
Key Event 2: Key Event 3: Key Event 9:
TABLE 13.5.
Key Event 1: Key Event 2: Key Event 3: Key Event 4: Key Event 9:
Metabolic activation (a) N-Hydroxylation (b) N-Esterification (gluconoride, acetyl, sulfate) (c) Hydrolysis to nitroniun ion DNA adduct formation (dG–C8, dA–C8, dG–N2) in pluripotent cells of target organs DNA mutation in critical gene(s) leading to cancer Cancer
Key Events in the Animal MOA for 1,3-Butadiene
Target cells (including bone marrow) exposed to ultimate DNA-reactive species. DNA adducts induced in tumor target tissues (liver, lung and testis), with dGNT and dAN6 being the most frequent. Mutations induced in target cells. Mutations induced in critical genes for cancer. Cancer.
(b) 4-Aminobiphenyl. A thorough case study has been developed by Cohen et al. (2006) for assessing the cancer MOA for 4-aminobiphenyl using the framework described above (Cohen et al. 2006). The key events identified in support of the MOA are presented in Table 13.4. Again, the human relevance is considered using the IPCS Human Relevance framework and is discussed in the Section 13.3, paragraph (b). (c) 1,3-Butadiene (Preston, 2007). The details for this example can be found in Preston (2007). An assessment was made of the key events defining a DNA-reactive MOA in rodents for 1,3-butadiene (with an emphasis on contrasting the much greater effectiveness in mice than in rats). These key events are summarized in Table 13.5. The production of these key events is sufficient to conclude that 1,3-butadiene acts via a DNA-reactive MOA. The human relevance is discussed in the original manuscript (Preston 2007) and in summary in the Section 13.3, paragraph(c).
13.3.
FRAMEWORK ANALYSIS: HUMAN RELEVANCE
Once a MOA for an animal model (most frequently rodent) has been developed, an assessment needs to be made to establish if this same MOA cannot be reasonably excluded or is plausible in humans. This has to be done for most chemical carcinogens because, as mentioned above, only a small minority of chemicals have been shown to be carcinogenic in humans. This plausibility was, until recently, conducted
13.3. FRAMEWORK ANALYSIS: HUMAN RELEVANCE
373
through a somewhat arbitrary approach largely involving a search for relevant data. More recently the process has been formalized through the development of a Human Relevance Framework (Meek et al. 2003) that builds upon the US EPA Cancer Guidelines (EPA 2005) and the IPCS MOA framework (Sonich-Mullin et al. 2001). The general approach is shown in Figure 13.1. The initial step is to use a weight of evidence approach that is based on key events, as discussed above, to establish the MOA in animals. On the assumption that there is sufficient evidence to establish the MOA, the key event approach is used to determine if the same MOA cannot reasonably be excluded is plausible in humans. The measure is “cannot reasonably be excluded” (Boobis et al. 2006) or “plausible” because there is generally much less data on key events for humans than rodents, and so it has been agreed that plausibility is sufficient. If the key events are reasonably excludable or are not plausible in humans, then the conclusion is that the chemical under review is rodentor species-specific. For example, it can be metabolized in a rodent kidney, but not in humans, to the ultimate carcinogenic species. On the assumption that the animal MOA is plausible in humans, it is necessary to take into account key toxicokinetic and toxicodynamic factors to further establish plausibility in humans. For example, is it possible in humans to achieve target tissue exposures to the parent compound or an active metabolite that are as high as those required in the animal model to produce the adverse outcome? If the conclusion is that there are no prohibitive kinetic or dynamic issues for plausibility of the animal MOA in humans, then the need is to proceed to the next steps of the risk assessment process: dose–response, human exposure analysis, and risk characterization. If there is clear evidence that there are kinetic or dynamic differences between the pertinent animal model and humans for the particular chemical being considered that would result in no response in humans at exposure levels much lower than those required in animals to produce the adverse outcome, then there is no need to continue the risk assessment for the outcome being assessed (i.e., cancer in the context of this review). This HRF has been applied to DNA-reactive and nonDNA-reactive carcinogens (Meek et al. 2003; Preston and Williams 2005) and to noncancer effects (Seed et al. 2005). It has also been applied to the three examples in Section 13.2.4.6 that were used to exemplify the use of the MOA framework and key events. (a) Dichloromethane (Preston and Williams 2005). A DNA-reactive MOA was demonstrated for dichloromethane carcinogenicity using key events 1–3 and 9. There were reported differences in metabolism, DNA damage, and genotoxicity between mice and rats, with rats being much less sensitive. When the same four key events were considered for human plausibility, it was found that humans were much more similar to rats as regards metabolism, DNA damage, and mutagenicity (Table 13.6). This led to a prediction that a DNA-reactive MOA was operational in mice for inducing tumors but that lower GSST1 levels and its distribution in humans (and rats), together with the weak genotoxicity and mutagenicity in vitro for human cells, would result in a lower carcinogenic potential in humans (and in rats) compared to mice. This conclusion was borne out by the available epidemiological data (and rat tumor studies).
374
CHAPTER 13 FRAMEWORK ANALYSIS FOR DETERMINING MODE OF ACTION
TABLE 13.6.
Concordance Analysis of Key Events in Humans for Dichloromethane
Key Events Metabolism by GSTTI-I (Key Event I)
DNA damage induced (Key Event 2)
Genotoxicity and mutagenicity (Key Event 3)
Carcinogenicity (Key Event 9)
Evidence in Animals
Evidence in Humans
In mice metabolism by GSTTI-I in target tissues for tumor formation In rats metabolism by GSTTI-I much lower in target tissues than for mice DNA–protein crosslink and single strand-breaks induced in mouse cells in vivo and in vitro but either not in rat cells or at a lower frequency Reduced DNA damage in presence of GST-depleting agent Mutagenic in bacteria Genotoxic/mutagenic in mouse cells in vitro and in vivo Less genotoxic or nongenotoxic in rat cells in vitro and in vivo Carcinogenic in mice by inhalation—lung and liver Carcinogenic, at a reduced level, in rats by inhalation—mammary gland
Enzyme is present, at lower levels than in mouse target tissues, metabolism by GSTT1-1 predicted by PBPK models DNA–protein crosslinks not observed in human hepatocytes in vitro Low levels of GSTT1-1 suggest that DNA damage induction is plausible but at low levels Inconsistent results for mutagenicity in humans cells in vitro
Human epidemiological data suggest that risks associated with dichloromethane exposure, if any, are small and limited to rare cancers
TABLE 13.7. Concordance Evaluation of Key Events of 4-Aminobiphenyl-Induced Urinary Bladder Carcinogenesis between Species
Key Event 1. Metabolic activation to reactive electrophile 2. DNA adduct formation 3. Mutagenesis 4. Carcinoma
Mouse
Dog
Human
+ + + +
+ + + +
+ + + +
(b) 4-Aminobiphenyl. A DNA-reactive MOA was demonstrated for mice using key events 1–3 and 9. A concordance evaluation of the key events for 4aminobiphenyl-induced bladder carcinogenesis between species (mice, dogs, humans) demonstrated that all these key events were either observed and/or plausible in humans (Table 13.7). Quantitative differences among species do exist, but they do not exclude the DNA-reactive MOA in mice and dogs from being operational in humans.
13.4. FUTURE DIRECTIONS
375
(c) 1,3-Butadiene. For 1,3-butadiene a DNA-reactive MOA was established based upon key events 1–4 and 9. Using the human relevance framework, it can be established that these key events are plausible in humans, even though they have not necessarily been confirmed for target cells or for specific endpoints (Preston 2007). This is based, in part, on (a) the mutagenic effectiveness of butadiene and its metabolites in many in vivo and in vitro assay systems and (b) the link between butadieneinduced DNA adducts (that have been observed in butadiene exposed humans) and mutation spectra. It is reasonable to conclude, based on kinetic considerations, that the levels of active metabolites that would reach the target tissues in a majority of humans are probably unlikely to be sufficiently high to induce mutations. In summary, having established an animal MOA and human relevance for this MOA, it is appropriate to address dose–response assessment, human exposure analysis, and risk characterization. Thus, the purpose of the human relevance framework is to establish which chemicals (or chemical mixtures) should be considered for a quantitative risk assessment and which do not require further consideration because they present a minimal risk or no risk to humans. Several thoroughly worked examples are presented in Meek et al. (2003).
13.4.
FUTURE DIRECTIONS
The framework analysis for determining MOA and human relevance appears to be well-established and can continue to provide sound risk assessment guidance. What will change over a relatively short timeframe will be the ability to collect information that can be used to develop key events for describing the MOA in both laboratory animals and humans. A major reason for this is that there is available the capability of using genome-wide approaches for assessing both responses to exposures to environmental chemicals and for describing diseases at a molecular level (i.e., DNA, RNA, and protein changes) (see Chapters 22 and 23) (Chen et al. 2008; Edwards and Preston 2008). This whole genome assessment capability has allowed for responses of cells or organisms to environmental chemical exposures and disease processes to be characterized in terms of key events and toxicity pathways. This in turn has allowed for the development of much more informative biological indicators of response that can be used as surrogates for adverse outcomes, cancer in the present context. It is important to emphasize that it is most effective to use several bioindicators of disease outcome for helping define the shape of the dose–response curve for cancer or for defining human relevance, because it is unlikely, based on underlying mechanism of tumor formation, that any single indicator can define the doseresponse characteristics. Despite this caution, it remains frequently the case in molecular epidemiology studies for single biomarkers or bioindicators to be assessed. Another level of enhanced effort will be in the area of epidemiology, both traditional and molecular, and human in vitro research in support of the identification of key events in humans and of the human relevance of an animal MoA. The recently published National Research Council (NRC) Report, Toxicity Testing in the 21st Century: A Vision and a Strategy (NRC 2007), provides a set of research options for enhancing the use of in vitro test systems with an emphasis on human cells. This
376
CHAPTER 13 FRAMEWORK ANALYSIS FOR DETERMINING MODE OF ACTION
appears to be a viable approach, with the proviso that maintaining some emphasis on in vivo laboratory animal studies remains essential. In this regard, there is an increasing emphasis on the use of informative animal models of human disease for establishing quantitative data for dose–response for a number of diseases, especially cancer (Jacks 2005). However, it needs to be emphasized that reduction in a reliance on extrapolation approaches will greatly reduce the uncertainty in a quantitative risk assessment. This can be achieved through the use of the informative and sensitive technologies that are available or are being developed for the detection and characterization of reliable disease markers (Conrad et al. 2008; Costa et al. 2008; Shim et al. 2008). It will be most interesting to follow the use of the Risk Assessment Framework described in this chapter for a broad range of chemical carcinogens and to see how new methods and data can enhance this use.
REFERENCES Block, T. M., Marrero, J., Gish, R. G., Sherman, M., London, W. T., Srivastava, S., and Wagner, P. D. (2008). The degree of readiness of selected biomarkers for the early detection of hepatocellular carcinoma: Notes from a recent workshop. Cancer Biomark 4, 19–33. Boobis, A. R., Cohen, S. M., Dellarco, V., McGregor, D., Meek, M. E., Vickers, C., Willcocks, D., and Farland, W. (2006). IPCS framework for analyzing the relevance of a cancer mode of action for humans. Crit Rev Toxicol 36, 781–792. Cahill, D. P., Kinzler, K. W., Vogelstein, B., and Lengauer, C. (1999). Genetic instability and darwinian selection in tumours. Trends Cell Biol 9, M57–M60. Chen, Y., Zhu, J., Lum, P. Y., Yang, X., Pinto, S., MacNeil, D. J., Zhang, C., Lamb, J., Edwards, S., Sieberts, S. K., Leonardson, A., Castellini, L. W., Wang, S., Champy, M. F., Zhang, B., Emilsson, V., Doss, S., Ghazalpour, A., Horvath, S., Drake, T. A., Lusis, A. J., and Schadt, E. E. (2008). Variations in DNA elucidate molecular networks that cause disease. Nature 452, 429–435. Cohen, S. M., Boobis, A. R., Meek, M. E., Preston, R. J., and McGregor, D. B. (2006). 4-Aminobiphenyl and DNA reactivity: Case study within the context of the 2006 IPCS Human Relevance Framework for Analysis of a cancer mode of action for humans. Crit Rev Toxicol 36, 803–819. Conrad, D. H., Goyette, J., and Thomas, P. S. (2008). Proteomics as a method for early detection of cancer: A review of proteomics, exhaled breath condensate, and lung cancer screening. J Gen Intern Med 23 (Suppl 1), 78–84. Costa, J. L., Meijer, G., Ylstra, B., and Caldas, C. (2008). Array comparative genomic hybridization copy number profiling: A new tool for translational research in solid malignancies. Semin Radiat Oncol 18, 98–104. Cullings, H. M., Fujita, S., Funamoto, S., Grant, E. J., Kerr, G. D., and Preston, D. L. (2006). Dose estimation for atomic bomb survivor studies: Its evolution and present status. Radiat Res 166, 219–254. Department of Health and Human Services (DHHS). (1982). The Health Consequences of Smoking: Cancer. A Report of the Surgeon General. Washington, DC, pp. 17–20 et seq. Edwards, S. W., and Preston, R. J. (2008). Systems biology and mode of action based risk assessment. Toxicol Sci. 106, 312–318. EPA (U.S. Environmental Protection Agency) (1999). Guidelines for Carcinogen Risk Assessment. Risk Assessment Forum. SAB review draft. U.S. Environmental Protection Agency, Washington, DC. http:// www.epa.gov/ncea/raf/crasab.htm. EPA (2000). Science Policy Council Handbook: Risk Characterization, Office of Science Policy, Office of Research and Development, US Environmental Protection Agency, Washington, D.C., EPA 100B-00-002, http://www.epa.gov/osa/spc/pdfs/rchandbk.pdf. EPA (2005). Guidelines for Carcinogen Risk Assessment, Risk Assessment Forum, US Environmental Protection Agency, Washington, D.C., EPA/630/P-03/001F, http://oaspub.epa.gov/eims/eimscomm. getfile?p_download_id=439797.
REFERENCES
377
Faustman, E. M., Ponce, R. A., Seeley, M. R., and Whittaker, S. G. (1997). Experimental approaches to evaluate mechanisms of developmental toxicity. In Handbook of Developmental Toxicology, Hood, R., ed. CRC Press, Boca Raton, FL, pp. 13–41. Fearon, E. R., and Vogelstein, B. (1990). A genetic model for colorectal tumorigenesis. Cell 61, 759–767. Gatenby, R. A., and Vincent, T. L. (2003). An evolutionary model of carcinogenesis. Cancer Res 63, 6212–6220. Hanahan, D., and Weinberg, R. A. (2000). The hallmarks of cancer. Cell 100, 57–70. Hill, A. B. (1965). The environment and disease: Association or causation? Proc R Soc Med 58, 295–300. ICRP (2008). ICRP Publication 103: Recommendations of the ICRP. Ann ICRP 37, 2–4. Jacks, T. (2005). Modeling cancer in the mouse. Harvey Lect 101, 1–19. Maley, C. C., Galipeau, P. C., Li, X., Sanchez, C. A., Paulson, T. G., and Reid, B. J. (2004). Selectively advantageous mutations and hitchhikers in neoplasms: p16 lesions are selected in Barrett’s esophagus. Cancer Res 64, 3414–3427. Meek, M. E., Bucher, J. R., Cohen, S. M., Dellarco, V., Hill, R. N., Lehman-McKeeman, L. D., Longfellow, D. G., Pastoor, T., Seed, J., and Patton, D. E. (2003). A framework for human relevance analysis of information on carcinogenic modes of action. Crit Rev Toxicol 33, 591–653. NRC (2006). Health Risks from Exposure to Low Levels of Ionizing Radiation: BEIR VII Phase 2, Board on Radiation Effects Research (BRER), National Research Council of the National Academies, Washington, D.C., http://books.nap.edu/openbook.php?record_id=11340&page=R1. NRC (2007). Toxicity Testing in the 21st Century: A Vision and a Strategy, Board on Environmental Studies and Toxicology (BEST), Institute for Laboratory Animal Research (ILAR), National Research Council of the National Academies, Washington, D.C., http://books.nap.edu/openbook.php? record_id=11970&page=R1. NTP (2005). Report on Carcinogens, 11 edition, National Toxicology Program, Public Health Service, U.S. Department of Health and Human Services, Research Triangle Park, NC, http://ntp.niehs.nih.gov/ index.cfm?objectid=32BA9724-F1F6-975E-7FCE50709CB4C932. Preston, D. L., Cullings, H., Suyama, A., Funamoto, S., Nishi, N., Soda, M., Mabuchi, K., Kodama, K., Kasagi, F., and Shore, R. E. (2008). Solid cancer incidence in atomic bomb survivors exposed in utero or as young children. J Natl Cancer Inst 100, 428–436. Preston, D. L., Ron, E., Tokuoka, S., Funamoto, S., Nishi, N., Soda, M., Mabuchi, K., and Kodama, K. (2007). Solid cancer incidence in atomic bomb survivors: 1958–1998. Radiat Res 168, 1–64. Preston, R. J. (2005). Extrapolations are the Achilles heel of risk assessment. Mutat Res 589, 153–157. Preston, R. J. (2007). Cancer risk assessment for 1,3-butadiene: Data integration opportunities. Chem Biol Interact 166, 150–155. Preston, R. J., and Williams, G. M. (2005). DNA-reactive carcinogens: Mode of action and human cancer hazard. Crit Rev Toxicol 35, 673–683. Seed, J., Carney, E. W., Corley, R. A., Crofton, K. M., DeSesso, J. M., Foster, P. M., Kavlock, R., Kimmel, G., Klaunig, J., Meek, M. E., Preston, R. J., Slikker, W., Jr., Tabacova, S., Williams, G. M., Wiltse, J., Zoeller, R. T., Fenner-Crisp, P., and Patton, D. E. (2005). Overview: Using mode of action and life stage information to evaluate the human relevance of animal toxicity data. Crit Rev Toxicol 35, 664–672. Shim, S. Y., Lim, D. K., and Nam, J. M. (2008). Ultrasensitive optical biodiagnostic methods using metallic nanoparticles. Nanomed 3, 215–232. Sonich-Mullin, C., Fielder, R., Wiltse, J., Baetcke, K., Dempsey, J., Fenner-Crisp, P., Grant, D., Hartley, M., Knaap, A., Kroese, D., Mangelsdorf, I., Meek, E., Rice, J. M., and Younes, M. (2001). IPCS conceptual framework for evaluating a mode of action for chemical carcinogenesis. Regul Toxicol Pharmacol 34, 146–152. Vincent, T. L., and Gatenby, R. A. (2008). An evolutionary model for initiation, promotion, and progression in carcinogenesis. Int J Oncol 32, 729–737. Weinberg, R. A. (2006). The Biology of Cancer, Garland Science, New York, pp. 1–850. Wiltse, J. A., and Dellarco, V. L. (2000). U.S. Environmental Protection Agency’s revised guidelines for carcinogen risk assessment: Evaluating a postulated mode of carcinogenic action in guiding dose– response extrapolation. Mutat Res 464, 105–115.
CH A P TE R
14
EXPERIMENTAL ANIMAL STUDIES AND CARCINOGENICITY Mary Elizabeth (Bette) Meek
14.1.
INTRODUCTION
The mainstay of experimental studies on carcinogenicity in animals has been the long-term combined chronic/cancer bioassay in rats and mice, which has been designed principally as a basis to identify hazard (i.e., what is the intrinsic potential of the substance to induce cancer?). Results of such assays are generally combined in a weight of evidence approach with those of short-term principally in vitro assays, which identify potential for interaction with DNA, including the propensity to cause mutation (see Chapters 10 and 11). Cancer bioassays in animals, screening assays for genotoxicity, or the two of them together are conducted at relatively high doses as a basis to identify hazard. These studies fail, however, to provide robust dose– response information or even a minimum amount of the kinetic and dynamic information in a mode of action context, which would most meaningfully contribute to estimation of human risk. The requirement of mandates worldwide to systematically assess much larger numbers of chemicals necessitates more efficient and effective toxicity testing. This includes intelligent testing strategies to focus early on endpoints of interest, to consider “chemical space” in targeted investigation, to prioritize shorter-term in vivo assays of a range of intermediate endpoints based on consideration of mode of action of the chemical(s), and to better tailor mutagenicity testing as a basis to consider mode of action for cancer. Acquisition in traditional cancer bioassays or shorter-term investigations of mechanistic data on key events to better inform modes of induction of tumors as a basis for more accurate extrapolation of risks between doses, species, and subgroups of the population is also a priority. These intermediate approaches to better tailor and target testing as a basis for more informative characterization of risk in humans are essential prerequisites to meeting a broader, longer-term strategy for greater reliance on computational modeling and in vitro data in humans
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
378
14.2. CURRENT STATUS OF HAZARD TESTING FOR CANCER FOR REGULATORY RISK ASSESSMENT
379
envisaged, for example, in the US National Research Council’s (NRC)’s report entitled Toxicity Testing in the 21st Century—A Vision and a Strategy (NRC 2007b). In this chapter, the nature of experimental animal testing for cancer will be reviewed, trends examined, and recommendations for more progressive testing strategies included, based on consideration of evolving regulatory pressures and scientific advances in the context specifically of their application in risk assessment. Interim and pragmatic strategies to advance common understanding in both the research and risk assessment communities in potential appropriate application of evolving data, as well as the implications for testing strategies, are also considered.
14.2. CURRENT STATUS OF HAZARD TESTING FOR CANCER FOR REGULATORY RISK ASSESSMENT 14.2.1. The Combined Chronic/Cancer Bioassay in Rats and Mice While a range of methods has been explored as a basis to identify substances with potential carcinogenic potential, none has been sufficiently developed and/or validated to be able to supplant the combined chronic/cancer bioassay in rats (24 months) and mice (18 months), which continues to be the mainstay of chemical carcinogenicity testing. It involves combined evaluation of potential carcinogenicity and noncancer chronic toxicity, by a highly standardized method, which has been widely adopted throughout the world. These studies have been conducted since the 1960s with only very limited development of their protocols. Reliance on the combined chronic/cancer bioassay has been predicated predominantly on the basis of positive results for compounds that are known human carcinogens. Indeed, positive results in one or more adequately investigated animal species have been observed for all known human carcinogens (Vanio and Wilbourn 1994). Given their predominance in hazard identification for cancer, this section focuses principally on the objective and design of the combined chronic/cancer bioassay in rats and mice. Specifics of the design of these bioassays are presented in Table 14.1. The combined chronic/cancer bioassay is carried out almost TABLE 14.1.
Conditions: Route: Experimental animals: Number of animals: Dose levels:
Duration of exposure:
Design of Carcinogenicity Studiesa
• Chemical identification of substance, its purity and chemical characteristics • Oral (gavage, diet, drinking water or capsules), inhalation, or dermal • Rat, mouse, hamster, dog, or monkey • 50 rodents per sex per group; for nonrodents usually not more than 7–20 animals per dose group • Control and at least three dose levels, more dose levels for quantitative risk assessment • Satellite groups may be added • Majority of expected lifespan • Inhalation: intermittent (e.g., 6 hr/day, 5 days/week) or continuous
(Continued)
380
CHAPTER 14 EXPERIMENTAL ANIMAL STUDIES AND CARCINOGENICITY
TABLE 14.1.
Examinations:
Tissues normally collectedb:
Results:
(Continued) • Physical measurements: Temperature, humidity, homogeneity and stability of test substance, food and water consumption, and, for inhalation studies, air flow, concentrations, particle size • Clinical observations: Body weight; changes in skin, fur, eyes, mucous membranes, occurrence of secretions and excretions, behavior, respiratory, circulatory, autonomic and central nervous systems, somatomotor activity, sensory reactivity to stimuli; assessment of grip strength and motor activity; ophthalmologic examinations (90 d/chronic) • Hematology: Hematocrit, hemoglobin concentration, erythrocyte count, total and differential leukocyte count, platelet count, measure of blood clotting time • Clinical biochemistry: Investigation of organ function, carbohydrate metabolism, electrolyte balance, serum salts (Ca, P, Na, K, Cl), serum enzymes (such as alanine aminotransferase, aspartate aminotransferase, alkaline phosphatase, gamma glutamyl transpeptidase, sorbitol dehydrogenase, ornithine decarboxylase), cholesterol, glucose, urea, creatinine, total protein, albumin, total bilirubin (may be extended to lipids, hormones, acid/base balance, methemoglobin, cholinesterase activity) • Urinalysis (not routinely in 28-day tests): Appearance, volume, osmolality or specific gravity, pH, protein, glucose, blood cells • Pathology: Gross necropsy including external surfaces, orifices, cranial, thoracic and abdominal cavities and contents, organ weights • Recommended for microscopic examinations: 1. All grossly visible tumors or lesions suspected of being tumors in all groups; 2. (a) All preserved organs and tissues of all animals that die or are killed during the study. (b) All preserved organs and tissues of animals of the highest dose group and controls. (c) If a significant difference is observed in hyperplastic, preneoplastic, or neoplastic lesions between the highest dose and control groups, microscopic examination of that particular organ or tissue of all animals in the study. (d) In case the results of the experiment indicate substantial alteration of the animals’ normal longevity or the induction of effects that might affect a neoplastic response, the next lower dose level should be examined as described above. (e) The incidence of tumors and other suspect lesions normally occurring in the strain of animals used (under the same laboratory conditions—i.e., historical control) is desirable for assessing the significance of changes observed in exposed animals. • Adrenal glands, aorta, bone (femur, sternum), bone marrow, brain, carcass, cecum, colon, cervix, duodenum, ear canal, epdidymis, esophagus, eyes and optic nerves, Harderian gland, heart, ileum, jejunum, kidney, larynx, liver, lymph nodes (mandibular and mesenteric), lungs, mammary glands, nose/turbinates, oviducts, ovaries, pancreas, parathyroid, pituitary gland, prostate, salivary gland, sciatic nerve, seminal vescile, skin, skeletal muscle, spinal chord, spleen, stomach, testes, thymus, thyroid glands, tongue, trachea, urinary bladder, uterus, vagina, and Zymbal gland • Information on carcinogenic properties, tumor incidences in relation to dose, latency period, tumor multiplicity, potential for metastasis
Source: Modified from Vermeire et al. (2007)a or Hamm (1994).b
14.2. CURRENT STATUS OF HAZARD TESTING FOR CANCER FOR REGULATORY RISK ASSESSMENT
381
exclusively in rats and mice (currently most often F344 rats and B6CF1 mice), where both sexes are exposed. In addition to tumors, preneoplastic lesions and other indications of chronic toxicity providing evidence of treatment-related effects are also investigated (EPA 2005a; Vermeire et al. 2007). The bioassay involves exposure of weanling or post-weanling animals (beginning generally at 6–8 weeks of age) for the majority of their lifespan (24 months in rats or 18 months in mice). Fifty animals per sex per group are exposed to at least three dose levels and controls. The dose and route of administration for the chronic/ cancer bioassay is based largely on data obtained from prechronic toxicity studies with the same compound, normally encompassing both 14-day and 90-day exposures at a wide range of doses with evaluation of histopathological endpoints similar to those in the chronic study. Traditionally, since small numbers of animals per group in cancer bioassays are used as surrogates for a much larger human population, doses have exceeded considerably those associated with most human exposures. The lowest dose is normally selected so as not to interfere with growth and development nor to cause effects, whereas the highest dose is selected to result in signs of toxicity. With the exception of macronutrients, the highest dose should not exceed a concentration of 5% of the diet or 1 g/kg body weight for oral gavage studies (OECD 1981). The study is designed to include one dose in addition to the control(s) that is not expected to elicit adverse effects. The intermediate dose is normally within the mid range between the high and the low doses. The middle and lowest doses are selected (adequately spaced) to characterize the shape of the dose–response curve as much as possible. It is not uncommon to add a satellite high-dose group (20 animals per sex) to induce frank toxicity and a satellite control group (10 animals per sex per group) to evaluate effects other than neoplasia (usually after 12 months experimentation). Caging, care, feed, and water supply (diet) must be optimum and wellcontrolled. The rate of exposure to the substance is normally comparable to the anticipated human exposure, with frequency dependent on the route. In oral studies, the substance is administered daily unless by gavage, in which event exposure is usually restricted to 5 times a week, as is also characteristic of inhalation studies, where exposure is generally limited to 6 hours per day. Careful daily clinical examination is required and appropriate action is taken to minimize loss of animals during the study due to autolysis or cannibalization. Body weights are measured daily during the first 13 weeks and once every 4 weeks thereafter. Food and water intake are determined weekly during the first 13 weeks and then quarterly for the remainder of the study. Blood tests are performed after 3, 6, 18, and 24 months on 20 animals per sex per group, and a differential blood count is performed on samples of animals from the highest dose group and the controls and is performed at lower dose levels when indicated. Urine analysis of 10 animals per sex per group is conducted at the same intervals. Every 6 months, clinical chemical analysis is conducted (see Table 14.1). At the end of the experiment, a 50% survival rate is expected for rats at 24 months and mice at 18 months. Complete gross examination is performed, and histopathological examination is carried out on all tissues and organs from the
382
CHAPTER 14 EXPERIMENTAL ANIMAL STUDIES AND CARCINOGENICITY
highest dose group and the control group (see Table 14.1). Where indicated, the tissues and organs of lower dose groups are examined and all tumors or tumor-like lesions are examined histopathologically. Statistical analysis is performed for each tumor type separately. The incidence of benign and malignant lesions of the same cell type, usually within a single tissue or organ, are considered separately but may be combined when scientifically defensible (McConnell et al. 1986). Trend tests such as the Cochran–Armitage test are those recommended for determining whether chance, rather than a treatment-related effect, is a plausible explanation for an apparent increase in tumor incidence (Cochran and Snedecor 1972). By convention, a statistically significant comparison is generally one for which p is less than 0.05 that the increased incidence is due to chance. Significance in either kind of test is sufficient to reject the hypothesis that chance accounts for the result.
14.2.2.
Perinatal Carcinogenicity Studies
Current standardized long-term carcinogenesis bioassays involve initiation of dosing of animals at 6–8 weeks of age and throughout the lifespan of the animal (18–24 months). This protocol has been modified in some cases to investigate the potential of the test agent to induce transplacental carcinogenesis or potential differences following perinatal and adult exposures. However, standard protocols to investigate these aspects have not been developed, a function (in part) of experience that exposure during the perinatal period rarely identifies carcinogens that are not detected in standard animal bioassays, although it may increase the incidence of a given type of tumor or reduce the latency period for tumor development.
14.2.3.
Limited In Vivo Studies
The cost and duration of the combined chronic/cancer bioassay has limited its conduct to small numbers of selected chemicals. As a result, several short-term methods aimed at increasing predictive accuracy to enable testing of larger numbers of chemicals have been developed in attempts to successfully correlate their results with evidence of carcinogenicity (or lack of carcinogenicity). This includes investigation of potential to promote tumor development, several model systems in transgenic and knockout models, and consideration of the predictive potential of traditional toxicity endpoints in shorter-term studies. Limited, medium-term in vivo studies have been developed to investigate the tumor enhancing properties of chemicals. These involve administration of a known initiator or a genotoxic carcinogen in a subcarcinogenic dose, followed by exposure to the substance being examined. Several organ systems have been investigated (Kroes 1987) in these assays such as skin, lung, stomach, mammary gland, kidney, thyroid, pancreas, intestines, and urinary bladder (Feron et al. 1999). Short- and medium-term assays in transgenic models as a basis to provide essential information about the predisposing factors to specific genetic alterations in carcinogenesis include the rat liver foci model, the XPA−/− and the p53+/− knockout mouse models, the Tg.AC and Tg.rasH2 transgenic mouse models, and the neonatal
14.3. APPLICATION IN RISK ASSESSMENT
383
mouse model (Vermeire et al. 2007). These models have a number of potential advantages in carcinogen identification, including reduction of both the necessary periods of exposure and numbers of animals. Assay length is generally in the range of 24–26 weeks, significantly shorter than the standard chronic/cancer rodent bioassay. Furthermore, with appropriate model selection, relevant to mode of action of the substance, it is possible to more accurately predict the human response, contributing directly to their relevance to risk assessment and regulatory decision making (Gulezian et al. 2000). The capacity of shorter-term in vivo assays to predict carcinogenicity by investigating traditional toxicity endpoints has also been investigated (Ashby and Tennant 1994). Allen et al. (2004) evaluated the correlation of prechronic liver lesions and liver tumor formation from studies performed by the U.S. National Toxicology Program (NTP) in mice (83 compounds) and rats (87 compounds). Lesions considered included hepatocellular necrosis, hepatocellular hypertrophy, hepatocellular cytomegaly, bile duct hyperplasia, and hepatocellular degeneration, along with increased liver weight. Results indicated that pooling of the prechronic data on hepatocellular necrosis, hepatocellular hypertrophy, and hepatocellular cytomegaly was predictive of carcinogenicity in the 2-year study (p < 0.05) (Allen et al. 2004). To study tumor enhancing properties, various in vitro tests have also been proposed (Yamasaki 1990). They are based on the determination of clinical properties common to a group of promoting agents, such as loss of cell-to-cell communication and outgrowth of partially transformed cells.
14.3.
APPLICATION IN RISK ASSESSMENT
Risk assessment (i.e., the characterization of the potential adverse effects of human exposures) is the requisite basis for the development and implementation of control measures that are protective of public health (i.e., risk management). Traditionally, risk assessment has been considered to be composed of four elements, namely hazard identification, dose–response assessment, exposure estimation, and risk characterization (NRC 1983), with the latter being a synthesis of relevant data from all of the component steps with a clear delineation of uncertainties and their implications for risk management (see Chapter 1). While traditionally chronic/cancer bioassays have been designed to address hazard identification (i.e., the intrinsic capacity of a substance to cause harm), there is an increasing need to revise testing guidelines to integrate more hierarchical and predictive mode of action based approaches that will make them much more relevant to hazard characterization and subsequently, risk characterization.
14.3.1.
Hazard Identification
Hazard identification as determined from an adequate assessment of data from chronic/cancer bioassays in animals has been reviewed in Maronpot (1994), Health
384
CHAPTER 14 EXPERIMENTAL ANIMAL STUDIES AND CARCINOGENICITY
Canada (HC) (HC 1994), Meek et al. (1994) and the U.S. Environmental Protection Agency (EPA) (EPA 2005a). In considering the adequacy of specific investigations as a basis to identify hazard in risk assessment, several features of study design are considered including the purity of the compound administered, the size of the study (i.e., numbers of exposed and control animals), whether the study was performed under Good Laboratory Practice standards, the relevance of the route of exposure to that of humans, duration of exposure, the number and suitability of the dose levels administered, the extent of examination of various toxicological endpoints, and the statistical analysis of the data (HC 1994; Meek et al. 1994). Criteria for the technical adequacy of animal carcinogenicity studies have been published (e.g., Chhabra et al. 1990; NTP 1984; OSTP 1986). For specific chemicals, all available studies of carcinogenicity in whole animals are considered, at least preliminarily, with those being judged to be wholly inadequate in protocol, conduct, or results being discarded. Current standards of adequacy as well as those that were contemporaneous with the study are consulted. Care is taken to include studies that provide some evidence bearing on carcinogenicity or that are relevant to interpretation of effects noted in other bioassays, even if these investigations have some limitations of protocol or conduct. The findings of long-term rodent bioassays are interpreted in conjunction with results of prechronic studies along with toxicokinetic studies and other pertinent information, if available. Evaluation of tumor effects takes into consideration both biological and statistical significance of the findings (EPA 2005a). Among the many criteria for consideration of technical adequacy of animal carcinogenicity studies is the appropriateness of dose selection. This has been particularly important where results are negative, since traditionally it has been considered that lack of a sufficiently high dose reduces the sensitivity of the studies. A scientific rationale for dose selection is normally articulated based on relevant toxicologic information from prechronic, mechanistic, and toxicokinetic studies. It has generally been considered that an adequate high dose would be one that produces some toxic effects without unduly affecting mortality from effects other than cancer or producing significant adverse effects on the nutrition and health of the test animals (NRC 1993; OECD 1981). If the test agent does not appear to cause any specific target organ toxicity or perturbation of physiological function, an adequate high dose can be specified in terms of a percentage reduction of body weight gain over the lifespan of the animals. The high dose would generally be considered inadequate if neither toxicity nor changes in weight gain is observed. On the other hand, significant increases in mortality from effects other than cancer generally have been considered to indicate that an adequate high dose has been exceeded. Other signs of treatment-related toxicity associated with an excessive high dose may include: (a) significant reduction of body weight gain (e.g., greater than 10%), (b) significant increases in abnormal behavioral and clinical signs, (c) significant changes in hematology or clinical chemistry, (d) saturation of absorption and detoxification mechanisms, or (e) marked changes in organ weight, morphology, and histopathology. Overt toxicity or qualitatively altered toxicokinetics due to excessively high doses may result in tumors secondary to toxicity. Moreover, a lack of
14.3. APPLICATION IN RISK ASSESSMENT
385
tumorigenic response at exposure levels that cause significant impairment of animal survival may not be acceptable as a basis to consider results as negative, in view of the reduced sensitivity of the bioassay. There has been continuing controversy about the use of the “maximum tolerated dose” as the highest dose in chronic/cancer bioassays. Essentially, many consider that observation of cancer under conditions that are often unique to the experimental conditions provides little relevant information even in the context of hazard identification for humans, which are exposed to much lower doses (often six orders of magnitude). For dietary studies, weight gain reductions are considered in the context of whether there may be an issue of palatability. In the case of inhalation studies with respirable particles, evidence of impairment of normal clearance of particles from the lung should be considered along with other signs of toxicity to the respiratory airways to determine whether the high exposure concentration has been appropriately selected (EPA 2001). For dermal studies, evidence of skin irritation may indicate that an adequate high dose has been reached (EPA 1989). Statistical versus biological significance is also necessarily taken into account in interpreting the results of cancer bioassays. A statistically significant response may or may not be biologically significant and vice versa. The selection of significance levels to distinguish positive results is a matter of policy based on a trade-off between the risks of false positives and false negatives; the value most commonly adopted is 5%. A result with a significance level of greater or less than 5% (or other selected value), then, is examined to see if it is consistent with other scientific information. A two-tailed test or a one-tailed test can be used. In either case, a rationale is provided. Statistical power affects the likelihood that a statistically significant result could reasonably be expected. This is especially important in studies or dose groups with small sample sizes or low dose rates. Consideration of the statistical power is often essential for reconciling positive and negative results from different studies. The impact of multiple comparisons should also be taken into account. Based on analysis of typical bioassays in which both sexes of two species were included, studies in which there is only one significant result that falls short of the 1% level for a common tumor should be treated with caution (EPA 2005a; Haseman 1983). While the statistical significance of tumor incidence is judged based principally on comparison in dosed versus concurrent control animals, consideration of historical control data provides additional insight concerning both statistical and biological significance of uncommon tumors types or those with high spontaneous incidence in particular strains (Haseman 1995; Tarone 1982). It can be particularly helpful for cases where there are small increases (not reaching statistical significance) in uncommon tumors in treated groups compared with concurrent controls. However, since they do not take into account differences in survival of animals among studies, caution must be exercised in the interpretation of ranges of historical responses, the most relevant of which are derived from studies in the same laboratory with the same supplier conducted within 2 or 3 years of the study under review. Moreover, the degree of confidence in historical control data is necessarily related to the number of studies in the database. Aspects that need to be addressed
386
CHAPTER 14 EXPERIMENTAL ANIMAL STUDIES AND CARCINOGENICITY
in comparisons with historical control data include genetic drift in the laboratory strains, differences in pathology examination at different times and in different laboratories (e.g., in criteria for evaluating lesions; variations in the techniques for the preparation or reading of tissue samples among laboratories), and comparability of animals from different suppliers. The description and nature of peer engagement in review of the combined chronic/cancer bioassay is also critical in evaluation of the outcome. It is necessitated because of the judgmental processes that accrue in cascading fashion during the evaluation of the study in relation to the evaluation of the pathology but also in relation to the relative strength of the experimental evidence provided by the study (Maronpot 1994). The consistency of the results of the principal studies are also considered in the assessment of the weight of evidence for an effect (e.g., have similar effects been observed in studies in other species or would such effects have been expected based on the structure or properties of the chemical?), taking into account traditional criteria for weight of evidence including consistency, specificity, and biological plausibility. The types, site, incidence, and severity of effects and the nature of the exposure– or dose–response relationship are also taken into account. In assessing potential to induce tumors in humans, aspects that add to the weight of evidence include (a) observation of uncommon tumor types, (b) occurrence of tumors at multiple sites by more than one route of administration in multiple strains, sexes, and species, and (c) progression of lesions from preneoplastic to benign to malignant, including metastases and comparatively short latency periods. Traditionally, weight of evidence descriptors such as “carcinogenic to humans,” “probably carcinogenic to humans,” and so on, for cancer hazard have been developed by a number of agencies, as a basis for distinguishing approaches to dose–response analysis in subsequent risk characterization and also as a basis to communicate hazard (see Chapters 1 and 3). Increasingly, however, there is trend to providing more narrative and accurate descriptors, which include reference to the conditions under which cancer is observed, as a basis to avoid misinterpretation.
14.3.2.
Hazard Characterization
Hazard characterization necessarily takes into account not only results of guidelines studies designed to identify hazard but additionally, mechanistic data. Characterization of hazard involves a weight of evidence determination (i.e., a comprehensive, integrated judgment of all relevant information supporting conclusions regarding a toxicological effect, including human relevance, which takes into account traditional criteria for weight of evidence). While the standard combined chronic/cancer bioassay is helpful in hazard identification, it contributes in a more limited extent to hazard characterization (i.e., the likelihood of causing adverse effects in humans). However, with some modification in the context of evolving integrated and hierarchical test strategies for groups of chemicals or individual substances, carcinogenicity bioassays have potential to contribute considerably additionally in this context. For example, as discussed
14.3. APPLICATION IN RISK ASSESSMENT
387
above, traditionally in combined chronic/cancer bioassays, exposure to the test material has been maximized. This is helpful in contributing to confidence in the context of hazard identification that the study is negative for carcinogenicity if there is no increase in tumor incidence at a toxic high dose nor toxicity or tumors at appropriately spaced lower doses. However, this necessitates appropriate caution in interpreting results associated with excessive dosage levels that confound the interpretation of study results. Studies that result in tumors only at excessive doses may be compromised and may or may not carry weight, depending on the interpretation in the context of other study results and lines of evidence, including those informing mode(s) of action. Mechanistic data for cancer are considered in hazard characterization in the context of “mode” of induction of toxic effects. A postulated mode of action is a biologically plausible sequence of key events leading to an observed effect supported by robust experimental observations and mechanistic data. It describes key cytological, genetic, and biochemical events—that is, those that are both measurable and necessary to the observed effect. Mode of action is contrasted with mechanism of action, which generally involves a much greater understanding of the molecular basis for an effect. In 2001, as part of its efforts to harmonize risk assessment practices, the International Programme on Chemical Safety (IPCS) (WHO/ILO/UNEP) published a framework for assessment of mode of action for carcinogenesis in laboratory animals (animal mode of action). This was based on consideration of specific aspects of data analysis developed much earlier by Sir Austin Bradford Hill as a basis for considering causality of observed associations in epidemiological studies (Hill 1965). Relevant factors include dose–response and temporal concordance between key and end events, consistency, biological plausibility, and coherence (SonichMullin et al. 2001). More recently, the IPCS framework has been expanded to address human relevance (Boobis et al. 2006, 2008), based on previous work of the International Life Sciences Institute (ILSI) (Meek et al. 2003; Seed et al. 2005). The human relevance framework (HRF), which was developed and refined originally through its application in case studies for principally nonDNA reactive carcinogens, has been extended more recently to DNA-reactive carcinogens, noncancer endpoints, different life stages, and combined exposures to multiple chemicals. Development of the HRF for mode of action has involved engagement of more than 150 scientists internationally. It has also been widely incorporated into international and supra-national guidance and is being applied in this context as a basis to increase transparency concerning uncertainty, promote consistency in decision-making, facilitate peer engagement, and identify critical research needs (EC 2003; EFSA 2006; IPCS 2006; JMPR 2006; OECD 2002; UNECE 2007). It has also been extensively, even routinely, adopted in risk assessments by the U.S. EPA (Dellarco and Baetcke 2005; EPA 2000a,b, 2005b, 2007), the United Kingdom (COC, 2004), HC [see, for example, Liteplo and Meek (2003)], and other governmental organizations. The Society of Toxicology’s 2006 awards for Best Paper in Fundamental and Applied Toxicology and Toxicological Sciences, provides evidence of peer recognition of the contribution of the framework (Green et al. 2005; Pastoor et al. 2005).
388
CHAPTER 14 EXPERIMENTAL ANIMAL STUDIES AND CARCINOGENICITY
In this framework for mode of action analysis, the weight of evidence for a hypothesized mode of action for a particular response observed in experimental animals is considered in the context of key events along the causal pathway. Once established in experimental animals, the HRF provides an analytical tool to enable the systematic evaluation of the data in order to consider its human relevance based often on consideration of more generic information, such as anatomical, physiological, and biochemical variations among species. In this manner, the framework encourages maximum use of both chemical-specific and more generic information. The International Agency for Research on Cancer (IARC) has identified key biochemical and histopathological events as a basis to characterize specific modes of action (IARC 1999a,b). Established mechanistic/key events are additionally being identified in recent IARC monographs for individual chemicals and groups (Straif et al. 2009). Information on mode of action is relevant not only to determine whether tumors observed in animals are relevant to humans but also to consider dose transitions and potentially susceptible subgroups. It is also critical as a basis to address whether or not there is likely to be site concordance of tumors between animals and humans. While there is evidence that growth control mechanisms at the level of the cell are homologous among mammals, there is no evidence supporting nor reason to believe that these mechanisms are site-concordant. Instead, information on likely variations between animals and humans in kinetic and dynamics, based on some understanding of mode of induction of tumors, will inform in the context of potential sites of cancer induction in humans. This information is essential to interpret (particularly) the significance of negative epidemiological data, taking into account the sensitivity of the study to detect cancers at most likely sites. In the development of relevant biomarkers in epidemiological studies, it is also critical to increase the utility of the latter as a basis for consideration of the risks to exposure to chemical in both the occupational and general environments. While their use in hazard identification is necessarily limited owing to limitations such as relevance of the mutation in one pathway to the specific tumor, careful selection and interpretation of data from transgenic models in a mode of action context has considerable potential to contribute to hazard characterization.
14.3.3. Dose–Response Analyses; Selection of Points of Departure While the dose–response relationship observed in cancer bioassays is commonly used as the basis for risk characterization for substances that are considered as carcinogens, the extent to which it meaningfully informs risk is limited by the small number of dose groups and the magnitude of the variation between exposure of humans and administered doses. The limited numbers of doses examined is necessarily a function of the costs associated with close-to-lifetime observation of groups of (commonly) 50 animals each. Normally, characterization of dose–response analyses as a basis for comparison with exposure estimation in risk characterization is based on only those tumors where available data indicate that the mode of action is relevant to humans. Tumors
14.3. APPLICATION IN RISK ASSESSMENT
389
from similar tissues of origin may be combined as a basis of analysis. Where data indicate that there are significant differences in absorption, distribution, metabolism, and elimination of the compound in different animal species, wherever possible, studies in which the species and strain of animal are most similar to humans in this regard are used and differences quantitatively taken into account to the extent possible, through physiologically based pharmacokinetic or biologically based modeling. Generally, mathematical models are used to extrapolate the data on the exposure– or dose–response relationship derived from carcinogenicity bioassays to estimate the risk at concentrations to which the general population is exposed in the absence of more biologically based kinetic or dynamic models. There are numerous uncertainties in such approaches, which often involve linear extrapolation of results over several orders of magnitude, commonly in the absence of relevant data on mode of action for tumor induction or differences in toxico-kinetics and -dynamics between the relevant experimental animal species and humans. Concentrations or doses associated with a negligible or de minimis level of risk (such as a lifetime cancer risk of 1 in 1,000,000) by low-dose extrapolation procedures are often compared with exposure to determine whether risks are acceptable. Selection of appropriate de minimis levels constitutes science policy (i.e., making a societal judgment about what level constitutes de minimis risk). There is no single “correct” value that adequately characterizes de minimis risk associated with a concentration or dose below which risks are acceptable and above which they are not; rather, the risk at low doses or concentrations is assumed to be a continuum, with reduction of exposure leading to an incremental reduction of risk and increases in exposure leading to incremental increases in risk (see Chapter 26). In addition, in view of the considerable uncertainties of current low-dose extrapolation procedures, specification of risks in terms of predicted incidence or numbers of excess deaths per unit of the population is highly inaccurate and open to misinterpretation, particularly without specification of the bounds of uncertainty (HC 1994; Meek et al. 1994). Indeed, low-dose risk estimates based on empirical modeling and extrapolation of the dose–response curve over ranges of as much as six orders of magnitude have meaning in a relative (i.e., one to another) rather than absolute sense. For assessment of substances under the Canadian Environmental Protection Act (CEPA)—for example, for compounds that are carcinogenic involving direct interaction with DNA, where data are judged sufficient—quantitative estimates of the carcinogenic potency are compared to (a) the estimated daily intake of the priority substance by the general population (or certain high-exposure subgroups) in Canada or (b) concentrations in specific relevant environmental media [referred to as the exposure/potency index (EPI)]. Potency is expressed as the concentration or dose, which induces a 5% increase in the incidence of, or deaths due to, tumors or heritable mutations considered being associated with exposure. The tolerable dose (TD)0.05 is not based on the confidence limit but, rather, is computed directly from the curve. This was considered to be appropriate in view of the stability of the data in the experimental range and to avoid unnecessarily conservative assumptions. Also, use of a point estimate or confidence limit does not affect the relative magnitude of the potency estimates for different compounds. The estimates of potency are
390
CHAPTER 14 EXPERIMENTAL ANIMAL STUDIES AND CARCINOGENICITY
generally restricted to effects for which there has been a statistically significant increase in incidence and a dose-response relationship, characterized using appropriate mathematical models (e.g., multistage).
14.4.
EVOLUTION OF TESTING STRATEGIES
Conduct of the chronic bioassay by a defined protocol in similar rodent strains permits direct comparisons among chemicals with diverse structural and/or biological properties. However, there are considerable limitations, which restrict the number of substances evaluated with such studies. These include the high cost of a full 2-year bioassay in both sexes of rats and mice and the length of time required for testing and interpretation. These limitations necessarily reduce the throughput of chemicals. The significant numbers of animals sacrificed in 2-year bioassays to consider limited numbers of chemicals are also inconsistent with increasing pressure worldwide to reduce, refine, and replace animal testing. As a result (in part) of these considerations and increasing regulatory pressures to prioritize and consider potential risks associated with much larger numbers of substances more efficiently, toxicity testing continues to evolve from the use of prescribed protocols of whole-animal bioassays to greater emphasis on understanding the underlying pathways that lead to carcinogenesis, or other endpoints. Notable in this context is the content of the U.S. NRC report on toxicity testing (NRC 2007b), which advocates the identification and use of toxicity pathways for both testing and in dose response modeling. Much of the testing envisioned in this report entails in vitro studies (particularly, using tests based on high-throughput assays). These assays aim to characterize cellular processes and toxicity pathways more accurately by testing different levels of cellular function, including (a) genomics, the study of genes and their function as a whole; (b) proteomics, the large-scale study of proteins and their function; and (c) metabolomics, the study of all metabolites in a biological system that are being used to describe toxicant responses (see Part V, this volume). Computational biology techniques can be applied to these “-omics” data to link toxicity pathways and to identify patterns characteristic of specific toxicants. A key challenge to the use of findings from such tests will be the extrapolation of findings from in vitro studies to better understand and estimate human risks. One of the most important contributions of this new strategy is that it attempts to integrate exciting developments in toxicogenomics to increase efficiency and relevance of toxicity testing to risk assessment (NRC 2007a). The advocated use of human cells or tissues has potential to eliminate the need for interspecies extrapolation, to increase efficiencies in testing, and to reduce the use of animals. However, a pragmatic and seemingly essential first step in addressing this reevaluation of adversity would be a recommendation to relate (a) early perturbations to apical endpoints in frameworks designed to systematically address consideration of key events in modes of action and (b) their subsequent implications for dose– response in risk assessment [see, for example, Meek (2008)]. This would be instrumental in advancing common understanding in both the research and risk assessment
14.5. DISCUSSION: CLOSING THE GAP BETWEEN HAZARD TESTING AND RISK ASSESSMENT
391
communities in potential appropriate application of data on early events in a toxicity pathway. Increasing experience in this context could provide the necessary basis for revisiting guidelines for toxicity testing. Moving forward, it will also be critical to link and integrate this vision with other ongoing activities in regulatory risk assessment, including pragmatic developments in several jurisdictions (in particular in Canada and Europe) to (a) address progressive regulatory requirements to efficiently consider much larger numbers of chemical substances and (b) address critical challenges in moving the regulatory community toward the use of this approach and better balance of the focus on hazard with that on exposure. This includes (a) tools developed to consider priorities from amongst the 23,000 compounds included on the Domestic Substances list under the CEPA (Meek and Armstrong 2007) and (b) intelligent or integrated hierarchical testing strategies being developed in Europe for implementation of the legislation for Registration, Evaluation, and Authorization and restriction of Chemical substances (REACH) (see Chapter 3) (Van Leeuwen et al. 2007). Objectives of initiatives under these programs include maximally drawing upon existing data on toxicity, as a basis to increase efficiency. The former also considered prioritization on the basis of much simpler and more discerning data and tools for the significantly potentially more influential component of risk assessment, namely exposure estimation. While the predictive capacity of current computational technologies such as (quantitative) structure–activity relationship analysis (including the threshold of toxicological concern) (see Chapter 4) (Renwick et al. 2003) is necessarily limited currently owing principally to the nature of available toxicological data, their meaningful consideration has important implications for the design of future toxicity testing strategies including focus on coverage of “chemical space” versus individual substances as a critical criterion to increase efficiency and focus on in vitro testing strategies for particular modes of action for specific endpoints. These approaches also require limited new resources and promote more effective and efficient use of existing data as a basis to meaningfully contribute to early risk management.
14.5. DISCUSSION: CLOSING THE GAP BETWEEN HAZARD TESTING AND RISK ASSESSMENT Ultimately, toxicological testing for cancer aims to predict possible adverse effects in humans when exposed to chemicals. Currently, it is designed principally to identify hazards at relatively high doses. As a result, data derived from animal studies are limited with regard to informing the potential risks to humans. Given the limited relevance of output for considerable investment of resources, the development of more predictive testing strategies is inevitable and essential. Fundamentally, the assumptions inherent in the use of the results of long-term chronic/cancer bioassays in rodents as a basis for assessment of risk in human populations consist of the following: (1) If the agent causes cancer in rodents, it can cause cancer in humans (interspecies extrapolation), and (2) if the agent significantly increases cancer incidence when administered at high dose, it will also cause cancer, albeit likely at lower incidence, at low doses (interdose extrapolation). Unfortunately, the considerable
392
CHAPTER 14 EXPERIMENTAL ANIMAL STUDIES AND CARCINOGENICITY
amount of data generated in regulatory testing is of less than optimum relevance for informing these default assumptions. In the U.S. NRC report on toxicity testing, four major objectives were identified for future testing strategies (NRC 2007b): “depth, providing the most accurate, relevant information possible for hazard identification and dose–response assessment; breadth, providing data on the broadest possible universe of chemicals, end points, and life stages; animal welfare, causing the least animal suffering possible and using the fewest animals possible; and conservation, minimizing the expenditure of money and time on testing and regulatory review.” In the context of risk assessment, there is a need to additionally evolve this thinking to focus on hazard characterization rather than hazard identification. Obstacles to meeting the above-mentioned objectives with the current toxicity-testing paradigm identified by the U.S. NRC include issues surrounding interspecies extrapolation along with the animal welfare objective noted above. In addition, the breadth and conservation objectives are inconsistent with the time-consuming and expensive process of toxicity testing in animals. Early evolution of animal testing for cancer is also essential to meet immediate regulatory pressures. This necessarily requires increased understanding of the objectives and application of testing data from relevant bioassays in risk assessment/ regulatory programs. In particular, much more iterative and integrated testing strategies which include early consideration of mode of action more relevant to risk characterization and assessment than to hazard identification are required. In fact, there is a need for a paradigm shift to move in a scientifically credible and transparent manner from that which requires extensive hazard (animal) testing to one in which a hypothesis- and risk-driven approach can be used to identify the most relevant in vivo information (Van Leeuwen et al. 2007). Thus, it is critically important to efficiently and credibly predict toxicity drawing upon available information as a basis to facilitate reasonable decisions as to whether experimental studies are required to refine risk assessment further. The underlying rationale is to (1) minimize animal testing through introduction of alternative methods, (2) apply shorter-term and less expensive methods before labor-intensive ones, (3) design studies to address hazard characterization relevant to risk assessment, (4) enable early consideration of potential for exposure as a key determinant of testing strategies and risk assessment, (5) maximize the use of up-to-date information from different sources in an integrated manner, (6) allow greater flexibility in introducing new tools and scientific knowledge, and (7) allow more robust and focused regulatory decisions using testing and nontesting approaches. For cancer, it will require, additionally, much greater emphasis on testing in a mode of action context for tumors rather than reliance on chronic/cancer bioassays and principally screening assays for genotoxicity. These screening methodologies typically include as a minimum a battery of three assays: (a) a test for gene mutation in bacteria, (b) an in vitro test for mutation and/or chromosomal damage in mammalian cells, and (c) an in vivo test for chromosomal damage using rodent hematopoietic cells. Performance of these studies satisfies the aim for which they were first developed (i.e., to identify genotoxic agents that might pose cancer risk in humans as a basis to determine whether or not there should be additional testing or development); however, these studies are less informative in the context of
REFERENCES
393
consideration of mode of action for tumor induction at much lower doses. Experimental studies on mutagenesis and carcinogenesis are usually confined to exposures covering one to two orders of magnitude, and these doses are frequently high to establish hazard identification. In contrast, risk assessment extrapolations frequently cover up to six orders of magnitude. Better data on the dose–response for mutations and the subsequent utilization of these data in quantitative risk assessment will provide critical scientific information on dose–response relationships that are highly relevant for carcinogenesis (Swenberg et al. 2008). As a basis to better integrate regulatory objectives in toxicity testing, weightof-evidence frameworks that provide for transparent integration of hazard and modeof-action information such as those developed by the IPCS/ILSI [see, for example, Meek (2008)] are critically important. In the context of testing strategies for carcinogenicity, they illustrate the types of information that would inform to the greatest extent in the context of risk assessment. As we move forward to develop more integrative test strategies, early assimilation of the information in a mode-of-action context as emphasized by these frameworks will be essential. It is expected to encourage collection of information on toxicokinetics and early toxicodynamic key events at interim periods in animals exposed in similar fashion (e.g., satellite groups in which biochemical and histopathological evaluations are performed). Encouragement of much more intelligent testing of this nature is not new [see, for example, Hamm (1994) and Hill (1994)]. This is consistent with testing strategies to more meaningfully inform hazard characterization based on existing information and integrated testing strategies, which take into account results of computational predictive approaches. Application of the HRF increases the transparency of delineation of the relative degrees of uncertainty associated with various options for consideration in assessment of risk for impacted populations. HRFs are also instrumental in acquiring transparency on critical data gaps that will further reduce uncertainty. They force distinction of choices made on the basis of science policy versus those that are science judgment related, including reliance on default, based on the erroneous premise that it is always health-protective (Meek and Doull 2009). They focus on early events in a toxicity pathway through relation of early perturbations to apical endpoints in frameworks designed to systematically address (a) consideration of key events in modes of action and (b) their subsequent implications for dose-response in risk assessment [see, for example, Meek (2008)].
REFERENCES Allen, D. G., Pearse, G., Haseman, J. K., and Maronpot, R. R. (2004). Prediction of rodent carcinogenesis: An evaluation of prechronic liver lesions as forecasters of liver tumors in NTP carcinogenicity studies. Toxicol Pathol 32, 393–401. Ashby, J., and Tennant, R. W. (1994). Prediction of rodent carcinogenicity for 44 chemicals: Results. Mutagenesis 9, 7–15. Boobis, A. R., Cohen, S. M., Dellarco, V., McGregor, D., Meek, M. E., Vickers, C., Willcocks, D., and Farland, W. (2006). IPCS framework for analyzing the relevance of a cancer mode of action for humans. Crit Rev Toxicol 36, 781–792.
394
CHAPTER 14 EXPERIMENTAL ANIMAL STUDIES AND CARCINOGENICITY
Boobis, A. R., Doe, J. E., Heinrich-Hirsch, B., Meek, M. E., Munn, S., Ruchirawat, M., Schlatter, J., Seed, J., and Vickers, C. (2008). IPCS framework for analyzing the relevance of a noncancer mode of action for humans. Crit Rev Toxicol 38, 87–96. Chhabra, R. S., Huff, J. E., Schwetz, B. S., and Selkirk, J. (1990). An overview of prechronic and chronic toxicity/carcinogenicity experimental study designs and criteria used by the National Toxicology Program. Environ Health Perspect 86, 313–321. COC (Committee on Carcinogenicity) (2004). Guidance on a Strategy for the Risk Assessment of Chemical Carcinogens. London, UK, Department of Health. Cochran, W. G., and Snedecor, G. W. (1972). Statistical Methods, 6th edition, Iowa State University Press, Ames, IA. Dellarco, V. L., and Baetcke, K. (2005). A risk assessment perspective: application of mode of action and human relevance frameworks to the analysis of rodent tumor data. Toxicol Sci 86, 1–3. EC (2003). Technical Guidance Document on Risk Assessment, European Commission Joint Research Centre, Italy, EUR 20418 EN/1. EFSA (2006). Opinion of the Scientific Panel on Plant Health, Plant Protection Products and Their Residues on the scientific principles in the assessment and guidance provided in the field of human toxicology between 2003 and 2006. EFSA J 346, 1–13, http://www.efsa.europa.eu/EFSA/Scientific_ Opinion/ppr_op_ej346_summary-tox_summary_en1.pdf?ssbinary=true. EPA (1989). Summary of the second workshop carcinogenesis bioassay with the dermal route, May 18–19, 1988, Research Triangle Park, NC, EPA/560/6-89/003. EPA (2000a). Atrazine: Hazard and Dose–Response Assessment and Characterization, FIFRA Scientific Advisory Panel Meeting June 27–29, 2000, Held at the Sheraton Crystal City Hotel, Arlington, Virginia, SAP Report No. 2000-05, 1–44, http://www.epa.gov/oscpmont/sap/meetings/2000/june27/finalatrazine.pdf. EPA (2000b). Review of the draft chloroform risk assessment, EPA-SAB-EC-00-009, 1–33, http:// yosemite.epa.gov/sab/sabproduct.nsf/D0E41CF58569B1618525719B0064BC3A/$File/ec0009.pdf. EPA (2001). OPPTS 870.8355 Combined chronic toxicity/carcinogenicity testing of respirable fibrous particles, EPA712-C-01-352, 1–15, http://www.epa.gov/opptsfrs/publications/OPPTS_Harmonized/ 870_Health_Effects_Test_Guidelines/Series/870-8355.pdf. EPA (2005a). Guidelines for carcinogen risk assessment, EPA/630/P-03/001F, 1–166. EPA (2005b). Science Issue Paper: Mode of carcinogenic action for cacodylic acid (Dimethylarsinic acid, DMA[v]) and Recommendations for dose response extrapolation, Prepared by: Health Effects Division, Office of Pesticides Programs, US Environmental Protection Agency. 1–201, http://www.epa.gov/ oppsrrd1/reregistration/cacodylic_acid/dma_moa.pdf. EPA (2007). Advisory on EPA’s assessments of carcinogenic effects of organic and inorganic arsenic: A report of the US EPA Science Advisory Board, EPA-SAB-07-008, 1–88. Feron, V. J., Schwartz, M., Krewski, D., and Hemminki, K. (1999). Long- and medium-term carcinogenicity studies in animals and short-term genotoxicity tests. In Quantitative Estimation and Prediction of Human Cancer Risks, Vol. 131, Moolgavkar, S., Krewski, D., Zeise, L., Cardis, E., and Moller, H. eds., International Agency for Research on Cancer, World Health Organization, Lyon, pp. 103–112. Green, T., Toghill, A., Lee, R., Waechter, F., Weber, E., Peffer, R., Noakes, J., and Robinson, M. (2005). Thiamethoxam induced mouse liver tumors and their relevance to humans. Part 2: Species differences in response. Toxicol Sci 86, 48–55. Gulezian, D., Jacobson-Kram, D., McCullough, C. B., Olson, H., Recio, L., Robinson, D., Storer, R., Tennant, R., Ward, J. M., and Neumann, D. A. (2000). Use of transgenic animals for carcinogenicity testing: Considerations and implications for risk assessment. Toxicol Pathol 28, 482–499. Hamm, T. E. (1994). Design of a long-term animal bioassay for carcinogenicity. In Handbook of Carcinogen Testing, 2nd edition, Milman, H. A., and Weisburger, E. K., eds., William Andrew Publishing/Noyes, Norwich, NY, pp. 1–893. Haseman, J. K. (1983). A reexamination of false-positive rates for carcinogenesis studies. Fundam Appl Toxicol 3, 334–339. Haseman, J. K. (1995). Data analysis: Statistical analysis and use of historical control data. Regul Toxicol Pharmacol 21, 52–59; discussion 81–86. Health Canada HC (1994). Human health risk assessment for priority substances, En40-215/41E, 1–41, http://www.hc-sc.gc.ca/ewh-semt/alt_formats/hecs-sesc/pdf/pubs/contaminants/approach/approacheng.pdf.
REFERENCES
395
Hill, A. B. (1965). The environment and disease: association or causation? Proc R Soc Med 58, 295–300. Hill, R. N. (1994). Regulatory implications: Perspective of the U.S. Environmental Protection Agency. In Handbook of Carcinogen Testing, 2nd edition, Milman, H. A., and Weisburger, E. K., eds., William Andrew Publishing/Noyes, Norwich, NY, pp. 1–893. IARC, ed. (1999a). Species Differences in Thyroid, Kidney and Urinary Bladder Carcinogenesis, International Agency for Research on Cancer, World Health Organization, Lyon. IARC, ed. (1999b). The Use of Short- and Medium-Term Tests for Carcinogens and Data on Genetic Effects in Carcinogenic Hazard Evaluation, International Agency for Research on Cancer, World Health Organization, Lyon. IPCS (2006). Tetrachloroethene. Concise International Chemical Assessment Document 68, 1–123, http:// www.who.int/ipcs/publications/cicad/cicad68.pdf. JMPR (2006). Report of the Joint Meeting of the FAO Panel of Experts on Pesticide Residues in Food and the Environment and WHO the Core Assessment Group, FAO Plant Production and Protection Paper, 187 (Thiacloprid), ftp://ftp.fao.org/docrep/fao/010/a0888e/a0888e00.pdf. Kroes, R. (1987). Contribution of toxicology towards risk assessment of carcinogens. Arch Toxicol 60, 224–228. Liteplo, R. G., and Meek, M. E. (2003). Inhaled formaldehyde: Exposure estimation, hazard characterization, and exposure-response analysis. J Toxicol Environ Health B Crit Rev 6, 85–114. Maronpot, R. R., ed. (1994). Considerations in the Evaluation and Interpretation of Long-Term Animal Bioassays for Carcinogenicity, William Andrew Publishing/Noyes, Norwich, NY. McConnell, E. E., Solleveld, H. A., Swenberg, J. A., and Boorman, G. A. (1986). Guidelines for combining neoplasms for evaluation of rodent carcinogenesis studies. J Natl Cancer Inst 76, 283–289. Meek, B., and Doull, J. (2009). Pragmatic challenges for the vision of toxicity testing in the 21st century in a regulatory context: Another Ames test? … or a new edition of “the Red Book”? Toxicol Sci 108, 19–21. Meek, M. E. (2008). Recent developments in frameworks to consider human relevance of hypothesized modes of action for tumours in animals. Environ Mol Mutagen 49, 110–116. Meek, M. E., and Armstrong, V. C. (2007). The assessment and management of industrial chemicals in Canada. In Risk Assessment of Chemicals: An Introduction, Van Leeuwen, K., and Vermeire, T., eds., Kluwer Academic Publishers, Dordrecht, the Netherlands, pp. 591–621. Meek, M. E., Bucher, J. R., Cohen, S. M., Dellarco, V., Hill, R. N., Lehman-McKeeman, L. D., Longfellow, D. G., Pastoor, T., Seed, J., and Patton, D. E. (2003). A framework for human relevance analysis of information on carcinogenic modes of action. Crit Rev Toxicol 33, 591–653. Meek, M. E., Newhook, R., Liteplo, R. G., and Armstrong, V. C. (1994). Approach to assessment of risk to human health for priority substances under the Canadian environmental protection act. J Environ Sci Health, Part C: Environ Carcinog Ecotoxicol Rev 12, 105–134. NRC (1983). Risk Assessment in the Federal Government: Managing the Process Working Papers, National Academies Press, http://www.nap.edu/openbook.php?isbn=POD115&page=R1, Washington, D.C. NRC (1993). Issues in Risk Assessment, National Academies Press, http://books.nap.edu/openbook. php?record_id=2078&page=R1, Washington, D.C. NRC (2007a). Applications of Toxicogenomic Technologies to Predictive Toxicology and Risk Assessment, National Academies Press, http://books.nap.edu/openbook.php?record_id=12037&page=R1, Washington, D.C. NRC (2007b). Toxicity Testing in the 21st Century: A Vision and a Stategy, National Academies Press, Washington, D.C., pp. 1–196. NTP (1984). Report of the NTP Ad Hoc Panel on chemical carcinogenisis testing and evaluation. Board of Scientific Counselors, National Toxicology Program, pp. 1–280. OECD (1981). Carcinogenicity studies. OECD Guideline for Testing of Chemicals 451, 1–17. OECD (2002). Guidance Notes for Analysis and Evaluation of Chronic Toxicity and Carcinogenicity Studies. OECD Series on Testing and Assessment No. 35 and Series on Pesticides No. 14, ENV/JM/ MONO(2002)19. http://www.olis.oecd.org/olis/2002doc.nsf/LinkTo/env-jm-mono(2002)19. OSTP (1986). Chemical carcinogens: a review of the science and its associated principles. U.S. Interagency Staff Group on Carcinogens. Environ Health Perspect 67, 201–282.
396
CHAPTER 14 EXPERIMENTAL ANIMAL STUDIES AND CARCINOGENICITY
Pastoor, T., Rose, P., Lloyd, S., Peffer, R., and Green, T. (2005). Case study: Weight of evidence evaluation of the human health relevance of thiamethoxam-related mouse liver tumors. Toxicol Sci 86, 56–60. Renwick, A. G., Barlow, S. M., Hertz-Picciotto, I., Boobis, A. R., Dybing, E., Edler, L., Eisenbrand, G., Greig, J. B., Kleiner, J., Lambe, J., Muller, D. J., Smith, M. R., Tritscher, A., Tuijtelaars, S., van den Brandt, P. A., Walker, R., and Kroes, R. (2003). Risk characterisation of chemicals in food and diet. Food Chem Toxicol 41, 1211–1271. Seed, J., Carney, E. W., Corley, R. A., Crofton, K. M., DeSesso, J. M., Foster, P. M., Kavlock, R., Kimmel, G., Klaunig, J., Meek, M. E., Preston, R. J., Slikker, W., Jr., Tabacova, S., Williams, G. M., Wiltse, J., Zoeller, R. T., Fenner-Crisp, P., and Patton, D. E. (2005). Overview: Using mode of action and life stage information to evaluate the human relevance of animal toxicity data. Crit Rev Toxicol 35, 664–672. Sonich-Mullin, C., Fielder, R., Wiltse, J., Baetcke, K., Dempsey, J., Fenner-Crisp, P., Grant, D., Hartley, M., Knaap, A., Kroese, D., Mangelsdorf, I., Meek, E., Rice, J. M., Younes, M., and International Programme on Chemical, S. (2001). IPCS conceptual framework for evaluating a mode of action for chemical carcinogenesis. Regul Toxicol Pharmacol 34, 146–152. Straif, K., Benbrahim-Tallaa, L., Baan, R., Grosse, Y., Secretan, B., El Ghissassi, F., Bouvard, V., Guha, N., Freeman, C., Galichet, L., and Cogliano, V., on behalf of the WHO International Agency for Research on Cancer Monograph Working Group (2009). A review of human carcinogens—part C: metals, arsenic, dusts, and fibres. Lancet Oncol 10, 453–454. Swenberg, J. A., Fryar-Tita, E., Jeong, Y. C., Boysen, G., Starr, T., Walker, V. E., and Albertini, R. J. (2008). Biomarkers in toxicology and risk assessment: Informing critical dose-response relationships. Chem Res Toxicol 21, 253–265. Tarone, R. E. (1982). The use of historical control information in testing for a trend in proportions. Biometrics 38, 215–220. UNECE (2007). Amendments to the Globally Harmonized System of classification and labelling of chemicals (GHS). United Nations, Geneva. Document ST/SG/AC.10/34/Add.3, http://www.unece.org/ trans/danger/publi/ghs/ghs_rev01/01amend_e.html. Van Leeuwen, C. J., Patlewicz, G. Y., and Worth, A. P. (2007). Intelligent testing strategies. In Risk Assessment of Chemicals: An Introduction, van Leeuwen, K., and Vermeire, T., eds., Kluwer Academic Publishers, Dordrecht, the Netherlands, pp. 467–504. Vanio, H., and Wilbourn, J., eds. (1994). International Perspectives on Carcinogenicity Testing—A Brief Overview. William Andrew Publishing/Noyes, Norwich, NY. Vermeire, V. G., Baars, B. J., Bessems, J. G. M., Blaauboer, B. J., Slob, W., and Muller, J. J. A. (2007). Toxicity testing for human health risk assessment. In Risk Assessment of Chemicals: An Introduction, Van Leeuwen, K., and Vermeire, T., eds., Kluwer Academic Publishers, Dordrecht, the Netherlands, pp. 467–504. Yamasaki, H. (1990). Gap junctional intercellular communication and carcinogenesis. Carcinogenesis 11, 1051–1058.
CH A P TE R
15
CANCER EPIDEMIOLOGY Herman J. Gibb Jessie P. Buckley
15.1.
INTRODUCTION
Cancer is the leading cause of death worldwide (WHO 2008) and the second leading cause of death in the United States (Kung et al. 2008). There were 7.9 million deaths and 11.3 million new cases of cancer worldwide in 2007 with the number of cases expected to increase over the next 20 years (WHO 2008). Lung, liver, stomach, colon, and breast cancer are the five leading global causes of cancer mortality. The World Health Organization (WHO) has estimated that 40% of cancer deaths worldwide are preventable (WHO 2008). Although overall cancer incidence and mortality in the United States is declining, it has been estimated that there were 1.4 million new cases of cancer and 565,650 cancer deaths in 2008 (NCI 2007a, 2008a). In addition, the incidence of several types of cancer is increasing: non-Hodgkin’s lymphoma, leukemia, multiple myeloma, liver cancer, pancreatic cancer, kidney cancer, thyroid cancer, esophageal cancer, testicular cancer in men, melanoma and cancers of the brain and bladder in women, and childhood cancer (NCI 2007a). Cancer epidemiology is designed to identify cancer risks in human populations and determine causal links between cancers and specific exposures. A vast number of etiological agents are of interest in cancer epidemiology, including environmental and occupational factors, infectious agents, lifestyle factors such as nutrition, smoking, or exercise, and genetic factors. Epidemiology studies investigate what makes one group of people at higher risk than another group, determine whether observed relationships are causal, and measure the strength of association between exposure and disease. In 1775, in what is probably the earliest reported evidence of occupationally associated cancer, Percival Pott reported that cancer of the scrotum was particularly prevalent and occurred at an unusually early age among chimney sweeps (Lilienfeld et al. 1967). Pott also postulated that some characteristic of chimney sweeps was relevant to the production of the disease. Following Pott’s observations, other studies reported increased risks of cancer among certain occupationally and environmentally
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
397
398
CHAPTER 15 CANCER EPIDEMIOLOGY
exposed populations, but it was not until the middle of the 20th century that a substantial body of knowledge on cancer risk factors began to accumulate. Also, about that time, the increase in mortality from lung cancer had become such an issue that the disease took on the characteristics of an epidemic, and it was apparent that something more than industrial causes was to blame. By 1950, reports were published on the relationship between lung cancer and cigarette smoking which led to the intensive study of the carcinogenic effects of tobacco and eventually to the active discouragement of cigarette smoking as an integral part of preventive medicine. Today virtually every type of cancer has received some attention from an epidemiologic perspective. Because of the variation in the incidence of different forms of cancer, it became recognized that environmental, occupational, and lifestyle factors play a major role in cancer risk. As early as 1964, the WHO declared that 75% of all human cancer was affected by extrinsic factors. In 1965, the International Agency for Research on Cancer (IARC) was created to focus on human cancer and the relationship of humans to their environment. Although in vitro and animal studies may indicate that a chemical is carcinogenic, human data are the highest standard of evidence for determining the association between exposure to a hazard and development of disease. Regulatory agencies such as the U.S. Environmental Protection Agency (EPA), IARC, and the National Research Council (NRC) have all stated that epidemiologic studies are the most convincing evidence that a hazard exists (EPA 2005; IARC 1999; NRC 1983). This chapter discusses (1) important issues relating to the study of cancer, (2) types of epidemiology studies, (3) the determination of causal association from epidemiologic evidence, and (4) the future for cancer epidemiology.
15.2. CONSIDERATIONS FOR THE EPIDEMIOLOGIC STUDY OF CANCER Cancer is not a single disease, but rather a process common to a very heterogeneous group of diseases, differing widely in etiology, in frequency, in pattern of occurrence, and in clinical manifestations, as well as in the diagnostic and therapeutic problems that they present (Lilienfeld et al. 1967). These considerations, while presenting considerable challenges to the epidemiologist, make the study of the disease fascinating in its intricacy.
15.2.1.
Demographics
The most commonly used demographic variables in epidemiology are age, race, and gender. Because these variables are relatively easy to study, a considerable amount of data has been amassed on these factors as they relate to cancer risk. 15.2.1.1. Age. Cancer has often been called a disease of old age. For most cancers, the incidence does increase with age, but that is not the case for all cancers. For example, the median age at diagnosis of testicular cancer is 34, and incidence
15.2. CONSIDERATIONS FOR THE EPIDEMIOLOGIC STUDY OF CANCER
399
TABLE 15.1. Incidence Rate per 100,000 of Selected Cancers by Age, Race/Ethnicity, and Sex—United States, 2001–2005 (SEER, 2008)
Age <1 1–4 5–9 10–14 15–19 20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 65–69 70–74 75–79 80–84 85+ Race/ Ethnicity White Black Asian/ Pacific Islander American Indian/ Alaska Native Hispanic
Lung and Bronchus
Colon and Rectum
Kidney and Renal Pelvis
Breasta
Ovarya
Testicularb
Prostateb
— — — — 0.1 0.3 0.5 1.1 3.6 10.9 25.5 51.5 102.3 184.9 283.7 378.8 433.9 408.6 294.9
— — — 0.1 0.3 0.8 1.9 3.9 7.8 14.3 27.7 53.5 78.5 119.1 179.0 237.6 305.3 364.2 392.1
1.7 1.8 0.4 0.1 0.2 0.3 0.7 1.6 3.4 6.7 11.7 18.5 28.3 40.0 51.0 57.3 64.3 60.9 47.0
— — — — 0.2 1.4 7.8 26.1 58.9 117.6 185.3 234.4 299.7 359.9 402.3 423.9 453.1 435.9 352.8
— — 0.3 0.7 1.5 1.7 2.1 3.1 5.1 8.9 15.8 22.2 28.8 37.2 41.3 47.4 53.3 57.8 51.0
0.9 0.4 — 0.2 3.6 9.4 12.5 12.9 11.2 9.3 6.3 4.2 2.5 1.5 1.4 0.9 0.9 1.1 1.3
— — — — — — — — 0.8 8.7 40.2 136.4 328.2 573.9 877.4 983.7 970.8 816.3 676.3
F
F
M
M
14.1 10.1 9.8
6.3 1.4 1.7
156.7 248.5 93.8
M
F
M
F
M
F
79.3 54.9 58.9 43.2 18.8 9.5 130.6 107.6 54.6 71.2 54.5 21.3 10.1 117.5 53.9 28.0 48.0 35.4 9.1 4.6 89.6
54.3 39.7 46.0 41.2 19.5 12.7
75.0
11.3
4.2
73.3
44.2 25.4 47.3 32.8 17.4
90.1
11.7
3.9
138.0
9.6
a
Among females (F) only.
b
Among males (M) only.
is highest in 25 to 39-year-olds (Table 15.1). Even for those cancers that increase with age, the rate of change can vary considerably. The incidence of prostate cancer in males varies 1000-fold from age group 35–39 to 80–84, while the incidence of kidney cancer varies less than 20-fold over the same age groups (Table 15.1).
400
CHAPTER 15 CANCER EPIDEMIOLOGY
15.2.1.2. Race/Ethnicity. Cancer risk may vary considerably by race or ethnic group. From 2001 to 2005, the incidence of prostate cancer among blacks (248.5 per 100,000) was 60% higher than the rate among whites, 70% higher than the rate among Hispanics, and over three times the rate for American Indians and Alaska Natives (Table 15.1). The incidence of both breast and ovarian cancer is highest among Caucasian women. There may also be racial/ethnic differences in the location of tumors. For example, the most common type of esophageal cancer among African Americans is squamous cell carcinoma in the upper esophagus, whereas the most common type of esophageal cancer in Caucasian Americans is adenocarcinoma in the lower esophagus (Mayo Clinic 2007). In addition to genetic factors, cancer risk may differ by race and ethnicity for reasons related to social constructs (e.g., socioeconomic status, nutrition, behavior, occupation) (Kaufman and Cooper 2008). 15.2.1.3. Gender. Differences in cancer risk are also observed by gender due to a number of factors, such as physiological, occupational, or lifestyle differences. The incidence of lung, colorectal, and kidney and renal pelvis cancer among males is much higher than the incidence among females in every major race/ethnicity group in the United States (Table 15.1). Since 1990, lung cancer incidence and mortality rates in the United States have been declining among males but not among females (NCI 2007a). 15.2.1.4. Histologic Type. Even different histologic types of cancer can vary by age, race/ethnicity, and gender. As Table 15.2 demonstrates, acute lymphocytic leukemia predominates in those younger than 20 years of age, while the incidence of chronic lymphocytic leukemia is most common in those over the age of 55. The incidence of acute lymphoctyic leukemia in Hispanics is about two and a half times that in blacks, while chronic lymphocytic leukemia incidence is higher in blacks than in Hispanics. The incidence of both acute and chronic myeloid leukemia increases with age. The incidence of all four histologic types is higher in males than in females.
15.2.2.
Other Variables
It is generally recognized that the greatest reduction in cancer risk can be achieved by a modification of lifestyle factors. Tobacco smoking is overwhelmingly the most important risk factor, followed by diet in adult life (including obesity) and infectious agents. Together, these three factors are believed to be causally associated with over 50% of cancer deaths worldwide. Environmental pollution, popularly believed to be associated with an increased risk of cancer, is estimated to be responsible for about 3% of cancer deaths worldwide (Lagiou et al. 2005). The variables discussed below are not intended to be a comprehensive list of factors that may affect cancer risk, but they provide the reader with an idea of the complexity of influences on the epidemiologic study of cancer. 15.2.2.1. Smoking. Lung cancer is the second most common form of cancer and the most common cause of cancer death in the United States (NCI 2008b). Cigarette smoking is associated with an increased risk of various cancers (Table 15.3). Tobacco is estimated to cause one-third of all cancer
15.2. CONSIDERATIONS FOR THE EPIDEMIOLOGIC STUDY OF CANCER
401
TABLE 15.2. Proportion of Leukemia Incidence by Age Group for Four Histological Types; Incidence of Leukemia by Race/Ethnicity and Gender for Four Histological Types—United States, 2001–2005 (SEER, 2008)
Acute Lymphocytic
Chronic Lymphocytic
Acute Myeloid
Chronic Myeloid
61.0 10.0 6.5 6.3 5.9 4.9 3.7 1.7
0.0 0.3 1.8 8.9 19.1 26.8 30.1 12.9
6.3 6.3 6.8 10.9 14.8 21.1 24.6 9.2
2.5 7.1 10.4 12.6 14.1 19.6 23.8 10.0
Agea <20 20–34 35–44 45–54 55–64 65–74 75–84 85+ Race/Ethnicityb White Black Asian/Pacific Islander American Indian/ Alaska Native Hispanic
M 2.0 0.9 1.6
F 1.5 0.8 1.3
M 6 4.4 1.2
F 3 2.2 0.6
M 4.6 3.8 3.7
F 3 2.6 2.5
M 2 2 1.3
F 1.2 1.1 0.7
1.9
1.2
1.9
—
1.9
2.4
—
—
2.4
2.0
2.7
1.4
3.5
2.6
1.5
1
a
Proportion of total cases.
b
Rate per 100,000 per year. M, male; F, female.
TABLE 15.3. Number of Deaths and Smoking Attributable Mortality (SAM) of Selected Cancers in Males and Females—United States, 1997–2001 (CDC, 2005)
Males
Females
Cause of Death (ICD-10 Code)
Deaths
SAM
%
Deaths
SAM
%
Lip, oral cavity, pharynx (C00-C14) Esophagus (C15) Stomach (C16) Pancreas (C25) Larynx (C32) Trachea, lung, bronchus (C33-C34) Cervix uteri (C53) Kidney, other urinary (C64-C65) Urinary bladder (C67) Acute myeloid leukemia (C92.0) Total
4,973 9,037 7,403 13,984 3,017 89,912
3,686 6,533 2,052 3,078 2,499 79,026
74.1 72.3 27.7 22.0 82.8 87.9
7,169 8,025 3,447 146,967
2,790 3,764 791 104,219
38.9 46.9 22.9 70.9
2,525 2,854 5,223 14,774 816 63,181 3,989 4,454 3,841 2,919 104,576
1,182 1,625 600 3,431 596 44,810 491 222 1,054 299 54,310
46.8 56.9 11.5 23.2 73.0 70.9 12.3 5.0 27.4 10.2 51.9
402
CHAPTER 15 CANCER EPIDEMIOLOGY
deaths in developed countries (Peto 2001). Environmental tobacco smoke is also associated with an increased risk of cancer, and it has been estimated that a nonsmoker ’s risk of lung cancer increases 20–30% by living with a smoker (NCI 2007b). 15.2.2.2. Diet. A diet high in fruits and vegetables may reduce the risk for several types of cancers, including those of the colon, esophagus, lung, mouth, and stomach. Alternatively, being overweight or obese is associated with increased risk of breast, colorectal, endometrial, esophageal, and kidney cancer and may also increase the risk of several other cancers (e.g., cervix, gallbladder, Hodgkin’s lymphoma, multiple myeloma, ovary, pancreas, prostate, thyroid). Alcohol is associated with an increased risk of breast and colorectal cancer (ACS 2007). 15.2.2.3. Infectious Agents. IARC has classified several viruses as known human carcinogens (Group 1), including Epstein–Barr virus; hepatitis B virus; hepatitis C virus; human immunodeficiency virus (HIV) type 1; human papilloma viruses (HPV) types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59 and 66; and human T-cell lymphotropic virus (HTLV) type I (IARC 2008). Kaposi’s sarcoma herpes virus/ human herpes virus 8 is probably carcinogenic to humans (Group 2A); and HIV type 2, HPV types 6 and 11, and HPV genus beta are possibly carcinogenic to humans (Group 2B) (IARC 2008). Simian virus (SV 40) has been associated with mesothelioma (Peto 2001). Helicobacter pylori, a chronic gastric bacterial infection that can cause gastric ulcers, is a major factor in the development of stomach cancer (IARC 1994). About 15% of cancers worldwide are believed to be due to infectious agents (Parkin et al. 1999). 15.2.2.4. Hormonal Differences. Hormonal differences are known to affect the risk of various cancers including endometrial, ovarian, and breast cancer. Use of oral contraceptives decreases the risk for endometrial cancer, but increased estrogen as a result of obesity or hormone replacement therapy, especially when not combined with progestin, is a strong risk factor for endometrial cancer (ACS 2008; Peto 2001). Studies show that women who have had children, who breastfeed, or who use hormonal contraceptives (like birth control pills) are less likely to develop ovarian cancer (Cleveland Clinic 2008).* Factors that increase breast cancer risk include being overweight or obese after menopause, use of estrogen-containing oral contraceptives, and use of postmenopausal hormone therapy (especially combined estrogen and progestin therapy) (ACS 2008; Peto 2001). Breast cancer risk has shown to be decreased by late menarche, early menopause, early first childbirth, and high parity (Peto 2001). *These factors decrease the number of times a woman ovulates, and studies suggest that reducing the number of ovulations during a woman’s lifetime may lower the risk of ovarian cancer.
15.3. EPIDEMIOLOGIC STUDY METHODS
403
TABLE 15.4. Inherited Genetic Mutations Associated with Increased Risk for Selected Cancers (ACS 2008; NCI 2008c–e)
Cancer Site
Inherited Genetic Mutation
Breast
Colon
Endometrium Lymphoma Multiple endocrine neoplasia type 2 Ovary Prostate
BRCA 1, BRCA2 Tumor suppressor genes: APC, AXIN2, TP53 (p53), STK11, PTEN, BMPR1A, and SMAD4 (DPC4) Repair/Stability genes: hMLH1, hMSH2, hMSH6, PMS2, MYH (MutYH), and BLM Oncogenes: KIT and PDGFRA Hereditary nonpolyposis colon cancer (HNPCC) Family history of lymphoma and certain common genetic variations in immune response genes RET proto-oncogene BRCA1, BRCA2 genes Hereditary Prostate Cancer 1, Prostate Cancer Predisposing Locus, Hereditary Prostate Cancer X, CAPB, ELAC2/HPC2, HPC20, 8p Loci, 8q, BRCA1 and BRCA2, KLF6
15.2.2.5. Inherited Genetic Alterations. About 5% of all cancers are strongly hereditary, in that an inherited genetic alteration confers a very high risk of developing one or more specific types of cancer (ACS 2008). Table 15.4 describes risks for various cancer sites and the inherited genetic alteration that has been associated with the increased risk for that site.
15.3.
EPIDEMIOLOGIC STUDY METHODS
The purpose of epidemiology is to study the distribution and determinants of disease in a population. Ethical considerations dictate that cancer epidemiology studies are usually observational as opposed to experimental studies.
15.3.1.
Types of Epidemiologic Studies
15.3.1.1. Cohort. Cohort studies examine the difference in disease occurrence over time between exposed persons and unexposed persons. Cohort studies may be prospective (exposure information is collected at the beginning of the study and continued until the end of the follow-up period) or retrospective (exposure information is collected from historical records) (IARC 1999). Cohort studies of cancer tend to be retrospective because most cancers have long latency periods between exposure and disease onset. This study design is commonly used for cancer investigations
404
CHAPTER 15 CANCER EPIDEMIOLOGY
TABLE 15.5. Five-Year Survival Rates for Selected Cancers—United States, 1996–2004 (SEER 2008)
Cancer Brain and other nervous system Breasta Colon and rectum Kidney and renal pelvis Liver and intrahepatic bile duct Lung and bronchus Lymphoma Ovarya Pancreas Prostateb Testicularb
Survival (%) 34.1 88.7 64.4 66.5 11.7 15.2 67.7 45.5 5.1 98.9 95.5
a
Among females (F) only.
b
Among males (M) only.
when exposure and medical records are readily available or the study population is fixed, such as in the case of an occupational cohort. Retrospective mortality studies are particularly common in cancer epidemiology. Cohort study designs are generally employed to study more common cancers such as lung or bladder cancer. In cohort studies, it is important that the groups being compared are as similar as possible with respect to all factors that may be related to the disease other than the factor which is under study. Mortality data are frequently used in cohort studies as opposed to incidence* data because of the relative ease of obtaining information on deaths. In particular, the advent of the National Death Index† has made mortality data more readily available. While all states maintain registries of incident cancer cases, many of the registries are relatively new and data quality can vary from state to state. Investigators conducting follow-up studies are required to comply with each state’s requirements for use of the data. For cancers such as pancreatic cancer where survival is poor, mortality data is an excellent surrogate for the risk of the disease.‡ For other cancers where the survival is much better, such as testicular cancer, mortality is a poor estimator of incidence. Table 15.5 describes the 5-year survival for several selected cancer sites for the period 1996–2004. As evident from the table, there is a considerable difference in survival rates for cancer of different sites. The 5-year survival rate for pancreatic cancer was only 5.1% compared to a survival rate of 98.9% for prostate cancer (SEER 2008). *Incidence is the number of new cases in a population for a given period of time, usually a year. † The National Death Index (NDI), a service provided by the National Center for Health Statistics, is a central computerized index of death record information on file in the state vital statistics offices. The NDI covers deaths in the United States beginning in 1979. ‡
Risk of disease is measured by the incidence and not the mortality.
15.3. EPIDEMIOLOGIC STUDY METHODS
405
To compare the incidence or mortality for a disease of interest between two populations, the data are adjusted to allow for differences between the two populations with respect to age, gender, race, and other variables. The adjustment may be direct or indirect. The reader is referred to WHO (1999) for a description of the two methods. The indirect method, which is more commonly used, generates measures of association known as standardized mortality ratios (SMRs) and standardized incidence ratios (SIRs). These ratios compare the mortality or incidence for the disease of interest in the population of interest with that which would be expected based on an external population.
15.3.1.2. Case–Control. Case–control studies of cancer generally compare persons with cancer (cases) to persons without cancer (controls) to identify factors or exposures that differ between groups. The case–control design allows researchers to investigate multiple exposures in relation to the disease of interest. Case–control studies are particularly useful to study rare tumors such as mesothelioma or pancreatic cancer since a cohort design may not produce enough cases to detect significant differences between groups. Another advantage to case–control studies over cohort studies is that the number of subjects necessary is much smaller than that needed for cohort studies. The most important concern of the case–control study is the selection of controls. Controls should have the same exposure distribution as the population from which the cases are drawn; otherwise there is the potential for selection bias. Cancers other than the cancer of interest have been used as controls in case-control studies. The advantages of the use of other cancers as controls has been described by Smith et al. (1988). The measure of association (risk) in a case–control study is known as the odds ratio, which is the odds of exposure in the cases divided by the odds of exposure in the controls. Similar to the discussion for the cohort study, odds ratios are adjusted for age, gender, race, and other variables.
15.3.1.3. Proportionate Mortality and Proportionate Incidence Studies. In a proportionate mortality or proportionate incidence study, one compares the proportion of deaths or incident cases due to a condition of interest with that expected based on deaths or incident cases in an external, usually the general, population. When the proportions of causes of death are compared, the ratio is known as the proportionate mortality ratio (PMR). The comparison of proportions of incident cases is known as the proportionate incidence ratio (PIR). Proportional measures can be misleading since a decrement in the proportion of deaths or incident cases due to a particular cause will de facto lead to an increase in the proportion due to another cause. The commonly held view with regard to PMRs is that they are good approximations to SMRs from cohort studies when the cohort’s all-causes combined SMR is equal to 1.0 (Checkoway et al. 1989). The odds ratio has also been used as a measure of association in PMR or PIR studies and may be a more appropriate measure of association for evaluating proportional measures (WHO 1999).
406
CHAPTER 15 CANCER EPIDEMIOLOGY
15.3.1.4. Cross-Sectional (Prevalence). Cross-sectional (prevalence) studies assess relationships between prevalence of exposure and disease at a single point in time or over a short period of time. Cross-sectional studies are generally conducted to learn about risk factors for diseases of slow onset and long duration for which medical care is often not sought until the disease has progressed to a relatively advanced stage such as osteoarthritis or chronic bronchitis. Cancer is generally of slow onset, but medical care is usually sought early. Regardless of the disease endpoint, a limitation of cross-sectional studies is that measurement of exposure and disease is made at the same point in time and it is impossible to ascertain whether exposure preceded disease. This is a particular problem in cross-sectional studies of cancer, since the putative exposure likely preceded the disease by a long period of time, even decades, and current exposures do not necessarily reflect those in the past. Also regardless of the endpoint studied, cross-sectional studies include a higher proportion of cases with long duration than would a study of incident cases. This bias is problematic when survival is poor, since cases with better survival will be overrepresented in the study population. Cross-sectional studies, while they may not be appropriate for the study of cancer etiology, are used for screening for cancer risk factors. Screening is a form of tertiary prevention in which asymptomatic people are classified as likely or unlikely to have the cancer of interest. Although screening programs do not prevent disease, they are designed and implemented to reduce cancer-related morbidity and mortality. Screening tests must be followed by diagnostic testing to determine whether an individual found likely to have cancer actually does. Screening tests are useful when (1) the cancer causes a significant health burden, has a long detectable preclinical period, and there is an effective treatment available, and (2) the test is reliable, inexpensive, and causes minimal pain. To maximize the utility of screening tests, they are often recommended for populations already known to have an increased risk of disease (e.g., breast cancer screening for women with a family history). 15.3.1.5. Molecular Epidemiology. Molecular epidemiology uses biomarkers to determine relationships between genetic and environmental risk factors and disease. Molecular epidemiology studies are generally cross-sectional in design, but molecular approaches can be employed in cohort or case–control studies as well. Biomarkers are cellular, biochemical, or molecular changes that can be measured in human samples, such as tissues, cells, or fluids (Hulka et al. 1990). Biomarkers have been classified into four groups: biomarkers of (1) internal dose, (2) biologically effective dose, (3) preclinical biologic effects, and (4) susceptibility (Perera and Weinstein 2000). The NRC of the US National Academy of Sciences (NAS) adopted this paradigm and expanded it to include another category: altered structure and function (NRC 1987; Vineis and Perera 2007). Biomarkers of exposure, such as urinary arsenic, have been the biomarkers most widely used. For most cancers, the long latency period between exposure and disease makes it difficult to use exposure biomarkers in etiologic studies. Therefore, molecular epidemiology is most useful for exposures that remain relatively constant over time (e.g., genetic traits) or when biologic samples are collected upon initiation of a cohort study and analyzed after sufficient time has passed for cancer to develop.
15.4. EVALUATION OF STUDIES AND THEIR RESULTS
407
15.3.1.6. Ecologic Analyses. While the cohort, case–control, and crosssectional studies consider individual exposure, ecologic studies examine disease occurrence within a group. This design examines population prevalence of cancer in relation to exposure levels in groups of people, or in the same group over time. Due to the lack of individual level data, it is unknown whether the persons with the disease are the same persons with the exposure of interest. The error of falsely applying associations in a population to the individual is termed ecological fallacy. Because these studies have more limitations and uncertainties than other observational studies, they are often used to generate hypotheses rather than to examine causal relationships. There are, however, some notable exceptions. Perhaps the most well known is the study of skin cancer in an arsenic-endemic area of Taiwan (Tseng et al. 1968; Tseng 1977); other ecologic studies of arsenic in drinking water (Chen et al. 1985; Smith et al. 1998; Wu et al. 1989; Hopenhayn-Rich et al. 1996, 1998) have also made significant contributions to establishing a causal association between ingested arsenic and an increased risk of cancer.
15.3.2.
Meta-analysis and Case Reports
15.3.2.1. Meta-Analysis. Meta-analyses (or pooled analyses) synthesize and compare studies that investigate similar health effects and risk factors. These analyses examine sources of heterogeneity and may clarify relationships between exposures and health effects. Inclusion and exclusion criteria and data analysis methods must be transparently described. Meta-analyses are not helpful when the relationship between exposure and disease is obvious, there are few studies, or the existing studies suffer from severe methodological faults (Blair et al. 1995). Dickersin (2002) described problems of meta-analysis and proposed various areas for improvement and study. 15.3.2.2. Case Reports. Case reports are not studies. They describe the exposures of an individual or group of persons to a substance or lifestyle that may be related to a disease or effect. For example, physicians may report a rare case of disease in a patient with a certain exposure. Case series are commonly reported for occupational or childhood cancers. Although case reports may be useful for hypothesis generating, particularly in the case of rare tumors, they are selective, often lack investigative follow-up, and are not considered evidence of a causal association. The EPA’s Guidelines for Carcinogen Risk Assessment state that “case reports are often anecdotal or highly selective in nature and generally of limited use for hazard assessment” (EPA 2005).
15.4. 15.4.1.
EVALUATION OF STUDIES AND THEIR RESULTS Quality of Studies
No single criterion can be used to judge an epidemiology study. Instead, many aspects of study design, conduct, and analysis must be evaluated to determine the quality and utility of an analytic or descriptive investigation. The EPA’s (2005)
408
CHAPTER 15 CANCER EPIDEMIOLOGY
Guidelines for Carcinogen Risk Assessment note the following considerations of well-performed studies: • • • • • • • •
Clear articulation of study objectives or hypothesis Proper selection and characterization of comparison groups Adequate characterization of exposure Sufficient length of follow-up for disease occurrence Valid ascertainment of the causes of cancer morbidity and mortality Proper consideration of bias and confounding factors Adequate sample size to detect an effect Clear, well-documented, and appropriate methodology for data collection and analysis • Adequate response rate and methodology for handling missing data • Complete and clear documentation of results
15.4.2.
Determining Causal Association
The subject of what evidence is necessary to conclude that an exposure is causally associated with disease has received much discussion over the years. In 1964, the seminal report to the Surgeon General on Smoking and Health [United States Department of Health, Education, and Welfare (DHEW 1964)] published criteria by which to evaluate whether an exposure was causally related to disease. These criteria were: consistency, strength of association, specificity of the association, temporal relationship of the association, and coherence of the association. In 1965, Sir Austin Bradford Hill expanded on these criteria in his “Environment and Disease: Association or Causation?” (Hill 1965). The article has been widely cited in journal articles, by health risk assessors, and by health risk assessment guidance including that of EPA (2005) and WHO (1999). In his article, Hill described what he referred to as aspects of an association between an environmental exposure and disease that should be considered before determining that the environmental exposure is causally associated with the disease. These “aspects” have commonly been referred to as criteria in the literature, although Hill never referred to them as such. The aspects that Hill described are: • Strength of Association. Hill provided examples demonstrating strength of association where there were high relative risks of disease (e.g., a 200-fold increase in the risk of scrotal cancer among chimney sweeps). He cautioned, however, that one must not dismiss a causal association merely on the grounds that the observed association appears to be slight. • Consistency. This aspect was described as associations that are repeatedly observed by different persons, in different places, circumstances, and times. Hill also stated, however, that there will be occasions when repetition is absent or impossible and a causal association is still credible.
15.4. EVALUATION OF STUDIES AND THEIR RESULTS
409
• Specificity. When the association is limited to specific workers and to particular sites and types of disease and there is no association between the work and other modes of dying, Hill argued that this is a strong argument in favor of causation. Again, however, he provided caution in that environmental exposures may be associated with more than one disease and that diseases may have more than one cause so that if specificity “is not apparent, we are not thereby necessarily left sitting irresolutely on the fence.” • Temporality. Does exposure to the environmental factor precede the disease? No cautions were offered by Hill on this aspect, and it is the one aspect that risk assessors all agree must occur for a causal association to exist. • Biological Gradient. This aspect is generally thought of by risk assessors as exposure response (i.e., the more exposure, the greater the response). Hill stated that often the difficulty in determining a biological gradient is the securing of satisfactory quantitative measures of exposure that will permit us to explore a dose–response. Most epidemiologists would readily agree. • Plausibility. By plausible, Hill was referring to biological plausibility but noted that biological plausibility depends on the biological knowledge of the day. For example, there was no biological knowledge to support (or refute) the observation by Percival Pott in the 18th century of an excess risk of scrotal cancer among chimney sweeps. • Coherence. Hill described coherence as a consistency with the known natural history and biology of the disease. He gives as an example the increase in lung cancer incidence correlating with a rise in cigarette consumption. Other examples provided could fit well under the plausibility aspect above and the distinction between the two is somewhat blurred. • Experiment. By experiment, Hill meant that if an exposure was withdrawn or reduced, there would be a corresponding elimination or reduction in disease risk. Hill stated that in this aspect the strongest support for causation may be revealed. • Analogy. Hill gives as an example of analogy that the effects of rubella or thalidomide should allow us to accept slighter but similar evidence with another drug or another viral disease in pregnancy. It is important to note Hill’s (1965) words on these nine aspects: Here then are nine different viewpoints from all of which we should study association before we cry causation. What I do not believe—and this has been suggested—that we can usefully lay down some hard-and-fast rules of evidence that must be obeyed before we can accept cause and effect. None of my nine viewpoints can bring indisputable evidence for or against the cause-and-effect hypothesis and none can be required as a sine qua non. What they can do, with greater or less strength, is to help us to make up our minds on the fundamental question—is there any other way of explaining the set of facts before us, is there any other answer equally, or more, likely than cause and effect?
Criteria for evaluating causal associations are helpful as guidance but should not be used as checklists. Lanes and Poole (1984) claim that attempts to codify
410
CHAPTER 15 CANCER EPIDEMIOLOGY
interpretation of epidemiologic studies by rote application of a checklist can only result in diminished understanding. Rothman and Greenland (2005) indicate that there may be a temptation to use causal criteria to simply buttress pet theories. Maclure (1985) argues that such criteria should be interpreted as criteria for refuting causal interpretations, rather than the inductive approach of drawing a conclusion from the application of the criteria to the results. Neither the Smoking and Health Report (DHEW 1964) nor Hill (1965) recommended the use of their “criteria” as checklists; nevertheless, in practice, there continue to be attempts to do so. Weed and Gorelic (1996) examined two series of review papers—one on alcohol and breast cancer, the other on vasectomy and prostate cancer—to examine how causal inference is practiced. The intent was to answer several questions including which causal criteria were used and what causal conclusions were drawn. The authors reported that sources of causal criteria were often not provided but that when the sources were identified, they were either Hill (1965) or the Smoking and Health (DHEW 1964) report. The authors claimed that the reviews often excluded and sometimes altered criteria without reason. The most frequently cited criteria were consistency, strength of association, dose–response, and biologic plausibility, whereas the criterion of temporality—which is considered to be a necessary causal condition—was infrequently used. Confounding and bias were often added considerations. Indeed, systematic bias is often greater than random error, and these biases must be assessed before drawing conclusions from the epidemiologic data. Lagiou et al. (2005) state that criteria for causality can be used in evaluating the results of a single epidemiologic study, but a firm conclusion from a single study is rarely possible. The criteria for causality are more frequently used for the assessment of evidence accumulated from several epidemiologic investigations and other biomedical investigations. This inductive process (evaluating a causal association from several studies) will often move away from the specifics of the studies to generalize to other situations. Lagiou et al. (2005) caution that regulatory agencies and policy makers may recommend standards, set limits, or authorize action even when the scientific evidence is inconclusive and that these procedures, while they serve public health by introducing a wide safety margin, should not be confused with the establishment of causation. The generalization of results of several epidemiologic studies to other situations is problematic. Environmental exposures, for example, will likely be considerably different than occupational exposures, both in intensity and in duration of exposure. Exposures to a particular substance will vary considerably among occupations as well, and thus a causal association found with respect to a particular substance may not apply to other occupational exposures to that substance. A substance found to be carcinogenic by one route of exposure may not be carcinogenic by another route of exposure for biologic reasons. While causality can be conclusively established between a particular exposure as an entity and a particular disease as an entity, it is not possible to conclusively establish such a link between an individual exposure and a particular disease of a certain individual. It is possible, however, to deductively infer that a specific individual’s illness more likely than not was caused by the specified exposure. Lagiou et al. (2005) describe several criteria for evaluating whether the disease in an individual could have been caused by a specified exposure.
15.5. SUBSTANCES CAUSALLY ASSOCIATED WITH CANCER
411
15.5. SUBSTANCES CAUSALLY ASSOCIATED WITH CANCER Epidemiologic data are the most useful in risk assessment because they evaluate risks for the species of concern (humans). Institutions such as the EPA, the National Toxicology Program (NTP), and IARC have used qualitative descriptors or classifications to describe the carcinogenic weight of evidence for different substances. Although the classifications differ somewhat, they all rely on human and animal evidence to determine carcinogenicity and give human data, for obvious reasons, the greatest weight. The EPA’s Integrated Risk Information System (IRIS), which evaluates the carcinogenicity of substances found in the environment,* has classified only 12 substances as known human carcinogens (Table 15.6). It is notable that these epidemiology studies are predominantly occupational, with only one substance (arsenic) having supporting environmental epidemiology studies. Human data are often obtained from studies of workers since exposure levels are higher than typically found in the environment or food. The most common cancer site among known carcinogens is the lung, likely because inhalation is the most common route of exposure in occupational settings. All of the results in humans are consistent with route-specific results in animal studies, the notable exception being arsenic. Most of the studies on which these evaluations are based are over 20 years old. As of April 2008, the IRIS program listed 74 substances as probable or likely carcinogens. Of these, diesel engine exhaust was determined to have “strong, but less than sufficient” epidemiologic evidence of an association with an increased lung cancer risk† (the evidence was not deemed sufficient due to concerns about use of historical exposure data and whether the studies adequately controlled for smoking and other confounders). For five other substances, limited human data were used as the basis of the classification. Four of these are considered probable‡ lung carcinogens (acrylonitrile, beryllium and compounds, cadmium, and formaldehyde) and one is considered a probable skin carcinogen (creosote). Forty-three substances have
*The EPA Guidelines for Carcinogen Risk Assessment classify substances using the following categories: carcinogenic to humans, likely to be carcinogenic to humans, suggestive evidence of carcinogenic potential, inadequate information to assess carcinogenic potential, or not likely to be carcinogenic to humans (EPA 2005). Prior to the 2005 Guidelines, the Agency used the 1999 Draft Guidelines for Carcinogen Risk Assessment; prior to that, it used the 1986 Guidelines for Carcinogen Risk Assessment. The classifications used in the 1999 Guidelines were: carcinogenic to humans; likely to be carcinogenic to humans; sufficient evidence of carcinogenicity, but not sufficient to assess human carcinogenic potential; data are inadequate for assessment of human carcinogenic potential; not likely to be carcinogenic to humans. The classifications used in the 1986 Guidelines were: (A) human carcinogen, (B) probable human carcinogen (included categories of B1 and B2. B1 was generally reserved for situations where there was limited human evidence of carcinogenicity), (C) possible human carcinogen, (D) not classifiable as to human carcinogenicity, (E) evidence of noncarcinogenicity for humans. † The classification of diesel engine exhaust followed EPA’s (1999) Draft Guidelines for Carcinogen Risk Assessment. ‡ The classification of acrylonitrile, beryllium and compounds, cadmium, formaldehyde, and creosote followed EPA’s (1986) Guidelines for Carcinogen Risk Assessment.
412 TABLE 15.6.
Substances Characterized as Carcinogenic to Humans by the U.S. EPA IRIS Program
Guideline Year
Epidemiology
Typea
Carcinogenic to Humans 1,3-Butadiene 1999 Arsenic, inorganic 1986
Sufficient Sufficient
O O, E
Inhalation Inhalation, oral, dermal
Asbestos
1986
Sufficient
O, E
Inhalation
Benzene
1986/1996
Sufficient
O
All routes
Benzidine Bis(chloromethyl) ether (BCME) Chloromethyl methyl ether (CMME) Chromium(VI) Coke oven emissions
1986 1986
Sufficient Sufficient
O O
1986
Sufficient
1996 1986
Nickel refinery dust
1986
Substance
Route
Cancer
Same Route and Exposure Positive in Animals?
Y (lymphoma) N
Not specified Not specified
Leukemia Lung cancer; also liver, kidney, bladder, and skin cancer Lung cancer, mesothelioma, gastrointestinal cancer Acute nonlymphocytic leukemia Bladder Lung cancer
O
Not specified
Lung cancerb
Y
Sufficient Sufficient
O O
Inhalation Not specified
Y Y (lung)
Sufficient
O
Not specified
Lung cancer Lung, trachea, bronchus, kidney, and prostate cancer Lung and nasal tumors
Y (lung and mesothelioma)
Y (hematopoietic) Y Y
Y (lung, some evidence)
Substance
Guideline Year
Epidemiology
Nickel subsulfide Vinyl chloride
1986 1986/1996
Sufficient Sufficient
Cancer
O O
Not specified Inhalation, oral (likely dermal)
Lung and nasal tumors Liver cancer, specifically angiosarcoma
Y (lung, some evidence) Y
Limited
O
Not specified
Lung cancer
1986/1996
Limited
O
Inhalation
Lung cancer
N (brain tumors by oral and inhalation) Y
1986 1986 1999
Limited Limited Strong
O O O
Inhalation Not specified Inhalation
Lung cancer Skin cancer Lung cancer
Y Y Y
1986
Limited
O, E
Not specified
Respiratory neoplasms
Y (nasal cavity squamous cell carcinoma)
a
E, environmental; O, occupational.
b
CMME is always contaminated with BCME; carcinogenicity may be due to either substance.
c
Same Route and Exposure Positive in Animals?
Route
Likely to be Carcinogenic to Humansc Acrylonitrile 1986 Beryllium and compounds Cadmium Creosote Diesel engine exhaust Formaldehyde
Typea
Only substances with supporting human data are listed.
413
414
CHAPTER 15 CANCER EPIDEMIOLOGY
some human data that are considered inadequate. Most commonly, these data are case reports or epidemiology studies with major methodological flaws. The remaining 25 probable or likely carcinogens have no human data whatsoever. The IARC has determined that more than 100 agents, groups of agents, mixtures, or exposure circumstances are carcinogenic to humans. In addition to chemical carcinogens, IARC has identified biological (e.g., aflatoxins, Helicobacter pylori, HPVs), lifestyle (e.g., ethanol consumption, smoking, ultraviolet radiation), and medicinal (e.g., estrogen therapy, oral contraceptives, tamoxifen) causes of cancer (IARC 2008). For all of these agents, human data were used to make the determination that they are known human carcinogens. The NTP, in its Eleventh Annual Report on Carcinogens (NTP 2005), reported that there were 54 known human carcinogens. These included the same 12 known human carcinogens identified by EPA’s IRIS as well as some of the biological, lifestyle, and medicinal agents evaluated by IARC. NTP’s known carcinogen list also included some environmental substances that at the time of this writing were either not evaluated by EPA (e.g., coal tars, erionite) or were currently being reevaluated by EPA (e.g., beryllium, ethylene oxide).
15.6.
FUTURE FOR CANCER EPIDEMIOLOGY
An increased awareness of the health effects of chemical exposures has led to a reduction of chemical exposures in the workplace and the environment. While there are undoubtedly substances that are yet to be causally associated with cancer, the number of substances like those described in Table 15.6 is unlikely to grow significantly in the next decade. So where does the future for cancer epidemiology lie?
15.6.1.
The Effect of Exposure at Different Ages
Most cancer epidemiology studies rely on cumulative dose as the dose metric. Under this convention for example, 1 μg/m3 for 10 years is equivalent to 10 μg/m3 for one year (i.e., both equal 10 μg/m3-years). Are both exposures the same, however? Does a high exposure for a short period of time present more risk than the equivalent dose spread over a longer period of time? Is the risk the same if exposure occurs at an older age as when it occurs at a younger age? The concept of late-stage and earlystage carcinogens suggest that some carcinogens will have a greater effect at an older age, and some carcinogens will have a greater effect at an early age (Castro et al. 2008; Gaylor 2007; Brown and Chu 1987). Using models and animal data, Bos et al. (2004) made estimates of risk following high dose rate exposures to genotoxic carcinogens for a young child and estimated considerably higher risks for children exposed to genotoxic carcinogens. Chen (2008) describes how the time from cessation of exposure to the onset of disease can be used to better characterize the role of the age of initiation and duration of exposure to uncover hidden biological implications of disease endpoints. Chen and Gibb (2003) describe methods to estimate how many cases of cancer are avoided at age t after cessation (or reduction)
15.6. FUTURE FOR CANCER EPIDEMIOLOGY
415
of exposure utilizing knowledge of the age of exposure to the carcinogen. The authors also estimate how long the effect of an exposure will last after exposure is terminated (or reduced). The exploration of the effects of dose rate at different ages and the effect that such exposures have following cessation of exposure will provide a better understanding of the mechanism of action of carcinogens. It will also provide regulators an understanding of the benefits to be achieved by reduction or elimination of carcinogenic exposure.
15.6.2.
Molecular Epidemiology
Molecular epidemiology has contributed to our understanding of chemical exposure (biomarkers of exposure), early changes preceding disease (biomarkers of effect), and the identification of susceptible subgroups (biomarkers of susceptibility). The majority of available biomarkers used in molecular epidemiology studies relate to agents that cause DNA damage and are mutagenic, but a large number of chemicals can enhance or inhibit the carcinogenic process through indirect genotoxic or epigenetic mechanisms (Perera and Weinstein 2000). New biomarkers are becoming available for epidemiologic studies including toxicogenomics, alterations in gene methylation and gene expression, proteomics, and metabonomics. Most of these newer biomarkers have not been validated, however, and their role in the causal paradigm is unclear (Vineis and Perera 2007). The integration of these exciting new discoveries with public health science makes the causal criterion, biologic plausibility, an increasingly important consideration in causal inference (Lagiou et al. 2005). In particular, at which levels is evidence relatively more important than others and at any given level, what is the best (i.e., strongest) type of evidence?
15.6.3.
Infectious Agents
Infectious agents are associated with increased risks of stomach cancer, liver cancer, bladder cancer, nasopharyngeal cancer, leukemia, non-Hodgkin’s lymphoma, and Hodgkin’s lymphoma. It has been estimated that worldwide 88% of cases with cancer of the cervix and vulva, 77% of cases with cancer of the liver, 42% of cases with cancer of the stomach, and 10% cases with lymphoma and leukemia are attributed to infectious agents. There is a higher percentage of cancer cases attributed to infection in developing countries (22.5%) than in developed countries, but the percentage in developed countries is not insignificant (6.8%) (Parkin et al. 1999). The interaction of viruses and chemical agents may be synergistic such as that of chronic HBV infection and aflatoxin exposure on the risk of hepatocellular carcinoma (Chen et al. 1997). For other infectious agents, indirect mechanisms have been proposed. For example, HIV is not oncogenic, but the immunodeficiency associated with infection results in an excess of cancers related to other agents. The bacteria and parasites appear to act via chronic inflammation caused by infection. Identification of these infectious agents, their mechanism of action, and what relationship they may have with other agents (e.g., carcinogens) appears to be another fertile area for study.
416
CHAPTER 15 CANCER EPIDEMIOLOGY
REFERENCES ACS (2007). Nutrition and Cancer. American Cancer Society. Available at http://www.cancer.org/ docroot/PRO/content/PRO_1_1x_Nutrition.pdf.asp?sitearea=PRO. Accessed January 19, 2009. ACS (2008). Cancer Facts and Figures. American Cancer Society. Available at http://www.cancer.org/ downloads/STT/2008CAFFfinalsecured.pdf. Accessed January 19, 2009. Blair, A., Burg, J., Foran, J., Gibb, H., Greenland, S., Morris, R., Raabe, G., Savitz, D., Teta, J., Wartenberg, D., Wong, O., and Zimmerman, R. (1995). Guidelines for application of meta-analysis in environmental epidemiology. Regul Toxicol Pharmacol 22, 189–197. Bos, P. M. J., Baars, B. J., and van Raaij, M. T. M. (2004). Risk assessment of peak exposure to genotoxic carcinogens: A pragmatic approach. Toxicol Lett 151 (1), 43–50. Brown, C. C., and Chu, K. C. (1987). Additive and multiplicative models and multistage carcinogenesis theory. Risk Anal 9 (1), 99–105. Castro, D. J., Löhr, C. V., Fischer, K. A., Perera, C. B., and Williams, D. E. (2008). Lymphoma and lung cancer in offspring born to pregnant mice dosed with dibenzo[a,l]pyrene: The importance of in utero vs. lactational exposure. Toxicol Appl Pharmacol 233 (3), 454–458. CDC (2005). Annual smoking-attributable mortality, years of potential life lost, and productivity losses— United States, 1997–2001. Centers for Disease Control and Prevention, MMWR Morb Mortal Wkly Rep 54 (25), 625–628. Checkoway, H., Pearce, N. E., and Crawford-Brown, D. J. (1989). Research methods in occupational epidemiology. Oxford University Press, New York. Chen, C. J., Yu, M. W., and Liaw, Y. F. (1997). Epidemiological characteristics and risk factors of hepatocellular carcinoma. J Gastroenterol Hepatol Suppl 12 (9/10), S294–S308. Chen, C. J., Chuang, Y. C., Lin, T. M., and Wu, H. Y. (1985). Malignant neoplasms among residents of a blackfoot disease-endemic area in Taiwan: High-arsenic artesian well water and cancers. Cancer Res 45, 5895–5899. Chen, C. W. (2008). Lingering effect: Epidemiological information useful for risk assessment. Regul Toxicol Pharmacol 52 (3), 242–247. Chen, C. W., and Gibb, H. J. (2003). Procedures for calculation cessation lag. Regul Toxicol Pharmacol 38 (2), 157–165. Cleveland Clinic. (2008). Menopause and Ovarian Cancer. Cleveland Clinic Center for Consumer Health Information. Available at http://www.clevelandclinic.org/health/healthinfo/docs/3000/3009. asp?index=10111. Accessed January 19, 2009. DHEW (1964). Smoking and Health. Surgeon General’s Advisory Committee on Smoking and Health, United States Public Health Service, Office of the Surgeon General, Public Health Service Publication No. 1103. Available at http://profiles.nlm.nih.gov/NN/B/C/X/B/. Accessed January 21, 2009. Dickersin, K. (2002). Systematic reviews in epidemiology: Why are we so far behind? Int J Epidemiol 31, 6–12. EPA (1986). Guidelines for carcinogen risk assessment. Federal Register 51 (185), 33992–34003. Available at http://cfpub.epa.gov/ncea/raf/recordisplay.cfm?deid=116283. Accessed January 19, 2009. EPA (1999). Guidelines for carcinogen risk assessment. NCEA-F-0644. Risk Assessment Forum, US Environmental Protection Agency, Washington, D.C. EPA (2005). Guidelines for carcinogen risk assessment. Risk Assessment Forum, US Environmental Protection Agency, Washington, D.C. EPA/630/P-03/001B, March 2005. Available at http://www.epa. gov/IRIS/cancer032505.pdf. Accessed January 19, 2009. Gaylor, D. (2007). Quantification of factors affecting uncertainties of cancer risk estimates. Society for Risk Analysis Annual Meeting, December 9–12, 2007, San Antonio, TX, Abstract No. T3-D.1. Hill, A. B. (1965). The environment and disease: association or causation? J R Soc Med 58, 295–300. Available at http://www.edwardtufte.com/tufte/hill. Accessed January 19, 2009. Hopenhayn-Rich, C., Biggs, M. L., and Smith, A. H. (1998). Lung and kidney cancer mortality associated with arsenic in drinking water in Córdoba, Argentina. Int J Epidemiol 27, 561–569. Hopenhayn-Rich, C., Biggs, M. L., Fuchs, A., Bergoglio, E. E., Tello, E. E., Nicolli, H., Smith, A. H. (1996). Bladder cancer mortality associated with arsenic in drinking water in Argentina. Epidemiology 7, 117–124.
REFERENCES
417
Hulka, B. S., Griffith, J. D., and Wilcosky, T. C. (1990). Biologic markers in epidemiology. Oxford University Press, New York. IARC (1994). Schistosomes, Liver Flukes, and Heliobacter pylori, IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, 61, IARC Press, Lyon, France. IARC (1999). Cancer Epidemiology: Principles and Methods, dos Santos Silva, I., ed., International Agency for Research on Cancer, World Health Organization. IARC Press, Lyon, France. IARC (2008). Overall evaluations of carcinogenicity to humans. Group 1: carcinogenic to humans. International Agency for Research on Cancer. Last update: March 28, 2008. Available at http:// monographs.iarc.fr/ENG/Classification/crthgr01.php. Accessed January 19, 2009. Kaufman, J. S., and Cooper, R. S. (2008). Race in epidemiology: New tools, old problems. Ann Epidemiol 18 (2), 119–123. Kung, H. C., Hoyert, D. L., Xu, J. Q., and Murphy, S. L. (2008). Deaths: Final data for 2005. National vital statistics reports, Vol. 56, No. 10, Hyattsville, MD, National Center for Health Statistics. Lagiou, P., Adami, H. O., and Trichopoulos, D. (2005). Causality in cancer epidemiology. Eur J Epidemiol 20, 565–574. Lanes, S. F., and Poole, C. (1984). “Truth in packaging?” The unwrapping of epidemiologic research. J Occup Med 26, 571–574. Lilienfeld, A. M., Pedersen, E., and Dowd, J. E. (1967). Cancer epidemiology: Methods of study. The Johns Hopkins Press, Baltimore, MD. Maclure, M. (1985). Popperian refutation in epidemiology. Am J Epidemiol 121 (3), 43–350. Mayo Clinic (2007). Esophageal cancer. Accessed May 9, 2008. Last updated May 11, 2007. Available at http://www.mayoclinic.com/health/esophageal-cancer/DS00500/DSECTION=3. Accessed January 19, 2009. NCI (2007a). Cancer Trends Progress Report—2007 Update, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, MD. Available at http:// progressreport.cancer.gov. Accessed January 19, 2009. NCI (2007b). Secondhand Smoke: Questions and Answers. National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, MD. Available at http://www.cancer. gov/cancertopics/factsheet/Tobacco/ETS. Accessed January 19, 2009. NCI (2008a). Defining Cancer. National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, MD. Available at http://www.cancer.gov/cancertopics/what-iscancer. Accessed January 19, 2009. NCI (2008b). A snapshot of lung cancer. Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, MD. Available at http://planning.cancer.gov/disease/LungSnapshot.pdf. Accessed January 20, 2009. NCI (2008c). Genetics of Colorectal Cancer. National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, MD. Available at http://www.cancer.gov/ cancertopics/pdq/genetics/colorectal/HealthProfessional/. Accessed January 19, 2009. NCI (2008d). Genetics of Medullary Thyroid Cancer. National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, MD. Available at http://www.cancer. gov/cancertopics/pdq/genetics/medullarythyroid/healthprofessional/. Accessed January 19, 2009. NCI (2008e). Genetics of Prostate Cancer. National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, MD. Available at http://www.cancer.gov/ cancertopics/pdq/genetics/prostate/healthprofessional/. Accessed January 19, 2009. NRC (1983). Risk Assessment in the Federal Government: Managing the Process, National Academies Press, Washington, D.C. NRC (1987). Biological markers in environmental health research. Environ Health Perspect 74, 3–9. NTP (2005). Report on Carcinogens, 11th edition, US Department of Health and Human Services, Public Health Service, National Toxicology Program. Available at http://ntp.niehs.nih.gov/index. cfm?objectid=32BA9724-F1F6-975E-7FCE50709CB4C932. Accessed January 19, 2009. Parkin, D. M., Pisani, P., Munoz, N., and Ferlay, J. (1999). The global burden of infection associated cancers. Cancer Surv 33, 5–33. Perera, F. P., and Weinstein, I. B. (2000). Molecular epidemiology: Recent advances and future directions. Carcinogenesis 21 (3), 517–524. Peto, J. (2001). Cancer epidemiology in the last century and the next decade. Nature 411, 390–395.
418
CHAPTER 15 CANCER EPIDEMIOLOGY
Rothman, K. J., and Greenland, S. (2005). Causation and causal inference in epidemiology. Am J Pub Health 95 (Suppl 1), S144–S150. SEER (2008). SEER Cancer Statistics Review, 1975–2005. Ries, L. A. G., Melbert, D., Krapcho, M., Stinchcomb, D. G., Howlader, N., Horner, M. J., Mariotto, A., Miller, B. A., Feuer, E. J., Altekruse, S. F., Lewis, D. R., Clegg, L., Eisner, M. P., Reichman, M., and Edwards, B. K., eds., National Cancer Institute: Bethesda, MD. Available at http://seer.cancer.gov/csr/1975_2005/. Accessed January 19, 2009. Smith, A. H., Pearce, N. E., and Callas, P. W. (1988). Cancer case-control studies with other cancers as controls. Int J Epidemiol 17 (2), 298–306. Smith, A. H., Goycolea, M., Haque, R., and Biggs, M. L. (1998). Marked increase in bladder and lung cancer mortality in a region of northern Chile due to arsenic in drinking water. Am J Epidemiol 147, 660–669. Tseng, W. P. (1977). Effects and dose–response relationships of skin cancer and blackfoot disease with arsenic. Environ Health Perspect 19, 109–119. Tseng, W. P., Chu, H. M., How, S. W., Fong, J. M., Lin, C. S., and Yeh, S. (1968). Prevalence of skin cancer in an endemic area of chronic arsenicism in Taiwan. J Natl Cancer Inst 40, 453–463. Vineis, P., and Perera, F. (2007). Molecular epidemiology and biomarkers in etiologic cancer research: The new in light of the old. Cancer Epidemiol Biomarkers Prev 16 (10), 1954–1965. Weed, D. L., and Gorelic, L. S. (1996). The practice of causal inference in cancer epidemiology. Cancer Epidemiol Biomarkers Prev 5, 303–311. WHO (1964). Prevention of cancer: Report of a WHO expert committee. World Health Organization. WHO Tech Rep Ser No. 276. WHO (1999). Principles for the assessment of risks to human health from exposure to chemicals. World Health Organization. Available at http://www.inchem.org/documents/ehc/ehc/ehc210.htm. Accessed January 19, 2009. WHO (2008). Cancer. Fact sheet N°297. World Health Organization. Available at http://www.who.int/ mediacentre/factsheets/fs297/en/. Accessed January 19, 2009. Wu, M. M., Kuo, T. L., Hwang, Y. H., and Chen, C. J. (1989). Dose–reponse relation between arsenic concentration in well water and mortality from cancers and vascular diseases. Am J Epidemiol 130, 1123–1132.
CH A P TE R
16
RODENT HEPATOCARCINOGENESIS James E. Klaunig
16.1.
INTRODUCTION
Defining and understanding the mechanisms by which chemical agents have the capability of inducing toxic effects in humans occupies one of the highest of priorities in the field of toxicology. A variety of data sources including epidemiological information, clinical study data, in vitro experimental information, and animal studies are frequently used in assessing human risk. Animals, in particular rodents, have been extensively used as testing surrogates for humans in the study of a chemical agent’s potential toxic properties. However, central to the use of rodents and other species in toxicity testing is understanding the relevance of the response(s) seen in the animal to humans. For the most part, the physiology and biochemistry of rodents is comparable to that of humans, and toxicity testing is generally predictive of potential human adverse effects. Rodent bioassays have been used since the 1950s to assess whether chemicals such as agricultural compounds, pharmaceuticals, industrial chemicals, and other products might cause cancer or other pathologies in humans. Thus, chronic bioassays have become the standard approach for the examination of the carcinogenic potential of chemicals where human exposure is anticipated. In recent years, the utilization of molecular and cellular studies in concert with the chronic and subchronic bioassays in rodents have increased our understanding of the similarities and the differences in the biological response of humans and rodents to chemical agents. While confirming many similar biological pathways common to all mammals, many mechanistic studies have also raised concerns regarding the appropriateness of extrapolating rodent tumor information to humans. In particular, the issue of dose selection in the chronic exposure studies, along with the relevance of carcinogenic effects seen in rodents after treatment with high doses of the agent being tested to human risk, has been questioned. The foundation of this predictability and application of rodent toxicity data to human risk assessment is dependent on the similar absorption, metabolism, distribution, and elimination processes in the model rodent species and in the human. In
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
419
420
CHAPTER 16 RODENT HEPATOCARCINOGENESIS
addition, the molecular targets for the chemical or metabolite interaction must be similar and have similar effects and actions in the comparative species. Understanding the species specific toxic and pathologic responses at the molecular and cellular level to specific chemical agents is essential in the prediction of potential human risk. The rodent liver is the most frequent site of tumor formation following chronic treatment with many pharmaceutical and chemical agents that induce a positive rodent carcinogenic response. The liver represents a major target organ for rodent carcinogens (Gold et al. 1991). It has been estimated that nearly half of the chemicals tested in the two-year chronic bioassay by the U.S. National Toxicology Program showed an increased incidence of liver cancer. Early pioneering work by Pitot (1993) and Farber (1984) showed the multistaged process that occurs in the liver is characterized by well-defined changes including the formation of initiated cells by genotoxic agents that then progress to preneoplastic focal lesions, which subsequently convert into neoplasms (Farber 1984; Pitot 1993). Rodent liver carcinogenesis has been extensively studied. Multiple model systems for assessing hepatocarcinogenesis in the rat and in the mouse have been developed and used to define and characterize the multistage nature of the cancer process (Pitot 1993). These stages, operationally defined as initiation, promotion, and progression, correlate with the mechanistic activity of the hepatocarcinogen (see Chapter 5) (Dragan et al. 1993). In addition, the sequential pathogenesis of the liver cancer process can be demonstrated linking the three stages to histopathology landmarks (preneoplastic foci, adenoma, and carcinoma) (Klaunig and Kamendulis 2008). These steps follow a temporal sequence of events that have been observed in a wide variety of target tissues (Figure 16.1). The defining characteristics and key events of each of these stages are shown in Table 16.1 and discussed in greater detail below.
16.1.1.
Initiation
The first stage of the cancer process involves initiation, a process that is defined as a stable, heritable change. This stage is a rapid, irreversible process that results in a carcinogen-induced mutational event (Kolaja et al. 1996). Chemical and physical agents that function at this stage are referred to as initiators or initiating agents.
Normal Cell Repair
DNA Damage
Initiation Figure 16.1.
Initiated Cell
Focal Lesion Apoptosis
Proliferation
Cancer Apoptosis
Proliferation
Promotion
Progression
Multistage rodent liver cancer showing the temporal nature of liver cancer.
16.1. INTRODUCTION
TABLE 16.1.
Initiation
Promotion
Progression
421
Multistage Nature of Hepatic Carcinogenesis
DNA interaction and modification Mutation Genotoxic One cell division necessary to lock-in mutation Modification is not enough to produce cancer Nonreversible Single treatment can induce mutation No direct DNA modification or damage Nongenotoxic No direct mutation Multiple cell divisions necessary Clonal expansion of the initiated cell population Increase in cell proliferation or decrease in cell death (apoptosis) Reversible Multiple treatments (prolonged treatment) necessary Threshold DNA modification Genotoxic event? Mutation, chromosome disarrangement Changes from preneoplasia to neoplasia Irreversible Number of treatments needed with compound unknown (may require only one treatment)
Initiating agents lead to genomic DNA changes including mutations and deletions. Chemical carcinogens that covalently bind to DNA and form adducts that result in mutations are initiating agents. Included among chemicals classified as initiating carcinogens are compounds such as polycyclic aromatic hydrocarbons, nitrosamines, biological agents, certain viruses, and physical agents such as X rays and UV light. Most chemical carcinogens that function at the initiation stage of the cancer process are indirect-acting genotoxic compounds that require metabolic activation in the target cell to produce the DNA-damaging event. For these compounds, the chemical must be taken into the target site and metabolized (in the case of an indirect genotoxic carcinogen). The ultimate form of the carcinogen is then able to bind to nuclear DNA, resulting in adduct formation. The initiating event becomes “fixed” when the DNA adducts or other damage to DNA are not correctly repaired or are incompletely repaired prior to DNA synthesis. This event can lead to inappropriate base pairing and/or formation of a mutation. Initiation by itself does not appear to be sufficient for neoplastic formation. Once initiated hepatocytes are formed, their fate has multiple potential outcomes: (1) The initiated cell can remain in a static nondividing state; (2) the initiated cell may acquire mutations incompatible with cell viability or normal function and thus the cell will be deleted through apoptotic mechanisms; or (3) the cell, stimulated by intrinsic factors and/or from chemical exposure, may undergo cell division resulting
422
CHAPTER 16 RODENT HEPATOCARCINOGENESIS
in the growth in the proliferation of the initiated cell. In some instances, typically following relatively high doses and/or repeated exposure to a genotoxic carcinogen, a chemical carcinogen may function as a complete carcinogen; that is, it is capable of progressing through all stages of the cancer process without further external chemical agent exposure.
16.1.2.
Promotion
Derived from either endogenous or exogenous stimuli of cell growth, the second stage of the carcinogenesis process involves the selective clonal expansion of an initiated cell to produce a preneoplastic lesion by enhancing cell replication in preneoplastic cells and/or inhibiting apoptosis in preneoplastic cells (Sonich-Mullin et al. 2001). This is referred to as the promotion stage of the carcinogenesis process. Both exogenous and endogenous agents that function at this stage are referred to as tumor promoters. Tumor promoters are not mutagenic and generally are not able to induce tumors by themselves; rather, they act through multiple mechanisms involving gene expression changes that result in sustained cell proliferation either through increases in cell proliferation and/or the inhibition of apoptosis. Nongenotoxic hepatic carcinogens frequently function at the tumor promotion stage. The growth of preneoplastic lesions requires repeated applications of or continuous exposure to tumor promoting compounds. While initial exposure to tumor promotion may result in an increase in cell proliferation and/or DNA synthesis in all tissues of the organ (normal and preneoplastic hepatocytes), this is usually a transient effect and with repeated exposure of the tumor promoter only the initiated cells continue to clonally expand and divide (Figure 16.1). Promotion is a reversible event whereby upon removal of the promoting agent, the preneoplastic focal cells may return back to the initiated cell. In addition, these agents demonstrate a well-documented threshold for their effects—below a certain dose or frequency of application; tumor promoters are unable to induce cell proliferation. Multiple chemical compounds as well as physical agents have been linked to the tumor promotion stage of the cancer process. Tumor promoters in general show organ-specific effects; for example, a tumor promoter of the liver, such as phenobarbital, will not function as a tumor promoter in the skin or other tissues.
16.1.3.
Progression
The final stage of the carcinogenesis process, progression, involves the conversion of the benign preneoplastic lesions into a neoplastic cancer. In this stage, due to an increase in DNA synthesis and cell proliferation in the preneoplastic lesions, additional genotoxic events may occur resulting in additional DNA damage including chromosomal damage such as aberrations and translocations. These events result in the transfer from preneoplastic, clonally derived cell populations into neoplastic cell populations. Agents that impact on the progression stage are usually genotoxic agents (Figure 16.1). By definition, the progression stage is an irreversible stage in that neoplasm formation, whether benign or malignant, occurs. With the formation of neoplasia, autonomous growth and/or lack of growth control is achieved.
16.2. MECHANISMS OF ACTION OF HEPATIC CARCINOGENS
423
Spontaneous progression can occur via spontaneous karyotypic changes that occur in mitotically active initiated cells during promotion. An accumulation of nonrandom chromosomal aberrations and karyotypic instability are hallmarks of progression. As such, chemicals that function as progressing agents are usually capable of causing genomic DNA damage and chromosomal abnormalities. Complete carcinogens have the ability to produce initiation, promotion, and progression and hence by definition possess both genotoxic and epigenetic properties.
16.2. MECHANISMS OF ACTION OF HEPATIC CARCINOGENS Utilizing the recently developed mode of action framework, it is possible to develop several defined modes of action (based on chemical activity) for rodent liver cancer development (Cohen et al. 2003; Meek et al. 2003). This classification is outlined in Table 16.2. Hepatic carcinogens can be divided into two broad categories based on their ability (or that of a metabolite) to interact with and mutate genomic DNA (genotoxic hepatic carcinogen) or function through nondirectly mutagenic mechanisms (nongenotoxic) (Table 16.3). For the purpose of this chapter, a description of the proposed modes and mechanisms of selected rodent hepatic carcinogens will be discussed.
TABLE 16.2.
Genotoxic and Nongenotoxic Hepatic Carcinogens
Genotoxic Carcinogens
Nongenotoxic Carcinogens
Mutagenic Can be complete carcinogens Tumorigenicity is dose responsive No theoretical threshold Nonmutagenic Threshold, reversible Tumorigenicity is dose responsive May function at tumor promotion stage No direct DNA damage Species, strain, tissue specificity
TABLE 16.3. Selected Genotoxic Hepatic Carcinogens
Aromatic amines N-Nitrosoamines Azo dyes Safrole Mycotoxins (e.g., Aflatoxin B1) Acetylaminofluorene
424
CHAPTER 16 RODENT HEPATOCARCINOGENESIS
16.2.1.
Genotoxic Agents
A number of hepatocarcinogenic compounds have been identified that function through a genotoxic mode of action. Table 16.4 lists selected rodent carcinogens that function through a genotoxic mode of action. All of these agents function through indirect-acting genotoxicity, and thus they require metabolism from the procarcinogenic form to an ultimate carcinogenic metabolite to be genotoxic (Figure 16.2). These compounds are also mutagenic in in vitro and in vivo mutagenesis assays and require metabolic enzyme addition to be mutagenic. As an example, aflatoxin B1 is metabolized to the ultimate DNA-reactive form of aflatoxin B1-2,3epoxide, which in turn forms an N7-guanine adduct that, if unrepaired, results in a mutation in the target hepatocyte. Aflatoxin B1 is a complete carcinogen that at a high- and long-enough dosing will result in liver tumors in rodents. Another well-
TABLE 16.4.
Proposed Mechanisms of Action for Selected Hepatic Carcinogens
Mechanism
Example Genotoxic Diethylnitrosamine Dimethylnitrosamine Aflatoxin B1 Safrole Acetylaminofluorene
Genotoxic
Nongenotoxic CAR
Phenobarbital
PPARα
Trichloroethylene Perchloroethylene DEHP Fibrates (e.g., clofibrate)
AhR
TCDD PCBs PBBs
Hormonal
Steroid and peptide hormones Phytoestrogens Tamoxifen
Receptor-mediated
Oxidative stress
Ethanol TCDD Lindane Dieldrin Metal overload (e.g., iron and copper)
Cytotoxicity
Chloroform Carbon tetrachloride
16.2. MECHANISMS OF ACTION OF HEPATIC CARCINOGENS Directepoxidation O
O
O
O
O
C
C
C
C
O
425
O O
O
O
O CH 3
Aflatoxin B1 (Pr)
H2 C
O
OC H 3
Aflatoxin B1-2,3-epoxide (Ut)
O
OH
C H2
CH 2
HC C
ester
HC
c
CH 2
O H2C
O
O
O H2C
Safrole
H 3C N N
O
l'-Hydroxy Safrole(Px)
CH 3 O
Dimethynitrosamine (Pr)
O
H2 C
H 3C N N
C H2O H
Safrolel' O-ester (Ut)
H 3C N
N O H HCOH C H3+ + N 2 + H 2O
O
Hydroxymethyl, methyl nitrosamine (Px)
Methyl carboniumion (Ut)
Figure 16.2. Structures of representative indirect acting genotoxic hepatic carcinogens and their metabolic derivatives, the proximate (Px) and ultimate (Ut) carcinogenic form, result from, the metabolism of the procarcinogenic form (Pr).
studied genotoxic hepatic carcinogen is diethylnitrosamine. The procarcinogenic form (diethylnitrosamine) is metabolized to the ultimate ethyl carbonium ion form that can form adducts with genomic DNA. The O-4-ethylthymine adduct that is formed is persistent and can result in misrepair and the formation of mutations in hepatocytes. Diethylnitrosamine is also a complete carcinogen at high dose and persistent dosing in rodent liver.
16.2.2. 16.2.2.1.
Nongenotoxic Mechanisms of Action Receptor-Mediated Modes of Action
16.2.2.1.1. Peroxisome Proliferator-Activated Receptor Alpha (PPARα Agonists). A variety of chemicals including pharmaceuticals, pesticides, plasticizers, and natural materials produce an increase in peroxisomes in rodent liver following exposure (see Chapter 17) (Reddy and Chu 1996; Reddy et al. 1976). Based on the observation of peroxisome proliferation, the chemical and pharmaceutical
426
CHAPTER 16 RODENT HEPATOCARCINOGENESIS
agents that induce this response have been collectively referred to as “peroxisome proliferators.” Peroxisomes are subcellular organelles found in the cytoplasm of mammalian cells, in particular the liver. The peroxisomes have important metabolic functions including the regulation of fatty acid oxidation (de Duve 1996; Hashimoto 1996). The induction of peroxisome proliferation involves an increase in the volume density of peroxisome along with increased fatty acid oxidation. A number of peroxisome proliferating chemicals are also carcinogenic in rodent liver. The discovery of the peroxisome proliferator-activated receptor (PPAR) led to further investigation for the mechanism of the cellular responses to peroxisome proliferators (Issemann and Green 1990). This nuclear hormone receptor, PPAR-alpha (PPARα), is a required component for the induction of peroxisome proliferation in rodent hepatocytes (Figure 16.3). PPARα plays a central role in lipid metabolism, and it acts as a transcription factor to modulate gene expression following ligand activation. This latter effect arises through the heterodimerization of PPAR and retinoid X receptor-alpha (RXRα), which results in binding to response elements (PPREs) and subsequent modulation of target gene transcription. Among the rat and mouse strains that have been studied to date, both genders develop tumors in response to exposure to a wide range of chemicals including di-(2-ethylhexyl)phthalate and other phthalates, chlorinated paraffins, chlorinated solvents, selected herbicides and insecticides, and hypolipidemic pharmaceuticals. Hepatic changes associated with peroxisome proliferator exposure include hepatomegaly, increases in the number and size of peroxisomes, and enhanced activity of peroxisomal marker enzymes including catalase, cytochrome P-450s (e.g., CYP4A1 and 4A3), and acyl coenzyme A (CoA) oxidase. Replicative hepatocytic DNA synthesis is also significantly increased in rodents following exposure to PPARα agonists. It is apparent that a functioning and activated PPARα is necessary for the induction of peroxisomes, cell 9-cis-Retinoic Acid
RXR
PPARα Agonist
PPAR
TCACCT n TCACCT (PPRE PPAR Response Element)
Enzyme Induction Peroxisome Proliferation Altered Lipid Metabolism
Figure 16.3.
Altered Cell Growth DNA Synthesis Induction Tumor Promotion
PPARα interaction with RXR receptors.
16.2. MECHANISMS OF ACTION OF HEPATIC CARCINOGENS
427
growth regulation, and lipid metabolic gene expression. There are differences in the ability of PPARα agonists to bind to and activate PPARα. As predictable, the relative binding affinity correlates well with the ability to activate PPARα target gene expression (Krey et al. 1997). PPAR exerts some of its anti-oxidant effects by blocking the association of NF-kappaB and AP1 (Ricote et al. 1998). The interactions of steroid hormone receptors with the AP-1 complex are well-recognized and may have important consequences in the regulation of growth regulatory gene expression (Mohamood et al. 1997; Zhang et al. 1991). In the absence of ligand, PPARα may bind to co-repressors, leading to decreases in receptor activity. Ligand binding induces a conformational change in PPARα that might favor binding to co-activators. The receptor/coactivator complex then can activate gene transcription by chromatin-modifying enzymes (i.e., histone acetyl transferases) (Horwitz et al. 1996). Administration of PPARα agonists causes numerous changes in gene expression; many of these genes are central to lipid metabolism. Peroxisomal, mitochondrial, and microsomal fatty acid metabolizing enzymes all are increased by PPARα agonists (Bocher et al. 2002). Liver mRNAs encoding proteins involved in lipid transport including fatty acid binding protein, fatty acid transporters, lipoprotein lipase, and apolipoproteins also are regulated by these chemicals (Bocher et al. 2002). In addition to modifying gene expression essential to lipid metabolism, PPARα agonists also can influence mRNAs encoding proteins that regulate cell proliferation or the acute phase response. PPARα is critical in the regulation of lipid metabolism, and this receptor likely regulates the cell proliferation/apoptosis underlying PPARα agonist-induced hepatocarcinogenesis. The precise role for some of these alterations in gene expression is not understood. For many of the target genes that regulate lipid metabolism, functional PPREs have been identified in their respective promoter region, including peroxisomal fatty acid metabolizing enzymes, fatty acid transporter, apolipoproteins, and lipoprotein lipase. Activation of PPARα is related causally to induction of rodent liver tumors. The weight of evidence is strong and the specificity is high because PPARα-null mice are refractory to tumor formation in response to a prototypical PPARα agonist (i.e., WY-14,643). This conclusion would be strengthened by the conduct of additional bioassays with other PPARα agonists. Regulation of peroxisomal acyl CoA oxidase in response to activation by PPARα agonists is associative with rodent tumor formation. PPARα agonists that cause high-level induction of palmitoyl CoA induce rodent liver tumors. Regulation of genes that mediate cell cycle progression, growth, and apoptosis in response to activation by PPARα agonists is deemed to be a causal event in the induction of rodent liver tumors. Largely through the use of PPARα knockout mice, the activation of PPARα by agonists is needed for these chemicals to induce peroxisome proliferation and tumorigenesis in rodents (Klaunig et al. 2003; Krey et al. 1997; Peters et al. 1998). Following this event is the induction of cell proliferation and suppression of apoptosis (James and Roberts 1996). Both of these events would then be expected to affect tumor development as these effects would enhance the rate of fixation of DNA damage in the genome, leading to changes in gene expression such as the silencing of tumor suppressor genes or increased expression of oncogenes, or suppression of apoptosis that may normally remove DNA-damaged, potentially tumorigenic cells. Since humans are exposed to
428
CHAPTER 16 RODENT HEPATOCARCINOGENESIS
a number of these chemicals, the relevance of this mode of action to human has been evaluated (Klaunig et al. 2003). Though the same events would be expected to occur in exposed humans, several species differences have been noted including a lack of induction of cell proliferation in nonhuman primates (Pugh et al. 2000), and the finding that PPARα in human liver is at least 10-fold lower compared with the rat or mouse (Tugwood et al. 1998). Based on these kinetic and dynamic differences between species, it has been concluded that tumors are not likely to occur in humans (Klaunig et al. 2003). 16.2.2.1.2. Aryl Hydrocarbon Receptor (AhR). Agonists of the AhR include polychlorinated and polybrominated biphenyls (PCBs and PBBs) and tetrachlorodibenzo-p-dioxin (TCDD). These agents and their binding to the AhR have been linked to liver tumor development. The AhR agonists appear to function at the tumor promotion stage of the cancer process (Pitot and Dragan 2001). The observed hepatic tumor response is AhR-dependent (Knutson and Poland 1982). After binding to the AhR, the xenobiotic and the bound AhR translocates to the nucleus, dimerizes with the Ah receptor nuclear translocator (ARNT), and binds to aryl hydrocarbon response elements (Nebert et al. 2000) (Figure 16.4). AhR-ARNT-dependent genes include members of the cytochrome P450 family, NAD(P)H:quinone oxidoreductase, UDP-glucuronosyltransferase, and glutathione transferase (Nebert et al. 2000). These are all genes involved in metabolic activation and/or the detoxification of chemical agents. It has been hypothesized that there are additional AhR-ARNTdependent genes. Studies with AhR knockout mice have displayed a decreased tumor response by AhR ligands (Nakatsuru et al. 2004). In contrast, overexpression of the AhR produced an increased incidence of liver tumors over control mice (Moennikes et al. 2004). Part of the effect of the activation of the AhR may be linked
Metabolism of ligand by CYP1 forms CYP1
AHR AHR ARNT
AHR ligand
HSP 90 HSP 90
AHR ARNT
AHR AIP
XRE
Increased CYP1A1, CYP1A2, transcription CYP1B1
Figure 16.4. Mechanisms of AhR agonist binding and resulting activity of gene transcription.
16.2. MECHANISMS OF ACTION OF HEPATIC CARCINOGENS
429
to the inhibition of apoptosis in preneoplastic hepatic foci, thus allowing for an increase in foci growth through decreased cell death (Stinchcombe et al. 1995). 16.2.2.1.3. Constitutive Androstane Receptor (CAR). Phenobarbital is a commonly studied hepatic carcinogen that is known to induce tumors by a nongenotoxic mechanism involving liver hyperplasia (Williams and Whysner 1996). One feature seen following phenobarbital exposure is the induction of P450 enzymes, particularly CYP2B (Nims and Lubet 1996). Since a number of diverse chemical agents are known to induce various members of the P450 system, the specificity of this effect to the carcinogenesis process has been questioned. Phenobarbital is the prototype of several rodent hepatocarcinogens (e.g., oxazepam, dichloro-diphenyltrichloroethane [DDT], etc.) that induce tumors by a nongenotoxic mechanism involving liver hyperplasia (Williams and Whysner 1996). This effect is due to activation of nuclear receptors, particularly the constitutive androstane receptor (CAR) (Figure 16.5). There is evidence from studies in knockout mice that CAR plays an essential role in the carcinogenicity of phenobarbital (Williams and Whysner 1996), it is uncertain whether CYP induction also leads to other pleiotropic responses or if P450 itself plays a role—for example, by generation of active oxygen species. Additional phenobarbital responses that are key in their tumorigenic effect include increased cell proliferation, inhibition of apoptosis, hypertrophy, and development of altered hepatic foci (Sonich-Mullin et al. 2001). These effects are all CARdependent (Wei et al. 2000; Yamamoto et al. 2004). Since phenobarbital has been used chronically in humans as a pharmacological agent, there is extensive experience in humans showing no increase in liver tumors. Therefore, phenobarbital-like compounds (that function through the CAR receptor) are unlikely to induce hepatic
CYP 2B6
CAR
P CAR RXR
Phenobarbital
CK
PP
P P CAR
PBREM
Figure 16.5.
Increased
CAR RXR
CYP2B6
transcription
Mechanisms of CAR binding and resulting activity of gene transcription.
430
CHAPTER 16 RODENT HEPATOCARCINOGENESIS
tumors in humans. Though CAR is expressed, and phenobarbital induces CYP enzymes in human liver, it may act more through the pregnane X receptor (PXR) than through CAR. However, there are exceptions such as phenytoin, which has been shown to induce CYP2B6 in humans via CAR (Wang et al. 2004). In addition, there is evidence that phenobarbital may activate human CAR, resulting in the induction of non-P450 genes, including UGT1A1. For example, liver size is increased in humans after prolonged treatment with phenobarbital (Pirttiaho et al. 1982). However, limited studies with human hepatocytes indicate that such cells are refractory to the hyperplastic and antiapoptotic effects of phenobarbital, which is needed for the carcinogenicity of this compound (Hasmall and Roberts 1999). Moreover, although the data for concordance analysis for phenobarbital are limited, there are convincing data showing that in patients receiving phenobarbital for many years, at doses producing plasma concentrations similar to those following a carcinogenic dose in rodents, there is no evidence of a hepatocarcinogenic effect (IARC 2001). There are a number of data gaps, including the extent to which this mode of action extends to other rodent hepatocarcinogens that are P450 inducers. Nevertheless, the situation with phenobarbital affords a somewhat unique opportunity, because extensive epidemiological data from the clinical applications of this drug can be used in a human relevance framework to help bridge some of the data gaps in an inverse direction from the traditional application of the framework. Caution is raised in classifying compounds that exhibit phenobarbital-like effects (e.g., cell proliferation, hypertrophy, CYP2B induction) as functioning as a Phenobarbital-type mechanism with evaluation of CAR receptor-dependent effects. If these criteria are met, for those compounds for which there are robust data for a phenobarbital-like mechanism, it can be concluded that the carcinogenic response is not relevant to humans. 16.2.2.1.4. Hormonal. Hormonally active agents including estrogen and androgens are inducers of rodent liver tumors. Chronic exogenous administration of hormonally active agents including synthetic estrogens and anabolic steroids can increase hepatic adenoma incidence in rats and in humans (IARC 1999; Li et al. 1992). Women taking high-estrogen-containing oral contraceptives displayed an increase in hepatic adenomas. These adenomas regressed after removal of the estrogen administration or progressed when administration of the estrogen contraceptive was continued (Edmondson et al. 1977). It appears that administration of the highestrogen-containing oral contraceptive requires eight years or more of continuous exposure to increase the hepatic tumor risk (Christopherson et al. 1978; Tavani et al. 1993). The rat liver following exposure to excess estrogen showed an increase in liver tumor incidence. Though the liver is not normally a target tissue for sex hormones, it is responsive to these hormones at higher concentrations. The key events in rodent liver carcinogenesis following exposure to estrogenic agents are perturbation of hormone level or function, altered cell proliferation of the preneoplastic hepatocytes, and development of hepatic neoplasms. This mode of action is receptor-mediated and receptor-dependent. Thus agents such as estrogen that function through this mechanism are threshold-dependent and dose-dependent. Hormonally active agents include biogenic amines, steroids, and peptide hormones that cause hepatic tissue specific changes following interaction and activation of the
16.2. MECHANISMS OF ACTION OF HEPATIC CARCINOGENS
431
hormone receptor. Phytoestrogens such as genistein, daidzein, glycitein, equol, and their metabolites, as well as tamoxifen, have been shown to possess estrogenic activity and, depending on the target tissue, dose, and duration of exposure, is either anticarcinogenic or carcinogenic. For example, tamoxifen is antiestrogenic in the mouse uterus with chronic administration but is carcinogenic in the rat liver. 16.2.2.1.5. Oxidative Stress and Hepatic Carcinogenesis. Experimental evidence has shown that increases in reactive oxygen in the cell, through either physiological modification or through chemical carcinogen exposure, contribute to carcinogenic processes (Guyton and Kensler 1993; Trush and Kensler 1991; Vuillaume 1987; Witz 1991). Reactive oxygen species can be produced by both endogenous and exogenous sources and are typically counterbalanced by antioxidants (Table 16.5). Antioxidant defenses are both enzymatic (e.g., superoxide dismutase, glutathione peroxidase, and catalase) and nonenzymatic (e.g., vitamin E, vitamin C, β-carotene, glutathione, etc.) (Abuja and Albertini 2001; Betteridge 2000). Importantly, many of these antioxidants are provided through dietary intake (Clarkson and Thompson 2000). Endogenous sources of reactive oxygen species include oxidative phosphorylation, P450 metabolism, peroxisomes, and inflammatory cell activation (Table 16.5). Within the mitochondria, a small percentage of oxygen is converted into the superoxide anion via one-electron reduction of molecular oxygen. Superoxide anion can be dismutated by superoxide dismutase to yield hydrogen peroxide (Barber and Harris 1994). In the presence of partially reduced metal ions, hydrogen peroxide is converted to the highly reactive hydroxyl radical through Fenton and Haber–Weiss
TABLE 16.5.
Reactive Oxygen Species Generation and Removal in the Cella
Endogenous Cellular oxidants Exogenous
Macrophage/ inflammatory cells
⋅ ⋅ O ⋅, NO, H O
Peroxisomes
H2O2
Redox cycling compounds
O2−
Metals (Fenton reaction)
H2O2 + Fe2+ → OH− + •OH + Fe3+
Mitochondria
O2− , H2O2, •OH
Cytochrome P450
O2− , H2O2
Radiation
Cellular antioxidants
Enzymatic Nonenzymatic
− 2
•
2
2,
OCl−
⋅
•
OH
⋅
Superoxide dismutase
2O2− + 2H + → O2 + H 2 O2
Catalase
H2O2 → O2 + 2H2O
Glutathione peroxidase
2GSH + 2H2O2 → GSSG + 2H2O
Vitamin E, vitamin C, glutathione (GSH), catechins, etc.
Antioxidants < Oxidants → Oxidative Damage (DNA, RNA, Lipid, Protein) a
Oxidants can be produced via both endogenous and exogenous sources. Antioxidants function to maintain the cellular redox balancing. However, excess production of oxidants and/or inadequate supplies of antioxidants result in damage to cellular biomolecules and may impact on neoplastic development.
432
CHAPTER 16 RODENT HEPATOCARCINOGENESIS
reactions (Betteridge 2000). Neutrophils, eosinophils, and macrophages represent another intracellular source of reactive oxygen species. Activated macrophages, through “respiratory burst,” elicit a rapid increase in oxygen uptake that gives rise to a variety of reactive oxygen species including superoxide anion, hydrogen peroxide, and nitric oxide. The release of cytokines and reactive oxygen intermediates from activated Kupffer cells (the resident macrophage of the liver) has been implicated in hepatotoxicologic and hepatocarcinogenic events; in particular, recent studies show that the Kupffer cell may function at the promotion stage of carcinogenesis (Rusyn et al. 1999). Reactive oxygen species can also be produced by cytochrome P450-mediated mechanisms including: (1) redox cycling in the presence of molecular oxygen; (2) peroxidase-catalyzed single-electron drug oxidations; and, (3) “futile cycling” of cytochromes P450 (Parke 1994). Ethanol, phenobarbital, and a number of chlorinated and nonchlorinated compounds such as dieldrin, TCDD, and lindane, are among the xenobiotics shown to increase reactive oxygen species through P450-mediated mechanisms (Klaunig et al. 1997). Chemicals classified as peroxisome proliferators (e.g., Clofibrate) represent chemicals that induce cytochrome P4504A and increase the formation of peroxisomes. As such, an increase in H2O2 production often accompanies exposure to peroxisome proliferating compounds (Rao and Reddy 1991). Through these or other yet-to-be-defined mechanisms, a number of chemical agents that induce cancer including chlorinated compounds, radiation, metal ions, barbiturates, and some peroxisome proliferating compounds, have been shown to induce reactive oxygen species formation and/or oxidative stress (Klaunig et al. 1997). Overproduction of reactive oxygen species can also result in damage to cellular macromolecules. In DNA, reactive oxygen species can produce single- or double-stranded DNA breaks, purine, pyrimidine, or deoxyribose modifications, and DNA crosslinks (von Sonntag 1987). Persistent DNA damage can result in either arrest or induction of transcription, induction of signal transduction pathways, replication errors, and genomic instability, all of which are seen in carcinogenesis. Oxidation of guanine at the C8 position, will result in the formation of 8-hydroxy2′-deoxyguanosine (8-OHdG), a commonly studied oxidative DNA adduct. 8-OHdG is mutagenic in bacterial and mammalian cells and produces G → T transversions, which are widely found in mutated oncogenes and tumor suppressor genes, and produces dose-related increases in cellular transformation (Zhang et al. 2000). 8-OHdG can also be found in the nucleotide pool, thus during DNA replication, 8-OHdG can incorporate into DNA opposite dC or dA on the template strand, resulting in A:T to C:G transversions (Demple and Harrison 1994). Oxidative damage to mitochondrial DNA and induction of mutations may also participate in carcinogenic processes; in fact, mutations in mitochondrial DNA have been identified in a number of cancers (Schumacher et al. 1973). The mitochondrial genome is relatively susceptible to oxidative base damage due to the following: (1) Mitochondrial DNA is in close proximity to the electron transport system, a major source of reactive oxygen species; (2) mitochondrial DNA is not protected by histones; and (3) DNA repair pathways exist in mitochondria to remove altered bases produced by oxidative-mediated reactions, but nucleotide excision repair is completely absent.
16.2. MECHANISMS OF ACTION OF HEPATIC CARCINOGENS
433
Aside from oxidized nucleic acids, radical-mediated damage to cellular biomembranes results in lipid peroxidation, a process that generates a variety of products including reactive electrophiles such as epoxides and aldehydes (Janero 1990). Malondialdehyde (MDA), a by product of lipid degradation, can react with cellular nucleophiles and can form MDA–MDA dimers. Both MDA and the MDA– MDA dimer are mutagenic in bacterial assays and the mouse lymphoma assay (Riggins and Marnett 2001). Reactive oxygen species production and oxidative stress can affect both cell proliferation and apoptosis (Cerutti 1985; Klaunig and Kamendulis 2004). H2O2 and superoxide anion can induce cell proliferation in several mammalian cell types (D’Souza et al. 1993). Conversely, high concentrations of reactive oxygen species can initiate apoptosis. It has clearly been demonstrated that low levels of reactive oxygen species and other free radicals influence the expression of a number of genes and signal transduction pathways. Many xenobiotics, by increasing cellular levels of oxidants, alter gene expression through activation of signaling pathways including cAMP-mediated cascades, calcium-calmodulin pathways, and transcription factors such as mitogen-activated protein (MAP) kinase, AP-1, and NF-κB pathways (Chang and Karin 2001). Activation of MAP kinases (extracellular signal-regulated kinases (ERK); c-Jun N-terminal kinases (JNK); and the p38 kinases) by chemical carcinogens results in modulation of proliferation, differentiation, and apoptosis, mainly through modulation of gene expression (Chang and Karin 2001). Similarly, activation of NFκB, a ubiquitously expressed transcription factor occurs in response to a wide variety of extracellular stimuli and is regulated, in part, by reactive oxygen species and the cellular redox status (Li and Karin 1998; Pineda-Molina et al. 2001). 16.2.2.1.6. Cytotoxicity. Cytotoxicity is a generally accepted mechanism of action for a number of nongenotoxic rodent hepatic carcinogens (Dietrich and Swenberg 1991). A hepatocyte cytotoxicant such as chloroform produces consistent hepatocyte death, leading to persistent regenerative growth (Butterworth 1990; Larson et al. 1994). The sustained induction of regenerative growth can result in the induction of spontaneous mutations in hepatocytes either through misrepair, which may cause cell initiation and preneoplastic foci, or through growth promotion of existing preneoplastic hepatocytes. In either case, the effect of the cytotoxic response is the induction of consistent chronic proliferation ultimately leading to hepatic tumors via further clonal expansion. Key events in this mechanism are the induction of considerable and chronic cytotoxicity and the resulting compensatory hyperplasia, which may lead to tumor formation. Hepatotoxicity can be demonstrated either by histopathology (detection of necrosis) or by a significant increase in liver serum enzymes (e.g., alanine transaminase [ALT] and aspartate transaminase [AST]). Using chloroform as an example, the metabolic activation of chloroform by CYP2E1 to produce the chloroform metabolites necessary to induce the hepatotoxicity is a requirement and a first step in the process. As with other nongenotoxic agents, a threshold, dose-dependent and nonlinear dose–response profile is observed for the cytotoxic hepatocarcinogenic agents. The dose and duration of exposure must be high and long enough to produce the substained cytotoxicity with a consistent regenerative compensatory hyperplasia (EPA 2001).
434
16.3.
CHAPTER 16 RODENT HEPATOCARCINOGENESIS
HUMAN RELEVANCE FRAMEWORK
In an effort to further characterize those mechanisms important in the development of cancer induced by a specific agent or similar agents in rodents and define if these same mechanisms of action apply to humans, a human relevance framework concept was developed by several international groups (Cohen et al. 2003; Meek et al. 2003). A central requirement of the framework is the identification of the mechanistic steps or mode of action by which the agent induces its toxic or carcinogenic response in rodents and humans (see Chapter 13). In the analysis of mode of action, key biochemical, cellular, and molecular events or steps are defined for the induction of the specific carcinogenic response. These key events can also be used to define and link the species, temporal, and dose–respon characteristics of a group of carcinogenic agents for a given mode of action. Following the establishment of the mode of action, an analysis of the biological plausibility of the key events and the determination of the relevance of the defined mode of action in an animal model for human cancer risk based on kinetic and dynamic parameters can be assessed. The scientific foundation for using rodents in safety bioassays is the premise that the toxic response seen with rodent exposure to chemical and pharmaceutical agents is scientifically transferable to human risk assessment. Much effort has been made during the past several decades to evaluate the mechanism by which chemicals cause cancer in rodents. Of prime importance in understanding the mechanism or mode of action of chemically induced neoplasms in rodents is the relevance of this process to human risk assessment. This in the past has been performed for individual chemicals with varying degrees of success. Owing to the variability in the approach to examining rodent cancer induction mode of action and its application to human cancer risk, a framework was developed by an International Life Sciences Risk Sciences Institute (ILSI-RSI) working group sponsored by the U.S. Environmental Protection Agency and Health Canada to provide a step-by-step scientifically based approach to this issue (Cohen et al. 2003; Meek et al. 2003). The human relevance framework provides a decision-tree-based approach to establishing a relationship between early cellular events and the development of cancer in rodents (mode of action), and the application of this mode of action for human hazard and risk assessment. This approach has also been established by the International Programme on Chemical Safety (IPCS) (Sonich-Mullin et al. 2001). Central to the framework approach is the determination of those key events in the mode of action that define the stepwise, temporal approach to the induction of cancer in rodents and the concordance analysis of these key events to the human. These measurable key events in the mode of action are explicitly defined and evaluated, allowing for the formulation of the mode of action as well as identifying any data gaps and uncertainties. The framework calls for addressing three specific questions: (1) Is the experimental weight of evidence for the establishment of a hepatic carcinogen sufficient to establish the mode of action in rodents? (2) Are these key events in the rodent hepatic carcinogen mode of action plausible in humans? (3) Taking into account kinetic and dynamic factors, are key events in the rodent mode of action of hepatic carcinogens plausible in humans?
REFERENCES
16.4.
435
SUMMARY
A large number of pharmaceuticals, natural products, pesticides, and other industrial compounds that have proven to have carcinogenic activity produce their effects in the rodent liver. Rodent liver carcinogenesis has been studied extensively during the past four decades. These studies have helped define the multistage nature of carcinogenic processes as well as helping to understand the cellular and molecular mechanisms by which hepatic carcinogens produce their effects. It is clear that besides genotoxicity, nongenotoxic effects including receptor-mediated processes, oxidative stress, and cytotoxicity can participate in liver carcinogenesis. As our understanding of the mechanisms of hepatic carcinogenesis continues, the application of these data to inform human risk assessment may be accomplished by evaluation with the human relevance framework.
REFERENCES Abuja, P. M., and Albertini, R. (2001). Methods for monitoring oxidative stress, lipid peroxidation and oxidation resistance of lipoproteins. Clin Chim Acta 306, 1–17. Barber, D. A., and Harris, S. R. (1994). Oxygen free radicals and antioxidants: A review. Am Pharm NS34, 26–35. Betteridge, D. J. (2000). What is oxidative stress? Metabolism 49, 3–8. Bocher, V., Pineda-Torra, I., Fruchart, J. C., and Staels, B. (2002). PPARs: Transcription factors controlling lipid and lipoprotein metabolism. Ann NY Acad Sci 967, 7–18. Butterworth, B. E. (1990). Consideration of both genotoxic and nongenotoxic mechanisms in predicting carcinogenic potential. Mutat Res 239, 117–132. Cerutti, P. A. (1985). Prooxidant states and tumor promotion. Science 227, 375–381. Chang, L., and Karin, M. (2001). Mammalian MAP kinase signalling cascades. Nature 410, 37–40. Christopherson, W. M., Mays, E. T., and Barrows, G. (1978). Hepatocellular carcinoma in young women on oral contraceptives. Lancet 2, 38–39. Clarkson, P. M., and Thompson, H. S. (2000). Antioxidants: What role do they play in physical activity and health? Am J Clin Nutr 72, 637S–646S. Cohen, S. M., Meek, M. E., Klaunig, J. E., Patton, D. E., and Fenner-Crisp, P. A. (2003). The human relevance of information on carcinogenic modes of action: Overview. Crit Rev Toxicol 33, 581–589. D’Souza, R. J., Phillips, H. M., Jones, P. W., Strange, R. C., and Aber, G. M. (1993). Interactions of hydrogen peroxide with interleukin-6 and platelet-derived growth factor in determining mesangial cell growth: Effect of repeated oxidant stress. Clin Sci (Lond) 85, 747–751. de Duve, C. (1996). The peroxisome in retrospect. Ann NY Acad Sci 804, 1–10. Demple, B., and Harrison, L. (1994). Repair of oxidative damage to DNA: Enzymology and biology. Annu Rev Biochem 63, 915–948. Dietrich, D. R., and Swenberg, J. A. (1991). Preneoplastic lesions in rodent kidney induced spontaneously or by non-genotoxic agents: Predictive nature and comparison to lesions induced by genotoxic carcinogens. Mutat Res 248, 239–260. Dragan, Y. P., Sargent, L., Xu, Y. D., Xu, Y. H., and Pitot, H. C. (1993). The initiation–promotion– progression model of rat hepatocarcinogenesis. Proc Soc Exp Biol Med 202, 16–24. Edmondson, H. A., Reynolds, T. B., Henderson, B., and Benton, B. (1977). Regression of liver cell adenomas associated with oral contraceptives. Ann Intern Med 86, 180–182. EPA (2001). Toxicological review of chloroform (CAS No. 67-66-3), In support of Summary Information on the Integrated Risk Information System (IRIS), EPA/635/R-01/001, 1–112, http://www.epa.gov/ ncea/iris/toxreviews/0025-tr.pdf. Farber, E. (1984). The multistep nature of cancer development. Cancer Res 44, 4217–4223.
436
CHAPTER 16 RODENT HEPATOCARCINOGENESIS
Gold, L. S., Slone, T. H., Manley, N. B., and Bernstein, L. (1991). Target organs in chronic bioassays of 533 chemical carcinogens. Environ Health Perspect 93, 233–246. Guyton, K. Z., and Kensler, T. W. (1993). Oxidative mechanisms in carcinogenesis. Br Med Bull 49, 523–544. Hashimoto, T. (1996). Peroxisomal beta-oxidation: enzymology and molecular biology. Ann NY Acad Sci 804, 86–98. Hasmall, S. C., and Roberts, R. A. (1999). The perturbation of apoptosis and mitosis by drugs and xenobiotics. Pharmacol Ther 82, 63–70. Horwitz, K. B., Jackson, T. A., Bain, D. L., Richer, J. K., Takimoto, G. S., and Tung, L. (1996). Nuclear receptor coactivators and corepressors. Mol Endocrinol 10, 1167–1177. IARC (1999). Hormonal contraception and post-menopausal hormonal therapy. In Cancer of the Liver and Gall-bladder, Vol. 72, IARC Monographs, IARC Press, Lyon, pp. 168–180. IARC (2001). Phenobarbital and its sodium salt. In Some Thyrotropic Agents, Vol. 79, IARC Monographs, IARC Press, Lyon, pp. 161–288. Issemann, I., and Green, S. (1990). Activation of a member of the steroid hormone receptor superfamily by peroxisome proliferators. Nature 347, 645–650. James, N. H., and Roberts, R. A. (1996). Species differences in response to peroxisome proliferators correlate in vitro with induction of DNA synthesis rather than suppression of apoptosis. Carcinogenesis 17, 1623–1632. Janero, D. R. (1990). Malondialdehyde and thiobarbituric acid-reactivity as diagnostic indices of lipid peroxidation and peroxidative tissue injury. Free Radic Biol Med 9, 515–540. Klaunig, J. E., Babich, M. A., Baetcke, K. P., Cook, J. C., Corton, J. C., David, R. M., DeLuca, J. G., Lai, D. Y., McKee, R. H., Peters, J. M., Roberts, R. A., and Fenner-Crisp, P. A. (2003). PPARalpha agonist-induced rodent tumors: modes of action and human relevance. Crit Rev Toxicol 33, 655–780. Klaunig, J. E., and Kamendulis, L. (2008). Chemical carcinogens. In Casarett and Doull’s Toxicology: The Basic Science of Poisons, 7th edition, Klaassen, C. D., ed., McGraw-Hill, New York, pp. 329–380. Klaunig, J. E., and Kamendulis, L. M. (2004). The role of oxidative stress in carcinogenesis. Annu Rev Pharmacol Toxicol 44, 239–267. Klaunig, J. E., Xu, Y., Bachowski, S., and Jiang, J. (1997). Free-radical oxygen-induced changes in chemical carcinogenesis. In Free Radical Toxicology, Wallace, K. B., ed., Taylor & Francis, London, pp. 375–400. Knutson, J. C., and Poland, A. (1982). Response of murine epidermis to 2,3,7,8-tetrachlorodibenzo-pdioxin: Interaction of the ah and hr loci. Cell 30, 225–234. Kolaja, K. L., Stevenson, D. E., Walborg, E. F., Jr., and Klaunig, J. E. (1996). Reversibility of promoter induced hepatic focal lesion growth in mice. Carcinogenesis 17, 1403–1409. Krey, G., Braissant, O., L’Horset, F., Kalkhoven, E., Perroud, M., Parker, M. G., and Wahli, W. (1997). Fatty acids, eicosanoids, and hypolipidemic agents identified as ligands of peroxisome proliferatoractivated receptors by coactivator-dependent receptor ligand assay. Mol Endocrinol 11, 779–791. Larson, J. L., Wolf, D. C., and Butterworth, B. E. (1994). Induced cytolethality and regenerative cell proliferation in the livers and kidneys of male B6C3F1 mice given chloroform by gavage. Fundam Appl Toxicol 23, 537–543. Li, J., Kirkman, H., and Li, S. (1992). Synthetic estrogens and liver cancer: Risk analysis of animal and human data. In Hormonal Carcinogenesis, Li, J., Nandi, S., and Li, S., eds., Springer-Verlag, New York, pp. 217–224. Li, N., and Karin, M. (1998). Ionizing radiation and short wavelength UV activate NF-kappaB through two distinct mechanisms. Proc Natl Acad Sci USA 95, 13012–13017. Meek, M. E., Bucher, J. R., Cohen, S. M., Dellarco, V., Hill, R. N., Lehman-McKeeman, L. D., Longfellow, D. G., Pastoor, T., Seed, J., and Patton, D. E. (2003). A framework for human relevance analysis of information on carcinogenic modes of action. Crit Rev Toxicol 33, 591–653. Moennikes, O., Loeppen, S., Buchmann, A., Andersson, P., Ittrich, C., Poellinger, L., and Schwarz, M. (2004). A constitutively active dioxin/aryl hydrocarbon receptor promotes hepatocarcinogenesis in mice. Cancer Res 64, 4707–4710. Mohamood, A. S., Gyles, P., Balan, K. V., Hollis, V. W., Eckberg, W. R., Asseffa, A., Han, Z., Wyche, J. H., and Anderson, W. A. (1997). Estrogen receptor, growth factor receptor and protooncogene
REFERENCES
437
protein activities and possible signal transduction crosstalk in estrogen dependent and independent breast cancer cell lines. J Submicrosc Cytol Pathol 29, 1–17. Nakatsuru, Y., Wakabayashi, K., Fujii-Kuriyama, Y., Ishikawa, T., Kusama, K., and Ide, F. (2004). Dibenzo[a,l]pyrene-induced genotoxic and carcinogenic responses are dramatically suppressed in aryl hydrocarbon receptor-deficient mice. Int J Cancer 112, 179–183. Nebert, D. W., Roe, A. L., Dieter, M. Z., Solis, W. A., Yang, Y., and Dalton, T. P. (2000). Role of the aromatic hydrocarbon receptor and [Ah] gene battery in the oxidative stress response, cell cycle control, and apoptosis. Biochem Pharmacol 59, 65–85. Nims, R. W., and Lubet, R. A. (1996). The CYP2B subfamily. In Cytochromes P450: Metabolic and Toxicological Aspects, Ioannides, C., ed., CRC Press, Boca Raton, FL, pp. 135–160. Parke, D. V. (1994). The cytochromes P450 and mechanisms of chemical carcinogenesis. Environ Health Perspect 102, 852–853. Peters, J. M., Aoyama, T., Cattley, R. C., Nobumitsu, U., Hashimoto, T., and Gonzalez, F. J. (1998). Role of peroxisome proliferator-activated receptor alpha in altered cell cycle regulation in mouse liver. Carcinogenesis 19, 1989–1994. Pineda-Molina, E., Klatt, P., Vazquez, J., Marina, A., Garcia de Lacoba, M., Perez-Sala, D., and Lamas, S. (2001). Glutathionylation of the p50 subunit of NF-kappaB: A mechanism for redox-induced inhibition of DNA binding. Biochemistry 40, 14134–14142. Pirttiaho, H. I., Sotaniemi, E. A., Pelkonen, R. O., and Pitkanen, U. (1982). Hepatic blood flow and drug metabolism in patients on enzyme-inducing anticonvulsants. Eur J Clin Pharmacol 22, 441–445. Pitot, H. C. (1993). The dynamics of carcinogenesis: Implications for human risk. Chemical Industry Institute of Toxicology (CIIT) Activities 13, 1–6. Pitot, H. C., III., and Dragan, Y. P. (2001). Chemical carcinogenesis. In Casarett and Doull’s Toxicology: The Basic Science of Poisons, 6th edition, Klaassen, C. D., ed., McGraw-Hill, New York, pp. 241–319. Pugh, G., Jr., Isenberg, J. S., Kamendulis, L. M., Ackley, D. C., Clare, L. J., Brown, R., Lington, A. W., Smith, J. H., and Klaunig, J. E. (2000). Effects of di-isononyl phthalate, di-2-ethylhexyl phthalate, and clofibrate in cynomolgus monkeys. Toxicol Sci 56, 181–188. Rao, M. S., and Reddy, J. K. (1991). An overview of peroxisome proliferator-induced hepatocarcinogenesis. Environ Health Perspect 93, 205–209. Reddy, J. K., and Chu, R. (1996). Peroxisome proliferator-induced pleiotropic responses: Pursuit of a phenomenon. Ann NY Acad Sci 804, 176–201. Reddy, J. K., Rao, S., and Moody, D. E. (1976). Hepatocellular carcinomas in acatalasemic mice treated with nafenopin, a hypolipidemic peroxisome proliferator. Cancer Res 36, 1211–1217. Ricote, M., Li, A. C., Willson, T. M., Kelly, C. J., and Glass, C. K. (1998). The peroxisome proliferatoractivated receptor-gamma is a negative regulator of macrophage activation. Nature 391, 79–82. Riggins, J. N., and Marnett, L. J. (2001). Mutagenicity of the malondialdehyde oligomerization products 2-(3′-oxo-1′-propenyl)-malondialdehyde and 2,4-dihydroxymethylene-3-(2,2-dimethoxyethyl) glutaraldehyde in Salmonella. Mutat Res 497, 153–157. Rusyn, I., Bradham, C. A., Cohn, L., Schoonhoven, R., Swenberg, J. A., Brenner, D. A., and Thurman, R. G. (1999). Corn oil rapidly activates nuclear factor-kappaB in hepatic Kupffer cells by oxidantdependent mechanisms. Carcinogenesis 20, 2095–2100. Schumacher, H. R., Szekely, I. E., Patel, S. B., and Fisher, D. R. (1973). Mitochondria: A clue to oncogenesis? Lancet 2, 327. Sonich-Mullin, C., Fielder, R., Wiltse, J., Baetcke, K., Dempsey, J., Fenner-Crisp, P., Grant, D., Hartley, M., Knaap, A., Kroese, D., Mangelsdorf, I., Meek, E., Rice, J. M., Younes, M., and International Programme on Chemical, S. (2001). IPCS conceptual framework for evaluating a mode of action for chemical carcinogenesis. Regul Toxicol Pharmacol 34, 146–152. Stinchcombe, S., Buchmann, A., Bock, K. W., and Schwarz, M. (1995). Inhibition of apoptosis during 2,3,7,8-tetrachlorodibenzo-p-dioxin-mediated tumour promotion in rat liver. Carcinogenesis 16, 1271–1275. Tavani, A., Negri, E., Parazzini, F., Franceschi, S., and La Vecchia, C. (1993). Female hormone utilisation and risk of hepatocellular carcinoma. Br J Cancer 67, 635–637. Trush, M. A., and Kensler, T. W. (1991). An overview of the relationship between oxidative stress and chemical carcinogenesis. Free Radic Biol Med 10, 201–209.
438
CHAPTER 16 RODENT HEPATOCARCINOGENESIS
Tugwood, J. D., Holden, P. R., James, N. H., Prince, R. A., and Roberts, R. A. (1998). A peroxisome proliferator-activated receptor-alpha (PPARalpha) cDNA cloned from guinea-pig liver encodes a protein with similar properties to the mouse PPARalpha: implications for species differences in responses to peroxisome proliferators. Arch Toxicol 72, 169–177. von Sonntag, C. (1987). New aspects in the free-radical chemistry of pyrimidine nucleobases. Free Radic Res Commun 2, 217–224. Vuillaume, M. (1987). Reduced oxygen species, mutation, induction and cancer initiation. Mutat Res 186, 43–72. Wang, H., Faucette, S., Moore, R., Sueyoshi, T., Negishi, M., and LeCluyse, E. (2004). Human constitutive androstane receptor mediates induction of CYP2B6 gene expression by phenytoin. J Biol Chem 279, 29295–29301. Wei, P., Zhang, J., Egan-Hafley, M., Liang, S., and Moore, D. D. (2000). The nuclear receptor CAR mediates specific xenobiotic induction of drug metabolism. Nature 407, 920–923. Williams, G. M., and Whysner, J. (1996). Epigenetic carcinogens: Evaluation and risk assessment. Exp Toxicol Pathol 48, 189–195. Witz, G. (1991). Active oxygen species as factors in multistage carcinogenesis. Proc Soc Exp Biol Med 198, 675–682. Yamamoto, Y., Moore, R., Goldsworthy, T. L., Negishi, M., and Maronpot, R. R. (2004). The orphan nuclear receptor constitutive active/androstane receptor is essential for liver tumor promotion by phenobarbital in mice. Cancer Res 64, 7197–7200. Zhang, H., Kamendulis, L. M., Jiang, J., Xu, Y., and Klaunig, J. E. (2000). Acrylonitrile-induced morphological transformation in Syrian hamster embryo cells. Carcinogenesis 21, 727–733. Zhang, X. K., Wills, K. N., Husmann, M., Hermann, T., and Pfahl, M. (1991). Novel pathway for thyroid hormone receptor action through interaction with jun and fos oncogene activities. Mol Cell Biol 11, 6016–6025.
CH A P TE R
17
MODE OF ACTION ANALYSIS AND HUMAN RELEVANCE OF LIVER TUMORS INDUCED BY PPARα ACTIVATION
J. Christopher Corton
17.1.
OVERVIEW
A number of therapeutic hypolipidemic agents and industrial chemicals cause peroxisome proliferation and induce liver tumors in rodents via activation of the nuclear receptor peroxisome proliferator-activated receptor alpha (PPARα).* Because of the increased understanding of the relationships between PPARα activation and hepatocarcinogenesis, the purpose of this chapter is to describe the state of the science on the rodent mode of action (MOA) of liver tumor induction and human relevance. A wealth of data supports the key events in the MOA, which lead to liver tumors. These include activation of PPARα, increases in oxidative stress, increases in nuclear factor kappa B (NF-kB) activation, perturbation of hepatocyte growth, and selective clonal expansion. While these key events in the rodent MOA are biologically plausible in humans, there is no evidence that suggests that PPARα activators could induce liver tumors in humans because of differences in PPARα expression and function between rodents and humans. Lines of evidence supporting this presumption include minimal or no effects on peroxisome proliferation, peroxisomal enzyme activity, increases in oxidative stress, NF-kB activation, hepatocellular proliferation, and liver tumors in humans and/or in species that are better human surrogates than mice and rats. Even when overexpressed in the mouse liver (humanized mice), human PPARα activation does not lead to cell proliferation or liver tumors. This analysis leads to the suggestion that the PPARα activator-induced rodent liver tumors are not relevant to humans. *The information in this document has been funded by the US Environmental Protection Agency. It has been subjected to review by the National Health and Environmental Effects Research Laboratory and approved for publication. Approval does not signify that the contents reflect the views of the Agency, nor does mention of trade names or commercial products constitute endorsement or recommendation for use.
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
439
440
17.2.
CHAPTER 17 HUMAN RELEVANCE OF LIVER TUMORS
INTRODUCTION
Peroxisomes are subcellular organelles found in the cytoplasm of mammalian cells that carry out important metabolic functions (de Duve 1996; Hashimoto 1996; Mannaerts and van Veldhoven 1996). Under a variety of altered physiological and metabolic states, peroxisomes are known to proliferate, most notably with increased concentrations of unsaturated and polyunsaturated fatty acids. Interest in the toxicology community was piqued when peroxisome proliferation was noted in rodent hepatocytes in response to the administration of certain xenobiotics (e.g., Hess et al. 1965; Reddy and Chu 1996; Reddy and Rao 1977). Based on the association between exposure and peroxisome proliferation, the chemical and pharmaceutical agents that induce this response have been collectively referred to as “peroxisome proliferators.” Due to the structural heterogeneity of these compounds, the mechanism of peroxisome proliferation was an enigma for many years. The seminal discovery of the nuclear receptor PPARα in 1990 (Issemann and Green 1990), followed by extensive work with the PPARα-null mouse model, has provided a molecular underpinning of the numerous biochemical, physiological, and molecular consequences of exposure to these compounds. The term “peroxisome proliferator” remains in broad use today primarily for historical reasons. In this chapter, the term “peroxisome proliferator” has been replaced with “PPARα activator” to denote the central role PPARα plays in mediating the pleiotropic effects of exposure. “Activator” is used in place of “agonist” because very few compounds have been assayed for direct binding to PPARα using biochemical assays. Thus, PPARα activators are those chemicals or their proximate metabolites that interact directly or indirectly with PPARα, initiating events that result in receptor activation. Although most chemicals likely act as classical agonists, there is evidence that other chemicals may activate PPARα secondary to increases in the availability of natural ligands through perturbation of lipid homeostasis. For example, perfluorooctanoic acid (PFOA) may induce PPARα activation indirectly through displacement of fatty acids from fatty acid binding protein (Luebker et al. 2002). PPARα activators are a unique class of chemical carcinogens that induce peroxisome proliferation and increase the incidence of liver tumors in rats and/or mice. These include several hypolipidemic drugs (e.g., WY-14,643, gemfibrozil, fenofibrate, bezafibrate, and ciprofibrate) and environmentally relevant compounds such as phthalates or their metabolites (e.g., di-(2-ethylhexyl) phthalate (DEHP), di-(2-ethylhexyl) adipate (DEHA), diisononyl phthalate (DINP), or 2-ethylhexanol (2-EH)), pesticides (e.g., 2,4-dichlorophenoxyacetic acid, diclofopmethyl, haloxyfop, lactofen, oxidiazon), solvents (e.g., perchloroethylene, trichloroethylene), and other industrial chemicals (e.g., HCFC-123, PFOA) (summarized in Klaunig et al. (2003)). In addition to liver tumors, many PPARα activators also induce testicular Leydig cell tumors as well as pancreatic acinar cell tumors in rats but not mice (also known as the “tumor triad”). Little progress has been made to refine the proposed modes of action for the pancreatic and testicular rat tumors as detailed in Klaunig et al. (2003). As such, the present chapter will focus on the mode of action of PPARα activator-induced liver tumors.
17.3. MODE OF ACTION ANALYSIS IN THE U.S. EPA RISK ASSESSMENT FRAMEWORK
441
17.3. MODE OF ACTION ANALYSIS IN THE U.S. EPA RISK ASSESSMENT FRAMEWORK The U.S. Environmental Protection Agency (EPA) conducts risk assessments on chemical carcinogens under the guidance provided in its cancer risk assessment guidelines (EPA 2005a,b). The U.S. EPA’s new cancer guidelines highlight the use of MOA data in the assessment of potential human carcinogens and provide a framework for critical analysis of MOA information to address the extent to which the available information supports a hypothesized MOA, whether alternative MOAs are also plausible, and whether there is confidence that the same inferences can be extended to human populations and lifestages that are not represented among the experimental data. In addition, the guidelines conclude that significant information should be developed to ensure that a scientifically justifiable MOA underlies the process leading to cancer at a given site. This approach has been further refined through guidance from the International Programme on Chemical Safety (IPCS) (Boobis et al. 2006) and has been extended to noncancer effects (Boobis et al. 2008). The definition of the term MOA is important in making the determination of the adequacy of information to support it and to test whether a database for a particular chemical is consistent with that MOA. In the guidelines, the MOA is defined as “a sequence of key events and processes, starting with interaction of an agent with a cell, proceeding through operational and anatomical changes, and resulting in cancer formation [pp. 1–10]” (EPA 2005a). A key event is defined as “an empirically observable precursor step that is itself a necessary element of the mode of action or is a biologically based marker for such an element [pp. 1–10]” (EPA 2005a). The MOA is contrasted with “mechanism of action, which implies a more detailed understanding of the events, often at the molecular level, than is meant by the mode of action [pp. 1–10]” (EPA 2005a). The framework for analyzing MOA begins with a summary description of the postulated MOA. The judgment of whether a postulated MOA is supported by available data takes into account all of the data in a weight of evidence (WOE) approach. MOA analysis must determine the links between the postulated key events and tumor induction including: (i) strength, consistency, and specificity of association, (ii) dose–response relationships between the key events and tumor induction, (iii) temporal relationships including the key events preceding tumor induction, (iv) biological plausibility and coherence of the key event and its relationship with the mode of action, and (v) alternative modes of action (Boobis et al. 2006; Boobis et al. 2008). The robustness of the proposed MOA for PPARα activatorinduced rodent liver tumors and relevance to humans are examined using this framework.
17.3.1. Summary of the Mode of Action and Human Relevance of Liver Tumors Induced by PPARα Activation Substantial scientific research on the role of PPARα in rodent hepatocarcinogenesis forms the basis for the cascade of key events that describes the MOA. Although the precise mechanism for the formation of liver tumors by a PPARα activator has not
442
CHAPTER 17 HUMAN RELEVANCE OF LIVER TUMORS
PPARα Activators Fatty Acids
PPARα Activation
Increases in Oxidative Stress
Activation of Nf-kB
Perturbation of Hepatocyte Fate
Increases in Preneoplastic Foci
Liver Tumors
Figure 17.1. Proposed mode of action (MOA) of rodent liver tumor induction by PPARα activators. PPARα activators including endogenous fatty acids activate the nuclear receptor PPARα, which then regulates the transcription of different classes of genes including lipid metabolizing enzymes involved in the therapeutic hypolipidemic effects of PPARα activators. Increased activity of oxidant generating enzymes leads to increases in oxidants and oxidative stress. Oxidative stress activates NF-kB (composed of p65 and p50 subunits) either directly by cell signaling or indirectly by increases in cytokine levels including TNFα, IL-1, and IL-6 released from activated Kupffer cells. NF-kB either directly or indirectly regulates genes involved in cell growth including those involved in cell proliferation and apoptosis. PPARα activator exposure increases cell proliferation and decreases apoptosis in the liver. Preneoplastic foci that arise either spontaneously or through a mechanism that involves oxidative stress-induced DNA damage exhibit increases in cell proliferation compared to the surrounding parenchyma. Additional mutational or epigenetic changes may occur leading to hepatocellular adenomas and carcinomas. This proposed MOA is an extension of that proposed earlier (Klaunig et al., 2003) incorporating the findings of recent studies. The MOA is an endogenous pathway that can be activated independently of chemical exposure by perturbations in fatty acid levels.
been established, key events for the MOA leading to liver tumors have been identified (Figure 17.1). These include: activation of PPARα, increases in oxidative stress, increases in NF-kB activation, perturbation of hepatocyte growth, and selective clonal expansion. This MOA is similar to one proposed earlier by an International Life Sciences Institute (ILSI) workgroup (Klaunig et al. 2003), but includes the addition of the NF-kB activation event based on more recent findings. Associated events that are observed with PPARα activators and liver tumor formation and that appear to be reliable markers that a chemical has activated PPARα include increased expression or activity of some peroxisomal genes (e.g., acyl-CoA oxidase encoding palmitoyl-CoA oxidase (PCO)) and peroxisome proliferation (i.e., an increase in the number and size of peroxisomes). PCO activity levels are correlated with cancer potency (Klaunig et al. 2003). The EPA cancer guidelines state, “If an hypothesized mode of action is sufficiently supported in the test animals, the sequence of key precursor events should be reviewed to identify critical similarities and differences between the test animals and humans [pp. 2–47].” (EPA 2005a). Despite the fact that PPARα activators induce liver tumors in rats and mice, the potential for PPARα activators to induce liver tumors in other species, including humans, is low. Tumor induction is unlikely because evidence obtained from in vivo and in vitro studies with hamsters, guinea pigs, nonhuman primates, and humans (i.e., cells in culture or liver biopsies) shows that, quantitatively, these other species are less likely to exhibit the key events upon PPARα activator exposure. Increases in liver-to-body weights and peroxisome proliferation are not evident in humans, although therapeutic hypolipidemic compounds lower triglyceride levels mediated by PPARα across species.
17.3. MODE OF ACTION ANALYSIS IN THE U.S. EPA RISK ASSESSMENT FRAMEWORK
443
There are several plausible explanations for the species-specific effects of PPARα activators: (1) Full-length PPARα protein is expressed at levels at least 10-fold greater in rodent liver than in human liver. (2) Humans but not rodents express an inactive form of PPARα in the liver which inhibits the active PPARα. (3) Even when expressed at levels similar to the mouse PPARα, human PPARα does not induce the cell-cycle machinery, cell proliferation, or liver cancer in mice. (4) The PPARα-responsive elements (PPREs) of some target genes, including acylCoA oxidase, have been shown to differ between rodents and humans. (5) Human epidemiological studies, although limited in duration, have not provided evidence of increased incidence of any type of cancer including liver neoplasms in humans. These data support the conclusion that induction of rodent liver tumors by this MOA may not be relevant to humans.
17.3.2.
Detailed Evaluation of the Rodent Mode of Action
The following sections provide an in-depth analysis of the proposed MOA for rat and mouse liver tumors induced by PPARα activators. This analysis is not intended to reflect an exhaustive review of the literature but rather a summation of key evidence. This review of the data did not consider alternative effects of PPARα activators, including mitochondrial effects, gap junction intercellular communication (GJIC), or methylation of DNA, because the data are weak or a direct causal link to liver tumor induction is lacking. The liver consists of the hepatic parenchyma (hepatocytes) and nonparenchymal cells (NPCs) including sinusoidal endothelial cells, Ito cells, and the dedicated hepatic macrophages known as the Kupffer cells. Kupffer cells can be “activated” by some liver toxicants leading to release of signaling molecules such as reactive oxygen species and cytokines (Roberts et al. 2007). Kupffer cells play important roles in responses to PPARα activators and are required for key events, which lead to liver tumors. Although the rodent PPARα activator mode of action is presented as a linear set of key events (Figure 17.1), some of the key events are likely dependent on interactions between hepatocytes and Kupffer cells. These relationships are described below. The evaluation of the key events in the MOA is discussed below followed by a discussion of mechanistic studies, which provide linkages between the key events. 17.3.2.1. PPARα Activation. PPARα is activated by many environmentally relevant chemicals as well as by endogenous fatty acids and their metabolites (Dreyer et al. 1992; Gottlicher et al. 1992; Issemann and Green 1990; Sher et al. 1993). Chemical-specific data show excellent correlations between PPARα activation, the key events in the MOA and liver cancer (see Table 17.1 for examples of five PPARα activators). PPARα activation is a causative key event in the PPARα activator MOA for liver tumor induction, as indicated by the following findings: (1) PPARα activation is consistently associated with exposure to PPARα activators in trans-activation assays (summarized in (Corton et al. 2000) and (Klaunig et al. 2003)); (2) the level of activation of PPARα in such assays is roughly proportional to the potency of the chemical as an inducer of liver tumor response (summarized
TABLE 17.1.
Occurrence of Key Events in the MOA after Exposure to PPARα Activators
444
PPARα Activation
Oxidative Stress
NF-kB Activation
WY-14,643
+26,27,29
+2,3,5,12,72
DEHP
+26,27,28
Clofibrate
+28,29
Nafenopin
+28,30
Ciprofibrate
+30
+2,6,7,8,9,16,18,72,75 −71,72 +8,10,14,15,20,40,41,59 −8,23,40 +9,15,21,24,51,76,84 −7,23,66 +9,22–24 −25,65 +17,18
Chemical
Increases in Transient Acute Cell Proliferation
Decreases in Acute Apoptosis
Increases in Chronic Cell Proliferation
Increases in Cell Proliferation in Preneoplastic Foci +73,74
Liver Tumors
+6,7,31,35
+55
+6,7,35
+31,32,42–45
+43
−10
+10 +47,48
+54,83
+7 −39 +80 +/−35 +34
+7,33,39,45 −13,87
+35
+1,4,11,19
+34,37,38
+35
+36
+35,49,62
+37
+50
Comments: In the table, a plus sign (+) indicates the chemical was found to lead to the key event, a minus sign (−) indicates the chemical was found not to lead to the key event, and plus/minus (+/−) indicates mixed results in the same study. PPARα activation was measured using trans-activation assays. NF-kB activation refers to binding of NF-kB (p65:p50 heterodimer) to the NF-kB response element in electrophoretic mobility shift assays. Acute cell proliferation was measured in the livers of treated mice or rats usually with 7 days or less of exposure. Apoptosis was mostly measured in primary hepatocytes given the low background in intact livers. However, three studies have measured apoptosis in rodent livers after exposure to a PPARα activator (Bursch et al. 1984; James et al. 1998b; Youssef et al. 2003). Chronic cell proliferation was measured in the livers of mice or rats exposed to PPARα activators, usually for >3 weeks. Noted in the references below: Studies were carried out with mice (M) or rats (R) or both species (M,R). The endpoint is indicated for studies that measured oxidative stress. If there are inconsistent effects, the possible origin of the inconsistency is indicated. In vitro studies are also noted. References: 1Calfee-Mason et al. (2004) (R); 2Fischer et al. (2002) (increase in TBARS but not conjugated dienes) (R); 3Rusyn et al. (2000b,c) (M,R); 4Nilakantan et al. (1998) (M); 5Rusyn et al. (1998) (R); 6Wada et al. (1992) (lipofuscin) (R); 7Marsman et al. (1992) (lipofuscin; trend for increase in cell proliferation by clofibrate) (R); 8Conway et al. (1989) (lipofuscin—positive for both WY-14,643, and DEHP but only WY-14,643 positive for conjugated dienes) (R); 9Reddy et al. (1982) (lipofuscin) (R); 10Cattley et al. (1987) (lipofuscin) (R); 11Tharappel et al. (2003) (M); 12Tharappel et al. (2001) (consistent changes with WY-14,643, but only one condition resulted in increases in NF-kB activation after gemfibrozil treatment) (R); 13Menegazzi et al. (1997) (R); 14Rao et al. (1987) (lipofuscin) (R); 15 Lake et al. (1987) (lipofuscin) (R); 16Rao et al. (1982) (lipofuscin) (R); 17Rao et al. (1991) (lipofuscin) (R); 18Goel et al. (1986) (lipid peroxidation and hydrogen peroxide) (R); 19Li et al. (1996) (R); 20 Hinton et al. (1986) (lipofuscin) (R); 21Stanko et al. (1995) (lipofuscin) (R); 22Lake et al. (1989a) (increases in oxidized glutathione and decreases in vitamin E) (R); 23Tomaszewski et al. (1990) (in vitro cultures; oxidized dienes) (R); 24Cai et al. (1995) (lipid peroxidation—trend increases for PFOA, nafenopin and clofibrate) (M); 26Bility et al. (2004) (in vitro trans-activation assays) (M); 27Corton and Lapinskas (2005) (review of in vitro trans-activation data) (M,R); 28Isseman and Green (1990) (in vitro trans-activation assays) (M); 29Gottlicher et al. (1992) (in vitro trans-activation assays) (R); 30Corton et al. (2000) (review) (M,R); 31Marsman et al. (1988) (R); 32Smith-Oliver and Butterworth (1987) (R); 33Tanaka et al. (1992) (R); 34Yeldandi et al. (1989) (chronic increases in cell proliferation) (R); 35 Lake et al. (1993) (R); 36Schulte-Hermann et al. (1981) (R); 37Chen et al. (1994) (R); 38Dwivedi et al. (1989) (M); 39Barrass et al. (1993) (R); 40Seo et al. (2004) (malondialdehyde) (R); 41Isenberg et al. (2001) (M,R); 42Isenberg et al. (2000) (M,R); 43Hasmall et al. (2000) (R, in vivo (DEHP) and in vitro (MEHP)); 44Soames et al. (1999) (R); 45Busser and Lutz (1987) (R); 47Reddy and Qureshi (1979) (R); 48 Svoboda and Azarnoff (1979) (R); 49Reddy and Rao (1977) (R); 50Rao et al. (1986) (R); 51Elliott and Elcombe (1987) (malondialdehyde—significant change for DEHP and clofibrate but trend increase for methyl clofenapate) (R); 54James and Roberts (1996) (in vitro) (M,R); 55Youssef et al. (2003) (R); 59Thottassery et al. (1992) (R); 62Abdellatif et al. (1990) (initiated with DEN, 2-acetylaminofluorene and carbon tetrachloride) (R); 66Nicholls-Grzemski et al. (2000) (TBARS) (M); 71Soliman et al. (1997) (F2-isoprostanes) (R); 72Fischer et al. (2002) (TBARS increase with treatment but conjugated dienes do not) (R); 73Marsman and Popp (1994) (R); 74Rose et al. (1999a) (R); 75O’Brien et al. (2001b) (decreases in vitamin E) (R); 76Qu et al. (2000); 80Price et al. (1992) (R); 83Bursch et al. (1984) (R, in vivo); 84 Dostalek et al. (2008) (M, increases in hydrogen peroxide, malondialdehyde, and urine F2-isoprostanes but not liver F2-isoprostanes); 87Ohmura et al. (1996) (R).
17.3. MODE OF ACTION ANALYSIS IN THE U.S. EPA RISK ASSESSMENT FRAMEWORK
445
in (Klaunig et al. 2003)); and (3) importantly, the majority of studies using PPARαnull mice do not show hepatocyte-specific changes associated with hepatocarcinogenesis (discussed below). There are examples where PPARα activation does not consistently lead to liver cancer, and these have been summarized in Klaunig et al. (2003). Weak PPARα activators (i.e., compounds that minimally induce markers of PPARα activation) would not necessarily increase liver tumor incidence, because a sufficient level of receptor activation is needed for induction of key events (Klaunig et al. 2003). Pharmacokinetic differences between susceptible and nonsusceptible rodents that lead to differences in tissue chemical concentration could also contribute to discrepancies between the ability of chemicals to activate PPARα in trans-activation assays and tumor induction. For example, trichloroacetate (TCA) exposure in mice leads to increases in PCO activity at doses similar to or below those that induce liver tumors whereas in rats TCA, even at high doses, only marginally increases PCO in the absence of increases in liver tumors (Corton 2008). PPARα regulates lipid homeostasis and peroxisome proliferation through the modulation of genes involved in fatty acid uptake, activation, and oxidation as well as peroxisome assembly (the Pex genes) (Desvergne et al. 1998; Desvergne and Wahli 1999; Schoonjans et al. 1996; Wahli et al. 1995). Collectively, these changes result in increased ability to metabolize fatty acids leading to the therapeutic lowering of lipid levels in mice, rats, Syrian hamsters, guinea pigs, monkeys, and humans. These changes have been shown to be PPARα-dependent in mice (summarized in (Peters et al. 2005)). Alterations in lipid metabolism and peroxisome proliferation genes are not thought to be involved in the hepatocarcinogenic effects of PPARα activators (Klaunig et al. 2003). 17.3.2.2. Role of Oxidative Stress in PPARα Activator-Induced Hepatocarcinogenesis. Linkages exist between increases in ROS and increased incidence of liver cancer by PPARα activators. Overproduction of oxidants might cause DNA damage leading to mutations and cancer (Reddy and Rao 1989; Yeldandi et al. 2000). In whole liver of both rats and mice, markers of oxidative stress were increased by PPARα activators (Table 17.1), as determined by measuring indices of lipid peroxidation (e.g., conjugated dienes, lipofuscin, malondialdehyde, F2isoprostanes, etc.), oxidized glutathione, or hydrogen peroxide. A few studies failed to detect increases in markers of oxidative stress, but these are difficult to interpret because other key events were not simultaneously analyzed (e.g., Huber et al. 1991, 1997). There were other studies in which one assay for oxidative stress was positive but another negative (e.g., Conway et al. 1989; Fischer et al. 2002). In spite of these minor discrepancies, the WOE demonstrates that PPARα activators increase oxidative stress. Possible sources of ROS in the livers of rodents exposed to PPARα activators include enzymes that generate and degrade hydrogen peroxide and other ROS. Hydrogen peroxide can oxidize DNA, lipids, and other molecules, and PPARα activators regulate the expression of many enzymes that produce hydrogen peroxide as a byproduct of metabolism including the peroxisomal, mitochondrial, and microsomal oxidases in hepatocytes such as fatty acyl-CoA oxidase (ACO) (Becuwe and
446
CHAPTER 17 HUMAN RELEVANCE OF LIVER TUMORS
Dauca 2005). Administration of PPARα activators can also lead to decreased levels of some enzymes, which degrade ROS that may contribute to the increases in oxidative stress upon exposure (Glauert et al. 1990; O’Brien et al. 2001a,b). The individual contributions of these enzymes to increases in oxidative stress and downstream key events leading to liver tumor induction has not been comprehensively addressed but will likely be complex. In one example, Reddy and co-workers originally proposed that peroxisomal ACO (Acox1) is the enzyme responsible for oxidative stress by PPARα activators (Nemali et al. 1988). However, ACO was later found to be dispensable for increases in oxidative stress. Control ACO-null mice exhibited the phenotype of wild-type mice exposed to PPARα activators including increases in oxidative stress and induction of liver tumors that are dependent on PPARα (Fan et al. 1998; Hashimoto et al. 1999). The role of other ACO family members (Acox2, Acox3) has not been determined in this Acox1-independent induction of oxidative stress and liver tumors. Extensive testing of PPARα activators has shown that these compounds do not consistently induce direct DNA damage. However, indirect DNA damage from oxidative stress has been hypothesized to be a common pathway for many nongenotoxic chemical carcinogens including PPARα activators (Klaunig et al. 1998). Relationships exist between chemical exposure, DNA damage, and cancer based on measurement of 8-hydroxy-2′-deoxyguanosine (8-OH-dG), a highly mutagenic lesion, in DNA isolated from livers of animals treated with PPARα activators (Kasai 1997; Qu et al. 2001; Takagi et al. 1990). However, subsequent studies showed that the increases in oxidative DNA damage may have originated in the way in which the genomic DNA was prepared (Cattley and Glover 1993; Sausen et al. 1995). Experiments measuring other indicators of DNA damage—that is, abasic sites or single strand-breaks in genomic DNA from rats and mice treated with WY-14,643 for one month—failed to show increases over controls (Rusyn et al. 2004). Only in the livers of wild-type but not PPARα-null mice treated with WY-14,643 for five months were there increases in abasic sites in genomic DNA (Woods et al. 2007a), indicating that exposure times longer than one month were necessary to observe increases in DNA damage. The relationship between the increases in abasic sites and subsequent tumor yield has not been determined. DNA repair mechanisms might compensate for increases in DNA damage and may explain the lack of consistent evidence for DNA damage from PPARα activator-induced oxidative stress. PPARα activators increased the expression of liver genes involved in the long-patch base excision DNA repair pathway in a timedependent manner; the degree of induction roughly correlated with the dose and carcinogenic potency of the PPARα activators tested (Rusyn et al. 2000a). Additionally, expression of enzymes that do not repair oxidative DNA damage was not changed. This induction of DNA base excision repair genes may be an indicator that DNA damage is occurring. Evidence that DNA damage caused by PPARα activator-induced oxidative stress is not involved in hepatocarcinogenesis comes from recent work with Ogg1null mice. Ogg1 encodes an 8-oxoguanine DNA glycosylase that repairs one of the major DNA lesions generated by ROS. Control Ogg1-null mice show elevated levels of oxidative DNA damage and exhibit increased spontaneous mutation rates in the
17.3. MODE OF ACTION ANALYSIS IN THE U.S. EPA RISK ASSESSMENT FRAMEWORK
447
absence of chemical exposure (Klungland et al. 1999). Ogg1-null mice, when exposed to WY-14,643 in the diet, did not show additional oxidative DNA damage but exhibited increased numbers and total volumes of preneoplastic lesions in the liver compared to similarly treated wild-type mice (Trapp et al. 2007). The authors concluded that the increase in preneoplastic lesions associated with WY-14,643 exposure did not arise from induced oxidative damage, but instead arose from the promotion of spontaneous mutations generated by endogenous oxidative DNA damage. Overall, PPARα activators increase the level of oxidative stress through multiple mechanisms. There is little direct evidence that increases in oxidative stress generated after PPARα activator exposure leads to direct or indirect DNA damage. The Ogg1-null mouse studies indicate that PPARα activators promote hepatocytes that have been spontaneously initiated. The WOE suggests that direct or oxidatively induced DNA damage is not part of the MOA. 17.3.2.3. Role of NF-kB in the PPARα Activator MOA. Central to the PPARα activator MOA is NF-kB activation. NF-kB transcription factors play critical roles in cancer development and progression (Arsura and Cavin 2005; Karin 2006). A wealth of data demonstrates that NF-kB is activated under conditions of inflammation and oxidative stress (Czaja 2007; Gloire et al. 2006). Consistent with this, studies with PPARα activators demonstrate linkages between oxidative stress and NF-kB activation. Activation is usually assessed by the ability of nuclear NF-kB (usually a heterodimer composed of p50 and p65 subunits) to bind to a NF-kB response element in an electrophoretic mobility shift assay (EMSA). In whole liver of both rats and mice, activity of NF-kB was increased by PPARα activators including WY-14,643, ciprofibrate, and gemfibrozil but not nafenopin (Table 17.1). The fact that nafenopin did not induce NF-kB may be due to differences in the EMSA procedures carried out by that lab (Menegazzi et al. 1997; Ohmura et al. 1996). NF-kB is activated in Kupffer cells and in hepatocytes at different times after exposure. After a single in vivo dose of WY-14,643, NF-kB activity was increased first in Kupffer cells (at 2 hours), and only ∼6 hours later was NF-kB activity increased in hepatocytes. Activation in hepatocytes never achieved the level observed in Kupffer cells (Rusyn et al. 1998). The increase in NF-kB activation in hepatocytes could be due to increases in mitogenic cytokines produced by Kupffer cells that activate signal transduction pathways ultimately impinging on NF-kB. Alternatively, NF-kB can be activated directly by a PPARα activator in the H4IIEC3 rat hepatoma cell line, responsive to the proliferative effects of PPARα activators (Li et al. 2000a). Increased NF-kB activity may be secondary to the action of hydrogen peroxidegenerating enzymes, such as ACO, since overexpression of ACO in COS-1 cells, in the presence of a hydrogen peroxide-generating substrate, can activate a NF-kBregulated reporter gene (Li et al. 2000b). 17.3.2.4. Alteration of Cell Proliferation/Apoptosis Balance by PPARα Activators. PPARα activators produce multiple tumor precursor effects including liver hyperplasia, and altered growth in preneoplastic foci. Increased cell replication induced by PPARα activators may increase the frequency of spontaneous mutations
448
CHAPTER 17 HUMAN RELEVANCE OF LIVER TUMORS
by increasing the frequency of errors in DNA repair or replication and can lead to silencing of tumor suppressor genes or increased expression of oncogenes (Cattley et al. 1998; Huber et al. 1991). Alternatively, PPARα activators can promote the growth of spontaneously initiated hepatocytes. All PPARα activators at a sufficient dose produce a strong, albeit transient, increase in replicative DNA synthesis during the first few days of exposure (Table 17.1). After this initial burst in replication, baseline levels of hepatocyte replication are approached while the liver remains enlarged. Many PPARα activators exhibit measurable sustained or chronic increases in cell proliferation, although the levels are much lower than that observed after acute exposures (Table 17.1). There are some PPARα activators that do not induce chronic cell proliferation; this may be due to the dose used in the experiment and because weak increases above variable background levels of cell proliferation are difficult to detect. PPARα activators promote the growth of chemically and spontaneously induced lesions through enhanced cell replication (Cattley et al. 1987; Isenberg et al. 1997; Marsman et al. 1988). Once early lesions are formed, continued exposure to PPARα activators causes a selective increase in DNA replication of up to ∼40% in these liver foci, while replication of hepatocytes in the normal surrounding liver is increased only slightly (Grasl-Kraupp et al. 1993). Furthermore, preneoplastic foci respond to the cell replicative effects, rather than the peroxisome proliferation effects of PPARα activators, suggesting that the growth stimulus but not the peroxisome proliferation effect is of particular significance for the carcinogenic action of this class of compounds (Grasl-Kraupp et al. 1993). Increases in cell proliferation alone are not sufficient to increase liver tumors. The response of mice transgenic for hepatocyte-specific expression of a constitutively activated form of PPARα (VP16PPARα) was compared to wild-type mice treated with WY-14,643 (Yang et al. 2007). Expression of VP16PPARα led to increases in hepatocyte proliferation in the absence of nonparenchymal cell proliferation, in contrast to WY-14,643 treatment in wild-type livers in which both hepatocytes and nonparenchymal cells exhibited increased replication. Importantly, chronic activation of VP16PPARα did not increase liver tumors (Yang et al. 2007). These results indicate that nonparenchymal cell activation is important for hepatocarcinogenesis and that PPARα-mediated hepatocyte proliferation by itself is not sufficient to induce liver cancer. Taken together, the results indicate that it is the combination of events in hepatocytes and NPC that are important for induction of liver tumors by PPARα activators. Nongenotoxic carcinogens, in general, and PPARα activators in particular suppress hepatocyte apoptosis. Suppression of apoptosis could inhibit the ability of the liver to remove DNA-damaged, pre-neoplastic hepatocytes (Bayly et al. 1994; James and Roberts 1996; Oberhammer and Qin 1995; Schulte-Hermann et al. 1981). Most of the evidence for apoptosis suppression comes from in vitro studies because of the difficulty in measuring the suppression of already low levels of apoptosis in vivo. Studies conducted in vitro show that the PPARα activators nafenopin, methylclofenapate, and WY-14,643 suppress spontaneous hepatocyte apoptosis as well as that induced by a negative regulator of liver growth, transforming growth factor beta 1 (TGFβ1) (Bayly et al. 1994; Oberhammer and Qin 1995) (Table 17.1).
17.3. MODE OF ACTION ANALYSIS IN THE U.S. EPA RISK ASSESSMENT FRAMEWORK
449
In addition, PPARα activators can suppress apoptosis in vitro induced by diverse stimuli such as DNA damage or ligation of Fas, a receptor related to the tumor necrosis factor alpha (TNFα) family of cell surface receptors (Gill et al. 1998). A limited number of in vivo studies also showed suppression of apoptosis after acute dosing with nafenopin, DEHP or WY-14,643 (Bursch et al. 1984; James et al. 1998b; Youssef et al. 2003). Suppression of apoptosis by PPARα activators occurs under acute exposure conditions when the liver is increasing in size. However, once a steady state of liver enlargement is reached, levels of apoptosis are likely to return to background levels or to levels which balance the low level of cell proliferation that occurs for some PPARα activators. Consistent with this, two reports suggest that chronic exposure of rats and mice to the PPARα activator WY-14,643 results in an increase in apoptosis (Burkhardt et al. 2001; Marsman et al. 1992). Furthermore, PPARα activators alter the ability of the liver to respond to apoptosis inducers. Sensitivity to two apoptosis inducers (Jo2 antibody and conconavalin A) was dramatically increased in wild-type but not PPARα-null mice exposed for 1 week to WY-14,643 (Xiao et al. 2006). Lastly, both cell proliferation and apoptosis increase in parallel in PPARα activator-induced tumors in the rat compared with normal surrounding tissue, suggesting that cell turnover is increased in tumorigenic lesions (GraslKraupp et al. 1997). To summarize, alterations in the balance between hepatocyte proliferation and apoptosis have been observed after exposure to multiple PPARα activators at different stages of carcinogenesis including under acute and chronic exposure conditions and in the preneoplastic and tumorigenic lesions. 17.3.2.5. Mechanisms of Cell Growth Alterations. Extensive work has been carried out to identify the mechanistic events that lead to alterations in cell growth by PPARα activators. There are a number of excellent reviews on the subject of signal transduction and downstream events that lead to alterations in cell growth (Burns and Vanden Heuvel 2007; Gonzalez and Shah 2008; Rusyn et al. 2006). Early studies focused on the regulation of individual growth genes that respond to growth promoting stimuli. More recent studies capitalized on technological advancements in assessing global changes in gene expression or assessing the role of individual genes/pathways in the intact animal using transgenic technologies. Many studies focused on growth factors derived from the Kupffer cell. Activated NPCs, particularly Kupffer cells, produce cytokines such as TNFα, interleukin-1α, and interleukin-1β (IL1α, IL1β). These cytokines affect the fate of neighboring hepatocytes. TNFα is able to increase hepatocyte proliferation and suppress apoptosis in cultured rodent hepatocytes (Holden et al. 2000; Rolfe et al. 1997). In intact animals hepatocyte growth can be prevented by injection of antibodies to either TNFα (Bojes et al. 1997; Rolfe et al. 1997) or TNFα receptor 1 (West et al. 1999). PPARα activators increased TNFα mRNA more than twofold (Bojes et al. 1997; Rolfe et al. 1997). Because increases in TNFα expression have not been consistently observed by others (Anderson et al. 2001; Holden et al. 2000), treatment with PPARα activators may not result in de novo TNFα expression, but
450
CHAPTER 17 HUMAN RELEVANCE OF LIVER TUMORS
rather bioactivation or release of preexisting TNFα protein from Kupffer cells (Holden et al. 2000). Other studies suggest that the cell proliferation response to PPARα activators is TNFα pathway-independent. Cell proliferation remained intact in TNFα-null and in TNFα receptor-null mice given a PPARα activator (Anderson et al. 2001; Lawrence et al. 2001). In addition, IL-1 receptor-null mice retained the ability to respond to the induction of hepatocyte proliferation to WY-14,643 (Corton et al., unpublished observations). There remains the possibility that loss of TNFα or IL-1 signaling results in compensation by other genes/pathways, including other cytokine-mediated pathways, because multiple growth modulators secreted by Kupffer cells have been suggested to play a role in hepatocyte proliferation after diethylnitrosamine (DEN) exposure (Maeda et al. 2005). Thus, studies with various nullizygous mice do not necessarily refute the role TNFα or IL-1 may play in PPARα activator-induced cell proliferation. MicroRNAs (miRNA) play important roles in complex processes such as development through the regulation of gene expression. Recent global analysis of the miRNA expression pattern after WY-14,643 exposure has uncovered a signaling pathway which culminates in increased expression of the c-Myc growth regulatory gene (Shah et al. 2007). Expression of let-7C, an miRNA important in cell growth, was down-regulated following acute or chronic treatment with WY-14,643 in wild-type mice. Because let-7C down-regulates the expression of c-Myc, the down-regulation of let-7C by WY-14,643 resulted in increased expression of cMyc. These molecular events did not occur in PPARα-null mice. These studies reveal a let-7C signaling cascade critical for PPARα activator-induced hepatocyte proliferation. Other growth signaling pathways may be involved in PPARα activator growth responses, but overall the data supporting their role is usually confined to gene expression data. Due to the lack of useful genetic models, there is little mechanistic data, which shows causal links between specific pathways and modulation of cell fate except for the role of PPARα and NF-kB activation (discussed below). 17.3.2.6. Genetic and Biochemical Inhibition Studies Support the MOA. Genetic and biochemical inhibition studies have highlighted the relationships between the key events of the PPARα activator MOA (Table 17.2). These studies showed that when a key event is inhibited genetically or biochemically, the downstream but not upstream event(s) are inhibited as well. Genetically modified mice have been useful to show the relationships between the key events in the PPARα MOA. PPARα-null mice provided critical evidence establishing the rodent MOA for PPARα activator-induced hepatocarcinogenesis. Evidence that a particular compound induces key events in wild-type mice but not in mice lacking PPARα would be considered strong support for a PPARα MOA for that particular compound. To date, three chronic bioassays have been conducted in these mice (Hays et al. 2005; Ito et al. 2007; Peters et al. 1997). A greater body of data exists in which precursor events for cancer have been assessed in wild-type and PPARα-null mice after acute or subacute exposures.
TABLE 17.2.
Effects of Inhibition of Key Events in the PPARα Activator MOA
Key Event Mechanism of Inhibition Genetic Inhibition PPARα-null Catalase transgenic P47Phox-null P50-null Biochemical Inhibition Antioxidants in diet Dexamethasone Glycine Methylpalmitate Diphenyleneiodonium
PPARα Activation
Oxidative Stress
NF-kB Activation
Alteration in Hepatocyte Growth
Clonal Expansion
↓ (by definition) NC11 NC1,27,29 NC8,9
↓1,27↑32
↓1 ↓11 ↓29, NC27 ↓8
↓2,3,4 ↓11 ↓29, NC27 ↓8,9
↓2,4
↓2,4↑32
↓9
↓9
NC6,10 NC14,15 ↓23 NC19,21 NC30 NC29
↓7,16↑10
↑10
↓5,12,13↑10
↓29, NC1
↓6,7,26 ↓17,18,24,25
↓20,22 ↓29
↓29
↓ ↓19, NC21 ↓30 ↓29
Liver Tumors
14,15,23
↓21
↓, inhibited; NC, no change; ↑, increases in the parameters measured. For studies in which antioxidants were co-treated with PPARα activators, the antioxidant is indicated in parentheses. References: 1Woods et al. (2007a); 2Peters et al. (1997); 3Peters et al. (1998); 4Hays et al. (2005); 5Rao et al. (1984) [ethoxyquin, 2(3)-tertbutyl-14-hydroxyanisole]; 6Calfee-Mason et al. (2004) (vitamin E); 7Li et al. (2000a) (in vitro studies with vitamin E-treated H4IIE3C cells); 8Tharappel et al. (2003); 9Glauert et al. (2006); 10Glauert et al. (1990) (vitamin E increases the number of tumors while depleting glutathione reserves); 11Nilakantan et al. (1998); 12Rao and Subbarao (1999) (dimethylthiourea); 13Rao and Subbarao (1997a) (deferoxamine—iron chelator); 14 Lawrence et al. (2001c); 15Rao and Subbarao (1997b) (dexamethasone); 16Stanko et al. (1995) (vitamin E); 17Ray and Prefontaine (1994); 18Widen et al. (2003); 19Rose et al. (1997a,b); 20Rose et al. (1999a) (superoxide production in Kupffer cells); 21Rose et al. (1999b); 22Rusyn et al. (2001) (free radicals in bile); 23Ohmura et al. (1996) (measured peroxisomal bifunctional enzyme as PPARα marker); 24Chang et al. (1997); 25De Bosscher et al. (2006) (review); 26Rusyn et al. (1998) (allopurinol); 27Woods et al. (2007b); 29Rusyn et al. (2000b,c); 30Rose et al. (1997b); 32Ito et al. (2007).
451
452
CHAPTER 17 HUMAN RELEVANCE OF LIVER TUMORS
Two studies assessed markers of oxidative stress in wild-type and PPARαnull mice. In the first study, abasic sites (i.e., sites that lack either a purine or a pyrimidine) in genomic DNA were used as a measure of oxidative stress. These sites were increased in wild-type but not PPARα-null mice after exposure to WY-14,643 for 5 months (Woods et al. 2007a). In the second study, electron spin resonance (ESR) identified increases in free radicals in the bile of wild-type but not PPARαnull mice after up to 3-week exposures to WY-14,643 or DEHP. NF-kB activation was observed in the livers of wild-type but not PPARα-null mice after exposure to WY-14,643 (Woods et al. 2007a,b). Using global gene expression profiling, alteration of gene expression by WY-14,643, PFOA, or ciprofibrate was almost completely abolished in PPARα-null mice at multiple time points (Anderson et al. 2004a,b; Corton et al. 2004; Rosen et al. 2008a,b; Sanderson et al. 2008; Woods et al. 2007c; Corton et al., unpublished). The up-regulation of the cell cycle components cyclin-dependent kinase (CDK)-1, CDK-2, CDK-4 and proliferating cell nuclear antigen (PCNA) proteins and CDK-1, CDK-4 and cyclin D1 mRNA was observed in wild-type but not PPARα-null mice fed WY-14,643 (Peters et al. 1998). Wild-type mice exhibited increased hepatocyte proliferation compared to untreated controls while no increases in hepatocyte proliferation were observed in PPARα-null mice after exposure to WY-14,643, DINP, PFOA, or trichloroethylene (Laughter et al. 2004; Peters et al. 1997, 1998; Valles et al. 2003; Wolf et al. 2008; Corton et al., unpublished). The ability of PPARα activators to suppress apoptosis was lost in hepatocytes isolated from PPARα-null mouse livers (Hasmall et al. 2000a). Importantly, chronic treatment with WY-14,643 or bezafibrate resulted in 100% incidence of hepatocellular neoplasia in wild-type mice while the PPARα-null mice were unaffected (Hays et al. 2005; Peters et al. 1997). An additional bioassay in which DEHP induced liver tumors in PPARα-null but not wild-type mice (Ito et al. 2007) is discussed below. Although the WY-14,643 and bezafibrate chronic exposure studies were carried out for relatively short exposure periods (up to a year), the PPARα-null mice did not exhibit any of the precursor events associated with carcinogenesis (Hays et al. 2005; Peters et al. 1997, 1998), making it unlikely that longer-term exposure would result in liver tumors. These studies demonstrate that all of the key events in the MOA are dependent on PPARα. Two transgenic mouse models have been used to determine the relationships between different sources of oxidative stress and downstream events. Catalase converts hydrogen peroxide to water and oxygen. In catalase-transgenic mice that exhibit increased liver expression and activity of catalase, there were decreased levels of NF-kB activation and decreased hepatocyte proliferation upon exposure to ciprofibrate (Nilakantan et al. 1998). NADPH oxidase in Kupffer cells plays an important role in generating superoxide radicals in response to Kupffer cell activators (De Minicis et al. 2006). NADPH oxidase is activated by PPARα activators and is important in cell proliferation after short-term PPARα activator exposure. Mice that lack one of the subunits of NADPH oxidase (the p47Phox-null mice) did not exhibit increases in oxidative stress, NF-kB activation, and hepatocyte proliferation after short-term PPARα activator exposure (Rusyn et al. 2000b,c). However, after exposure of mice to WY-14,643 for three weeks, there were increases in indicators of oxidative stress (including PCO activity), NF-kB activation and cell proliferation,
17.3. MODE OF ACTION ANALYSIS IN THE U.S. EPA RISK ASSESSMENT FRAMEWORK
453
independent of the status of the p47Phox gene; these key events were dependent on PPARα (Woods et al. 2007a,b). Longer-term exposure may allow bypass of p47Phox dependence including increases in oxidative stress through activation of enzymes that produce hydrogen peroxide. NF-kB activation is involved in modulation of hepatocyte fate in response to inducers of oxidative stress [e.g., Maeda et al. (2005)] including PPARα activators. Wild-type mice and mice deficient in the p50 subunit of NF-kB (p50-null mice) were fed a diet with or without 0.01% ciprofibrate for 10 days. NF-kB DNA binding activity was increased after ciprofibrate treatment in wild-type mice but not p50-null mice. The apoptotic index was low in wild-type mice in the presence or absence of ciprofibrate. Consistent with NF-kB acting as a negative regulator of apoptosis (Arsura and Cavin 2005; Karin 2006), apoptosis was higher in untreated p50-null mice compared to wild-type mice (Tharappel et al. 2003). Apoptosis was reduced in p50-null mice after ciprofibrate feeding but was still higher than wild-type levels. The untreated p50-null mice had a higher level of hepatic cell proliferation, as measured by bromodeoxyuridine (BrdU) labeling, than did untreated wild-type mice possibly as a mechanism to compensate for the higher levels of apoptosis. However, ciprofibrate-fed p50-null mice had lower levels of cell proliferation than comparatively treated wild-type mice (Tharappel et al. 2003). A chronic (38-week) exposure study provides direct evidence that NF-kB activation is necessary for hepatocarcinogenesis induced by a PPARα activator (Glauert et al. 2006). Wild-type mice receiving only DEN developed a low incidence of tumors (25%). The majority of wild-type mice receiving both DEN + WY-14,643 developed tumors (63%). However, no tumors were seen in the DEN or DEN + WY14,643-treated p50-null mice, demonstrating that the p50 subunit of NF-kB was required for the promotion of hepatic tumors by WY-14,643. Treatment with DEN + WY-14,643 increased both cell proliferation and apoptosis in wild-type and p50-null mice. Consistent with the tumor levels, cell proliferation and apoptosis were lower in the p50-null mice than in wild-type mice (Glauert et al. 2006). This study shows direct dependence on the p50 subunit of NF-kB for liver tumor induction by a PPARα activator. Biochemical inhibition studies using compounds that inhibit oxidative stress or inflammation also highlight linkages of the key events in the PPARα MOA. In these studies, animals were pretreated with the inhibitor before PPARα activator exposure or co-treated with a PPARα activator and the inhibitor. The free radical scavenger and xanthine oxidase inhibitor allopurinol inhibited the activation of NF-kB in the livers of WY-14,643-treated rats (Rusyn et al. 1998). In in vitro studies, the anti-oxidants vitamin E or N-acetylcysteine blocked the ability of NF-kB to activate a reporter gene in ciprofibrate-treated HIIE3C cells (Li et al. 2000b). Co-treatment with ciprofibrate and one of two anti-oxidants, 2(3)-tert-butyl-14hydroxyanisole or ethoxyquin, decreased the incidence and size of liver tumors compared to ciprofibrate treatment alone (Rao et al. 1984). Studies using either dimethylthiourea or deferoxamine as antioxidants decreased the incidence of liver tumors in rats fed the PPARα activator ciprofibrate (Rao and Subbarao 1997a, 1999). When co-treating rats with the PPARα activator ciprofibrate and the antioxidant vitamin E, the levels of the antioxidant glutathione were paradoxically
454
CHAPTER 17 HUMAN RELEVANCE OF LIVER TUMORS
depleted, and the animals exhibited increased tumor numbers (Glauert et al. 1990). In other studies vitamin E inhibited clofibrate-induced increases in lipofuscinlike products and ciprofibrate-induced increases in NF-kB activation in the absence of effects on markers of PPARα activation (Calfee-Mason et al. 2004; Stanko et al. 1995). Inhibition of key events by compounds that alter inflammatory states including Kupffer cell activation has been observed in multiple studies. The glucocorticoid receptor agonist dexamethasone is an anti-inflammatory agent that decreases the ability of NF-kB to be activated under a variety of inflammatory conditions (Chang et al. 1997; De Bosscher et al. 2006; Ray and Prefontaine 1994; Widen et al. 2003). Dexamethasone decreased PPARα activator-induced hepatocyte proliferation after acute exposures (Lawrence et al. 2001; Ohmura et al. 1996; Rao and Subbarao 1997b) while having either no effect (Lawrence et al. 2001; Rao and Subbarao 1997b) or modest decreases (Ohmura et al. 1996) on markers of PPARα activation. Compounds that inhibit Kupffer cell activation (e.g., glycine, methylpalmitate) or inhibit NADPH oxidase (e.g., diphenyleneiodonium) inhibited increases in oxidative stress and NF-kB activation after exposure to PPARα activators but had no effects on markers of PPARα activation (Rose et al. 1997a,b, 1999a,b; Rusyn et al. 2000b,c, 2001). While pretreatment with diphenyleneiodonium, glycine or methylpalmitate decreased acute cell proliferation (Rose et al. 1997a,b, 1999b; Rusyn et al. 2000b,c), glycine had no effect on chronic cell proliferation but did decrease the size and number of tumors (Rose et al. 1999a). Taken together, these biochemical and genetic inhibition studies demonstrate the linkages of the key events in the PPARα activator MOA. 17.3.2.7. Some PPARα Activators Exhibit Complex MOAs. Before a PPARα activator MOA can be defined as the primary MOA, alternative MOA(s) must be considered. Comparison of wild-type and PPARα-null mice have provided opportunities to determine if additional key events are necessary in addition to PPARα activation. In one example, PFOA was analyzed for liver effects in wild-type and PPARα-null mice. At two doses tested (1 and 3 mg/kg/day), PPARα-null mice lacked increases in cell proliferation but retained increases in liver to body weights. At the highest dose tested (10 mg/kg/day), PPARα-null mice exhibited increases in cell proliferation (Wolf et al. 2008). Microarray analysis using full-genome gene chips showed that PFOA altered ∼85% of the total number of genes in a PPARαdependent manner at 3 mg/kg/day. The PPARα-independent genes exhibited signatures of activation of other nuclear receptors. In particular, the PPARα-independent genes significantly overlapped with those regulated by the constitutive activated receptor (CAR), which regulates cell growth and xenobiotic metabolism genes including Cyp2b family members (Rosen et al. 2008a,b). These CAR signature genes were more robustly regulated in PFOA-treated PPARα-null mice compared to wildtype mice. These findings indicate that CAR activation may be a key event in the transcriptional and cell proliferation effects in PPARα-null mice. In wild-type mice, there were relatively minor alterations of CAR signature genes compared to the strong changes in PPARα-dependent genes indicating that CAR plays a minor role in mediating PFOA effects in wild-type mice (Rosen et al. 2008b).
17.3. MODE OF ACTION ANALYSIS IN THE U.S. EPA RISK ASSESSMENT FRAMEWORK
455
The carcinogenic effects of DEHP were examined in wild-type and PPARαnull mice for 22 months (Ito et al. 2007). A low level of liver tumors was observed in PPARα-null mice but not in wild-type mice. These data suggest that an additional biological event may be operating in DEHP-induced rodent liver tumors.* The tumors in PPARα-null mice most likely arose through a mechanism that is not dominant in wild-type mice. Wild-type and PPARα-null mice did not exhibit equivalent levels of tumor induction. There were no statistically significant increases in liver tumors in the wild-type mice under these exposure conditions, indicating that the biological effects of exposure were not equivalent in these two strains. Expression of growth control genes showed responses in PPARα-null mice but not in wild-type mice at equivalent doses. In follow-up work from the same lab (Takashima et al. 2008), transcript profiling and reverse transcription polymerase chain reaction (RTPCR) showed highly dissimilar changes in gene expression in the liver tumors from the two strains. These data indicate that although DEHP can induce marginal increases in liver tumors in PPARα-null mice, the MOA is different from that in wild-type mice. DEHP is a inducer of Cyp2b family members in wild-type mice (Currie et al. 2005; Ren et al., 2010) suggesting that in the absence of PPARα, DEHP activates CAR, as the rate-limiting key event resulting in increases in liver tumors by a CAR-dependent pathway. In summary, chemicals may produce similar PPARα-independent effects defined in part as effects observed in PPARα-null mice. These effects may suggest additional key events that become the main control points in the absence of PPARα. A determination of the relative contribution of each proposed key event would require comparison of signature genes and biomarkers representing each key event in the two strains. 17.3.2.8. The PPARα Activator MOA Is Chemical-Independent. Mode of action is a series of key events that together result in an adverse health effect such as a liver tumor and as such is chemical-independent (Boobis et al. 2008; Holsapple et al. 2006; Meek 2008). Consistent with this the MOA for PPARα activators is an endogenous series of events that can occur independent of chemical exposure. Livers from ACO-null mice exhibit severe steatosis, increases in markers of PPARα activation (i.e., genes involved in β- and ω-fatty acid oxidation), increases in hydrogen peroxide levels, increases in cell proliferation and liver tumors (Fan et al. 1998). The increases in the markers of PPARα were shown to be PPARα-dependent as the changes were abolished in a double ACO-/PPARα-null mouse (Hashimoto et al. 1999). Microarray analysis of the tumors spontaneously induced in ACO-null mice showed extensive similarity with the liver tumors induced by the PPARα activator ciprofibrate, indicating the mechanism leading to the induction of the tumors was similar (Meyer et al. 2003). Additional mouse models nullizygous for other genes involved in fatty acid oxidation exhibit phenotypes indicative of constitutive PPARα activation (Jia et al. 2003). A mouse model of hepatitis C virus (HCV)-induced *It should be noted that the authors combined different types of liver tumors in their analysis, a nonstandard method of analyzing tumor data leaving open the possibility that the increase in tumor response is actually not statistically significant.
456
CHAPTER 17 HUMAN RELEVANCE OF LIVER TUMORS
hepatocellular carcinoma (HCC) which overexpresses the HCV core protein was used to show that induction of oxidative stress, increases in cell proliferation, and liver tumor induction were PPARα-dependent (Tanaka et al. 2008a,b). The authors conclude that there “is the absolute requirement of persistent PPARα activation for the development of HCV core protein-induced steatosis and HCC” (Tanaka et al. 2008b). All of these mouse models exhibit disruption of fatty acid transport and metabolism resulting in increases in endogenous activators of PPARα including fatty acids (Fan et al. 1998; Tanaka et al. 2008a,b). Taken together, the PPARα MOA is operational in the absence of chemical exposure. Chemical PPARα activators will persistently activate this MOA resulting in liver tumors. 17.3.2.9. Species Differences in Responsiveness of Key Events in the PPARα MOA. Studies conducted in numerous test species indicate that while some rodents (mice and rats) are highly responsive to PPARα activatorinduced hepatocarcinogenicity and associated responses, other species (e.g., Syrian hamsters, dogs, guinea pigs, New and Old World primates, and humans) are less sensitive (Ashby et al. 1994; Bentley et al. 1993; Cattley et al. 1998; Doull et al. 1999). This difference is likely based in large part on differing levels of PPARα expression among species. In a side-by-side comparison, mice had ∼10-fold more PPARα expression than guinea pigs and ∼3-fold more than Syrian hamsters (Choudhury et al. 2004). Humans exhibited ≥10-fold lower expression than mice and rats (described in greater detail below). Thus, guinea pigs may be the more relevant model for PPARα activator effects in the human liver based solely on expression levels of the full-length active PPARα. Table 17.3 summarizes PPARα MOA key events in responsive species (rats and mice summarized from Table 17.1) compared to Syrian hamsters, guinea pigs, Cynomolgus monkeys, and humans. Due to the relative paucity of data for key events, other endpoints associated with exposure to PPARα activators are included (i.e., liver weight to body weight, hypolipidemic effects). Syrian hamsters and guinea pigs exhibit a partial PPARα activator response even though they are considered “nonresponsive species” compared to rats and mice. Fatty acid metabolism genes/proteins are only weakly activated after PPARα activator exposure in the livers of these species. Diminished responsiveness in guinea pigs is not due to a defective PPARα because when overexpressed in cell lines, PPARα from guinea pigs activates reporter genes to levels comparable to rats and mice (Bell et al. 1998; Macdonald et al. 1999; Tugwood et al. 1998). PPARα activators WY14,643 or methylclofenapate decrease triglycerides and very low density lipoproteins (VLDL) in Syrian hamsters and guinea pigs. Five out of the six PPARα activators examined increased liver to body weights in Syrian hamsters, but only one chemical (i.e., perfluorodecanoic acid) out of seven examined increased liver to body weights in guinea pigs; but for perfluorodecanoic acid, there was conflicting evidence of increases. WY-14,643 does not activate NF-kB in hamsters, indicating that this response is species-specific. Differences were also seen between species in relationship to cell proliferation. Studies measuring changes in cell proliferation in Syrian hamsters showed no response, a weak response, or inconsistent responses. Multiple studies showed guinea pigs did not exhibit increases in cell proliferation to four
TABLE 17.3.
Species Differences in Responses to PPARα Activators
Response
PPARα Activation Species
Rats
Relative PPARα expression Likely similar to mice
Mice
10
Syrian hamster
3
Guinea pig
1
Hypolipidemic Effect (Decreases in Triglycerides or VLDL-Triglycerides)
Increases in Liver Weight
Oxidative Stress
NF-kB Activation
Increases in Acute Cell Proliferation
Decreases in Apoptosis
Liver Tumors
Chemical
See table 1 for chemical and reference See table 1 for chemical and reference Nafenopin
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
−1,2,17
+17
−1
WY-14,643 DEHP Methyl clofenapate Ciprofibrate Bezafibrate Methylclofenapate Ciprofibrate WY-14,643 Nafenopin
+1,7,8 (+)4,27 +8 +8,22 +24 −8,21 +8,18 −15,22 +9,11,16 −8 +16,23 −10,12
+1,2,23
+1,2,23 +7,9 +7,9
+9,11 +9,11
+1,7,8,9 +4,27 +7,8,9 +8,22 −24 −8 −8,22 −8 −12,23
−5
−1,8 (+)4 (+)28 +25 −8 −8 −8,25 −8 −8 −12,14,17
−1 −46
−13
+32
457
(Continued)
TABLE 17.3.
Species Differences in Responses to PPARα Activators (Continued)
458
Response
PPARα Activation
Cynomolgus monkey
Humans30
?
≤1
Fenofibrate Perfluorodecanoic acid Bezafibrate DEHP DINP Clofibrate Fenofibrate Ciprofibrate See references for compound used
Hypolipidemic Effect (Decreases in Triglycerides or VLDL-Triglycerides)
Increases in Liver Weight
Oxidative Stress
NF-kB Activation
Increases in Acute Cell Proliferation
−19 −20,26
+26 −20
−24 −3
−24 −3
−3
−3 −3 −6 +6 −31
−3 −3 −6 −6 −41–44,45
− +6 +6 (+)33 +34 −35–39,45 3
−6 +40
−6 −6,33
Decreases in Apoptosis
Liver Tumors
−12,42,43,45
Comments: PPARα activation is a summary of trans-activation data as well as response of markers such as ACO and CYP4A gene, along with protein and enzymatic activity, which are indicators of PPARα activation and are dependent on level of PPARα expression. The endpoint examined in these studies is indicated below. + indicates a strong response, (+) indicates a weak response, and − indicates no response. Spaces left blank indicate no data available. It should be noted that the table does not include PCO data from monkey species other than Cynomolgus monkeys; other monkey data (which is almost universally negative) are summarized in Klaunig et al. (2003). PCO, palmitoyl-CoA oxidase. References: 1Lake et al. (1993) (ACO); 2Price et al. (1992) (ACO); 3Pugh et al. (2000) (peroxisomal fatty acid beta-oxidation); 4Isenberg et al. (2000); 5Tharappel et al. (2001); 6Hoivik et al. (2004) (lipofuscin, peroxisome number, PCO); 7Choudhury et al. (2004) (CYP4A increases); 8Lake et al. (2000) (peroxisome proliferation, CYP4A and carnitine acetyl transferase); 9Choudhury et al. (2000) (trans-activation assay); 10Macdonald et al. (1999) (trans-activation assay); 11Bell et al. (1998) (trans-activation assay); 12Hasmall et al. (1998) (nafenopin); 13Plant et al. (1998) (in vitro apoptosis assay); 14 Elcock et al. (1998) (in vitro); 15Caira et al. (1998) (multifunctional protein, ACO, thiolase); 16Tugwood et al. (1998) (trans-activation assay); 17James and Roberts (1996); 18Pacot et al. (1996) (ACO increases only 1.6-fold); 19Cornu-Chagnon et al. (1995) (ACO in vitro); 20Chinje et al. (1994) (CYP4A); 21Bell et al. (1993) (CYP4A13); 22Makowska et al. (1992) (ACO, CYP4A); 23Lake et al. (1989b) (ACO, CYP4A); 24Watanabe et al. (1989) (slight increases in ACO); 25Styles et al. (1988); 26Van Rafelghem et al. (1987) (peroxisome proliferation); 27Lake et al. (1987) (ACO in vivo and in vitro); 28 Styles et al. (1990); 30Compounds used to treat humans or human primary hepatocytes are indicated in the references; 31Gariot et al. (1987) (fenofibrate); 32James and Roberts (1996); 33Cariello et al. (2005) (fatty acid β-oxidation genes); 34Hanefeld et al. (1983) (clofibrate); 35Hanefeld et al. (1980) (clofibrate); 36De La Iglesia et al. (1982) (gemfibrozil); 37Blumcke et al. (1983) (fenofibrate); 38Gariot et al. (1983) (fenofibrate); 39Bentley et al. (1993) (review); 40Klaunig et al. (2003) (review); 41Perrone et al. (1998) (clofibric acid, diprofibrate); 42Goll et al. (1999) (clofibrate, ciprofibrate, bezafibrate, nafenopin, DEHP); 43Hasmall et al. (1999) (MEHP, MINP, primary metabolite of DINP); 44Hasmall et al. (2000) (MEHP); 45Shaw et al. (2002) (MINP); 46Schmezer et al. (1988).
17.3. MODE OF ACTION ANALYSIS IN THE U.S. EPA RISK ASSESSMENT FRAMEWORK
459
chemicals. Syrian hamsters exhibited suppression of apoptosis after exposure to nafenopin, and guinea pigs exhibited suppression of apoptosis with nafenopin but no change with methylclofenapate. Cancer bioassays performed in Syrian hamsters with nafenopin, WY-14,643 and DEHP were negative (Lake et al. 1993; Schmezer et al. 1988). In summary, although Syrian hamsters and to a lesser extent guinea pigs exhibit changes in endpoints associated with PPARα activation (hypolipidemic effects and changes in fatty acid metabolizing enzymes), they do not exhibit consistent changes in the key events associated with the PPARα activator MOA for liver cancer in rats and mice. In vitro and in vivo data on Cynomolgus monkeys (Table 17.3) and other species of monkeys (i.e., marmoset and Rhesus) indicate that the key events in the PPARα activator MOA are relatively nonresponsive in monkeys. Palmitoyl-CoA oxidase activity was evaluated in monkeys after in vivo exposure to a variety of PPARα activators [e.g., bezafibrate, clofibrate, DEHP, mono-2-ethylhexyl phthalate (MEHP), fenofibrate, nafenopin, and LY171883], and changes were minimal or nonexistent relative to controls (Klaunig et al. 2003). Moreover, Cynomolgus monkeys exposed to DEHP, DINP, or clofibrate failed to exhibit an increase in cell proliferation (Doull et al. 1999; Pugh et al. 2000). Cynomolgus monkeys treated for two weeks with clinically relevant doses of the PPARα activators fenofibrate or ciprofibrate exhibited increases in the number of hepatic peroxisomes (Hoivik et al. 2004). In this study ciprofibrate but not fenofibrate increased liver to body weights in the absence of hepatocyte proliferation. In a follow-up to this study, transcript profiling was used to characterize the genes altered by ciprofibrate exposure (Cariello et al. 2005). Many genes involved in fatty acid metabolism and mitochondrial oxidative phosphorylation were up-regulated, reflecting the known hypolipidemic effects of exposure. However, the magnitude of induction in the β-oxidation pathway was substantially less in monkeys compared to mice and rats. Consistent with the lack of hepatocyte proliferation, there were a number of key regulatory genes that were down-regulated, including members of the JUN, MYC, and NF-kB families. In contrast, JUN and MYC gene expression were up-regulated after PPARα activator treatment in rats (Hsieh et al. 1991). No transcriptional signal for DNA damage or oxidative stress was observed. Lastly, marmosets exposed for 6.5 years to clofibrate at relatively high doses (94 mg/kg/day or higher) did not develop liver tumors over the duration of this study (Tugwood et al. 1996).* Taken together, the key events after PPARα activation in the rodent MOA for liver tumors were not observed in primates treated with PPARα activators. Humans are generally nonresponsive to the effects of PPARα activators. Liver weights were not increased in patients treated with fenofibrate (Gariot et al. 1987). Liver biopsies from humans treated with hypolipidemic drugs or primary human hepatocytes treated with PPARα activators were almost uniformly negative for peroxisome proliferation [reviewed in Bentley et al. (1993)]. In only one out of five studies was there a statistically significant increase in peroxisome number (∼50%), but there was no corresponding increase in volume of peroxisomes (Blumcke et al. 1983; De La Iglesia et al. 1982; Gariot et al. 1983; Hanefeld et al. 1980, 1983). *It should be noted that the duration of this study did not represent a lifetime exposure.
460
CHAPTER 17 HUMAN RELEVANCE OF LIVER TUMORS
Exposure to PPARα activators alters different PPARα gene targets in rodents and humans, including the ACO gene. Unlike the large increases in the expression of marker mRNAs and proteins that are found in rodent primary hepatocytes treated with PPARα activators in vitro, very minor increases, if any, are observed in human primary hepatocytes (Bichet et al. 1990; Cornu-Chagnon et al. 1995; Duclos et al. 1997; Elcombe 1985; Elcombe et al. 1996; Goll et al. 1999; Hasmall et al. 2000; Hasmall et al. 1999; Perrone et al. 1998). ACO mRNA in liver samples from 48 patients treated with one of several fibrates (bezafibrate, fenofibrate or gemfibrozil) was not induced despite significant induction of hepatic apolipoprotein A-I mRNA and lowering of serum lipids following treatment (Roglans et al. 2002). The relatively weak increases in ACO observed in human primary hepatocytes are in stark contrast to the robust inductions observed in the livers of mice and rats exposed to PPARα activators [summarized in Klaunig et al. (2003)]. In summary, there is no evidence that the ACO gene exhibits more than minor inductions in humans. Species differences in sensitivity to PPARα activators may be explained in part by differences in the structure of the promoter regions that regulate the expression of target genes. The lack of ACO induction in human livers and primary human hepatocytes may be attributable to an inactive PPRE. Evidence that a functional PPRE exists in the human ACO gene promoter (Varanasi et al. 1996), was challenged by subsequent studies which showed that the PPRE is inactive in in vitro transactivation assays and that the sequence differs from that originally reported at three positions (Woodyatt et al. 1999). Little heterogeneity exists within the human ACO PPRE because the same altered PPRE sequence was found in all 22 unrelated humans that were investigated as well as in the human hepatocellular carcinoma cell line HepG2 (Woodyatt et al. 1999). A nonfunctional PPRE in the ACO promoter would be consistent with studies showing little, if any induction of the ACO gene/protein expression upon exposure to PPARα activators in human primary hepatocytes. PPARα ligands do not induce cell proliferation or suppress apoptosis in human hepatocytes in vitro (Goll et al. 1999; Hasmall et al. 1999; Perrone et al. 1998; Williams and Perrone 1996). Many of these studies included a positive control to ensure that human hepatocytes were of sufficient quality to mount a positive growth response. In comparison, rat or mouse primary hepatocytes exposed to PPARα activators exhibit up to 8-fold induction in cell proliferation (summarized in (Klaunig et al. 2003)). There are no data on human hepatocyte proliferation in vivo, although in vivo and in vitro data from nonhuman primates show cell proliferation is not induced by PPARα activators [Table 17.3 and reviewed in Doull et al. (1999)]. In summary, available data suggest that PPARα activators are unlikely to alter apoptosis and proliferation in human hepatocytes. 17.3.2.10. Molecular Basis of Species Differences. In the following section, the properties of PPARα and associated responses in the livers of rodents and primates are compared with an emphasis on human data, if available. The weight of evidence demonstrates that humans respond to PPARα activators differently than rodents in that many of the typical markers of PPARα activator exposure associated with hepatocarcinogenesis in rodents are absent in humans. Differences in the
17.3. MODE OF ACTION ANALYSIS IN THE U.S. EPA RISK ASSESSMENT FRAMEWORK
TABLE 17.4.
461
Properties of Rodent (Rat and Mouse) PPARα Versus Human PPARα in Liver
Property Allelic variants
Rodent (Rat/Mouse) None identified
Human L162V
V227A “6/29”
Truncated PPARα (deleted exon 6) Inducibility by environmentally relevant ligands
Below 10% of total PPARα Chemical-specific range of responsiveness
Basal expression of PPARα Regulation of hypolipidemic response
High in liver Intact
10–50% of total PPARα Some differences with rodent activation noted leading to decreased activation <∼10% of mice based on one study Intact
Regulation of liver growth
Intact
No evidence
Impact on Responsiveness to PPARα Activators in Humans Compared to Mice and Rats Exhibits greater ligandinduced activity at higher doses compared to the wild-type receptor; found at high frequencies in some populations Decreased responsiveness; rare variant Decreased responsiveness; acts as a dominant negative; very rare variant Decreased responsiveness Equal or decreased responsiveness
Much lower responsiveness No difference in endpoint but different genes may be regulated in the different species No response in humans because of fundamental differences in spectrum of genes regulated; hPPARα does not regulate cell proliferation in mice
properties of PPARα, including structure, function, and expression, determine the underlying basis for human–rodent differences in the biological effects of PPARα activators. The properties of mouse and rat PPARα versus human PPARα in liver are summarized in Table 17.4. Allelic Variants of Human PPARα. The human PPARα (hPPARα) is indistinguishable from the rodent PPARα in overall structure (Desvergne et al. 1998; Mukherjee et al. 1994; Sher et al. 1993; Tugwood et al. 1996), but a number of
462
CHAPTER 17 HUMAN RELEVANCE OF LIVER TUMORS
allelic variants of hPPARα have been isolated which possess properties different from the original cloned hPPARα. The L162V variant containing an amino acid change in the DNA-binding domain is found at an allelic frequency of ∼0.025–0.073 in ethnically diverse populations (Flavell et al. 2000; Lacquemant et al. 2000; Tai et al. 2002). In North Indians, this allele is found at high frequencies (0.745) (Sapone et al. 2000). The hPPARα L162V variant exhibits no response to low doses of WY but greater ligand-induced activity (up to ∼4-fold) at higher doses compared to the wild-type receptor (Flavell et al. 2000; Sapone et al. 2000). Humans carrying this variant exhibit greater decreases in total serum cholesterol to the hypolipidemic, bezafibrate (Flavell et al. 2000). Three different Asian populations carry a hPPARα variant (V227A) within the hinge region between the DNA binding and ligand binding domains at frequencies of 0.003–0.051 (Chan et al. 2006; YamakawaKobayashi et al. 2002). This allele has been associated with decreases in serum cholesterol and triglycerides in a Japanese population (Yamakawa-Kobayashi et al. 2002) and in Chinese women (Chan et al. 2006). Because of increased interactions with a co-repressor, nuclear receptor corepressor (NCoR), this variant exhibits decreased responsiveness to PPARα activators (Liu et al. 2008). The hPPARα-6/29 variant containing four amino acid substitutions is a dominant negative that binds to a PPRE but cannot be activated by PPARα activators (James et al. 1998a). The hPPARα-6/29 variant is likely very rare, because it was not detected in any of the 173 human subjects from two studies (Roberts 1999; Sapone et al. 2000). Overall, some PPARα allelic heterogeneity exists in human populations, but no variants have been identified that are more sensitive to low, environmentally relevant doses of PPARα activators than the “wild-type” human receptor. The field would benefit from a side-by-side comparison of wild-type and hPPARα variants in trans-activation assays to determine dose-response relationships of PPARα activators. Differences in Ligand Inducibility. Human PPARα is not more sensitive than rodent PPARα to chemical activation. Most compounds activate the rodent receptor better or exhibit no differences between species. A number of environmentally relevant chemicals and hypolipidemic agents were able to activate rat or mouse PPARα at lower concentrations or to higher absolute levels than hPPARα in sideby-side trans-activation studies. These PPARα activators include WY-14,643 (Keller et al. 1997; Maloney and Waxman 1999; Takacs and Abbott 2007), PFOA (Maloney and Waxman 1999), perfluorooctanesulfonate (Shipley et al. 2004; Takacs and Abbott 2007), and a number of phthalate ester metabolites [Bility et al. (2004) and summarized in Corton and Lapinskas (2005)]. Some PPARα activators show no differences between activation of the mouse and human PPARα, including TCA, dichloroacetate, 2-ethylhexanoic acid (Maloney and Waxman 1999), a number of phthalates (Bility et al. 2004), clofibrate (Keller et al. 1993), and PFOA (Vanden Heuvel et al. 2006). Only perfluorooctanesulfonamide (Shipley et al. 2004) was shown to modestly activate the human but not the rodent PPARα at one lower dose (25 μM in human versus 34 μM mouse). Overall, the data indicate that hPPARα is no more sensitive than the mouse or rat PPARα to significant activation by environmentally relevant PPARα activators.
17.3. MODE OF ACTION ANALYSIS IN THE U.S. EPA RISK ASSESSMENT FRAMEWORK
463
Expression of the PPARα Gene and Protein. PPARα expression is the factor most often cited for determining species-specific differences in PPARα activator responsiveness. Palmer et al. (1998) used EMSA to determine the level of PPARα that binds to a PPRE from the CYP4A6 gene (Palmer et al. 1998). In seven lysates from individual human livers in which PPARα could be detected by the assay, the amounts were ∼10-fold lower than those detected in the livers of CD-1 or BALB/cByJ mice and for the remainder of the 13 individual human livers, the amounts were below detection (>20-fold less than mouse liver). A 3-fold variation in the expression of the full-length PPARα mRNA between human samples was noted. The data indicate that hPPARα in liver is expressed at levels far below that expressed in rodent liver. Additional studies evaluating expression and function of PPARα in human liver are needed to more definitively determine the relative expression of PPARα in rodents and humans. Such studies would benefit from better assessment of the degree of protein and mRNA degradation in the samples. Truncated PPARα. A truncated PPARα variant has been identified in a number of labs and is called hPPARα-8/14 (Tugwood et al. 1996), hPPARSV (Palmer et al. 1998), PPARαtr (Gervois et al. 1999), and PPARα2 (Hanselman et al. 2001). This truncated form lacks exon 6 due to alternative splicing, resulting in a hPPARα lacking the hinge region and ligand binding domain. This form acts as a dominant negative, inhibiting the ability of the wild-type receptor to activate transcription, possibly by titrating out limiting amounts of co-activators (Gervois et al. 1999). The level of the mRNA of this form ranges from 10–50% of full-length hPPARα mRNA (Gervois et al. 1999; Hanselman et al. 2001; Palmer et al. 1998; Roberts et al. 2000) similar to Cynomolgus monkeys (Hanselman et al. 2001). In comparison, this level is below 10% in mice and rats (Hanselman et al. 2001). A more definitive role for this truncated form awaits studies in which the levels of full-length and truncated hPPARα forms are simultaneously measured with well-characterized hPPARα target genes in primary human hepatocytes exposed to PPARα activators. Differences in Transcriptional Networks Controlled by Human and Rodent PPARα. There is overwhelming evidence that the transcriptional networks controlled by PPARα are different between humans and rodents and underlie species-specific differences in key events in the PPARα MOA. Humans and rodents do share hypolipidemic effects of PPARα activators but may achieve this beneficial effect through regulation of different gene sets. A number of genes are likely responsible for the therapeutic hypolipidemic effects of PPARα activators in humans. Many of these genes have functional PPREs that are transcriptionally regulated by human PPARα, including apolipoprotein (apo) C-III (Hertz et al. 1995), lipoprotein lipase (Schoonjans et al. 1996), apo A-I (Vu-Dac et al. 1994), apo A-II (Vu-Dac et al. 1995), and carnitine palmitoyl transferase-I (Mascaro et al. 1998). Human PPARα activation of apolipoprotein A-II and lipoprotein lipase transcription and suppression of apolipoprotein C-III expression are key to lowering serum triglycerides (Auwerx et al. 1996; Staels et al. 1997; Vu-Dac et al. 1995). Human apolipoprotein C-III can be down-regulated by fibrates in cultured human hepatocytes in the absence of changes in PPARα target genes encoding peroxisomal enzymes
464
CHAPTER 17 HUMAN RELEVANCE OF LIVER TUMORS
including ACO, bifunctional enzyme, and thiolase (Lawrence et al. 2001b). Furthermore, stably transfected HepG2 cells expressing either human or murine PPARα at levels similar to rodent liver respond to fibrates by increased expression of 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA) synthase and carnitine palmitoyl transferase-I (CPT-I) but lack the typical robust induction of typical PPARα targets—that is, ACO, bifunctional enzyme, or thiolase (Hsu et al. 2001; Lawrence et al. 2001a; Tachibana et al. 2005). In a global analysis of gene expression, genes of the cytosolic, microsomal, and mitochondrial pathways involved in fatty acid transport and metabolism were up-regulated by clofibrate in both rodent and human hepatocyte cultures, whereas genes of the peroxisomal pathway of lipid metabolism were up-regulated only in rodents (Richert et al. 2003). Thus, PPARα activation may lower lipid levels in humans and rodents through regulation of different sets of genes. The human PPARα does not possess all of the functions of the rodent PPARα including the ability to regulate cell proliferation. Two mouse strains have been created which express the hPPARα in the absence of mPPARα (“humanized” hPPARα mice). In the TRE-hPPARα mouse, PPARα is under the control of a liverspecific promoter and is preferentially expressed in hepatocytes (Cheung et al. 2004); the cellular location of hPPARα expression in the humanized PPARα mouse corresponds to the location of mPPARα expression in wild-type mice—that is, in hepatocytes but not Kupffer cells (Peters et al. 2000). The hPPARαPAC mouse contains a 211-kilobase region encoding the regulatory and structural regions of the human PPARα gene. The hPPARα is expressed in the same tissues as those of the mouse PPARα (Yang et al. 2008). The humanized PPARα mouse strains do not respond to a PPARα activator (WY-14,643) in the same manner as wild-type mice even though both strains express hPPARα to levels comparable to mPPARα in wild-type mice. The humanized mice exhibit increases in peroxisome proliferation, decreases in serum total triglycerides and normal activation of lipid metabolism genes including those involved in peroxisome proliferation. However, these mice do not exhibit increased expression of cell cycle genes or increased hepatocyte proliferation in response to a PPARα activator as do wild-type mice (Cheung et al. 2004; Morimura et al. 2006; Yang et al. 2008). In a 38- to 44-week exposure study with the PPARα activator WY-14,643, the TRE-hPPARα mice were also resistant to PPARα activator-induced liver cancer. Wild-type mice but not humanized mice exhibited a significant increase in liver tumors despite the fact that the humanized mice were exposed 6 weeks longer than the wild-type mice to the compound (Morimura et al. 2006). These studies show that hPPARα is pharmacologically active but does not regulate the full spectrum of responses necessary for hepatocarcinogenesis in rodents. The molecular basis for differences between mouse and human PPARα may be differences in the ability of the receptors to interact with transcriptional coactivators or to regulate miRNA cascades. Co-activators convey the transcriptional activation of the ligand-induced nuclear receptor to the transcriptional machinery. Elegant biochemical and crystallographic analyses have shown key interactions between co-activators and the ligand binding domains of nuclear receptors including PPAR family members (Li et al. 2008; Xu and Li 2008). The mouse and rat PPARα
17.3. MODE OF ACTION ANALYSIS IN THE U.S. EPA RISK ASSESSMENT FRAMEWORK
465
ligand binding domains (LBD) do possess amino acid differences with human PPARα LBD (Mukherjee et al. 1994; Sher et al. 1993; Tugwood et al. 1996). Amino acid differences in the LBD between mice and humans may uncouple receptor coactivator interactions in humans required for cell proliferation and gene regulation, while retaining those important in lipid metabolism gene regulation. Alternatively, differences in miRNA regulation may contribute to species differences, as the ability to regulate the let-7c cascade is lost in humanized mice in response to a PPARα activator (Yang et al. 2008). Further studies are needed to define the specific mechanistic basis for species differences. 17.3.2.11. Summary of Key Data that Support the MOA. The PPARα MOA describes the sequence of events beginning with PPARα activation and leading to an increased incidence of liver tumors in rats and mice. This MOA exists independent of exposure to any particular chemical but has been shown to be triggered by chemicals collectively referred to as PPARα activators. The overall WOE supports a MOA that involves five key events. First, PPARα activators activate PPARα. Second, PPARα activation leads to alterations in the expression of genes that regulate oxidative stress and increases in oxidative stress. Third, oxidative stress activates the transcription factor NF-kB. Fourth, NF-kB activation leads to increased cell proliferation and decreased apoptosis in the liver. Fifth, sustained growth signaling upon chronic exposure causes clonal expansion of initiated cells leading to preneoplastic foci and tumors—that is, hepatocellular adenomas and carcinomas. Table 17.5 summarizes the specificity and WOE of the PPARα activator MOA. The WOE strongly supports the MOA due to the large number of studies that have been carried out since the discovery of peroxisome proliferation by these chemicals in 1965 (Hess et al. 1965). PPARα activation is by definition specific, because this key event is distinct from other initiating events such as CAR activation or increases in cytotoxicity. The other key events by themselves are considered to
TABLE 17.5.
PPARα Activators: Mode of Action (MOA) Key Events
Causal Key Eventa 1. 2. 3. 4. 5.
Activation of PPARα Increases in oxidative stress NF-kB activation Perturbation of cell growth and survival Clonal expansion of preneoplastic foci
Specificityb
Evidencec
High Low Low Low Low
Strong Strong Strong Strong Strong
Causal key event is a required step for PPARα MOA, based on empirical evidence.
a
Specificity of each key event to PPARα-induced rodent hepatic tumors is considered high if it is unique to this MOA and low if not. The key events other than PPARα activation by themselves are considered to have low specificity, because these events are observed with other carcinogens. However, the key events when linked are considered to have high specificity because they are dependent on PPARα. b
c
Evidence was determined to be strong if several studies support that key event as part of the MOA, preferably with multiple PPARα activators from multiple laboratories, with limited evidence of contradiction. Evidence is considered weak if only a single study with a single PPARα activator from a single laboratory supports that key event or if a significant amount of contradiction appears in the literature.
466
CHAPTER 17 HUMAN RELEVANCE OF LIVER TUMORS
TABLE 17.6.
Comparative Analysis of Rodent and Human Data—Liver Tumors
Causal Key Events
Plausible in Humans?
Taking into Account Kinetic and Dynamic Factors, Is the Key Event Plausible in Humans?
1. Activation of PPARα 2. Increases in oxidative stress
Yes
Yes
Yes
Unknown
3. NF-kB activation
Yes
Unknown
4. Perturbation of cell growth and survival
Yes
Not likely
5. Selective clonal expansion of preneoplastic foci 6. Liver tumors
Yes
Not likely
Yes
Not likely
Comments PPARα is a target of human hypolipidemic drugs. Gene products that produce oxidative stress in rodents exist in humans but are not induced to the same extent in humans or monkeys. More traditional methods of measuring oxidative stress have not been used. NF-kB exists in humans but has not been measured in human liver or primary hepatocytes after exposure to PPARα activators. Not seen in independent studies of human hepatocytes in vitro; not measured in vivo; not seen in nonhuman primates in vivo or in vitro; not seen in hamsters or guinea pigs. No response in nonhuman primates. Not measured in livers of humans exposed to PPARα activators; no tumors in hamsters with expression of PPARα intermediate between mice/rats and humans.
have low specificity, because these events are observed with other carcinogens. However, the key events when linked are considered to have high specificity because they are dependent on PPARα. Table 17.1 provides examples of chemical-specific data evaluating whether the key events occur after exposure to five different PPARα activators.
REFERENCES
467
Evidence showing the mechanistic linkages between the key events of the MOA is summarized in Table 17.2. Studies that inhibit key events by genetic or biochemical means reveal such relationships because inhibition of one event blocks downstream events. Additional support for the PPARα MOA comes from a comparison of responses in rats and mice to “nonresponsive” species such as Syrian hamsters, guinea pigs, and monkeys. These data are summarized in Table 17.3. Overall, these data show that while all species exhibit a hypolipidemic response and alterations in lipid metabolism and transport genes, Syrian hamsters, guinea pigs and monkeys exhibit little, if any, changes in oxidative stress markers, NF-kB activation, and alterations of hepatocyte growth or tumor response (Klaunig et al. 2003).
17.4. RELEVANCE OF PPARα ACTIVATOR-INDUCED RODENT LIVER TUMOR RESPONSE TO HUMANS Although humans have been regularly exposed to PPARα activators through administration of hypolipidemic pharmaceuticals, epidemiological studies have not provided evidence of increased incidence of liver neoplasms in humans exposed to PPARα activators for up to 13 years [summarized in Klaunig et al. (2003)]. Species comparisons of key events and other endpoints relevant to the PPARα MOA show that mice and rats are much more responsive than humans (Table 17.6) and other species (e.g., hamsters, guinea pigs, and primates) (Table 17.3). Experimental evidence suggests that the differences in responsiveness among species may be due to differences in promoter structure and/or function of PPARα target genes, sensitivity of PPARα to activation, the expression level of full-length and dominant negative forms of PPARα, and species differences in the ability of PPARα to alter expression of genes involved in cell fate (Table 17.4). Overall, the weight of evidence suggests that although the rodent MOA is plausible in humans, humans would not be expected to respond with a hepatocarcinogenic effect from chronic exposure consistent with the original conclusion by an ILSI workgroup (Klaunig et al. 2003).
REFERENCES Abdellatif, A. G., Preat, V., Vamecq, J., Nilsson, R., and Roberfroid, M. (1990). Peroxisome proliferation and modulation of rat liver carcinogenesis by 2,4-dichlorophenoxyacetic acid, 2,4,5-trichlorophenoxyacetic acid, perfluorooctanoic acid and nafenopin. Carcinogenesis 11, 1899–1902. Anderson, S. P., Dunn, C., Laughter, A., Yoon, L., Swanson, C., Stulnig, T. M., Steffensen, K. R., Chandraratna, R. A., Gustafsson, J. A., and Corton, J. C. (2004a). Overlapping transcriptional programs regulated by the nuclear receptors peroxisome proliferator-activated receptor alpha, retinoid X receptor, and liver X receptor in mouse liver. Mol Pharmacol 66, 1440–1452. Anderson, S. P., Dunn, C. S., Cattley, R. C., and Corton, J. C. (2001). Hepatocellular proliferation in response to a peroxisome proliferator does not require TNFalpha signaling. Carcinogenesis 22, 1843–1851.
468
CHAPTER 17 MODE OF ACTION ANALYSIS AND HUMAN RELEVANCE
Anderson, S. P., Howroyd, P., Liu, J., Qian, X., Bahnemann, R., Swanson, C., Kwak, M. K., Kensler, T. W., and Corton, J. C. (2004b). The transcriptional response to a peroxisome proliferator-activated receptor alpha agonist includes increased expression of proteome maintenance genes. J Biol Chem 279, 52390–52398. Arsura, M., and Cavin, L. G. (2005). Nuclear factor-kappaB and liver carcinogenesis. Cancer Lett 229, 157–169. Ashby, J., Brady, A., Elcombe, C. R., Elliott, B. M., Ishmael, J., Odum, J., Tugwood, J. D., Kettle, S., and Purchase, I. F. (1994). Mechanistically-based human hazard assessment of peroxisome proliferator-induced hepatocarcinogenesis. Hum Exp Toxicol 13 (Suppl 2), S1–S117. Auwerx, J., Schoonjans, K., Fruchart, J. C., and Staels, B. (1996). Transcriptional control of triglyceride metabolism: fibrates and fatty acids change the expression of the LPL and apo C-III genes by activating the nuclear receptor PPAR. Atherosclerosis 124 (Suppl), S29–S37. Barrass, N. C., Price, R. J., Lake, B. G., and Orton, T. C. (1993). Comparison of the acute and chronic mitogenic effects of the peroxisome proliferators methylclofenapate and clofibric acid in rat liver. Carcinogenesis 14, 1451–1456. Bayly, A. C., Roberts, R. A., and Dive, C. (1994). Suppression of liver cell apoptosis in vitro by the non-genotoxic hepatocarcinogen and peroxisome proliferator nafenopin. J Cell Biol 125, 197–203. Becuwe, P., and Dauca, M. (2005). Comparison of cytotoxicity induced by hypolipidemic drugs via reactive oxygen species in human and rodent liver cells. Int J Mol Med 16, 483–492. Bell, A. R., Savory, R., Horley, N. J., Choudhury, A. I., Dickins, M., Gray, T. J., Salter, A. M., and Bell, D. R. (1998). Molecular basis of non-responsiveness to peroxisome proliferators: the guinea-pig PPARalpha is functional and mediates peroxisome proliferator-induced hypolipidaemia. Biochem J 332 (Pt 3), 689–693. Bell, D. R., Plant, N. J., Rider, C. G., Na, L., Brown, S., Ateitalla, I., Acharya, S. K., Davies, M. H., Elias, E., Jenkins, N. A., and et al. (1993). Species-specific induction of cytochrome P-450 4A RNAs: PCR cloning of partial guinea-pig, human and mouse CYP4A cDNAs. Biochem J 294 (Pt 1), 173–180. Bentley, P., Calder, I., Elcombe, C., Grasso, P., Stringer, D., and Wiegand, H. J. (1993). Hepatic peroxisome proliferation in rodents and its significance for humans. Food Chem Toxicol 31, 857–907. Bichet, N., Cahard, D., Fabre, G., Remandet, B., Gouy, D., and Cano, J. P. (1990). Toxicological studies on a benzofuran derivative. III. Comparison of peroxisome proliferation in rat and human hepatocytes in primary culture. Toxicol Appl Pharmacol 106, 509–517. Bility, M. T., Thompson, J. T., McKee, R. H., David, R. M., Butala, J. H., Vanden Heuvel, J. P., and Peters, J. M. (2004). Activation of mouse and human peroxisome proliferator-activated receptors (PPARs) by phthalate monoesters. Toxicol Sci 82, 170–182. Blumcke, S., Schwartzkopff, W., Lobeck, H., Edmondson, N. A., Prentice, D. E., and Blane, G. F. (1983). Influence of fenofibrate on cellular and subcellular liver structure in hyperlipidemic patients. Atherosclerosis 46, 105–116. Bojes, H. K., Germolec, D. R., Simeonova, P., Bruccoleri, A., Schoonhoven, R., Luster, M. I., and Thurman, R. G. (1997). Antibodies to tumor necrosis factor alpha prevent increases in cell replication in liver due to the potent peroxisome proliferator, WY-14,643. Carcinogenesis 18, 669–674. Boobis, A. R., Cohen, S. M., Dellarco, V., McGregor, D., Meek, M. E., Vickers, C., Willcocks, D., and Farland, W. (2006). IPCS framework for analyzing the relevance of a cancer mode of action for humans. Crit Rev Toxicol 36, 781–792. Boobis, A. R., Doe, J. E., Heinrich-Hirsch, B., Meek, M. E., Munn, S., Ruchirawat, M., Schlatter, J., Seed, J., and Vickers, C. (2008). IPCS framework for analyzing the relevance of a noncancer mode of action for humans. Crit Rev Toxicol 38, 87–96. Burkhardt, S., Mellert, W., Reinachet, M., and Bahnemann, R. (2001). Zonal evaluation of proliferation and apoptosis in the liver of mice reveals new mechanistic data for phenobarbital (PB), Wyeth 14,643 (WY), and chloroform (CH). The Toxicologist 60, 286 (Abstract #1362). Burns, K. A., and Vanden Heuvel, J. P. (2007). Modulation of PPAR activity via phosphorylation. Biochim Biophys Acta 1771, 952–960. Bursch, W., Lauer, B., Timmermann-Trosiener, I., Barthel, G., Schuppler, J., and Schulte-Hermann, R. (1984). Controlled death (apoptosis) of normal and putative preneoplastic cells in rat liver following withdrawal of tumor promoters. Carcinogenesis 5, 453–458.
REFERENCES
469
Busser, M. T., and Lutz, W. K. (1987). Stimulation of DNA synthesis in rat and mouse liver by various tumor promoters. Carcinogenesis 8, 1433–1437. Cai, Y., Appelkvist, E. L., and DePierre, J. W. (1995). Hepatic oxidative stress and related defenses during treatment of mice with acetylsalicylic acid and other peroxisome proliferators. J Biochem Toxicol 10, 87–94. Caira, F., Clemencet, M. C., Cherkaoui-Malki, M., Dieuaide-Noubhani, M., Pacot, C., Van Veldhoven, P. P., and Latruffe, N. (1998). Differential regulation by a peroxisome proliferator of the different multifunctional proteins in guinea pig: cDNA cloning of the guinea pig D-specific multifunctional protein 2. Biochem J 330 (Pt 3), 1361–1368. Calfee-Mason, K. G., Spear, B. T., and Glauert, H. P. (2004). Effects of vitamin E on the NF-kappaB pathway in rats treated with the peroxisome proliferator, ciprofibrate. Toxicol Appl Pharmacol 199, 1–9. Cariello, N. F., Romach, E. H., Colton, H. M., Ni, H., Yoon, L., Falls, J. G., Casey, W., Creech, D., Anderson, S. P., Benavides, G. R., Hoivik, D. J., Brown, R., and Miller, R. T. (2005). Gene expression profiling of the PPAR-alpha agonist ciprofibrate in the cynomolgus monkey liver. Toxicol Sci 88, 250–264. Cattley, R. C., Conway, J. G., and Popp, J. A. (1987). Association of persistent peroxisome proliferation and oxidative injury with hepatocarcinogenicity in female F-344 rats fed di(2-ethylhexyl)phthalate for 2 years. Cancer Lett 38, 15–22. Cattley, R. C., DeLuca, J., Elcombe, C., Fenner-Crisp, P., Lake, B. G., Marsman, D. S., Pastoor, T. A., Popp, J. A., Robinson, D. E., Schwetz, B., Tugwood, J., and Wahli, W. (1998). Do peroxisome proliferating compounds pose a hepatocarcinogenic hazard to humans? Regul Toxicol Pharmacol 27, 47–60. Cattley, R. C., and Glover, S. E. (1993). Elevated 8-hydroxydeoxyguanosine in hepatic DNA of rats following exposure to peroxisome proliferators: relationship to carcinogenesis and nuclear localization. Carcinogenesis 14, 2495–2499. Chan, E., Tan, C. S., Deurenberg-Yap, M., Chia, K. S., Chew, S. K., and Tai, E. S. (2006). The V227A polymorphism at the PPARA locus is associated with serum lipid concentrations and modulates the association between dietary polyunsaturated fatty acid intake and serum high density lipoprotein concentrations in Chinese women. Atherosclerosis 187, 309–315. Chang, C. K., Llanes, S., and Schumer, W. (1997). Effect of dexamethasone on NF-kB activation, tumor necrosis factor formation, and glucose dyshomeostasis in septic rats. J Surg Res 72, 141–145. Chen, H., Huang, C. Y., Wilson, M. W., Lay, L. T., Robertson, L. W., Chow, C. K., and Glauert, H. P. (1994). Effect of the peroxisome proliferators ciprofibrate and perfluorodecanoic acid on hepatic cell proliferation and toxicity in Sprague-Dawley rats. Carcinogenesis 15, 2847–2850. Cheung, C., Akiyama, T. E., Ward, J. M., Nicol, C. J., Feigenbaum, L., Vinson, C., and Gonzalez, F. J. (2004). Diminished hepatocellular proliferation in mice humanized for the nuclear receptor peroxisome proliferator-activated receptor alpha. Cancer Res 64, 3849–3854. Chinje, E., Kentish, P., Jarnot, B., George, M., and Gibson, G. (1994). Induction of the CYP4A subfamily by perfluorodecanoic acid: the rat and the guinea pig as susceptible and non-susceptible species. Toxicol Lett 71, 69–75. Choudhury, A. I., Chahal, S., Bell, A. R., Tomlinson, S. R., Roberts, R. A., Salter, A. M., and Bell, D. R. (2000). Species differences in peroxisome proliferation; mechanisms and relevance. Mutat Res 448, 201–212. Choudhury, A. I., Sims, H. M., Horley, N. J., Roberts, R. A., Tomlinson, S. R., Salter, A. M., Bruce, M., Shaw, P. N., Kendall, D., Barrett, D. A., and Bell, D. R. (2004). Molecular analysis of peroxisome proliferation in the hamster. Toxicol Appl Pharmacol 197, 9–18. Conway, J. G., Tomaszewski, K. E., Olson, M. J., Cattley, R. C., Marsman, D. S., and Popp, J. A. (1989). Relationship of oxidative damage to the hepatocarcinogenicity of the peroxisome proliferators di(2-ethylhexyl)phthalate and Wy-14,643. Carcinogenesis 10, 513–519. Cornu-Chagnon, M. C., Dupont, H., and Edgar, A. (1995). Fenofibrate: metabolism and species differences for peroxisome proliferation in cultured hepatocytes. Fundam Appl Toxicol 26, 63–74. Corton, J. C. (2008). Evaluation of the role of peroxisome proliferator-activated receptor alpha (PPARalpha) in mouse liver tumor induction by trichloroethylene and metabolites. Crit Rev Toxicol 38, 857–875.
470
CHAPTER 17 MODE OF ACTION ANALYSIS AND HUMAN RELEVANCE
Corton, J. C., Anderson, S. P., and Stauber, A. (2000). Central role of peroxisome proliferator-activated receptors in the actions of peroxisome proliferators. Annu Rev Pharmacol Toxicol 40, 491–518. Corton, J. C., Apte, U., Anderson, S. P., Limaye, P., Yoon, L., Latendresse, J., Dunn, C., Everitt, J. I., Voss, K. A., Swanson, C., Kimbrough, C., Wong, J. S., Gill, S. S., Chandraratna, R. A., Kwak, M. K., Kensler, T. W., Stulnig, T. M., Steffensen, K. R., Gustafsson, J. A., and Mehendale, H. M. (2004). Mimetics of caloric restriction include agonists of lipid-activated nuclear receptors. J Biol Chem 279, 46204–46212. Corton, J. C., and Lapinskas, P. J. (2005). Peroxisome proliferator-activated receptors: mediators of phthalate ester-induced effects in the male reproductive tract? Toxicol Sci 83, 4–17. Currie, R. A., Bombail, V., Oliver, J. D., Moore, D. J., Lim, F. L., Gwilliam, V., Kimber, I., Chipman, K., Moggs, J. G., and Orphanides, G. (2005). Gene ontology mapping as an unbiased method for identifying molecular pathways and processes affected by toxicant exposure: Application to acute effects caused by the rodent non-genotoxic carcinogen diethylhexylphthalate. Toxicol Sci 86, 453–469. Czaja, M. J. (2007). Cell signaling in oxidative stress-induced liver injury. Semin Liver Dis 27, 378–389. De Bosscher, K., Vanden Berghe, W., and Haegeman, G. (2006). Cross-talk between nuclear receptors and nuclear factor kappaB. Oncogene 25, 6868–6886. de Duve, C. (1996). The peroxisome in retrospect. Ann N Y Acad Sci 804, 1–10. De La Iglesia, F. A., Lewis, J. E., Buchanan, R. A., Marcus, E. L., and McMahon, G. (1982). Light and electron microscopy of liver in hyperlipoproteinemic patients under long-term gemfibrozil treatment. Atherosclerosis 43, 19–37. De Minicis, S., Bataller, R., and Brenner, D. A. (2006). NADPH oxidase in the liver: defensive, offensive, or fibrogenic? Gastroenterology 131, 272–275. Desvergne, B., A, I. J., Devchand, P. R., and Wahli, W. (1998). The peroxisome proliferatoractivated receptors at the cross-road of diet and hormonal signalling. J Steroid Biochem Mol Biol 65, 65–74. Desvergne, B., and Wahli, W. (1999). Peroxisome proliferator-activated receptors: nuclear control of metabolism. Endocr Rev 20, 649–688. Dostalek, M., Hardy, K. D., Milne, G. L., Morrow, J. D., Chen, C., Gonzalez, F. J., Gu, J., Ding, X., Johnson, D. A., Johnson, J. A., Martin, M. V., and Guengerich, F. P. (2008). Development of oxidative stress by cytochrome P450 induction in rodents is selective for barbiturates and related to loss of pyridine nucleotide-dependent protective systems. J Biol Chem 283, 17147–17157. Doull, J., Cattley, R., Elcombe, C., Lake, B. G., Swenberg, J., Wilkinson, C., Williams, G., and van Gemert, M. (1999). A cancer risk assessment of di(2-ethylhexyl)phthalate: Application of the new U.S. EPA Risk Assessment Guidelines. Regul Toxicol Pharmacol 29, 327–357. Dreyer, C., Krey, G., Keller, H., Givel, F., Helftenbein, G., and Wahli, W. (1992). Control of the peroxisomal beta-oxidation pathway by a novel family of nuclear hormone receptors. Cell 68, 879–887. Duclos, S., Bride, J., Ramirez, L. C., and Bournot, P. (1997). Peroxisome proliferation and beta-oxidation in Fao and MH1C1 rat hepatoma cells, HepG2 human hepatoblastoma cells and cultured human hepatocytes: effect of ciprofibrate. Eur J Cell Biol 72, 314–323. Dwivedi, R. S., Alvares, K., Nemali, M. R., Subbarao, V., Reddy, M. K., Usman, M. I., Rademaker, A. W., Reddy, J. K., and Rao, M. S. (1989). Comparison of the peroxisome proliferator-induced pleiotropic response in the liver of nine strains of mice. Toxicol Pathol 17, 16–26. Elcock, F. J., Chipman, J. K., and Roberts, R. A. (1998). The rodent nongenotoxic hepatocarcinogen and peroxisome proliferator nafenopin inhibits intercellular communication in rat but not guinea-pig hepatocytes, perturbing S-phase but not apoptosis. Arch Toxicol 72, 439–444. Elcombe, C. R. (1985). Species differences in carcinogenicity and peroxisome proliferation due to trichloroethylene: a biochemical human hazard assessment. Arch Toxicol Suppl 8, 6–17. Elcombe, C. R., Bell, D. R., Elias, E., Hasmall, S. C., and Plant, N. J. (1996). Peroxisome proliferators: species differences in response of primary hepatocyte cultures. Ann N Y Acad Sci 804, 628–635. Elliott, B. M., and Elcombe, C. R. (1987). Lack of DNA damage or lipid peroxidation measured in vivo in the rat liver following treatment with peroxisomal proliferators. Carcinogenesis 8, 1213–1218.
REFERENCES
471
EPA (2005a). Guidelines for carcinogen risk assessment. EPA/630/P-03/001F, 1–166, http://oaspub.epa. gov/eims/eimscomm.getfile?p_download_id=439797. EPA (2005b). Supplemental guidance for assessing susceptibility from early-life exposure to carcinogens. EPA/630/R-03/003F, 1–126, http://oaspub.epa.gov/eims/eimscomm.getfile?p_download_id=459042. Fan, C. Y., Pan, J., Usuda, N., Yeldandi, A. V., Rao, M. S., and Reddy, J. K. (1998). Steatohepatitis, spontaneous peroxisome proliferation and liver tumors in mice lacking peroxisomal fatty acyl-CoA oxidase. Implications for peroxisome proliferator-activated receptor alpha natural ligand metabolism. J Biol Chem 273, 15639–15645. Fischer, J. G., Glauert, H. P., Yin, T., Sweeney-Reeves, M. L., Larmonier, N., and Black, M. C. (2002). Moderate iron overload enhances lipid peroxidation in livers of rats, but does not affect NF-kappaB activation induced by the peroxisome proliferator, Wy-14,643. J Nutr 132, 2525–2531. Flavell, D. M., Pineda Torra, I., Jamshidi, Y., Evans, D., Diamond, J. R., Elkeles, R. S., Bujac, S. R., Miller, G., Talmud, P. J., Staels, B., and Humphries, S. E. (2000). Variation in the PPARalpha gene is associated with altered function in vitro and plasma lipid concentrations in Type II diabetic subjects. Diabetologia 43, 673–680. Gariot, P., Barrat, E., Drouin, P., Genton, P., Pointel, J. P., Foliguet, B., Kolopp, M., and Debry, G. (1987). Morphometric study of human hepatic cell modifications induced by fenofibrate. Metabolism 36, 203–210. Gariot, P., Barrat, E., Mejean, L., Pointel, J. P., Drouin, P., and Debry, G. (1983). Fenofibrate and human liver. Lack of proliferation of peroxisomes. Arch Toxicol 53, 151–163. Gervois, P., Torra, I. P., Chinetti, G., Grotzinger, T., Dubois, G., Fruchart, J. C., Fruchart-Najib, J., Leitersdorf, E., and Staels, B. (1999). A truncated human peroxisome proliferator-activated receptor alpha splice variant with dominant negative activity. Mol Endocrinol 13, 1535–1549. Gill, J. H., James, N. H., Roberts, R. A., and Dive, C. (1998). The non-genotoxic hepatocarcinogen nafenopin suppresses rodent hepatocyte apoptosis induced by TGFbeta1, DNA damage and Fas. Carcinogenesis 19, 299–304. Glauert, H. P., Beaty, M. M., Clark, T. D., Greenwell, W. S., Tatum, V., Chen, L. C., Borges, T., Clark, T. L., Srinivasan, S. R., and Chow, C. K. (1990). Effect of dietary vitamin E on the development of altered hepatic foci and hepatic tumors induced by the peroxisome proliferator ciprofibrate. J Cancer Res Clin Oncol 116, 351–356. Glauert, H. P., Eyigor, A., Tharappel, J. C., Cooper, S., Lee, E. Y., and Spear, B. T. (2006). Inhibition of hepatocarcinogenesis by the deletion of the p50 subunit of NF-kappaB in mice administered the peroxisome proliferator Wy-14,643. Toxicol Sci 90, 331–336. Gloire, G., Legrand-Poels, S., and Piette, J. (2006). NF-kappaB activation by reactive oxygen species: fifteen years later. Biochem Pharmacol 72, 1493–1505. Goel, S. K., Lalwani, N. D., and Reddy, J. K. (1986). Peroxisome proliferation and lipid peroxidation in rat liver. Cancer Res 46, 1324–1330. Goll, V., Alexandre, E., Viollon-Abadie, C., Nicod, L., Jaeck, D., and Richert, L. (1999). Comparison of the effects of various peroxisome proliferators on peroxisomal enzyme activities, DNA synthesis, and apoptosis in rat and human hepatocyte cultures. Toxicol Appl Pharmacol 160, 21–32. Gonzalez, F. J., and Shah, Y. M. (2008). PPARalpha: mechanism of species differences and hepatocarcinogenesis of peroxisome proliferators. Toxicology 246, 2–8. Gottlicher, M., Widmark, E., Li, Q., and Gustafsson, J. A. (1992). Fatty acids activate a chimera of the clofibric acid-activated receptor and the glucocorticoid receptor. Proc Natl Acad Sci USA 89, 4653–4657. Grasl-Kraupp, B., Huber, W., Timmermann-Trosiener, I., and Schulte-Hermann, R. (1993). Peroxisomal enzyme induction uncoupled from enhanced DNA synthesis in putative preneoplastic liver foci of rats treated with a single dose of the peroxisome proliferator nafenopin. Carcinogenesis 14, 2435–2437. Grasl-Kraupp, B., Ruttkay-Nedecky, B., Mullauer, L., Taper, H., Huber, W., Bursch, W., and SchulteHermann, R. (1997). Inherent increase of apoptosis in liver tumors: implications for carcinogenesis and tumor regression. Hepatology 25, 906–912. Hanefeld, M., Kemmer, C., and Kadner, E. (1983). Relationship between morphological changes and lipid-lowering action of p-chlorphenoxyisobutyric acid (CPIB) on hepatic mitochondria and peroxisomes in man. Atherosclerosis 46, 239–246.
472
CHAPTER 17 MODE OF ACTION ANALYSIS AND HUMAN RELEVANCE
Hanefeld, M., Kemmer, C., Leonhardt, W., Kunze, K. D., Jaross, W., and Haller, H. (1980). Effects of p-chlorophenoxyisobutyric acid (CPIB) on the human liver. Atherosclerosis 36, 159–172. Hanselman, J. C., Vartanian, M. A., Koester, B. P., Gray, S. A., Essenburg, A. D., Rea, T. J., Bisgaier, C. L., and Pape, M. E. (2001). Expression of the mRNA encoding truncated PPAR alpha does not correlate with hepatic insensitivity to peroxisome proliferators. Mol Cell Biochem 217, 91–97. Hashimoto, T. (1996). Peroxisomal beta-oxidation: enzymology and molecular biology. Ann N Y Acad Sci 804, 86–98. Hashimoto, T., Fujita, T., Usuda, N., Cook, W., Qi, C., Peters, J. M., Gonzalez, F. J., Yeldandi, A. V., Rao, M. S., and Reddy, J. K. (1999). Peroxisomal and mitochondrial fatty acid beta-oxidation in mice nullizygous for both peroxisome proliferator-activated receptor alpha and peroxisomal fatty acyl-CoA oxidase. Genotype correlation with fatty liver phenotype. J Biol Chem 274, 19228–19236. Hasmall, S. C., James, N. H., Macdonald, N., Gonzalez, F. J., Peters, J. M., and Roberts, R. A. (2000a). Suppression of mouse hepatocyte apoptosis by peroxisome proliferators: role of PPARalpha and TNFalpha. Mutat Res 448, 193–200. Hasmall, S. C., James, N. H., Macdonald, N., Soames, A. R., and Roberts, R. A. (2000b). Species differences in response to diethylhexylphthalate: suppression of apoptosis, induction of DNA synthesis and peroxisome proliferator activated receptor alpha-mediated gene expression. Arch Toxicol 74, 85–91. Hasmall, S. C., James, N. H., Macdonald, N., West, D., Chevalier, S., Cosulich, S. C., and Roberts, R. A. (1999). Suppression of apoptosis and induction of DNA synthesis in vitro by the phthalate plasticizers monoethylhexylphthalate (MEHP) and diisononylphthalate (DINP): a comparison of rat and human hepatocytes in vitro. Arch Toxicol 73, 451–456. Hasmall, S. C., James, N. H., Soames, A. R., and Roberts, R. A. (1998). The peroxisome proliferator nafenopin does not suppress hepatocyte apoptosis in guinea-pig liver in vivo nor in human hepatocytes in vitro. Arch Toxicol 72, 777–783. Hays, T., Rusyn, I., Burns, A. M., Kennett, M. J., Ward, J. M., Gonzalez, F. J., and Peters, J. M. (2005). Role of peroxisome proliferator-activated receptor-alpha (PPARalpha) in bezafibrate-induced hepatocarcinogenesis and cholestasis. Carcinogenesis 26, 219–227. Hertz, R., Bishara-Shieban, J., and Bar-Tana, J. (1995). Mode of action of peroxisome proliferators as hypolipidemic drugs. Suppression of apolipoprotein C-III. J Biol Chem 270, 13470–13475. Hess, R., Staubli, W., and Riess, W. (1965). Nature of the hepatomegalic effect produced by ethylchlorophenoxy-isobutyrate in the rat. Nature 208, 856–858. Hinton, R. H., Mitchell, F. E., Mann, A., Chescoe, D., Price, S. C., Nunn, A., Grasso, P., and Bridges, J. W. (1986). Effects of phthalic acid esters on the liver and thyroid. Environ Health Perspect 70, 195–210. Hoivik, D. J., Qualls, C. W., Jr., Mirabile, R. C., Cariello, N. F., Kimbrough, C. L., Colton, H. M., Anderson, S. P., Santostefano, M. J., Morgan, R. J., Dahl, R. R., Brown, A. R., Zhao, Z., Mudd, P. N., Jr., Oliver, W. B., Jr., Brown, H. R., and Miller, R. T. (2004). Fibrates induce hepatic peroxisome and mitochondrial proliferation without overt evidence of cellular proliferation and oxidative stress in cynomolgus monkeys. Carcinogenesis 25, 1757–1769. Holden, P. R., Hasmall, S. C., James, N. H., West, D. R., Brindle, R. D., Gonzalez, F. J., Peters, J. M., and Roberts, R. A. (2000). Tumour necrosis factor alpha (TNFalpha): role in suppression of apoptosis by the peroxisome proliferator nafenopin. Cell Mol Biol (Noisy-le-grand) 46, 29–39. Holsapple, M. P., Pitot, H. C., Cohen, S. M., Boobis, A. R., Klaunig, J. E., Pastoor, T., Dellarco, V. L., and Dragan, Y. P. (2006). Mode of action in relevance of rodent liver tumors to human cancer risk. Toxicol Sci 89, 51–56. Hsieh, L. L., Shinozuka, H., and Weinstein, I. B. (1991). Changes in expression of cellular oncogenes and endogenous retrovirus-like sequences during hepatocarcinogenesis induced by a peroxisome proliferator. Br J Cancer 64, 815–820. Hsu, M. H., Savas, U., Griffin, K. J., and Johnson, E. F. (2001). Identification of peroxisome proliferatorresponsive human genes by elevated expression of the peroxisome proliferator-activated receptor alpha in HepG2 cells. J Biol Chem 276, 27950–27958. Huber, W., Kraupp-Grasl, B., Esterbauer, H., and Schulte-Hermann, R. (1991). Role of oxidative stress in age dependent hepatocarcinogenesis by the peroxisome proliferator nafenopin in the rat. Cancer Res 51, 1789–1792.
REFERENCES
473
Huber, W. W., Grasl-Kraupp, B., Stekel, H., Gschwentner, C., Lang, H., and Schulte-Hermann, R. (1997). Inhibition instead of enhancement of lipid peroxidation by pretreatment with the carcinogenic peroxisome proliferator nafenopin in rat liver exposed to a high single dose of corn oil. Arch Toxicol 71, 575–581. Isenberg, J. S., Kamendulis, L. M., Ackley, D. C., Smith, J. H., Pugh, G., Jr., Lington, A. W., McKee, R. H., and Klaunig, J. E. (2001). Reversibility and persistence of di-2-ethylhexyl phthalate (DEHP)- and phenobarbital-induced hepatocellular changes in rodents. Toxicol Sci 64, 192–199. Isenberg, J. S., Kamendulis, L. M., Smith, J. H., Ackley, D. C., Pugh, G., Jr., Lington, A. W., and Klaunig, J. E. (2000). Effects of Di-2-ethylhexyl phthalate (DEHP) on gap-junctional intercellular communication (GJIC), DNA synthesis, and peroxisomal beta oxidation (PBOX) in rat, mouse, and hamster liver. Toxicol Sci 56, 73–85. Isenberg, J. S., Kolaja, K. L., Ayoubi, S. A., Watkins, J. B., 3rd, and Klaunig, J. E. (1997). Inhibition of WY-14,643 induced hepatic lesion growth in mice by rotenone. Carcinogenesis 18, 1511–1519. Issemann, I., and Green, S. (1990). Activation of a member of the steroid hormone receptor superfamily by peroxisome proliferators. Nature 347, 645–650. Ito, Y., Yamanoshita, O., Asaeda, N., Tagawa, Y., Lee, C. H., Aoyama, T., Ichihara, G., Furuhashi, K., Kamijima, M., Gonzalez, F. J., and Nakajima, T. (2007). Di(2-ethylhexyl)phthalate induces hepatic tumorigenesis through a peroxisome proliferator-activated receptor alpha-independent pathway. J Occup Health 49, 172–182. James, N. H., Gill, J. H., Brindle, R., Woodyatt, N. J., Macdonald, N., Rolfe, M., Hasmall, S. C., Tugwood, J. D., Holden, P. R., and Roberts, R. A. (1998a). Peroxisome proliferator-activated receptor (PPAR) alpha-regulated growth responses and their importance to hepatocarcinogenesis. Toxicol Lett 102–103, 91–96. James, N. H., and Roberts, R. A. (1996). Species differences in response to peroxisome proliferators correlate in vitro with induction of DNA synthesis rather than suppression of apoptosis. Carcinogenesis 17, 1623–1632. James, N. H., Soames, A. R., and Roberts, R. A. (1998b). Suppression of hepatocyte apoptosis and induction of DNA synthesis by the rat and mouse hepatocarcinogen diethylhexylphlathate (DEHP) and the mouse hepatocarcinogen 1,4-dichlorobenzene (DCB). Arch Toxicol 72, 784–790. Jia, Y., Qi, C., Zhang, Z., Hashimoto, T., Rao, M. S., Huyghe, S., Suzuki, Y., Van Veldhoven, P. P., Baes, M., and Reddy, J. K. (2003). Overexpression of peroxisome proliferator-activated receptor-alpha (PPARalpha)-regulated genes in liver in the absence of peroxisome proliferation in mice deficient in both L- and D-forms of enoyl-CoA hydratase/dehydrogenase enzymes of peroxisomal beta-oxidation system. J Biol Chem 278, 47232–47239. Karin, M. (2006). Nuclear factor-kappaB in cancer development and progression. Nature 441, 431–436. Kasai, H. (1997). Analysis of a form of oxidative DNA damage, 8-hydroxy-2′-deoxyguanosine, as a marker of cellular oxidative stress during carcinogenesis. Mutat Res 387, 147–163. Keller, H., Devchand, P. R., Perroud, M., and Wahli, W. (1997). PPAR alpha structure-function relationships derived from species-specific differences in responsiveness to hypolipidemic agents. Biol Chem 378, 651–655. Keller, H., Dreyer, C., Medin, J., Mahfoudi, A., Ozato, K., and Wahli, W. (1993). Fatty acids and retinoids control lipid metabolism through activation of peroxisome proliferator-activated receptor-retinoid X receptor heterodimers. Proc Natl Acad Sci USA 90, 2160–2164. Klaunig, J. E., Babich, M. A., Baetcke, K. P., Cook, J. C., Corton, J. C., David, R. M., DeLuca, J. G., Lai, D. Y., McKee, R. H., Peters, J. M., Roberts, R. A., and Fenner-Crisp, P. A. (2003). PPARalpha agonist-induced rodent tumors: modes of action and human relevance. Crit Rev Toxicol 33, 655–780. Klaunig, J. E., Xu, Y., Isenberg, J. S., Bachowski, S., Kolaja, K. L., Jiang, J., Stevenson, D. E., and Walborg, E. F., Jr. (1998). The role of oxidative stress in chemical carcinogenesis. Environ Health Perspect 106 (Suppl 1), 289–295. Klungland, A., Rosewell, I., Hollenbach, S., Larsen, E., Daly, G., Epe, B., Seeberg, E., Lindahl, T., and Barnes, D. E. (1999). Accumulation of premutagenic DNA lesions in mice defective in removal of oxidative base damage. Proc Natl Acad Sci USA 96, 13300–13305.
474
CHAPTER 17 MODE OF ACTION ANALYSIS AND HUMAN RELEVANCE
Lacquemant, C., Lepretre, F., Pineda Torra, I., Manraj, M., Charpentier, G., Ruiz, J., Staels, B., and Froguel, P. (2000). Mutation screening of the PPARalpha gene in type 2 diabetes associated with coronary heart disease. Diabetes Metab 26, 393–401. Lake, B. G., Evans, J. G., Cunninghame, M. E., and Price, R. J. (1993). Comparison of the hepatic effects of nafenopin and WY-14,643 on peroxisome proliferation and cell replication in the rat and Syrian hamster. Environ Health Perspect 101 (Suppl 5), 241–247. Lake, B. G., Evans, J. G., Gray, T. J., Korosi, S. A., and North, C. J. (1989a). Comparative studies on nafenopin-induced hepatic peroxisome proliferation in the rat, Syrian hamster, guinea pig, and marmoset. Toxicol Appl Pharmacol 99, 148–160. Lake, B. G., Gray, T. J., Korosi, S. A., and Walters, D. G. (1989b). Nafenopin, a peroxisome proliferator, depletes hepatic vitamin E content and elevates plasma oxidised glutathione levels in rats. Toxicol Lett 45, 221–229. Lake, B. G., Kozlen, S. L., Evans, J. G., Gray, T. J., Young, P. J., and Gangolli, S. D. (1987). Effect of prolonged administration of clofibric acid and di-(2-ethylhexyl)phthalate on hepatic enzyme activities and lipid peroxidation in the rat. Toxicology 44, 213–228. Lake, B. G., Rumsby, P. C., Price, R. J., and Cunninghame, M. E. (2000). Species differences in hepatic peroxisome proliferation, cell replication and transforming growth factor-beta1 gene expression in the rat, Syrian hamster and guinea pig. Mutat Res 448, 213–225. Laughter, A. R., Dunn, C. S., Swanson, C. L., Howroyd, P., Cattley, R. C., and Corton, J. C. (2004). Role of the peroxisome proliferator-activated receptor alpha (PPARalpha) in responses to trichloroethylene and metabolites, trichloroacetate and dichloroacetate in mouse liver. Toxicology 203, 83–98. Lawrence, J. W., Li, Y., Chen, S., DeLuca, J. G., Berger, J. P., Umbenhauer, D. R., Moller, D. E., and Zhou, G. (2001a). Differential gene regulation in human versus rodent hepatocytes by peroxisome proliferator-activated receptor (PPAR) alpha. PPAR alpha fails to induce peroxisome proliferationassociated genes in human cells independently of the level of receptor expresson. J Biol Chem 276, 31521–31527. Lawrence, J. W., Wollenberg, G. K., and DeLuca, J. G. (2001b). Tumor necrosis factor alpha is not required for WY14,643-induced cell proliferation. Carcinogenesis 22, 381–386. Lawrence, J. W., Wollenberg, G. K., Frank, J. D., and DeLuca, J. G. (2001c). Dexamethasone selectively inhibits WY14,643-induced cell proliferation and not peroxisome proliferation in mice. Toxicol Appl Pharmacol 170, 113–123. Li, Y., Glauert, H. P., and Spear, B. T. (2000a). Activation of nuclear factor-kappaB by the peroxisome proliferator ciprofibrate in H4IIEC3 rat hepatoma cells and its inhibition by the antioxidants N-acetylcysteine and vitamin E. Biochem Pharmacol 59, 427–434. Li, Y., Kovach, A., Suino-Powell, K., Martynowski, D., and Xu, H. E. (2008). Structural and biochemical basis for the binding selectivity of peroxisome proliferator-activated receptor gamma to PGC-1alpha. J Biol Chem 283, 19132–19139. Li, Y., Leung, L. K., Glauert, H. P., and Spear, B. T. (1996). Treatment of rats with the peroxisome proliferator ciprofibrate results in increased liver NF-kappaB activity. Carcinogenesis 17, 2305– 2309. Li, Y., Tharappel, J. C., Cooper, S., Glenn, M., Glauert, H. P., and Spear, B. T. (2000b). Expression of the hydrogen peroxide-generating enzyme fatty acyl CoA oxidase activates NF-kappaB. DNA Cell Biol 19, 113–120. Liu, M. H., Li, J., Shen, P., Husna, B., Tai, E. S., and Yong, E. L. (2008). A natural polymorphism in peroxisome proliferator-activated receptor-alpha hinge region attenuates transcription due to defective release of nuclear receptor corepressor from chromatin. Mol Endocrinol 22, 1078–1092. Luebker, D. J., Hansen, K. J., Bass, N. M., Butenhoff, J. L., and Seacat, A. M. (2002). Interactions of fluorochemicals with rat liver fatty acid-binding protein. Toxicology 176, 175–185. MacDonald, N., Holden, P. R., and Roberts, R. A. (1999). Addition of peroxisome proliferator-activated receptor alpha to guinea pig hepatocytes confers increased responsiveness to peroxisome proliferators. Cancer Res 59, 4776–4780. Maeda, S., Kamata, H., Luo, J. L., Leffert, H., and Karin, M. (2005). IKKbeta couples hepatocyte death to cytokine-driven compensatory proliferation that promotes chemical hepatocarcinogenesis. Cell 121, 977–990.
REFERENCES
475
Makowska, J. M., Gibson, G. G., and Bonner, F. W. (1992). Species differences in ciprofibrate induction of hepatic cytochrome P450 4A1 and peroxisome proliferation. J Biochem Toxicol 7, 183–191. Maloney, E. K., and Waxman, D. J. (1999). trans-Activation of PPARalpha and PPARgamma by structurally diverse environmental chemicals. Toxicol Appl Pharmacol 161, 209–218. Mannaerts, G. P., and van Veldhoven, P. P. (1996). Functions and organization of peroxisomal betaoxidation. Ann NY Acad Sci 804, 99–115. Marsman, D. S., Cattley, R. C., Conway, J. G., and Popp, J. A. (1988). Relationship of hepatic peroxisome proliferation and replicative DNA synthesis to the hepatocarcinogenicity of the peroxisome proliferators di(2-ethylhexyl)phthalate and [4-chloro-6-(2,3-xylidino)-2-pyrimidinylthio]acetic acid (Wy-14,643) in rats. Cancer Res 48, 6739–6744. Marsman, D. S., Goldsworthy, T. L., and Popp, J. A. (1992). Contrasting hepatocytic peroxisome proliferation, lipofuscin accumulation and cell turnover for the hepatocarcinogens Wy-14,643 and clofibric acid. Carcinogenesis 13, 1011–1017. Marsman, D. S., and Popp, J. A. (1994). Biological potential of basophilic hepatocellular foci and hepatic adenoma induced by the peroxisome proliferator, Wy-14,643. Carcinogenesis 15, 111–117. Mascaro, C., Acosta, E., Ortiz, J. A., Marrero, P. F., Hegardt, F. G., and Haro, D. (1998). Control of human muscle-type carnitine palmitoyltransferase I gene transcription by peroxisome proliferatoractivated receptor. J Biol Chem 273, 8560–8563. Meek, M. E. (2008). Recent developments in frameworks to consider human relevance of hypothesized modes of action for tumours in animals. Environ Mol Mutagen 49, 110–116. Menegazzi, M., Carcereri-De Prati, A., Suzuki, H., Shinozuka, H., Pibiri, M., Piga, R., Columbano, A., and Ledda-Columbano, G. M. (1997). Liver cell proliferation induced by nafenopin and cyproterone acetate is not associated with increases in activation of transcription factors NF-kappaB and AP-1 or with expression of tumor necrosis factor alpha. Hepatology 25, 585–592. Meyer, K., Lee, J. S., Dyck, P. A., Cao, W. Q., Rao, M. S., Thorgeirsson, S. S., and Reddy, J. K. (2003). Molecular profiling of hepatocellular carcinomas developing spontaneously in acyl-CoA oxidase deficient mice: comparison with liver tumors induced in wild-type mice by a peroxisome proliferator and a genotoxic carcinogen. Carcinogenesis 24, 975–984. Morimura, K., Cheung, C., Ward, J. M., Reddy, J. K., and Gonzalez, F. J. (2006). Differential susceptibility of mice humanized for peroxisome proliferator-activated receptor alpha to Wy-14,643-induced liver tumorigenesis. Carcinogenesis 27, 1074–1080. Mukherjee, R., Jow, L., Noonan, D., and McDonnell, D. P. (1994). Human and rat peroxisome proliferator activated receptors (PPARs) demonstrate similar tissue distribution but different responsiveness to PPAR activators. J Steroid Biochem Mol Biol 51, 157–166. Nemali, M. R., Usuda, N., Reddy, M. K., Oyasu, K., Hashimoto, T., Osumi, T., Rao, M. S., and Reddy, J. K. (1988). Comparison of constitutive and inducible levels of expression of peroxisomal betaoxidation and catalase genes in liver and extrahepatic tissues of rat. Cancer Res 48, 5316–5324. Nicholls-Grzemski, F. A., Belling, G. B., Priestly, B. G., Calder, I. C., and Burcham, P. C. (2000). Clofibrate pretreatment in mice confers resistance against hepatic lipid peroxidation. J Biochem Mol Toxicol 14, 335–345. Nilakantan, V., Spear, B. T., and Glauert, H. P. (1998). Liver-specific catalase expression in transgenic mice inhibits NF-kappaB activation and DNA synthesis induced by the peroxisome proliferator ciprofibrate. Carcinogenesis 19, 631–637. O’Brien, M. L., Cunningham, M. L., Spear, B. T., and Glauert, H. P. (2001a). Effects of peroxisome proliferators on glutathione and glutathione-related enzymes in rats and hamsters. Toxicol Appl Pharmacol 171, 27–37. O’Brien, M. L., Twaroski, T. P., Cunningham, M. L., Glauert, H. P., and Spear, B. T. (2001b). Effects of peroxisome proliferators on antioxidant enzymes and antioxidant vitamins in rats and hamsters. Toxicol Sci 60, 271–278. Oberhammer, F. A., and Qin, H. M. (1995). Effect of three tumour promoters on the stability of hepatocyte cultures and apoptosis after transforming growth factor-beta 1. Carcinogenesis 16, 1363–1371. Ohmura, T., Ledda-Columbano, G. M., Piga, R., Columbano, A., Glemba, J., Katyal, S. L., Locker, J., and Shinozuka, H. (1996). Hepatocyte proliferation induced by a single dose of a peroxisome proliferator. Am J Pathol 148, 815–824.
476
CHAPTER 17 MODE OF ACTION ANALYSIS AND HUMAN RELEVANCE
Pacot, C., Petit, M., Rollin, M., Behechti, N., Moisant, M., Deslex, P., Althoff, J., Lhuguenot, J. C., and Latruffe, N. (1996). Difference between guinea pig and rat in the liver peroxisomal response to equivalent plasmatic level of ciprofibrate. Arch Biochem Biophys 327, 181–188. Palmer, C. N., Hsu, M. H., Griffin, K. J., Raucy, J. L., and Johnson, E. F. (1998). Peroxisome proliferator activated receptor-alpha expression in human liver. Mol Pharmacol 53, 14–22. Perrone, C. E., Shao, L., and Williams, G. M. (1998). Effect of rodent hepatocarcinogenic peroxisome proliferators on fatty acyl-CoA oxidase, DNA synthesis, and apoptosis in cultured human and rat hepatocytes. Toxicol Appl Pharmacol 150, 277–286. Peters, J. M., Aoyama, T., Cattley, R. C., Nobumitsu, U., Hashimoto, T., and Gonzalez, F. J. (1998). Role of peroxisome proliferator-activated receptor alpha in altered cell cycle regulation in mouse liver. Carcinogenesis 19, 1989–1994. Peters, J. M., Cattley, R. C., and Gonzalez, F. J. (1997). Role of PPAR alpha in the mechanism of action of the nongenotoxic carcinogen and peroxisome proliferator Wy-14,643. Carcinogenesis 18, 2029–2033. Peters, J. M., Cheung, C., and Gonzalez, F. J. (2005). Peroxisome proliferator-activated receptor-alpha and liver cancer: where do we stand? J Mol Med 83, 774–785. Peters, J. M., Rusyn, I., Rose, M. L., Gonzalez, F. J., and Thurman, R. G. (2000). Peroxisome proliferatoractivated receptor alpha is restricted to hepatic parenchymal cells, not Kupffer cells: implications for the mechanism of action of peroxisome proliferators in hepatocarcinogenesis. Carcinogenesis 21, 823–826. Plant, N. J., Horley, N. J., Dickins, M., Hasmall, S., Elcombe, C. R., and Bell, D. R. (1998). The coordinate regulation of DNA synthesis and suppression of apoptosis is differentially regulated by the liver growth agents, phenobarbital and methylclofenapate. Carcinogenesis 19, 1521–1527. Price, R. J., Evans, J. G., and Lake, B. G. (1992). Comparison of the effects of nafenopin on hepatic peroxisome proliferation and replicative DNA synthesis in the rat and Syrian hamster. Food Chem Toxicol 30, 937–944. Pugh, G., Jr., Isenberg, J. S., Kamendulis, L. M., Ackley, D. C., Clare, L. J., Brown, R., Lington, A. W., Smith, J. H., and Klaunig, J. E. (2000). Effects of di-isononyl phthalate, di-2-ethylhexyl phthalate, and clofibrate in cynomolgus monkeys. Toxicol Sci 56, 181–188. Qu, B., Halliwell, B., Ong, C. N., Lee, B. L., and Li, Q. T. (2000). Caloric restriction prevents oxidative damage induced by the carcinogen clofibrate in mouse liver. FEBS Lett 473, 85–88. Qu, B., Li, Q. T., Wong, K. P., Tan, T. M., and Halliwell, B. (2001). Mechanism of clofibrate hepatotoxicity: mitochondrial damage and oxidative stress in hepatocytes. Free Radic Biol Med 31, 659–669. Rao, M. S., Lalwani, N. D., Scarpelli, D. G., and Reddy, J. K. (1982). The absence of gamma-glutamyl transpeptidase activity in putative preneoplastic lesions and in hepatocellular carcinomas induced in rats by the hypolipidemic peroxisome proliferator Wy-14,643. Carcinogenesis 3, 1231–1233. Rao, M. S., Lalwani, N. D., Watanabe, T. K., and Reddy, J. K. (1984). Inhibitory effect of antioxidants ethoxyquin and 2(3)-tert-butyl-4-hydroxyanisole on hepatic tumorigenesis in rats fed ciprofibrate, a peroxisome proliferator. Cancer Res 44, 1072–1076. Rao, M. S., and Subbarao, V. (1997a). The effect of deferoxamine on ciprofibrate-induced hepatocarcinogenesis in the rat. In Vivo 11, 495–498. Rao, M. S., and Subbarao, V. (1997b). Effect of dexamethasone on ciprofibrate-induced cell proliferation and peroxisome proliferation. Fundam Appl Toxicol 35, 78–83. Rao, M. S., and Subbarao, V. (1999). Inhibition of ciprofibrate-induced hepatocarcinogenesis in the rat by dimethylthiourea, a scavenger of hydroxyl radical. Oncol Rep 6, 1285–1288. Rao, M. S., Tatematsu, M., Subbarao, V., Ito, N., and Reddy, J. K. (1986). Analysis of peroxisome proliferator-induced preneoplastic and neoplastic lesions of rat liver for placental form of glutathione S-transferase and gamma-glutamyltranspeptidase. Cancer Res 46, 5287–5290. Rao, M. S., Thangada, S., and Subbarao, V. (1991). Peroxisome proliferation in neoplastic nodules and hepatocellular carcinomas induced by ciprofibrate in the rat. Exp Pathol 41, 44–49. Rao, M. S., Usuda, N., Subbarao, V., and Reddy, J. K. (1987). Absence of gamma-glutamyl transpeptidase activity in neoplastic lesions induced in the liver of male F-344 rats by di-(2-ethylhexyl)phthalate, a peroxisome proliferator. Carcinogenesis 8, 1347–1350.
REFERENCES
477
Ray, A., and Prefontaine, K. E. (1994). Physical association and functional antagonism between the p65 subunit of transcription factor NF-kappa B and the glucocorticoid receptor. Proc Natl Acad Sci USA 91, 752–756. Reddy, J. K., and Chu, R. (1996). Peroxisome proliferator-induced pleiotropic responses: pursuit of a phenomenon. Ann NY Acad Sci 804, 176–201. Reddy, J. K., Lalwani, N. D., Reddy, M. K., and Qureshi, S. A. (1982). Excessive accumulation of autofluorescent lipofuscin in the liver during hepatocarcinogenesis by methyl clofenapate and other hypolipidemic peroxisome proliferators. Cancer Res 42, 259–266. Reddy, J. K., and Qureshi, S. A. (1979). Tumorigenicity of the hypolipidaemic peroxisome proliferator ethyl-alpha-p-chlorophenoxyisobutyrate (clofibrate) in rats. Br J Cancer 40, 476–482. Reddy, J. K., and Rao, M. S. (1977). Malignant tumors in rats fed nafenopin, a hepatic peroxisome proliferator. J Natl Cancer Inst 59, 1645–1650. Reddy, J. K., and Rao, M. S. (1989). Oxidative DNA damage caused by persistent peroxisome proliferation: its role in hepatocarcinogenesis. Mutat Res 214, 63–68. Ren, H., Aleksunes, L. M., Wood, C., Vallanat, B., George, M. H., Klaassen, C. D., and Corton, J. C. (2010). Characterization of peroxisome proliferator-activated receptor alpha–independent effects of PPARalpha activators in the rodent liver: di-(2-ethylhexyl) phthalate also activates the constitutiveactivated receptor. Toxicol Sci 113(1), 45–59. Richert, L., Lamboley, C., Viollon-Abadie, C., Grass, P., Hartmann, N., Laurent, S., Heyd, B., Mantion, G., Chibout, S. D., and Staedtler, F. (2003). Effects of clofibric acid on mRNA expression profiles in primary cultures of rat, mouse and human hepatocytes. Toxicol Appl Pharmacol 191, 130–146. Roberts, R. A. (1999). Peroxisome proliferators: mechanisms of adverse effects in rodents and molecular basis for species differences. Arch Toxicol 73, 413–418. Roberts, R. A., Ganey, P. E., Ju, C., Kamendulis, L. M., Rusyn, I., and Klaunig, J. E. (2007). Role of the Kupffer cell in mediating hepatic toxicity and carcinogenesis. Toxicol Sci 96, 2–15. Roberts, R. A., James, N. H., Hasmall, S. C., Holden, P. R., Lambe, K., Macdonald, N., West, D., Woodyatt, N. J., and Whitcome, D. (2000). Apoptosis and proliferation in nongenotoxic carcinogenesis: species differences and role of PPARalpha. Toxicol Lett 112–113, 49–57. Roglans, N., Bellido, A., Rodriguez, C., Cabrero, A., Novell, F., Ros, E., Zambon, D., and Laguna, J. C. (2002). Fibrate treatment does not modify the expression of acyl coenzyme A oxidase in human liver. Clin Pharmacol Ther 72, 692–701. Rolfe, M., James, N. H., and Roberts, R. A. (1997). Tumour necrosis factor alpha (TNF alpha) suppresses apoptosis and induces DNA synthesis in rodent hepatocytes: a mediator of the hepatocarcinogenicity of peroxisome proliferators? Carcinogenesis 18, 2277–2280. Rose, M. L., Cattley, R. C., Dunn, C., Wong, V., Li, X., and Thurman, R. G. (1999a). Dietary glycine prevents the development of liver tumors caused by the peroxisome proliferator WY-14,643. Carcinogenesis 20, 2075–2081. Rose, M. L., Germolec, D., Arteel, G. E., Schoonhoven, R., and Thurman, R. G. (1997a). Dietary glycine prevents increases in hepatocyte proliferation caused by the peroxisome proliferator WY-14,643. Chem Res Toxicol 10, 1198–1204. Rose, M. L., Germolec, D. R., Schoonhoven, R., and Thurman, R. G. (1997b). Kupffer cells are causally responsible for the mitogenic effect of peroxisome proliferators. Carcinogenesis 18, 1453–1456. Rose, M. L., Rivera, C. A., Bradford, B. U., Graves, L. M., Cattley, R. C., Schoonhoven, R., Swenberg, J. A., and Thurman, R. G. (1999b). Kupffer cell oxidant production is central to the mechanism of peroxisome proliferators. Carcinogenesis 20, 27–33. Rosen, M. B., Abbott, B. D., Wolf, D. C., Corton, J. C., Wood, C. R., Schmid, J. E., Das, K. P., Zehr, R. D., Blair, E. T., and Lau, C. (2008a). Gene profiling in the livers of wild-type and PPARalpha-null mice exposed to perfluorooctanoic acid. Toxicol Pathol 36, 592–607. Rosen, M. B., Lee, J. S., Ren, H., Vallanat, B., Liu, J., Waalkes, M. P., Abbott, B. D., Lau, C., and Corton, J. C. (2008b). Toxicogenomic dissection of the perfluorooctanoic acid transcript profile in mouse liver: evidence for the involvement of nuclear receptors PPAR alpha and CAR. Toxicol Sci 103, 46–56. Rusyn, I., Asakura, S., Pachkowski, B., Bradford, B. U., Denissenko, M. F., Peters, J. M., Holland, S. M., Reddy, J. K., Cunningham, M. L., and Swenberg, J. A. (2004). Expression of base excision DNA
478
CHAPTER 17 MODE OF ACTION ANALYSIS AND HUMAN RELEVANCE
repair genes is a sensitive biomarker for in vivo detection of chemical-induced chronic oxidative stress: identification of the molecular source of radicals responsible for DNA damage by peroxisome proliferators. Cancer Res 64, 1050–1057. Rusyn, I., Denissenko, M. F., Wong, V. A., Butterworth, B. E., Cunningham, M. L., Upton, P. B., Thurman, R. G., and Swenberg, J. A. (2000a). Expression of base excision repair enzymes in rat and mouse liver is induced by peroxisome proliferators and is dependent upon carcinogenic potency. Carcinogenesis 21, 2141–2145. Rusyn, I., Kadiiska, M. B., Dikalova, A., Kono, H., Yin, M., Tsuchiya, K., Mason, R. P., Peters, J. M., Gonzalez, F. J., Segal, B. H., Holland, S. M., and Thurman, R. G. (2001). Phthalates rapidly increase production of reactive oxygen species in vivo: role of Kupffer cells. Mol Pharmacol 59, 744–750. Rusyn, I., Peters, J. M., and Cunningham, M. L. (2006). Modes of action and species-specific effects of di-(2-ethylhexyl)phthalate in the liver. Crit Rev Toxicol 36, 459–479. Rusyn, I., Rose, M. L., Bojes, H. K., and Thurman, R. G. (2000b). Novel role of oxidants in the molecular mechanism of action of peroxisome proliferators. Antioxid Redox Signal 2, 607–621. Rusyn, I., Tsukamoto, H., and Thurman, R. G. (1998). WY-14 643 rapidly activates nuclear factor kappaB in Kupffer cells before hepatocytes. Carcinogenesis 19, 1217–1222. Rusyn, I., Yamashina, S., Segal, B. H., Schoonhoven, R., Holland, S. M., Cattley, R. C., Swenberg, J. A., and Thurman, R. G. (2000c). Oxidants from nicotinamide adenine dinucleotide phosphate oxidase are involved in triggering cell proliferation in the liver due to peroxisome proliferators. Cancer Res 60, 4798–4803. Sanderson, L. M., de Groot, P. J., Hooiveld, G. J., Koppen, A., Kalkhoven, E., Muller, M., and Kersten, S. (2008). Effect of synthetic dietary triglycerides: a novel research paradigm for nutrigenomics. PLoS ONE 3, e1681. Sapone, A., Peters, J. M., Sakai, S., Tomita, S., Papiha, S. S., Dai, R., Friedman, F. K., and Gonzalez, F. J. (2000). The human peroxisome proliferator-activated receptor alpha gene: identification and functional characterization of two natural allelic variants. Pharmacogenetics 10, 321–333. Sausen, P. J., Lee, D. C., Rose, M. L., and Cattley, R. C. (1995). Elevated 8-hydroxydeoxyguanosine in hepatic DNA of rats following exposure to peroxisome proliferators: relationship to mitochondrial alterations. Carcinogenesis 16, 1795–1801. Schmezer, P., Pool, B. L., Klein, R. G., Komitowski, D., and Schmahl, D. (1988). Various short-term assays and two long-term studies with the plasticizer di(2-ethylhexyl)phthalate in the Syrian golden hamster. Carcinogenesis 9, 37–43. Schoonjans, K., Peinado-Onsurbe, J., Lefebvre, A. M., Heyman, R. A., Briggs, M., Deeb, S., Staels, B., and Auwerx, J. (1996). PPARalpha and PPARgamma activators direct a distinct tissue-specific transcriptional response via a PPRE in the lipoprotein lipase gene. EMBO J 15, 5336–5348. Schulte-Hermann, R., Ohde, G., Schuppler, J., and Timmermann-Trosiener, I. (1981). Enhanced proliferation of putative preneoplastic cells in rat liver following treatment with the tumor promoters phenobarbital, hexachlorocyclohexane, steroid compounds, and nafenopin. Cancer Res 41, 2556– 2562. Seo, K. W., Kim, K. B., Kim, Y. J., Choi, J. Y., Lee, K. T., and Choi, K. S. (2004). Comparison of oxidative stress and changes of xenobiotic metabolizing enzymes induced by phthalates in rats. Food Chem Toxicol 42, 107–114. Shah, Y. M., Morimura, K., Yang, Q., Tanabe, T., Takagi, M., and Gonzalez, F. J. (2007). Peroxisome proliferator-activated receptor alpha regulates a microRNA-mediated signaling cascade responsible for hepatocellular proliferation. Mol Cell Biol 27, 4238–4247. Shaw, D., Lee, R., and Roberts, R. A. (2002). Species differences in response to the phthalate plasticizer monoisononylphthalate (MINP) in vitro: a comparison of rat and human hepatocytes. Arch Toxicol 76, 344–350. Sher, T., Yi, H. F., McBride, O. W., and Gonzalez, F. J. (1993). cDNA cloning, chromosomal mapping, and functional characterization of the human peroxisome proliferator activated receptor. Biochemistry 32, 5598–5604. Shipley, J. M., Hurst, C. H., Tanaka, S. S., DeRoos, F. L., Butenhoff, J. L., Seacat, A. M., and Waxman, D. J. (2004). trans-activation of PPARalpha and induction of PPARalpha target genes by perfluorooctane-based chemicals. Toxicol Sci 80, 151–160.
REFERENCES
479
Smith-Oliver, T., and Butterworth, B. E. (1987). Correlation of the carcinogenic potential of di(2-ethylhexyl)phthalate (DEHP) with induced hyperplasia rather than with genotoxic activity. Mutat Res 188, 21–28. Soames, A. R., Cliffe, S., Pate, I., and Foster, J. R. (1999). Quantitative analysis of the lobular distribution of S-phase in rat liver following dietary administration of di(2-ethylhexyl)phthalate. Toxicol Pathol 27, 436–440. Soliman, M. S., Cunningham, M. L., Morrow, J. D., Roberts, L. J., 2nd, and Badr, M. Z. (1997). Evidence against peroxisome proliferation-induced hepatic oxidative damage. Biochem Pharmacol 53, 1369– 1374. Staels, B., Schoonjans, K., Fruchart, J. C., and Auwerx, J. (1997). The effects of fibrates and thiazolidinediones on plasma triglyceride metabolism are mediated by distinct peroxisome proliferator activated receptors (PPARs). Biochimie 79, 95–99. Stanko, R. T., Sekas, G., Isaacson, I. A., Clarke, M. R., Billiar, T. R., and Paul, H. S. (1995). Pyruvate inhibits clofibrate-induced hepatic peroxisomal proliferation and free radical production in rats. Metabolism 44, 166–171. Styles, J. A., Kelly, M., Pritchard, N. R., and Elcombe, C. R. (1988). A species comparison of acute hyperplasia induced by the peroxisome proliferator methylclofenapate: involvement of the binucleated hepatocyte. Carcinogenesis 9, 1647–1655. Styles, J. A., Kelly, M. D., Pritchard, N. R., and Elcombe, C. R. (1990). Acute hyperplasia and peroxisome proliferation induced by methylclofenapate: a species comparison and implications for liver carcinogenesis. Prog Clin Biol Res 331, 385–393. Svoboda, D. J., and Azarnoff, D. L. (1979). Tumors in male rats fed ethyl chlorophenoxyisobutyrate, a hypolipidemic drug. Cancer Res 39, 3419–3428. Tachibana, K., Kobayashi, Y., Tanaka, T., Tagami, M., Sugiyama, A., Katayama, T., Ueda, C., Yamasaki, D., Ishimoto, K., Sumitomo, M., Uchiyama, Y., Kohro, T., Sakai, J., Hamakubo, T., Kodama, T., and Doi, T. (2005). Gene expression profiling of potential peroxisome proliferator-activated receptor (PPAR) target genes in human hepatoblastoma cell lines inducibly expressing different PPAR isoforms. Nucl Recept 3, 3. Tai, E. S., Demissie, S., Cupples, L. A., Corella, D., Wilson, P. W., Schaefer, E. J., and Ordovas, J. M. (2002). Association between the PPARA L162V polymorphism and plasma lipid levels: the Framingham Offspring Study. Arterioscler Thromb Vasc Biol 22, 805–810. Takacs, M. L., and Abbott, B. D. (2007). Activation of mouse and human peroxisome proliferatoractivated receptors (alpha, beta/delta, gamma) by perfluorooctanoic acid and perfluorooctane sulfonate. Toxicol Sci 95, 108–117. Takagi, A., Sai, K., Umemura, T., Hasegawa, R., and Kurokawa, Y. (1990). Significant increase of 8-hydroxydeoxyguanosine in liver DNA of rats following short-term exposure to the peroxisome proliferators di(2-ethylhexyl)phthalate and di(2-ethylhexyl)adipate. Jpn J Cancer Res 81, 213–215. Takashima, K., Ito, Y., Gonzalez, F. J., and Nakajima, T. (2008). Different mechanisms of DEHP-induced hepatocellular adenoma tumorigenesis in wild-type and Ppar alpha-null mice. J Occup Health 50, 169–180. Tanaka, K., Smith, P. F., Stromberg, P. C., Eydelloth, R. S., Herold, E. G., Grossman, S. J., Frank, J. D., Hertzog, P. R., Soper, K. A., and Keenan, K. P. (1992). Studies of early hepatocellular proliferation and peroxisomal proliferation in Sprague-Dawley rats treated with tumorigenic doses of clofibrate. Toxicol Appl Pharmacol 116, 71–77. Tanaka, N., Moriya, K., Kiyosawa, K., Koike, K., and Aoyama, T. (2008a). Hepatitis C virus core protein induces spontaneous and persistent activation of peroxisome proliferator-activated receptor alpha in transgenic mice: implications for HCV-associated hepatocarcinogenesis. Int J Cancer 122, 124–131. Tanaka, N., Moriya, K., Kiyosawa, K., Koike, K., Gonzalez, F. J., and Aoyama, T. (2008b). PPARalpha activation is essential for HCV core protein-induced hepatic steatosis and hepatocellular carcinoma in mice. J Clin Invest 118, 683–694. Tharappel, J. C., Cunningham, M. L., Spear, B. T., and Glauert, H. P. (2001). Differential activation of hepatic NF-kappaB in rats and hamsters by the peroxisome proliferators Wy-14,643, gemfibrozil, and dibutyl phthalate. Toxicol Sci 62, 20–27.
480
CHAPTER 17 MODE OF ACTION ANALYSIS AND HUMAN RELEVANCE
Tharappel, J. C., Nalca, A., Owens, A. B., Ghabrial, L., Konz, E. C., Glauert, H. P., and Spear, B. T. (2003). Cell proliferation and apoptosis are altered in mice deficient in the NF-kappaB p50 subunit after treatment with the peroxisome proliferator ciprofibrate. Toxicol Sci 75, 300–308. Thottassery, J., Winberg, L., Youssef, J., Cunningham, M., and Badr, M. (1992). Regulation of perfluorooctanoic acid–induced peroxisomal enzyme activities and hepatocellular growth by adrenal hormones. Hepatology 15, 316–322. Tomaszewski, K. E., Heindel, S. W., Jenkins, W. L., and Melnick, R. L. (1990). Induction of peroxisomal acyl CoA oxidase activity and lipid peroxidation in primary rat hepatocyte cultures. Toxicology 65, 49–60. Trapp, C., Schwarz, M., and Epe, B. (2007). The peroxisome proliferator WY-14,643 promotes hepatocarcinogenesis caused by endogenously generated oxidative DNA base modifications in repairdeficient Csbm/m/Ogg1-/- mice. Cancer Res 67, 5156–5161. Tugwood, J. D., Aldridge, T. C., Lambe, K. G., Macdonald, N., and Woodyatt, N. J. (1996). Peroxisome proliferator-activated receptors: structures and function. Ann NY Acad Sci 804, 252–265. Tugwood, J. D., Holden, P. R., James, N. H., Prince, R. A., and Roberts, R. A. (1998). A peroxisome proliferator-activated receptor-alpha (PPARalpha) cDNA cloned from guinea-pig liver encodes a protein with similar properties to the mouse PPARalpha: implications for species differences in responses to peroxisome proliferators. Arch Toxicol 72, 169–177. Valles, E. G., Laughter, A. R., Dunn, C. S., Cannelle, S., Swanson, C. L., Cattley, R. C., and Corton, J. C. (2003). Role of the peroxisome proliferator-activated receptor alpha in responses to diisononyl phthalate. Toxicology 191, 211–225. Van Rafelghem, M. J., Mattie, D. R., Bruner, R. H., and Andersen, M. E. (1987). Pathological and hepatic ultrastructural effects of a single dose of perfluoro-n-decanoic acid in the rat, hamster, mouse, and guinea pig. Fundam Appl Toxicol 9, 522–540. Vanden Heuvel, J. P., Thompson, J. T., Frame, S. R., and Gillies, P. J. (2006). Differential activation of nuclear receptors by perfluorinated fatty acid analogs and natural fatty acids: a comparison of human, mouse, and rat peroxisome proliferator-activated receptor-alpha, -beta, and -gamma, liver X receptorbeta, and retinoid X receptor-alpha. Toxicol Sci 92, 476–489. Varanasi, U., Chu, R., Huang, Q., Castellon, R., Yeldandi, A. V., and Reddy, J. K. (1996). Identification of a peroxisome proliferator-responsive element upstream of the human peroxisomal fatty acyl coenzyme A oxidase gene. J Biol Chem 271, 2147–2155. Vu-Dac, N., Schoonjans, K., Kosykh, V., Dallongeville, J., Fruchart, J. C., Staels, B., and Auwerx, J. (1995). Fibrates increase human apolipoprotein A-II expression through activation of the peroxisome proliferator-activated receptor. J Clin Invest 96, 741–750. Vu-Dac, N., Schoonjans, K., Laine, B., Fruchart, J. C., Auwerx, J., and Staels, B. (1994). Negative regulation of the human apolipoprotein A-I promoter by fibrates can be attenuated by the interaction of the peroxisome proliferator-activated receptor with its response element. J Biol Chem 269, 31012–31018. Wada, N., Marsman, D. S., and Popp, J. A. (1992). Dose-related effects of the hepatocarcinogen, Wy-14,643, on peroxisomes and cell replication. Fundam Appl Toxicol 18, 149–154. Wahli, W., Braissant, O., and Desvergne, B. (1995). Peroxisome proliferator activated receptors: transcriptional regulators of adipogenesis, lipid metabolism and more. Chem Biol 2, 261–266. Watanabe, T., Horie, S., Yamada, J., Isaji, M., Nishigaki, T., Naito, J., and Suga, T. (1989). Species differences in the effects of bezafibrate, a hypolipidemic agent, on hepatic peroxisome-associated enzymes. Biochem Pharmacol 38, 367–371. West, D. A., James, N. H., Cosulich, S. C., Holden, P. R., Brindle, R., Rolfe, M., and Roberts, R. A. (1999). Role for tumor necrosis factor alpha receptor 1 and interleukin-1 receptor in the suppression of mouse hepatocyte apoptosis by the peroxisome proliferator nafenopin. Hepatology 30, 1417–1424. Widen, C., Gustafsson, J. A., and Wikstrom, A. C. (2003). Cytosolic glucocorticoid receptor interaction with nuclear factor-kappa B proteins in rat liver cells. Biochem J 373, 211–220. Williams, G. M., and Perrone, C. (1996). Mechanism-based risk assessment of peroxisome proliferating rodent hepatocarcinogens. Ann NY Acad Sci 804, 554–572. Wolf, D. C., Moore, T., Abbott, B. D., Rosen, M. B., Das, K. P., Zehr, R. D., Lindstrom, A. B., Strynar, M. J., and Lau, C. (2008). Comparative hepatic effects of perfluorooctanoic acid and WY 14,643 in PPAR-alpha knockout and wild-type mice. Toxicol Pathol 36, 632–639.
REFERENCES
481
Woods, C. G., Burns, A. M., Bradford, B. U., Ross, P. K., Kosyk, O., Swenberg, J. A., Cunningham, M. L., and Rusyn, I. (2007a). WY-14,643 induced cell proliferation and oxidative stress in mouse liver are independent of NADPH oxidase. Toxicol Sci 98, 366–374. Woods, C. G., Burns, A. M., Maki, A., Bradford, B. U., Cunningham, M. L., Connor, H. D., Kadiiska, M. B., Mason, R. P., Peters, J. M., and Rusyn, I. (2007b). Sustained formation of alpha-(4-pyridyl-1oxide)-N-tert-butylnitrone radical adducts in mouse liver by peroxisome proliferators is dependent upon peroxisome proliferator-activated receptor-alpha, but not NADPH oxidase. Free Radic Biol Med 42, 335–342. Woods, C. G., Kosyk, O., Bradford, B. U., Ross, P. K., Burns, A. M., Cunningham, M. L., Qu, P., Ibrahim, J. G., and Rusyn, I. (2007c). Time course investigation of PPARalpha- and Kupffer cell-dependent effects of WY-14,643 in mouse liver using microarray gene expression. Toxicol Appl Pharmacol 225, 267–277. Woodyatt, N. J., Lambe, K. G., Myers, K. A., Tugwood, J. D., and Roberts, R. A. (1999). The peroxisome proliferator (PP) response element upstream of the human acyl CoA oxidase gene is inactive among a sample human population: significance for species differences in response to PPs. Carcinogenesis 20, 369–372. Xiao, S., Anderson, S. P., Swanson, C., Bahnemann, R., Voss, K. A., Stauber, A. J., and Corton, J. C. (2006). Activation of peroxisome proliferator-activated receptor alpha enhances apoptosis in the mouse liver. Toxicol Sci 92, 368–377. Xu, H. E., and Li, Y. (2008). Ligand-dependent and -independent regulation of PPAR gamma and orphan nuclear receptors. Sci Signal 1, pe52. Yamakawa-Kobayashi, K., Ishiguro, H., Arinami, T., Miyazaki, R., and Hamaguchi, H. (2002). A Val227Ala polymorphism in the peroxisome proliferator activated receptor alpha (PPARalpha) gene is associated with variations in serum lipid levels. J Med Genet 39, 189–191. Yang, Q., Ito, S., and Gonzalez, F. J. (2007). Hepatocyte-restricted constitutive activation of PPAR alpha induces hepatoproliferation but not hepatocarcinogenesis. Carcinogenesis 28, 1171–1177. Yang, Q., Nagano, T., Shah, Y., Cheung, C., Ito, S., and Gonzalez, F. J. (2008). The PPAR alphahumanized mouse: a model to investigate species differences in liver toxicity mediated by PPAR alpha. Toxicol Sci 101, 132–139. Yeldandi, A. V., Milano, M., Subbarao, V., Reddy, J. K., and Rao, M. S. (1989). Evaluation of liver cell proliferation during ciprofibrate-induced hepatocarcinogenesis. Cancer Lett 47, 21–27. Yeldandi, A. V., Rao, M. S., and Reddy, J. K. (2000). Hydrogen peroxide generation in peroxisome proliferator-induced oncogenesis. Mutat Res 448, 159–177. Youssef, J. A., Bouziane, M., and Badr, M. Z. (2003). Age-dependent effects of nongenotoxic hepatocarcinogens on liver apoptosis in vivo. Mech Ageing Dev 124, 333–340.
CH A P TE R
18
ALPHA2U-GLOBULIN NEPHROPATHY AND CHRONIC PROGRESSIVE NEPHROPATHY AS MODES OF ACTION FOR RENAL TUBULE TUMOR INDUCTION IN RATS, AND THEIR POSSIBLE INTERACTION Edward A. Lock Gordon C. Hard
18.1.
INTRODUCTION
It is now well established that a low incidence of renal tubule tumors (RTT) can be produced by certain chemicals in male rats through a mechanism involving proximal tubule accumulation of a rat-specific protein with subsequent sustained compensatory cell regeneration. The protein, α2u-globulin (α2u-g), occurs at much lower concentrations in female rats, and not at all in mice. Hence, chemically induced RTT arising in female rats or male or female mice cannot be explained by this mechanism (Swenberg et al. 1989; Hard et al. 1993; Hard 1998; Lehman-McKeeman et al. 1998; Meek et al. 2003; Lock and Hard 2004). Male rats—and, to a lesser extent, female rats—are also predisposed to developing chronic progressive nephropathy (CPN), and this age-related, spontaneous disease entity appears to convey a slightly increased risk for development of atypical tubule hyperplasia, a preneoplastic lesion, and RTT later in life (Hard 1998, 2002; Seely et al. 2002; Lock and Hard 2004; Hard and Khan 2004). This chapter will do the following: (1) briefly discuss the mechanisms whereby these responses are observed in rat kidney, (2) provide examples of chemicals falling into these two classes of activity, (3) discuss some of the areas of potential conflict
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
482
18.2. CHEMICALS THAT INCREASE THE INCIDENCE OF RENAL TUBULE TUMORS
483
when endeavoring to classify chemicals into one or the other of these classes, (4) discuss the potential for interaction between these two pathways in leading to tumor development, and (5) discuss the relevance of these findings to humans.
18.2. CHEMICALS THAT INCREASE THE INCIDENCE OF RENAL TUBULE TUMORS IN MALE RATS BY AN α2U-GLOBULIN MODE OF ACTION Conventional male rats, but not female rats, are physiologically proteinuric because of the high urinary excretion of a low-molecular-weight protein (LMW), α2u-g. This protein (molecular weight 18–20 kilodaltons) is synthesized mainly in the liver of the male rat, where hepatic mRNA for α2u-g represents about 1% of total hepatic mRNA. Female and male rats synthesize this protein in much smaller amounts in other locations, such as secondary sex glands, salivary glands, and lachrymal glands (Lock et al. 1987; MacInnes et al. 1986; Mancini et al. 1989). In male rats, α2u-g is freely filtered at the glomerulus into the tubular lumen, with about 40% being excreted in the urine and the remainder endocytosed by cells in the P2 segment of the proximal tubule, where the protein undergoes catabolism within cellular phagolysosomes (Neuhaus et al. 1981; Lehman-McKeeman et al. 1998). Female rats excrete several hundred times less α2u-g in their urine than do males (Vandoren et al. 1983). The function of the urinary protein in male rats appears to be for territorial scent marking, with the protein acting as a scent carrier. A structurally similar protein is found in mouse urine, which has been studied extensively by scientists interested in olfaction and animal ecology (Novotny 2003; Brennan and Kendrick 2006). These rodent proteins have the ability to bind pheromones. A number of strongly odoriferous compounds have been identified bound to mouse urinary protein (MUP), such as dehydro-exo-brevicomin and 2-sec-butyl-4,5-dihydrothiazole, which potentiate aggression in male mice, and the sesquiterpenes α-farnesene and β-farnesene, which signal dominance in males. Other ligands include 2-heptanone, a fairly common metabolic product, and 6-hydroxy-6-methyl-3-heptanone. Though female rats possess the entire complement of hepatic α2u-g genes, estrogen is a very effective repressor of the expression of these genes in the liver (Roy et al. 1975). Masculinisation of female rats will increase the expression of α2u-g, but not to the same levels as in males (Roy and Neuhaus 1967). In male rats, α2u-g expression is regulated by a complex interaction of testosterone, glucocorticoids, insulin, thyroid hormone, and growth hormone, with gene expression being maximal in hormonally intact, sexually mature male rats. Hence, in immature male rats the protein is either absent or present at a low concentration, while in older rats it tends to decline as testosterone levels wane, being absent or at a very low level in 12- to 18-month-old male rats (Roy et al. 1983). A number of chemicals of diverse structure have been shown to produce a specific form of nephropathy in male rats but not in female rats or mice of either sex. The histological features of this syndrome, which has been called “α2u-globulin nephropathy” or “hyaline droplet nephropathy,” are the excessive accumulation of eosinophilic, hyaline droplets in epithelial cells of the P2 segment, an increase in
484
CHAPTER 18 ALPHA2U-GLOBULIN NEPHROPATHY AND CPN
granular casts at the junction of the outer and inner stripes of the outer medulla, and evidence of early exacerbation of CPN. After many months to 2 years of treatment, linear mineralization caused by the accumulation of calcium hydroxyapatite in the thin limbs of Henle is noted in the papilla, and there may be a low incidence of renal tubule hyperplasia and RTT, but in male rats only. By this stage, exacerbation of CPN is more pronounced (Swenberg et al. 1989; Hard et al. 1993). Among the chemicals in this class tested in 2-year carcinogenicity bioassays by the U.S. National Toxicology Program (NTP), the solvent, decalin (NTP 2005), has produced the highest incidence of RTT, at 30%. The food constituent, d-limonene (NTP 1990) also produced a relatively high incidence, at 22%. Other examples of these male rat-specific renal carcinogens include unleaded gasoline and certain jet fuels, the dry-cleaning agent tetrachloroethylene, and the insect repellant 1,4-dichlorobenzene. The initiating step in the mechanism is the noncovalent binding of the xenobiotic chemical or its metabolite to α2u-g (Lock et al. 1987; Lehman-McKeeman et al. 1989). This reversible binding interferes with the intra-renal lysosomal degradation of the protein, by prolonging the naturally very long half-life of 5–8 hr by about 30% (Lehman-McKeeman et al. 1990). The latter estimate of 30% is derived from in vitro experimentation, but serves to pinpoint what is undoubtedly a more severe problem in vivo. In the milieu of the functioning nephron where many LMW proteins are competing for lysosomal catabolism in the P2 tubules at the same time, the prolongation of α2u-g half-life is likely to be very much longer, causing lysosomal congestion. This results in the accumulation of the protein chemical complex in the P2 segment, which is visible microscopically as hyaline droplets. Under fluorescence microscopy, or following Mallory–Heidenhain staining, many of the droplets can be seen to contain large, polyangular crystalline forms (Hard 2008). Lysosomal overload of tubule cells results in single cell detachment into the lumen, and these exfoliated cells pass down the tubule in the filtrate. They become lodged at the junction where the wider lumen of the P3 tubule narrows into the thin descending limb of Henle to form granular casts, which markedly dilate the affected portion of tubule. The cells probably accumulate at this juncture because they are engorged with poorly digestible protein of crystalline nature. In turn, there is a compensatory cell proliferation in the cortex where the cell loss occurred, which persists as long as exposure to the chemical continues (Short et al. 1989), but presumably not beyond the age when liver synthesis of α2u-g has ceased. Thus, the renal tubule injury is a consequence of the perturbation of a physiological process and not due to a direct action of the chemical or a metabolite. A number of key studies have substantiated the tight correlation between the accumulation of α2u-g and the increase in hyaline droplets. The most compelling evidence for the role of this protein in the nephropathy comes from studies with genetically defective rats and transgenic mice. The male NCI Black–Reiter (NBR) rat lacks mRNA for α2u-g in the liver (Chatterjee et al. 1989) and consequently does not develop the nephropathy when challenged with chemicals such a d-limonene and lindane (Dietrich and Swenberg 1990, 1991a), propylene glycol mono-t-butyl ether (Doi et al. 2004), and decalin (NTP 2005). As mentioned earlier, wild-type mice excrete large amounts of a urinary protein (MUP), which is structurally related
18.2. CHEMICALS THAT INCREASE THE INCIDENCE OF RENAL TUBULE TUMORS
485
to α2u-g; however, they do not develop a nephropathy upon exposure to these chemicals. This appears to be due to two main factors: (1) MUP does not bind ligands such as d-limonene-1,2-epoxide, and (2) MUP is not reabsorbed by the proximal tubule (Lehman-McKeeman and Caudill 1992b). Interestingly, MUP binds 2-secbutyl-4,5-dihydrothiazole, which gives the protein and urine its odoriferous smell. This ligand can also bind to α2u-g with a Ki = 2.3 μM and is able to displace dlimonene-1,2-epoxide from its binding site (Lehman-McKeeman et al. 1998). Administration of 2-sec-butyl-4,5-dihydrothiazole to male rats, as anticipated, produces an increase in hyaline droplet formation in the P2 segment of the proximal tubules and increases the concentration of α2u-g in the kidney (Lehman-McKeeman et al. 1998). The crystal structure of α2u-g and its complex with d-limonene-1,2epoxide, at 2.9-Å resolution, has been published and the binding site for these ligands compared with the corresponding mouse protein (Chaudhuri et al. 1999). Transgenic mice modified to express α2u-g exhibit hyaline droplet formation on challenge with d-limonene, whereas wild-type mice do not (Lehman-McKeeman and Caudill 1994). Of importance to human risk assessment are studies showing that other members of the lipocalin superfamily of proteins, including human-derived α1-acid glycoprotein, rat-derived retinol-binding protein, human protein-1, and bovine β-lactoglobulin, do not bind either d-limonene-1,2-epoxide or 2,4,4trimethyl-2-pentanol, both high-affinity ligands for α2u-g. These proteins were, however, able to bind their own ligands such as (a) retinol by retinol binding protein and (b) progesterone by α1-acid glycoprotein. It therefore appears that under conditions where members of the α2u-g superfamily of proteins are known to bind to established, physiological ligands, those proteins do not interact with hyaline droplet inducing agents (Lehman-McKeeman and Caudill 1992a). The link between hyaline droplet nephropathy, renal tubule cell proliferation, and renal tumors comes from studies by Swenberg and co-workers. They demonstrated that unleaded gasoline produced a sustained increase in cell proliferation in renal cortical tubule cells throughout and beyond the period of chemical exposure (Short et al. 1989). In an initiation/promotion model using ethyl hydroxyethylnitrosamine (EHEN) as the initiating agent, they also showed that in contrast to male Fischer 344 rats, male NBR rats did not respond to d-limonene with an increase in renal hyperplasia and did not develop renal tumors beyond that of the background with EHEN alone (Dietrich and Swenberg 1991b). This series of studies has provided a mechanistic basis for the production of male-rat specific renal tumors, by a nongenotoxic mechanism that has no relevance to humans (Hard et al. 1993; Dietrich 1995; Lehman-McKeeman et al. 1998; IARC 1999). Criteria have been defined by regulatory and authoritative bodies, such as the U.S. Environmental Protection Agency (EPA) (EPA 1991) and the International Agency for Research on Cancer (IARC) (IARC 1999), to enable chemicals to be placed in this class. The essential evidence required for establishing a role for α2u-g nephropathy in renal carcinogenesis is as follows: 1. The renal tumors occur only in male rats. 2. Acute exposure to the chemical causes hyaline droplet accumulation in proximal convoluted tubules.
486
CHAPTER 18 ALPHA2U-GLOBULIN NEPHROPATHY AND CPN
3. The protein accumulating in hyaline droplets should be identified as α2u-g. 4. Hallmark histopathologic lesions, including granular casts at subchronic timepoints and linear papillary mineralization at chronic stages, should be observed. 5. There should be an absence of hyaline droplets and other typical histopathological changes in female rats and mice of both sexes. 6. The chemical should be negative in short-term tests for genotoxicity. Additional supporting evidence includes demonstration of (a) reversible binding of the chemical or a metabolite to α2u-g, (b) a sustained increase in cell proliferation in the proximal tubules (P2 segment), and (c) a dose–response relationship between hyaline droplet severity and renal tumor incidence (IARC 1999). The α2u-g hypothesis has stood the test of time, although not without challenge and debate [see Melnick (1993), Borghoff et al. (1993), Ashby (1996), Huff (1996), Melnick et al. (1997), and Dietrich (1997)]. This is due, in part, to the fact that only a small number of chemicals have been shown to fulfill all of the necessary criteria, leading some workers to seek alternative mechanisms of action (Melnick 1992; Kohn and Melnick 1999; Doi et al. 2007). Amongst the few chemicals that have been shown to sufficiently meet the required criteria are: d-limonene (Hard and Whysner 1994), 1,4-dichlorobenzene (Barter and Sherman 1999), and decalin (Dill et al. 2003). Concerns have also been raised because chemicals—for example, 1-(aminomethyl)cyclohexaneacetic acid (GABA-pentane)(Dominick et al. 1991); Stoddard solvent IIC (Doi et al. 2007); p-nitrobenzoic acid (Williams et al. 2001), and lindane (Dietrich and Swenberg 1990)—produce hyaline droplet formation representing accumulation of α2u-g in male rats without an increase in RTT. In some of these cases, the answer may lie in the fact that the severity of the hyaline droplet accumulation has not been sufficient to produce extensive cell degeneration/regeneration. Additionally, if the ligand has a rather low affinity for α2u-g, the protein–chemical complex may not be stable enough to slow lysosomal degradation and lead to cell loss. In other words, chemicals with the potential to bind to α2u-g and cause some degree of hyaline droplet accumulation will not be of equal potency, but will instead show a range of activity from weak to strong. Perhaps only the chemicals with strong activity produce sufficiently sustained regenerative conditions to lead to tumour development. Kohn and Melnick (1999) have attempted to construct a physiologically based pharmacokinetic model using datasets (sometimes incomplete) produced by scientists at the Chemical Industries Institute of Toxicology (CIIT), who worked with the α2u-g ligand 2,4,4-trimethyl-2-pentanol in male rats. The CIIT scientists had shown that male but not female rats administered trimethylpentane developed hyaline droplets and increased α2u-g in the renal P2 segment and that a metabolite of trimethylpentane (i.e., 2,4,4-trimethyl-2-pentanol) was reversibly bound to α2u-g (Charbonneau et al. 1987a; Lock et al. 1987). Subsequent studies showed that male rats administered unleaded gasoline also had 2,4,4-trimethyl-2-pentanol bound to renal α2u-g, suggesting that this metabolite of aliphatic hydrocarbons present in unleaded gasoline was the likely source leading to the renal injury (Charbonneau et al. 1987b). Kohn and Melnick (1999) found that they were unable to model the
18.2. CHEMICALS THAT INCREASE THE INCIDENCE OF RENAL TUBULE TUMORS
487
experimental findings by just reducing the rate of lysosomal proteolysis of α2u-g. However, if they built in a transient increase in the hepatic synthesis of the protein and a consequent increase in its secretion from the liver, they were able to reproduce the time-course data for blood and renal 2,4,4-trimethyl-2-pentanol concentrations and for α2u-g (Kohn and Melnick 1999). They concluded that in addition to a decreased proteolysis of the protein, some increased hepatic synthesis of the protein was required to explain the findings. This seems plausible because one might expect some feedback mechanisms to exist to switch on hepatic synthesis should plasma levels transiently drop due to removal of the chemically bound form from the circulation. These authors also suggested that increased lysosomal activity and the generation of toxic metabolites of trimethylpentane within the tubule cells may have contributed to the nephrotoxicity observed. However, there are a number of key findings that suggest that any toxic metabolite accumulation is not sufficient to cause nephrotoxicity, as illustrated by the lack of renal tubule necrosis in female rats exposed to these chemicals where the metabolism is broadly similar to that of male rats (Charbonneau et al. 1987a). Furthermore, male rats that do not express the protein, such as the NBR strain, do not show any evidence of renal injury when exposed to these chemicals. Moreover, an in vitro study with isolated proximal tubule cells exposed to high concentrations of one model compound showed no cytotoxicity (Wilke et al. 1993). A recent comparative study (Doi et al. 2007) attempted to clarify the relationship between α2u-g nephropathy and RTT development in male rats by reevaluating the data from four chemicals tested by the U.S. NTP, namely, d-limonene (NTP 1990), decalin (NTP 2005), propylene glycol monobutyl ether (PGMBE)(NTP 2004c), and Stoddard solvent IIC (SS IIC)(NTP 2004b). Doi et al. (2007) examined the reported hyaline droplet formation, α2u-g concentration in the kidney, presence of granular casts in the outer medulla, and the extent of renal tubule cell proliferation following 3 months of exposure at five dose levels. They also reexamined the kidneys of 30 animals following 2 years exposure for severity of linear mineralization of the renal papilla, CPN, renal tubule hyperplasia, and incidence of neoplastic lesions. This was done for two dose levels for d-limonene, four for decalin, and three for PGMBE and SS IIC. Because the Doi et al. (2007) results represented a sampling of the total group numbers from the original NTP studies, their incidences of hyperplastic foci and RTT do not reflect numerically the NTP 2-year bioassay data for these four chemicals. All compounds produced an increase in the renal content of α2u-g; for dlimonene, the protein was increased 2- to 2.75-fold over the dose range following 14 doses over 21 days. With decalin, the maximum increase was 4-fold; however, the control value was very low and was not in line with the previous data or with that expected in mature adult male rat kidneys. For PGMBE, again the control value was low with the maximum increase being about 2.25-fold. For SS IIC, the increase was 1.5- to 2-fold over the dose range after 3 months of exposure. The quantitation of hyaline droplet accumulation showed a dose-related increase with these chemicals with a severity band of 2 for the controls and 3–4 for the treated animals. For dlimonene, the severity band was 2–3, the lower response presumably reflecting the shorter duration of exposure. Thus d-limonene and decalin showed a more marked
488
CHAPTER 18 ALPHA2U-GLOBULIN NEPHROPATHY AND CPN
increase in α2u-g, while the response was somewhat lower with PGMBE and SS IIC. Tubule regeneration (defined as clusters of basophilic tubules in the cortex with increased nuclear density and occasional mitotic figures) was observed in the kidneys of all dosed rats showing a 6-fold increase with d-limonene, an 8-fold increase with decalin, an 11-fold increase with SS IIC, while with PGMBE the response was very flat, increasing 2- to 3-fold across the dose range. Reexamination of the data on renal cell turnover (which had been conducted by proliferating cell nuclear antigen [PCNA] immunostaining) following PGMBE and decalin exposure showed a 1.5- to 2-fold increase over control values at 13 weeks. Renal cell turnover had not been determined with d-limonene, while studies with SS IIC had used bromodeoxyuridine (BrDU) infusion with mini-pumps, but the results were not presented in the NTP report. Though there was evidence of renal tubule regeneration with all compounds, the presence of granular casts in the outer medulla was more marked in the dlimonene and decalin treated male rats than in SS IIC-treated rats, while in the PGMBE-treated rats there was only minimal evidence of casts across the dose range. Thus, all four compounds showed many of the hallmarks of α2u-g nephropathy, but the size and severity of the responses measured appeared to be most marked with d-limonene and decalin (Doi et al. 2007). After 2 years of exposure, Doi et al. (2007) noted an increase in linear mineralization of the papilla associated with all compounds in all dose groups. Again, the response was more severe with d-limonene and decalin. Similarly, the severity of CPN exacerbation was more marked with d-limonene, decalin, and PGMBE, while there was no increase in severity with SS IIC. The occurrence of renal tubule hyperplasia, lesions considered to be preneoplastic, was low with all four chemicals. Finally, the incidence of renal tubule adenoma or carcinoma was statistically significantly increased with d-limonene and decalin, with a small but not significant increase in adenomas for PGMBE. Only one adenoma was seen with SS IIC compared to no preneoplastic/neoplastic lesions in any of the four control groups. The Doi et al. (2007) reevaluation is in agreement with other studies showing that it is possible to have α2u-g nephropathy without an increased incidence of RTT (e.g., SS IIC). These authors pointed out that α2u-g accumulation at 3 months and linear mineralization of the papilla at the end of the 2-year study correlated somewhat to the tumor response, while the severity of CPN was, in general, in better agreement with the tumor response. In summary, the position still stands with regard to a link between (a) renal α2u-g accumulation coupled with tubule cell regeneration and (b) the later development of RTT only in male rats. What subsequent work has shown is that the severity of the protein accumulation and the extent of cellular repair can determine the final outcome. In particular, the severity of granular cast formation is a reflection of the degree of tubule cell loss in the cortex coupled with necessity for tubule regeneration. Likewise, the later development of linear papillary mineralisation appears to be an indicator of the amount of preceding tubule cell injury and granular cast formation. It should be noted that granular casts seem to be more easily visualized (or more numerous) in sagittal kidney sections than in transverse sections. If the carcinogenicity study provides only transverse sections, the severity of granular cast formation may be underestimated. Again, sagittal kidney sections often do not
18.3. CHEMICALS INCREASING THE INCIDENCE OF RENAL TUMORS
489
transect the papilla, in which case an absence or low incidence of linear papillary mineralisation may be misleading. Nevertheless, if these histopathological hallmarks of α2u-g nephropathy (granular casts and papillary mineralisation) are severe, they appear to predict development of RTTs. However, it is becoming increasingly clear that the severity of exacerbated CPN may also be a contributory factor.
18.3. CHEMICALS INCREASING THE INCIDENCE OF RENAL TUMORS THROUGH EXACERBATION OF SPONTANEOUS CHRONIC PROGRESSIVE NEPHROPATHY (CPN) CPN is a very common, age-related, spontaneous renal disease affecting conventional strains of rat used in safety evaluation studies—and in particular, the most commonly used strains Fischer 344 and Sprague–Dawley (Gray 1977; Peter et al. 1986; Goldstein et al. 1988; Montgomery and Seely 1990; Hard and Khan 2004). The incidence and severity of CPN can represent a confounding factor in subchronic toxicity and chronic carcinogenicity bioassays, especially if the kidney is the target organ for toxicity (Wolf and Mann 2005). CPN occurs in both sexes of rat, but, because of hormonal factors, it occurs at a higher incidence and with progressively greater severity in males than in females. It is a generally accepted dogma that the disease is due to increased glomerular permeability resulting from protein hyperfiltration. Notwithstanding, the precise basis for the disease is still poorly understood and controversial. It is known that a number of factors, primarily diet-related, can influence the incidence and severity of CPN. Reducing the protein content of the diet is protective, while increasing the protein content exacerbates the disease (Rao et al. 1993). Modification of other dietary components also ameliorates the disease, although restriction of caloric intake is more powerful than any other dietary manipulation (Bertani et al. 1989; Masoro and Yu 1989; Keenan et al. 2000). Long-term administration of androgen can make females more sensitive (Tanaka et al. 1995), indicating that it is the presence of male sex steroids that is associated with the risk of developing CPN, rather than the absence of estrogens (Baylis 1994). It has been recommended that preclinical studies using the common strains of laboratory rat should be conducted under conditions of dietary restriction (Keenan et al. 2000). The reluctance to adopt this approach has primarily been due to the lack of historical background data on lesion incidences with this regimen, compared to the extensive information on conventional studies. It is encouraging to note that the U.S. NTP, which is responsible for the testing of chemicals in the United States, converted some 15 years ago to a rodent diet that was formulated with lower protein content (14%) and higher fiber and fat, namely NTP-2000. Subsequent studies with this diet have reported a reduction in the severity of CPN, renal cortico-medullary tubule mineralization, and cardiomyopathy, without having any major effects on growth or bodyweight (Rao et al. 2001; Rao 2002). It is not the intention in this chapter to discuss the pathology of CPN in rats in detail. This has been done in several comprehensive reviews to which the reader
490
CHAPTER 18 ALPHA2U-GLOBULIN NEPHROPATHY AND CPN
is referred (Gray 1977; Barthold 1979; Hirokawa 1975; Hard and Khan 2004; Peter et al. 1986). Suffice it to say that CPN is first seen at the light microscope level as occasional, small, and discrete foci of basophilic tubules with thickened basement membranes, located in the cortex. This change is associated with eosinophilic hyaline casts in the same tubule downstream in the medulla. With progression, more tubules become affected and foci merge into areas of tubule alteration, accompanied by frank glomerulosclerosis and minor interstitial infiltration of mononuclear inflammatory cells. Thus, the histological hallmarks of the disease are basophilic tubules, thickened basement membranes, hyaline cast formation, and glomerulosclerosis (Peter et al. 1986; Hard and Khan 2004; Hard and Seely 2005). Studies examining cell proliferation in the kidney have shown that CPN is both a degenerative and regenerative disease. The regenerative aspect is supported by studies with 3 H-thymidine, BrDU or PCNA, showing that certain tubules within CPN have a high rate of cell proliferative activity. Because of the increased proliferative activity, it appears that CPN can be a weak risk factor for the spontaneous development of RTT in rats, particularly in males (Hard and Khan 2004). For 90-day toxicity and 2-year carcinogenicity studies performed in accordance with Good Laboratory Practice standards, it is necessary to grade the severity of CPN, which should treat the component lesions as one disease entity. Traditionally, this has been done on a scale of 0–4 by estimating the percentage of parenchyma affected by CPN, recognizing minimal, mild, moderate, and marked stages of CPN. One of the present authors (G. C. Hard) has used a much broader scale (Table 18.1) to (a) enable discrimination of differences between control and treatment groups at an early stage of CPN development in 90-day studies and (b) increase the statistical power of associating CPN grade with tumor incidence at later stages. This is aided by having separate grades for advanced CPN, including end-stage kidney, which signals imminent renal failure and death (Hard and Khan 2004). The schema is described in detail here for the purpose of supporting statements in the succeeding sections. It is not our intention to recommend its application, because the 0–4 grading system is quite adequate for general use.
TABLE 18.1.
The 8-Grade Scale for Semiquantitating CPN
Grade of Lesion Progression
Stage of CPN
0 1 2 3 4 5 6 7 8
Nil Minimal Mild Low-moderate Mid-moderate High-moderate Low-severe High-severe End-stage
Description No CPN lesions Lesions are focal Progressive increase in number of foci from minimal to moderate Foci too numerous to count Foci coalesce into areas Majority of outer parenchyma involved No, or almost no normal parenchyma remains
18.4. CHEMICALS INCREASING RTT INCIDENCE
491
18.4. CHEMICALS INCREASING RTT INCIDENCE THROUGH A MODE OF ACTION INVOLVING EXACERBATION OF CPN Evidence is emerging that certain chemicals can interact with CPN to increase the incidence of CPN-related proliferative lesions (Hard et al. 1997; Hard 2002). Seely et al. (2002) investigated the relationship between CPN severity and the occurrence of renal tumors in male Fischer rats from the NTP database. They found a slight but statistically significant increase in CPN severity in rats with RTT compared to agematched control males without tumors, suggestive of a positive correlation between these two states in untreated animals. This study also revealed that there had been a decrease in the mean incidence of RTT in control male rats from the U.S. NTP’s carcinogenicity bioassays, since the NTP-2000 diet was introduced. The postulate that chemically exacerbated CPN could increase the risk of RTT development was initially investigated by histopathological reevaluations of the U.S. NTP’s carcinogenicity bioassays on two chemicals, hydroquinone (Hard et al. 1997) and ethyl benzene (Hard 2002). With hydroquinone, reevaluation demonstrated that the compound caused exacerbation of CPN such that almost 40% of the high-dose males had end-stage renal disease compared to only 7% in the control males. All of the foci of atypical tubule hyperplasia and adenomas were observed to be arising within areas of CPN and all occurred in rats with either end-stage renal disease or the next highest grade of CPN (high severe). In addition, all of the tumors were adenomas with nearly half being incipient or marginal lesions (Hard et al. 1997). Ethyl benzene showed even more persuasive evidence of renal tumor association with advanced CPN. Exacerbation of end-stage kidney disease involved 68% of the high-dose males versus 12% of control males, while the high-dose females also showed a modest 8% increase compared to none in the controls. The tumors occurred in areas of the parenchyma involved in the CPN process similar to the situation seen with hydroquinone. Though three of the tumors were graded as carcinomas, a high proportion were small or marginal lesions, borderline between atypical hyperplasia and adenoma. In control rats with end-stage CPN, there was an equivalent incidence of renal proliferative lesions (atypical tubule hyperplasia and RTT) as in treated rats with end-stage kidney. Statistical analysis confirmed a highly significant correlation between (a) atypical tubule hyperplasia combined with RTT and (b) severity grade of CPN. Furthermore, when the tumor incidence data were adjusted for end-stage CPN, there was no statistically significant difference between control and treated groups of rats (Hard 2002). Because hormonal influences predispose male rats to CPN more than females, RTT increases linked to chemical exacerbation of CPN are more frequently observed in males than in females, although females can occasionally be prone to this association. It has been proposed that very specific criteria for renal tumor induction need to be met in order to conclude that an increase in RTT incidence has occurred solely through chemical exacerbation of CPN. First and foremost, the chemical must have been shown to exacerbate CPN to a very advanced grade of severity, involving high severe (grade 7) or end-stage kidney (grade 8), or grade 4 in the conventional 0–4
492
CHAPTER 18 ALPHA2U-GLOBULIN NEPHROPATHY AND CPN
grading system, in comparison to control groups in a 2-year carcinogenicity study. The tumors should occur at a very low incidence representing a marginal increase only. For the most part, the tumors should be minimal grade lesions conforming to small adenomas, or lesions borderline between atypical tubule hyperplasia and adenoma. The tumors and any precursor foci of atypical hyperplasia must be located within CPN-affected parenchyma, and they should usually be observed only toward the end of two-year studies. Very importantly, careful microscopic examination of the renal parenchyma not involved in the CPN process should reveal an absence of cytotoxicity that would suggest alternative modes of action. For example, based on many in vitro studies, it has been speculated that hydroquinone might produce RTT via a free radical mechanism (Ramachandiran et al. 2002). However, in the U.S. NTP’s 2-year bioassay of this chemical, the associated renal tumors occurred only in tissue affected by CPN, with no histopathologic evidence of cytotoxicity in renal tubules unaffected by CPN (Hard et al. 1997). Recently, all chemicals or mixtures tested in the U.S. NTP’s carcinogenicity bioassay program, which had some association with an increased incidence of RTT, were placed into categories based on available mechanistic information (Lock and Hard 2004). Ethyl benzene and hydroquinone, discussed above, were the only two chemicals placed in a category of tumor induction consequent upon exacerbated CPN. However, another category included 16 chemicals that induced RTT through an unknown mechanism, but all of these were associated with CPN exacerbation. Some of the 16 chemicals are probably candidates for inclusion in the above CPNrelated category, but the studies would require careful histological reevaluation to determine presence of the necessary criteria. Subsequently, the 2-year study of quercetin (one of the 16 chemicals mentioned above) has been reevaluated, and detailed histological examination showed that the pathology met the criteria proscribed for a CPN pathway (Hard et al. 2007). Quercetin produced a modest increase in renal tubule tumors in male rats, which correlated with CPN exacerbated to endstage (grade 8) in 20% of the high-dose male rats, versus only 2% seen in the control group at 2 years. No renal tumors were present in female rats, which correlated with a lack of exacerbation of CPN. The tumors were mainly adenomas, either borderline lesions with atypical tubule hyperplasia or of a small size. The occurrence of tumors and foci of atypical tubule hyperplasia was predominantly in rats with advanced CPN and located in tissue affected by CPN. Again, there was an absence of any cellular alterations indicative of chemical toxicity in parenchyma that was not involved in the CPN process (Hard et al. 2007). Thus, it is becoming increasingly suggestive that chemical exacerbation of CPN to severe stages (grades 7 and 8) may account for the marginally increased incidence of RTT (small adenomas or lesions borderline between atypical tubule hyperplasia and adenoma) seen in male rats in some carcinogenicity bioassays. In the U.S. NTP’s database of 2-year carcinogenicity studies, the kidney is the second most frequent site for chemically associated tumor induction in male rats, involving almost exclusively, tumors of renal tubule origin (NTP 2004a). This is partly due to the number of chemicals acting through the α2u-g mode of action, but there are an equal number of marginal renal carcinogens that potentially involve only an exacerbation of CPN (Hard et al. 2007; Lock and Hard 2004).
18.5. EXAMPLES WHERE α2U-G AND CPN MODES OF ACTION MAY BE ACTING IN CONCERT
493
18.5. EXAMPLES WHERE THE α2U-G AND EXACERBATED CPN MODES OF ACTION MAY BE ACTING IN CONCERT As discussed in preceding sections, it appears that there are two processes whereby chemicals can produce a small increased incidence of RTT in male rats. These processes are not mutually exclusive, and it is likely that a combination of indirect cytotoxicity via an α2u-g mechanism and exacerbation of CPN could both contribute to a small increase in renal tumor incidence. In fact, α2u-g nephropathy and CPN are linked from an early time-point in their development. Very recently, in a histopathological survey for renal changes in 43 of the U.S. NTP’s 90-day studies that had been conducted over a 10-year period from 1991 to 2001, it was observed that all cases of hyaline droplet nephropathy likely to be due to α2u-g-binding were associated with early exacerbation of CPN (G. S. Travlos and G. C. Hard, unpublished observations). Where the 90-day study was complemented by a 2-year study of the same chemical, CPN had become further exacerbated to advanced stages. In a 1993 survey of chemicals considered to be acting through an α2u-g mode of action at that time, Hard et al. (1993) noted that linear mineralisation of the papilla had been recorded in the male rats from the 2-year studies in all cases of chemicals suspected of acting through this mechanism. Furthermore, attention was drawn to the increase in severity of CPN by 2-years with each one of these chemicals (Hard et al. 1993). The list of chemicals included d-limonene, α-methylbenzyl alcohol, dimethyl methyl phosphonate, 1,4-dichlorobenzene, isophorone, and hexachloroethane (Hard et al. 1993). Thus, there is an intimate association of α2u-g nephropathy with exacerbating spontaneous nephropathy throughout the course of α2u-g nephropathy disease progression. When comparing the U.S. NTP’s 90-day and 2-year results for d-limonene, decalin, SS IIC, and PGMBE, Doi et al. (2007) also demonstrated that the severity of CPN appeared to be somewhat more predictive of renal tumor outcome than hallmark histological markers of a α2u-g response, such as granular cast formation and linear papillary mineralization. In fact, they suggested that α2u-g nephropathy may simply contribute to a weak background tumorigenic stimulus provided by age-related CPN. Though this may have been the case with weak responders like SS IIC and PGMBE, it is most likely that a strong α2u-g response overrides a CPN mode of action, which is a weak promoter of RTT. The renal tumor response by d-limonene and decalin must certainly represent an α2u-g mode of action because these two chemicals are the most potent of the α2u-g-binding class, inducing frequent granular casts at 3 months, severe linear papillary mineralization at 2-years, and RTT induction ranging from a 22% to 30% incidence—that is, much higher tumor incidences than seen where exacerbated CPN is the sole mode of action. One example of a chemical that induces the pathological hallmarks of α2u-g nephropathy in subchronic studies, but shows little in the way of chronic markers at 2-years, is methyl-tert-butyl ether (MTBE). Studies in rats administering MTBE (usually by vapor inhalation) at various time-points of 10 days (Prescott-Mathews et al. 1997), 14 days (Robinson et al. 1990), 28 days (Bird et al. 1997), and 90 days (Lington et al. 1997; Robinson et al. 1990) all recorded (in male rats only) an
494
CHAPTER 18 ALPHA2U-GLOBULIN NEPHROPATHY AND CPN
increase in hyaline droplet nephropathy typical of α2u-g nephropathy. One study at 10 days identified additionally a concomitant increase in proximal tubule cell proliferation, along with a significant increase in α2u-g concentration as measured by enzyme-linked immunoabsorbent assay (ELISA). There was a strong positive correlation between cell proliferation and α2u-g concentration, with MTBE exposure levels (Prescott-Mathews et al. 1997). In these various subchronic studies the α2u-g response was regarded as positive but weak. MTBE was tested for carcinogenicity in a 2-year study by vapor inhalation in Fischer 344 rats (Chun et al. 1992), although the results of this bioassay were not officially published until 1997 (Bird et al. 1997). Modest increases in RTT at the two highest doses of exposure were reported (Bird et al. 1997). Subsequently, at the request of a sponsor, one of the authors of the present chapter has reevaluated the renal histopathology from this 2-year study (Hard 2006). Though mineralization had been reported in many mid- and high-dose male rats by Chun et al (1992), on reevaluation this was found to be mostly basement membrane mineralization of proximal tubules that was a consequence of terminal disease progression and not a direct result of compound exposure. In fact, only one male rat demonstrated linear papillary mineralization typical of the long-term effects of α2u-g nephropathy. In contrast, 86% of the high-dose males and 55% of mid-dose males had end-stage CPN, compared to only 8% for the control males. Furthermore, 85% of the treatment-related adenomas occurred in rats with end-stage CPN (severity grade 8), and the remainder occurred in male rats with grade 7 CPN (i.e., very advanced CPN). So in the case of MTBE, although α2u-g nephropathy is an important early force, it appears that CPN exacerbation takes over as the main mode of action underlying later RTT development. However, a contribution from the α2u-g mechanism in the process cannot be excluded. Tert butyl alcohol (TBA), the primary metabolite of MTBE (McGregor 2006), also cannot be excluded from the short list of chemicals in which both modes of action may be operative. In the 13-week study conducted by the U.S. NTP, hyaline droplet accumulation, associated with angular crystalline structures in some droplets, was observed in male rats, but not in female rats or mice of either sex (NTP 1995). Borghoff et al. (2001) showed that the accumulating protein was immunoreactive for α2u-g (Borghoff et al. 2001). Williams and Borghoff (2001) demonstrated that TBA was capable of binding reversibly to α2u-g, thus fulfilling an important biochemical criterion for an α2u-g mode of action (Williams and Borghoff 2001). In the 2-year study, demonstration of a statistically significant increase in RTT required step-sectioning of the remaining wet tissue to produce an additional 7–8 kidney sections for examination. A tumor increase was observed in the mid-dose males, but a lower incidence in the high-dose males was probably influenced by the high early death rate inflicted mainly by end-stage CPN. Linear mineralization of the papilla was present in the majority of high-dose males and was also present in many of the mid-dose males, but to a lesser degree of severity. TBA therefore represents a case where both the α2u-g and exacerbated CPN modes of action seem likely to play a role in tumor development, with no clear distinction between the two. Methyl isobutyl ketone (MIBK), an industrial solvent, undergoes metabolism to form 4-methyl-2-pentanol, which is the sort of chemical structure that may bind to α2u-g. This suggestion is consistent with short-term inhalation studies with MIBK
18.6. RELEVANCE OF RAT A2U-GLOBULIN NEPHROPATHY AND CPN TO HUMANS
495
where hyaline droplet formation was observed in the kidneys of male rats exposed to 500 and 2000 ppm and epithelial regeneration of proximal convoluted tubules at 2000 ppm (Phillips et al. 1987). In the 2-year inhalation study, the incidence of CPN was increased at the top dose of 1800 ppm and the severity was increased at all dose levels. Demonstration of a statistically significant increase in RTT required stepsectioning, a combination of single and step sections giving a 26% increase at the top dose of 1800 ppm in male rats only (Stout et al. 2008). Linear mineralization of the papilla was present in 58% of the top dose and 44% of the mid-dose males. Thus, MIBK may also be a case where both α2u-g and exacerbated CPN modes of action play a role in tumor development. A recent short-term study with MIBK has confirmed the identity of the accumulating protein to be α2u-g (Borghoff et al. 2009).
18.6. RELEVANCE OF RAT A2U-GLOBULIN NEPHROPATHY AND CPN TO HUMANS From the discussion above, it should be clear that the induction of tumors in α2u-g nephropathy is a male-rat-specific phenomenon, which does not occur in female rats or male or female mice, or in rats where the gene for hepatic synthesis of α2u-g is absent. In addition, the chemicals that bind to α2u-g have been shown not to bind to human members of the lipocalin superfamily of proteins. Thus, it is now recognized that provided the chemical of interest meets the criteria set by the various regulatory or authoritative bodies, such as the U.S. EPA and IARC, then chemicals producing a low incidence of RTT in male rats by this mode of action should be judged as having no relevance for hazard assessment in humans. With respect to a mode of action involving exacerbation of CPN, Hard et al. (2009) have made a detailed comparison of this spontaneous rat disease with the various types of nephropathy that afflict humans. Humans are affected by several different nephropathies of known etiology, but there is no entity in humans that shows the combination or pattern of histological features that characterize CPN. In particular, CPN is not an inflammatory or vascular disease, nor does it have an immunological or autoimmune basis, and hematuria and glucosuria are not clinical findings. Relative to the various human causes of end-stage renal disease, the pattern of histological features confers uniqueness on CPN such that it can be concluded as having no strict counterpart in humans. Furthermore, no chemical that exacerbates CPN in rats is known to cause an increase in severity of any human renal disease (Hard et al. 2009). As a consequence of this reasoning, chemicals that exacerbate rat CPN in carcinogenicity bioassays, linked to a marginal but sometimes statistically significant increase in RTT incidence in treated rats, can be regarded as having no relevance for extrapolation to humans (Hard et al. 2009). Furthermore, in the view of Hard et al. (2009), because so many physiological factors influence the severity of CPN, chemically induced exacerbation of this spontaneous disease process might be regarded as an adverse event and not necessarily as an expression of chemical toxicity. In the few cases where a chemical may be judged as acting through modes of action involving both α2u-g nephropathy and exacerbation of CPN, the judgment
496
CHAPTER 18 ALPHA2U-GLOBULIN NEPHROPATHY AND CPN
should be that any RTT increase has no relevance to humans, as would be the case for each mode of action when considered separately.
REFERENCES Ashby, J. (1996). Alpha 2 mu-globulin nephropathy in white ravens. Environ Health Perspect 104, 1264–1267. Barter, J. A., and Sherman, J. H. (1999). An evaluation of the carcinogenic hazard of 1,4-dichlorobenzene based on internationally recognized criteria. Regul Toxicol Pharmacol 29, 64–79. Barthold, S. W. (1979). Chronic progressive nephropathy in aging rats. Toxicol Pathol 7, 1–6. Baylis, C. (1994). Age-dependent glomerular damage in the rat. Dissociation between glomerular injury and both glomerular hypertension and hypertrophy. Male gender as a primary risk factor. J Clin Invest 94, 1823–1829. Bertani, T., Zoja, C., Abbate, M., Rossini, M., and Remuzzi, G. (1989). Age-related nephropathy and proteinuria in rats with intact kidneys exposed to diets with different protein content. Lab Invest 60, 196–204. Bird, M. G., Burleigh-Flayer, H. D., Chun, J. S., Douglas, J. F., Kneiss, J. J., and Andrews, L. S. (1997). Oncogenicity studies of inhaled methyl tertiary-butyl ether (MTBE) in CD-1 mice and F-344 rats. J Appl Toxicol 17 (Suppl 1), S45–S55. Borghoff, S. J., Hard, G. C., Berdasco, N. M., Gingell, R., Green, S. M., and Gulledge, W. (2009). Methyl isobutyl ketone (MIBK) induction of alpha2u-globulin nephropathy in male, but not female rats. Toxicology 258, 131–138. Borghoff, S. J., Lehman-McKeeman, L. D., Short, B. G., Hard, G. C., and Swenberg, J. A. (1993). Critique of R. Melnick’s “An alternative hypothesis on the role of chemically induced protein droplet (alpha 2u-globulin) nephropathy in renal carcinogenesis”. Regul Toxicol Pharmacol 18, 357–364. Borghoff, S. J., Prescott, J. S., Janszen, D. B., Wong, B. A., and Everitt, J. I. (2001). Alpha 2u-globulin nephropathy, renal cell proliferation, and dosimetry of inhaled tert-butyl alcohol in male and female F-344 rats. Toxicol Sci 61, 176–186. Brennan, P. A., and Kendrick, K. M. (2006). Mammalian social odours: Attraction and individual recognition. Philos Trans R Soc Lond B Biol Sci 361, 2061–2078. Charbonneau, M., Lock, E. A., Strasser, J., Cox, M. G., Turner, M. J., and Bus, J. S. (1987a). 2,2,4-Trimethylpentane-induced nephrotoxicity. I. Metabolic disposition of TMP in male and female Fischer 344 rats. Toxicol Appl Pharmacol 91, 171–181. Charbonneau, M., Short, B. G., Lock, E. A., and Swenberg, J. A. (1987b). Mechanism of petroleum induced sex-specific protein droplet nephropathy and renal cell proliferation in Fischer-344 rats: Relevance to humans. Trace Subst Environ Health 21, 263–273. Chatterjee, B., Demyan, W. F., Song, C. S., Garg, B. D., and Roy, A. K. (1989). Loss of androgenic induction of alpha 2u-globulin gene family in the liver of NIH black rats. Endocrinology 125, 1385–1388. Chaudhuri, B. N., Kleywegt, G. J., Bjorkman, J., Lehman-McKeeman, L. D., Oliver, J. D., and Jones, T. A. (1999). The structures of alpha 2u-globulin and its complex with a hyaline droplet inducer. Acta Crystallogr D Biol Crystallogr 55, 753–762. Chun, J. S., Burleigh-Flayer, H. D., and Kintigh, W. J. (1992). Methyl tertiary butyl ether: Vapor inhalation oncogenicity study in Fischer 344 rats. Bushy Run Research Center, Export, Pennsylvania. Dietrich, D. R. (1995). Alpha 2u-globulin: Species- and sex-specific protein synthesis and excretion, association with chemically induced renal toxicity and neoplasia in the male rat and relevance in human cancer risk assessment. Rev Biochem Toxicol 11, 115–180. Dietrich, D. R. (1997). Doubting nongenotoxic mechanisms of renal cancer: comparing apples and oranges in the alpha2u-globulin hypothesis. Environ Health Perspect 105, 898–902. Dietrich, D. R., and Swenberg, J. A. (1990). Lindane induces nephropathy and renal accumulation of alpha 2u-globulin in male but not in female Fischer 344 rats or male NBR rats. Toxicol Lett 53, 179–181.
REFERENCES
497
Dietrich, D. R., and Swenberg, J. A. (1991a). NCI-Black-Reiter (NBR) male rats fail to develop renal disease following exposure to agents that induce alpha-2u-globulin (alpha 2u) nephropathy. Fundam Appl Toxicol 16, 749–762. Dietrich, D. R., and Swenberg, J. A. (1991b). The presence of alpha 2u-globulin is necessary for dlimonene promotion of male rat kidney tumors. Cancer Res 51, 3512–3521. Dill, J. A., Lee, K. M., Renne, R. A., Miller, R. A., Fuciarelli, A. F., Gideon, K. M., Chan, P. C., Burka, L. T., and Roycroft, J. H. (2003). Alpha 2u-globulin nephropathy and carcinogenicity following exposure to decalin (decahydronaphthalene) in F344/N rats. Toxicol Sci 72, 223–234. Doi, A. M., Hill, G., Seely, J., Hailey, J. R., Kissling, G., and Bucher, J. R. (2007). Alpha 2u-globulin nephropathy and renal tumors in national toxicology program studies. Toxicol Pathol 35, 533–540. Doi, A. M., Roycroft, J. H., Herbert, R. A., Haseman, J. K., Hailey, J. R., Chou, B. J., Dill, J. A., Grumbein, S. L., Miller, R. A., Renne, R. A., and Bucher, J. R. (2004). Inhalation toxicology and carcinogenesis studies of propylene glycol mono-t-butyl ether in rats and mice. Toxicology 199, 1–22. Dominick, M. A., Robertson, D. G., Bleavins, M. R., Sigler, R. E., Bobrowski, W. F., and Gough, A. W. (1991). Alpha 2u-globulin nephropathy without nephrocarcinogenesis in male Wistar rats administered 1-(aminomethyl)cyclohexaneacetic acid. Toxicol Appl Pharmacol 111, 375–387. EPA (1991). Alpha2u-globulin: Association with chemically induced renal toxicity and neoplasia in the male rat. EPA/625/3-91/091F, 1–118. Goldstein, R. S., Tarloff, J. B., and Hook, J. B. (1988). Age-related nephropathy in laboratory rats. FASEB J 2, 2241–2251. Gray, J. E. (1977). Chronic progressive nephrosis in the albino rat. CRC Crit Rev Toxicol 5, 115–144. Hard, G. C. (1998). Mechanisms of chemically induced renal carcinogenesis in the laboratory rodent. Toxicol Pathol 26, 104–112. Hard, G. C. (2002). Significance of the renal effects of ethyl benzene in rodents for assessing human carcinogenic risk. Toxicol Sci 69, 30–41. Hard, G. C. (2006). Expert review of kidney histopathology in the 2-year carcinogenicity study of methyl tert butyl ether (MTBE) administered to Fischer 344 rats by vapor inhalation. Report prepared for Lyondell Chemical Company, October 10th, Houston, Texas. Hard, G. C. (2008). Some aids to histological recognition of hyaline droplet nephropathy in ninety-day toxicity studies. Toxicol Pathol 36, 1014–1017. Hard, G. C., Johnson, K. J., and Cohen, S. M. (2009). A comparison of rat chronic progressive nephropathy with human renal disease—Implications for human risk assessment. Crit Rev Toxicol 39, 332–346. Hard, G. C., and Khan, K. N. (2004). A contemporary overview of chronic progressive nephropathy in the laboratory rat, and its significance for human risk assessment. Toxicol Pathol 32, 171–180. Hard, G. C., Rodgers, I. S., Baetcke, K. P., Richards, W. L., McGaughy, R. E., and Valcovic, L. R. (1993). Hazard evaluation of chemicals that cause accumulation of alpha 2u-globulin, hyaline droplet nephropathy, and tubule neoplasia in the kidneys of male rats. Environ Health Perspect 99, 313–349. Hard, G. C., and Seely, J. C. (2005). Recommendations for the interpretation of renal tubule proliferative lesions occurring in rat kidneys with advanced chronic progressive nephropathy (CPN). Toxicol Pathol 33, 641–649. Hard, G. C., Seely, J. C., Betz, L. J., and Hayashi, S. M. (2007). Re-evaluation of the kidney tumors and renal histopathology occurring in a 2-year rat carcinogenicity bioassay of quercetin. Food Chem Toxicol 45, 600–608. Hard, G. C., and Whysner, J. (1994). Risk assessment of d-limonene: An example of male rat-specific renal tumorigens. Crit Rev Toxicol 24, 231–254. Hard, G. C., Whysner, J., English, J. C., Zang, E., and Williams, G. M. (1997). Relationship of hydroquinone-associated rat renal tumors with spontaneous chronic progressive nephropathy. Toxicol Pathol 25, 132–143. Hirokawa, K. (1975). Characterization of age-associated kidney disease in Wistar rats. Mech Ageing Dev 4, 301–316. Huff, J. (1996). Response: Alpha-2-mu-globulin nephropathy, posed mechanisms, and white ravens. Environ Health Perspect 104, 1264–1267.
498
CHAPTER 18 ALPHA2U-GLOBULIN NEPHROPATHY AND CPN
IARC (1999). Species differences in thyroid, kidney and urinary bladder carcinogenesis. IARC Sci Publ 147, 1–225. Keenan, K. P., Coleman, J. B., McCoy, C. L., Hoe, C. M., Soper, K. A., and Laroque, P. (2000). Chronic nephropathy in ad libitum overfed Sprague–Dawley rats and its early attenuation by increasing degrees of dietary (caloric) restriction to control growth. Toxicol Pathol 28, 788–798. Kohn, M. C., and Melnick, R. L. (1999). A physiological model for ligand-induced accumulation of alpha 2u globulin in male rat kidney: Roles of protein synthesis and lysosomal degradation in the renal dosimetry of 2,4,4-trimethyl-2-pentanol. Toxicology 136, 89–105. Lehman-McKeeman, L. D., and Caudill, D. (1992a). Alpha 2u-globulin is the only member of the lipocalin protein superfamily that binds to hyaline droplet inducing agents. Toxicol Appl Pharmacol 116, 170–176. Lehman-McKeeman, L. D., and Caudill, D. (1992b). Biochemical basis for mouse resistance to hyaline droplet nephropathy: Lack of relevance of the alpha 2u-globulin protein superfamily in this male ratspecific syndrome. Toxicol Appl Pharmacol 112, 214–221. Lehman-McKeeman, L. D., and Caudill, D. (1994). d-Limonene induced hyaline droplet nephropathy in alpha 2u-globulin transgenic mice. Fundam Appl Toxicol 23, 562–568. Lehman-McKeeman, L. D., Caudill, D., Rodriguez, P. A., and Eddy, C. (1998). 2-sec-Butyl-4,5dihydrothiazole is a ligand for mouse urinary protein and rat alpha 2u-globulin: Physiological and toxicological relevance. Toxicol Appl Pharmacol 149, 32–40. Lehman-McKeeman, L. D., Rivera-Torres, M. I., and Caudill, D. (1990). Lysosomal degradation of alpha 2u-globulin and alpha 2u-globulin-xenobiotic conjugates. Toxicol Appl Pharmacol 103, 539–548. Lehman-McKeeman, L. D., Rodriguez, P. A., Takigiku, R., Caudill, D., and Fey, M. L. (1989). dLimonene-induced male rat-specific nephrotoxicity: evaluation of the association between d-limonene and alpha 2u-globulin. Toxicol Appl Pharmacol 99, 250–259. Lington, A. W., Dodd, D. E., Ridlon, S. A., Douglas, J. F., Kneiss, J. J., and Andrews, L. S. (1997). Evaluation of 13-week inhalation toxicity study on methyl t-butyl ether (MTBE) in Fischer 344 rats. J Appl Toxicol 17 (Suppl 1), S37–S44. Lock, E. A., Charbonneau, M., Strasser, J., Swenberg, J. A., and Bus, J. S. (1987). 2,2,4-Trimethylpentaneinduced nephrotoxicity. II. The reversible binding of a TMP metabolite to a renal protein fraction containing alpha 2u-globulin. Toxicol Appl Pharmacol 91, 182–192. Lock, E. A., and Hard, G. C. (2004). Chemically induced renal tubule tumors in the laboratory rat and mouse: Review of the NCI/NTP database and categorization of renal carcinogens based on mechanistic information. Crit Rev Toxicol 34, 211–299. MacInnes, J. I., Nozik, E. S., and Kurtz, D. T. (1986). Tissue-specific expression of the rat alpha 2u globulin gene family. Mol Cell Biol 6, 3563–3567. Mancini, M. A., Majumdar, D., Chatterjee, B., and Roy, A. K. (1989). Alpha 2u-globulin in modified sebaceous glands with pheromonal functions: Localization of the protein and its mRNA in preputial, meibomian, and perianal glands. J Histochem Cytochem 37, 149–157. Masoro, E. J., and Yu, B. P. (1989). Diet and nephropathy. Lab Invest 60, 165–167. McGregor, D. (2006). Methyl tertiary-butyl ether: Studies for potential human health hazards. Crit Rev Toxicol 36, 319–358. Meek, M. E., Bucher, J. R., Cohen, S. M., Dellarco, V., Hill, R. N., Lehman-McKeeman, L. D., Longfellow, D. G., Pastoor, T., Seed, J., and Patton, D. E. (2003). A framework for human relevance analysis of information on carcinogenic modes of action. Crit Rev Toxicol 33, 591–653. Melnick, R. L. (1992). An alternative hypothesis on the role of chemically induced protein droplet (alpha 2u-globulin) nephropathy in renal carcinogenesis. Regul Toxicol Pharmacol 16, 111–125. Melnick, R. L. (1993). Critique does not validate assumptions in the model on alpha 2u-globulin and renal carcinogenesis. Regul Toxicol Pharmacol 18, 365–368. Melnick, R. L., Kohn, M. C., and Huff, J. (1997). Weight of evidence versus weight of speculation to evaluate the alpha2u-globulin hypothesis. Environ Health Perspect 105, 904–906. Montgomery, C. A., and Seely, J. C. (1990). Kidney. In Pathology of the Fischer Rat: Reference and Atlas, Boorman, G. A., Eustis, S. L., Elwell, M. R., and Montogomery, C., eds., Academic Press, San Diego, pp. 127–152. Neuhaus, O. W., Flory, W., Biswas, N., and Hollerman, C. E. (1981). Urinary excretion of alpha 2 muglobulin and albumin by adult male rats following treatment with nephrotoxic agents. Nephron 28, 133–140.
REFERENCES
499
Novotny, M. V. (2003). Pheromones, binding proteins and receptor responses in rodents. Biochem Soc Trans 31, 117–122. NTP (1990). Toxicology and carcinogenesis studies of d-limonene (CAS No. 5989-27-5) in F344/N rats and B6C3F1 mice (Gavage Studies). NTP TR 347; NIH Publication No. 90-2802, 1–165. NTP (1995). NTP technical report on the toxicology and carcinogenesis studies of t-butyl alcohol (CAS No. 75-65-0) in F344/N rats and B6C3F1 mice (Drinking Water Studies). NTP TR 436; NIH Publication No. 95-3167, 1–313. NTP (2004a). Chemicals associated with site-specific tumor induction in kidney tubular cell. Accessed from http://ntp-server.niehs.nih.gov/htdocs/sites/psite_cnt.html on 18 May 2004. NTP (2004b). NTP technical report on the toxicology and carcinogenesis of Stoddard Solvent IIC (CAS No. 64742-88-7) in F344/N rats and B6C3F1 mice (Inhalation Studies). NTP TR 519; NIH Publication No. 04-4453, 1–274. NTP (2004c). NTP technical report on the toxicology and carcinogenesis studies of propylene glycol mono-t-butyl ether (CAS No. 57018-52-7) in F344/N rats and B6C3F1 mice and a toxicology study of propylene glycol mono-t-butyl ether in male NBR rats (Inhalation Studies). NTP TR 515; NIH Publication No. 04-4449, 1–306. NTP (2005). NTP technical report on the toxicology and carcinogenesis studies of decalin (CAS No. 91-17-8) in F344/N rats and B6C3F1 mice and a toxicology study of decalin in male NBR rats (Inhalation Studies). NTP TR 513; NIH Publication No. 05-4447, 1–316. Peter, C. P., Burek, J. D., and van Zwieten, M. J. (1986). Spontaneous nephropathies in rats. Toxicol Pathol 14, 91–100. Phillips, R. D., Moran, E. J., Dodd, D. E., Fowler, E. H., Kary, C. D., and O’Donoghue, J. (1987). A 14week vapor inhalation toxicity study of methyl isobutyl ketone. Fundam Appl Toxicol 9, 380–388. Prescott-Mathews, J. S., Wolf, D. C., Wong, B. A., and Borghoff, S. J. (1997). Methyl tert-butyl ether causes alpha2u-globulin nephropathy and enhanced renal cell proliferation in male Fischer-344 rats. Toxicol Appl Pharmacol 143, 301–314. Ramachandiran, S., Huang, Q., Dong, J., Lau, S. S., and Monks, T. J. (2002). Mitogen-activated protein kinases contribute to reactive oxygen species-induced cell death in renal proximal tubule epithelial cells. Chem Res Toxicol 15, 1635–1642. Rao, G. N. (2002). Diet and kidney diseases in rats. Toxicol Pathol 30, 651–656. Rao, G. N., Edmondson, J., and Elwell, M. R. (1993). Influence of dietary protein concentration on severity of nephropathy in Fischer-344 (F-344/N) rats. Toxicol Pathol 21, 353–361. Rao, G. N., Morris, R. W., and Seely, J. C. (2001). Beneficial effects of NTP-2000 diet on growth, survival, and kidney and heart diseases of Fischer 344 rats in chronic studies. Toxicol Sci 63, 245–255. Robinson, M., Bruner, R. H., and Olson, G. R. (1990). Fourteen- and ninety-day oral toxicity studies of methyl tertiary-butyl ether in Sprague–Dawley rats. J Am Coll Toxicol 9, 525–540. Roy, A. K., McMinn, D. M., and Biswas, N. M. (1975). Estrogenic inhibition of the hepatic synthesis of alpha2u globulin in the rat. Endocrinology 97, 1501–1508. Roy, A. K., Nath, T. S., Motwani, N. M., and Chatterjee, B. (1983). Age-dependent regulation of the polymorphic forms of alpha 2u-globulin. J Biol Chem 258, 10123–10127. Roy, A. K., and Neuhaus, O. W. (1967). Androgenic control of a sex-dependent protein in the rat. Nature 214, 618–620. Seely, J. C., Haseman, J. K., Nyska, A., Wolf, D. C., Everitt, J. I., and Hailey, J. R. (2002). The effect of chronic progressive nephropathy on the incidence of renal tubule cell neoplasms in control male F344 rats. Toxicol Pathol 30, 681–686. Short, B. G., Burnett, V. L., and Swenberg, J. A. (1989). Elevated proliferation of proximal tubule cells and localization of accumulated alpha 2u-globulin in F344 rats during chronic exposure to unleaded gasoline or 2,2,4-trimethylpentane. Toxicol Appl Pharmacol 101, 414–431. Stout, M. D., Herbert, R. A., Kissling, G. E., Suarez, F., Roycroft, J. H., Chhabra, R. S., and Bucher, J. R. (2008). Toxicity and carcinogenicity of methyl isobutyl ketone in F344N rats and B6C3F1 mice following 2-year inhalation exposure. Toxicology 244, 209–219. Swenberg, J. A., Short, B., Borghoff, S., Strasser, J., and Charbonneau, M. (1989). The comparative pathobiology of alpha 2u-globulin nephropathy. Toxicol Appl Pharmacol 97, 35–46. Tanaka, A., Kyokuwa, M., Mori, T., and Kawashima, S. (1995). Acceleration of renal dysfunction with ageing by the use of androgen in Wistar/Tw rats. In Vivo 9, 495–502.
500
CHAPTER 18 ALPHA2U-GLOBULIN NEPHROPATHY AND CPN
Vandoren, G., Mertens, B., Heyns, W., Van Baelen, H., Rombauts, W., and Verhoeven, G. (1983). Different forms of alpha 2u-globulin in male and female rat urine. Eur J Biochem 134, 175–181. Wilke, A. V., Dorman, D. C., and Borghoff, S. J. (1993). Alpha2u-globulin (alpha2u) and 2,4,4-trimethyl2-pentanol (TMP-2-OH) toxicity in proximal tubule cells. The Toxicologist 13, 439. Williams, K. D., Dunnick, J., Horton, J., Greenwell, A., Eldridge, S. R., Elwell, M., and Sills, R. C. (2001). P-Nitrobenzoic acid alpha2u nephropathy in 13-week studies is not associated with renal carcinogenesis in 2-year feed studies. Toxicol Pathol 29, 507–513. Williams, T. M., and Borghoff, S. J. (2001). Characterization of tert-butyl alcohol binding to alpha2uglobulin in F-344 rats. Toxicol Sci 62, 228–235. Wolf, D. C., and Mann, P. C. (2005). Confounders in interpreting pathology for safety and risk assessment. Toxicol Appl Pharmacol 202, 302–308.
CH A P TE R
19
URINARY TRACT CALCULI AND BLADDER TUMORS Samuel M. Cohen Lora L. Arnold Shugo Suzuki
19.1.
INTRODUCTION
Urinary bladder cancer in humans has been associated with exposure to chemicals since the observation by Rehn in 1895 of an increased incidence of bladder cancer in aniline dye workers in Germany (Cohen 1998; Cohen et al. 2000; Johansson and Cohen 1997). As a result of extensive research during the past century, numerous chemicals and mixtures have been identified as causative factors for human bladder cancer, most notably cigarette smoking, which accounts for approximately one-half of the bladder cancer cases in the United States (Cohen et al. 2000; Johansson and Cohen 1997). Most of the chemicals that have been identified as human bladder carcinogens documented by the International Agency for Research on Cancer (IARC) are DNA reactive chemicals, many of which are aromatic amines, such as 2-naphthylamine, 4-aminobiphenyl, and benzidine (IARC 1987, 2008). Aromatic amines are also believed to be the major component of cigarette smoke associated with the high incidences of bladder cancer (Cohen et al. 2000). Most of the chemicals that are known to be human bladder carcinogens also induce increased incidences of bladder cancer in rodent models, although many aromatic amines increase tumors of other organs in addition to the bladder—and occasionally organs other than the bladder, such as the liver and mammary gland (IARC 1987, 2008). Rodents, most commonly rats and mice, have been extensively utilized during the past 50 years as model systems to screen for chemical carcinogens in general, including those with potential effects on the urinary bladder. As a consequence of this testing, numerous chemicals have been identified as carcinogens toward the urinary bladder in rats and/or mice, including several DNA reactive chemicals, as well as several chemicals that are not metabolically activated to reactive electrophiles and thus do not form DNA adducts (Cohen 1998; Gold et al. 2001; NTP 2008). These chemicals are classified as non-DNA-reactive (Cohen 1998).
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
501
502
CHAPTER 19 URINARY TRACT CALCULI AND BLADDER TUMORS
Chemicals induce bladder cancer in animal models and in humans by a limited number of modes of action (Cohen 1998). For human carcinogens, the most common mode of action is DNA reactivity leading to adduct formation, mutations, and ultimately cancer. However, in animal models, numerous bladder carcinogens are nonDNA-reactive. These chemicals are able to increase the incidence of bladder cancer in these animal models by increasing cell proliferation. Increased cell proliferation of the urothelium can be induced either by direct stimulation of cell proliferation (mitogenesis) or by inducing cytotoxicity with consequent regenerative proliferation and hyperplasia. Cytotoxicity can be induced either by the formation of a reactive, cytotoxic metabolite of the administered chemical (occasionally the parent chemical itself), which is excreted in the urine at sufficiently high concentrations to produce cytotoxicity, or by the formation of urinary solids which act as corrosive, cytotoxic agents to the urothelium, again with consequent regenerative proliferation. There has also been a suggestion that extreme modifications of urinary composition, such as urinary pH, volume, or osmolality, can also produce a cytotoxic effect. However, these extremes in urinary composition frequently are associated with the formation of urinary solids, so it remains unclear whether the urinary compositional changes themselves are actually the cytotoxic stimulus. For all of these non-DNA-reactive chemicals, the evidence strongly supports a dose–response relationship that is nonlinear and likely to involve a threshold (Cohen 1998). A nonlinear dose response and the presence of a true threshold, based on biologic and physical chemical mechanistic considerations, are best demonstrated using examples of cytotoxicity involving the formation of urinary solids (Cohen et al. 2002; IARC 1999; RBCWG 1995). Urinary solids only form when there is sufficient material in the urine to exceed the level of solubility for that agent based on its chemical and physical properties, including its interaction with other components in the urine. This is the clearest example of a true threshold, based on mechanistic considerations. This is of considerable importance since many of the substances that form urinary solids in the urine, whether in rodents or in humans, involve substances that are essential for life, such as calcium, phosphate, and cysteine, in addition to other natural and synthetic chemicals (Table 19.1). As long as the exposure does not lead to a urinary concentration that exceeds the solubility of the chemical, the substance remains in solution and does not pose a toxicological risk. Numerous chemicals have been identified in rodent models which lead to the production of urinary solids, especially in rats. Urinary solids also occur in humans (McPherson et al. 2006). In rodents, urinary solids are frequently associated not only with cytotoxicity and regenerative proliferation, but also with an increased incidence of bladder tumors when the substance is administered in a standard two-year bioassay or longer (Cohen et al. 2002; IARC 1999; RBCWG 1995). The possible risk to humans of such substances is the subject of this chapter.
19.2. DIRECT AND INDIRECT FORMATION OF URINARY SOLIDS There are essentially three types of urinary solids: precipitate (amorphous material), crystals, and calculi. The distinction between crystals and calculi, especially in
19.2. DIRECT AND INDIRECT FORMATION OF URINARY SOLIDS
TABLE 19.1.
503
Substances Leading to the Formation of Urinary Solids
Endogenous Substances (Normal Urinary Constituents) Calcium carbonate Calcium oxalate Calcium phosphate Magnesium ammonium phosphate Urates and uric acid Xanthine Cystine Hippuric acid Tyrosine Uracil Bilirubin Cholesterol Hematin Hemosiderin Vitamin C
Synthetic Chemicals and Pharmaceuticals Sulfonamides Carbonic anhydrase inhibitors HIV protease inhibitors Ampicillin and amoxicillin Radiographic media (meglumine diatrizoate) β3-Adrenoceptor agonists Glafenic acid Terephthalic acid Dimethyl terephthalate Biphenyl Melamine Fosetyl-A1 Sulfosulfuron
rodent urine, is somewhat artificial, based on size (Dominick et al. 2006). Amorphous precipitate and crystals of various types are present normally in the urine of most mammalian species, including humans (Cohen 1998; Cohen et al. 2002; Dominick et al. 2006; IARC 1999; McPherson et al. 2006; RBCWG 1995). In rodents, the most common type of crystalline material is magnesium ammonium phosphate (struvite) crystals (Cohen 1995, 1998). These are also the most common crystals in human urine, but other types of crystals commonly occur in human urine, such as calcium oxalate and calcium phosphate, in addition to several other crystals formed from substances that are normally present in the urine (McPherson et al. 2006). Formation of urinary solids, either (a) qualitatively different from those normally seen in the urine or (b) with an increase in the number or size of solids normally detected in the urine, can be produced either directly by the administered substance or indirectly (Cohen 1998) (Figure 19.1). Direct formation is defined as formation of the urinary solid by the administered substance or by one of its metabolites. An example is the formation of either (a) melamine-containing crystals and calculi in rats administered melamine (Meek et al. 2003) or (b) uracil-containing crystals and calculi in rats (Shirai et al. 1989) or mice (Sakata et al. 1988) administered uracil at high levels of the diet. Indirect formation of urinary solids occurs when administration of a substance, whether natural or synthetic, to the animal causes an increased concentration of substances that are normally present in the urine, resulting in concentrations in the urine in excess of the solubility of the naturally occurring substance and also leading to formation of urinary solids (Cohen 1998). Examples include (a) the administration of extremely high doses of sodium salts (such as saccharin, ascorbate, chloride, or bicarbonate) leading to formation of calcium phosphate-containing amorphous precipitate in the urine (Cohen 1998; IARC 1999) or (b) the administration of PPARγ or dual α/γ agonists leading to formation of calcium phosphate- and calcium
504
A
CHAPTER 19 URINARY TRACT CALCULI AND BLADDER TUMORS
Chemical
High urinary concentration(s)
Urinary solids
Metabolite
B
Chemical
Altered urinary composition
High concentration of normal urinary constituent(s)
Alteration in endogenous intermediary metabolism
Inherited disorder or surgical procedure
Figure 19.1. Alternative processes for formation of urinary tract solids. (A) Direct formation of solids composed of chronically administered parent chemical or metabolite(s). (B) Indirect formation of urinary tract solids composed of chemicals normally present in the urine. Formation occurs because of significant alterations in urine composition secondary to altered urinary physiology, alteration of normal intermediary metabolism, or secondary to an inherited metabolic disorder (e.g., gout, oxalosis) or surgical procedure (e.g., porta caval shunt).
oxalate-containing crystals and calculi (Dominick et al. 2006). Formation of urinary solids from substances normally present in the urine can be produced not only by administering substances exogenously, but also by changes in normal intermediary metabolism. For example, hyperparathyroidism leads to increased urinary calcium levels and the formation of calcium-containing crystals and calculi (McPherson et al. 2006). Similarly, in patients with gout, associated with hyperuricemia, urate crystals increasingly form in the urine; and if present at sufficiently high levels, urate calculi can form (McPherson et al. 2006). A model of this in rodents involves the surgical formation of a portacaval shunt that produces marked alterations in uric acid metabolism, eventually leading to the formation of urate-containing crystals and calculi (Clayson et al. 1995). Numerous other examples of endogenous alterations in metabolism leading to formation of urinary tract calculi have been identified. In many of the rodent models involving formation of urinary solids, there is associated cytotoxicity, regenerative proliferation, and ultimately the induction of tumors of the urothelium, usually of the urinary bladder but occasionally of the kidney pelvis or ureters (Shirai et al. 1989). Cytotoxicity and regenerative
19.3. URINARY FACTORS INFLUENCING THE FORMATION OF URINARY SOLIDS
505
proliferation is dependent on several factors, including the amount and size of urinary solid and its surface features (rough-surfaced calculi are more abrasive than smooth calculi) (Clayson 1974; Clayson and Cooper 1970; Clayson et al. 1995). Examples have been identified where the presence of a crystal formed from an administered substance, such as sulfosulfuron, is not cytotoxic, but the calculi composed of the same substance are cytotoxic (Arnold et al. 2001). The relationship of urinary solids to bladder carcinogenesis is not associated with the actual chemical composition of the solid, but is reliant entirely on the physical properties of the solid (Clayson 1974; Clayson et al. 1995; DeSesso 1989; IARC 1999; RBCWG 1995). This has been demonstrated by implanting pellets of various substances, including paraffin wax, cholesterol, glass, stainless steel, and wood, directly into the lumen of the mouse or rat bladder (Bryan 1969; Clayson and Cooper 1970; DeSesso 1989). In a classic experiment reported by Jull (1979) involving implantation of paraffin wax pellets into the mouse bladder, he was able to demonstrate that the incidence of bladder tumors after one year was approximately 10% whereas by two years it was approximately 50% (Jull 1979). This was a critical study, because the method involving pellet implantation had been used by numerous investigators until that time as a way of directly exposing the urothelium to known substances by incorporating the substance into the pellet and then implanting the pellet into the bladder to determine a possible direct effect of the substance on the urothelium without involving metabolism (Bryan 1969; Clayson 1974; Clayson and Cooper 1970; DeSesso 1989). The experiment by Jull (1979), as well as research by others, proved that the effects on the rodent urothelium were due to the pellet itself rather than the chemical, although the speed with which the chemical could be leached from the pellet greatly affected the surface characteristics of the pellet and thereby affected the incidence of tumors being induced by the pellet (DeSesso 1989).
19.3. URINARY FACTORS INFLUENCING THE FORMATION OF URINARY SOLIDS The critical factor leading to the formation of urinary solids is the solubility of the substance in the urine. However, this can be influenced by several factors normally present in the urine which can vary considerably, not only between species but also within a given animal or human based on variations in food and water consumption, type of diet, hydration, and alterations in metabolism (Cohen 1995, 1998; McPherson et al. 2006; Pearle and Lotan 2007). The composition of urine varies considerably with diet and drinking, which results in a marked diurnal variation in the composition of the urine (Fisher et al. 1989). This has been demonstrated in rats for several parameters, but is essentially true for all and reflects the role of the urine as an excretory pathway. The relationship of this diurnal variation to food and water consumption can be demonstrated by reversing the light–dark hours in an animal room. Rodents are nocturnal in their eating pattern, so the urine composition varies based on the light and dark variations of their environment. Variations in the urine of humans also are dependent on food
506
CHAPTER 19 URINARY TRACT CALCULI AND BLADDER TUMORS
and water consumption (Pearle and Lotan 2007). This diurnal variation in the composition of urine also results in significant biological variation of the urothelium itself on a diurnal basis, including the mitotic rate (Tiltman and Friedell 1972). Several major factors can affect the solubility of various solutes, depending on whether they are ionic or not (Cohen 1995, 1998; Pearle and Lotan 2007). A major determinant is urinary pH. For example, in rodents it appears that a urinary pH of 6.5 or greater is necessary for precipitation of calcium salts to occur (Cohen 1998; IARC 1999). In contrast, urate-containing crystals form at low, acidic pH (McPherson et al. 2006; Pearle and Lotan 2007). The formation of these crystals can be prevented by treatments that alter urinary pH. For example, in rats, urinary pH can be increased by administration of a variety of carbonates or bicarbonates, or it can be lowered by administration of ammonium chloride (Cohen 1995, 1998; Cohen et al. 2002; Dominick et al. 2006; IARC 1999; McPherson et al. 2006; RBCWG 1995). Various rodent diets also affect the urinary pH. For example, Purina 5002 diet results in a nearly neutral urinary pH whereas Altromin 1531 produces markedly alkaline urine (Cohen et al. 1994). In contrast, AIN-76A a semi-synthetic diet, which is dependent on the presence of casein as the protein source, produces markedly acidic urine (Okamura et al. 1991). By converting the protein source to albumin, the urinary pH can be increased. These factors also are pertinent to the human situation, where individuals with calcium-containing calculi are given various treatments, such as the consumption of cranberry juice, to acidify the urine (Cohen 1998; Pearle and Lotan 2007), whereas the treatment of gout, prior to the availability of allopurinol, usually was associated with administration of substances that alkalinized the urine (McPherson et al. 2006; Pearle and Lotan 2007). Obviously, a major determinant of the solubility of the substances is their concentration in the urine. If the substances contain calcium, magnesium, phosphate, or other ions, the concentration of these ions in the urine obviously will contribute most significantly to the potential for crystallization and precipitation. However, in addition to pH and concentration, several other factors can contribute to the potential of the substances to actually precipitate in the urine (Cohen 1995; Cohen et al. 2002; Dominick et al. 2006; IARC 1999; McPherson et al. 2006; Pearle and Lotan 2007; RBCWG 1995). Many of these salts are present in the urine as supersaturated solutions, predominantly because of the presence of substances in the urine which act to keep several of these ions in solution rather than precipitate. For example, citrate is a major chelating substance in the urine for calcium and magnesium salts, particularly calcium (Dominick et al. 2006; Pearle and Lotan 2007). Lowering urinary citrate levels can contribute to the potential for precipitation of calciumcontaining salts, without altering the actual levels of the calcium ion. Also, there are several proteins in the urine which bind to calcium and keep it in solution. This includes albumin, a protein referred to as Tamm–Horsfall protein (Marengo et al. 2002; Pearle and Lotan 2007), and others. Thus, merely measuring the concentration of the various substances in the urine that constitute the urinary solid does not represent the entire picture of the potential for crystallization of any given substance. Urine is a complex mixture, and the potential for solubilization versus crystallization is extremely complex. All of these factors need to be taken into account when assessing the potential for the
19.4. COLLECTION OF URINE FOR DETECTION OF URINARY SOLIDS
507
formation of urinary solids as a potential mode of action for the relationship of bladder urothelial proliferation and tumorgenicity.
19.4. COLLECTION OF URINE FOR DETECTION OF URINARY SOLIDS The method used for the collection of urine to detect urinary solids is particularly sensitive to a variety of artifacts and variations in treatment (Cohen et al. 2007). Most of all, it is essential that the animals not be fasted or go without water during the period of collection of urine. Since the excretion of the substances that are included in formation of the urinary solids is dependent on their consumption, fasting the animals changes the urine composition considerably and can lead to a condition in which the solids are no longer formed. Furthermore, urinary solids can be rapidly excreted in the urine and are not retained; so if they are not being constantly formed anew, they will not be detected. This includes urinary tract calculi. Some calculi will be small enough that they will be excreted in the urine, or dissolve with the lowering of the concentration of the solute itself. Furthermore, many of these calculi are actually quite soluble in urine, such as uracil, and rapidly solubilize in the urine. It is essential that urine collection also be performed taking into account the propensity for artifactual changes, as well as, the potential for solubilization or further crystallization of the urine while standing (Cohen et al. 2007). Thus, collecting urine overnight on ice greatly increases the propensity of solids to form in the urine with the lowering of temperature. Also, if the urine is collected and stored in the refrigerator before examination, urinary solids can be artifactually formed. Allowing time for urinary solids to sit in the urine before being examined can lead to their solubilization. The same caution pertains to urine or tissues placed in aqueous fixatives. Artifactual crystallization or solubilization can occur depending on the specific circumstances of the process. Because of the potential for these artifacts, we strongly recommend the collection of fresh void specimens from animals with immediate examination of the urinary sediment for possible presence of urinary solids. Examination of the urinary sediment by light microscopy can detect many of the usual types of crystals and calculi. However, passing urine through a Millipore filter and then examining it by scanning electron microscopy is a much more sensitive method (Cohen et al. 2007). With attached energy dispersive X-ray spectroscopy, the elemental composition of the urinary solid can also be determined. Likewise, collection of fresh void urine specimens for examination of the urine for chemical composition is also strongly recommended (Cohen et al. 2007). It is the actual concentration of the substances, not the overall amount being excreted or its ratio to some normalizing substances such as creatinine that is the critical variable when evaluating urine for the potential formation of solids. The procedures just described for urine collection for evaluation of the presence of urinary solids is in marked contrast to the way that urine is typically collected for assessment of renal function. Furthermore, we have found that examination of rodent urine with dipsticks can also lead to misleading results, particularly with respect to measurement
508
CHAPTER 19 URINARY TRACT CALCULI AND BLADDER TUMORS
of urinary pH, protein, and the potential for hematuria (Cohen 1995; Cohen et al. 2007). We strongly recommend examination of urine immediately after collection using a microelectrode to determine pH, rather than using a dipstick. Also, for rodent urine, the Bradford method for assaying protein is preferable to biuret-based methodology (Cohen 1995).
19.5. INTERSPECIES COMPARISON OF URINE COMPOSITION Though the substances present in the urine are generally the same between species of mammals, there are marked quantitative differences as well as some critical qualitative differences. A major difference between rodent and human urine is the osmolality (Cohen 1995). Because of extremely high concentrations of ions and urea, rodent urine has a high osmolality, generally in the range of 1500–2500 mosmol. In contrast, human urine generally has an osmolality in the range of 200–300 mosmol, similar to blood. It can become considerably more dilute, but even under conditions of dehydration it rarely attains an osmolality above 600 or 700. It has theoretically been calculated that human urine, based on renal physiology, cannot attain a concentration greater than approximately 1000–1200 mosmol. Also, there are significant differences in protein concentrations between rodents and humans, particularly for male rats and male mice. Male rats and mice excrete unique proteins, α2u-globulin and mouse urinary protein (MUP), respectively, for which there are no human homologues (Hard 1995; Meek et al. 2003; Olson et al. 1990). This is particularly noteworthy in rats, since various substances can bind to the α2u-globulin and greatly affect renal and urinary tract function. These proteins lead to extremely high protein concentrations in the urine, generally in the range of several milligrams per milliliter, in contrast to human urine in which there is usually micrograms of protein per milliliter. Furthermore, both rats and mice gradually develop an aging nephropathy, which leads to increased excretion of albumin with age. This is particularly noteworthy in rats, both males and females.
19.6. URINARY SOLID CARCINOGENESIS IN RODENTS Numerous substances administered to rats and/or mice lead to formation of urinary solids with consequent cytotoxicity, regenerative proliferation, and ultimately the formation of tumors. A variety of specific mechanisms have been demonstrated for the formation of these solids (Clayson et al. 1995; Cohen 1998; IARC 1999; RBCWG 1995). Most readily understood is the direct formation of crystals and calculi composed of the administered substance or metabolite(s). Thus, dietary administration of melamine (Meek et al. 2003), uracil (Shirai et al. 1989), or sulfosulfuron (Arnold et al. 2001) at extremely high levels leads to the formation of crystals composed predominantly of these substances. If the exposure levels do not produce urinary
19.6. URINARY SOLID CARCINOGENESIS IN RODENTS
509
concentrations sufficient to lead to the formation of crystals and/or calculi composed of these substances, then urinary solids are not formed, there is no evidence of cytotoxicity, and no consequent regenerative proliferation or tumorigenicity occurs. Tumors are only produced when the administered dose is sufficient to produce the formation of solids in the urinary tract. This is based on physical chemical properties and is the mechanistic basis for a clearly defined threshold for the carcinogenicity for these substances. Additionally, it appears that these substances have to be administered for an adequate period of time to lead to a detectable incidence of tumors. Thus, short-term administration, even if there is formation of calculi and extensive proliferation, is inadequate for generating a statistically significant incidence of tumors in a standard 2-year bioassay (Cohen 1998; Shirai et al. 1989). Upon ceasing administration of the test substance, any crystals or calculi that have formed gradually are dissolved and/or excreted, and the proliferative response ends. The removal of the hyperplastic, rapidly proliferating urothelial cells appears to be primarily by apoptosis. The urothelium returns to a normal morphologic and cell kinetic state within a few weeks after the disappearance of the urinary solid (Shirai et al. 1989). A recently described example of indirect crystalluria and calculus formation has been demonstrated for the PPAR α/γ agonist, muraglitazar (Dominick et al. 2006). Administration of muraglitazar leads to the formation of urinary calcium phosphate-containing precipitate, as well as calcium and magnesium-containing crystals and calcium-containing calculi. These solids appear to be due to the inhibition by muraglitazar of citrate synthesis leading to hypocitratemia and consequent hypocitraturia. As indicated above, citrate is the major chelating substance for calcium in the urine; and with the decrease of urinary citrate, calcium- and (to some extent) magnesium-containing crystals are able to form in the urine. For reasons that are not entirely clear, precipitation of calcium-containing crystals in the urine requires a pH greater than or equal to 6.5. Co-administration of ammonium chloride in the diet with muraglitazar treatment leads to significant acidification of the urine, generally at pH less than 6.0; this nearly completely inhibits the formation of urinary solids, with complete inhibition of urinary tract cytotoxicity, regenerative proliferation, and tumorigenicity. Numerous substances such as calcium, magnesium, and phosphate, administered to rats and/or mice, have been demonstrated to lead to formation of urinary solids and are listed in Table 19.1. This table includes not only a large number of natural, essential ingredients in our diet, but also a number of substances that are formed from normal intermediary metabolism, such as carbonate, oxalate, cystine, urate, and uracil, which are present in normal urine. Numerous synthetic chemicals also produce urinary solids when administered at very high doses, including agrichemicals (such as sulfosulfuron and Fosetyl-A1), industrial chemicals (such as melamine), and pharmaceuticals (such as sulfonamides, carbonic anhydrase inhibitors, and HIV protease inhibitors). Sodium saccharin and numerous other sodium salts of moderately strong acids, such as ascorbate, glutamate, bicarbonate, aspartate, citrate, and others, administered at very high levels in the diet (≥25,000 ppm) to rats (males > females), result in the production of large amounts of an amorphous calcium phosphate-containing
510
CHAPTER 19 URINARY TRACT CALCULI AND BLADDER TUMORS
precipitate in the urine (Cohen 1998; IARC 1999; RBCWG 1995). Calcium phosphate is an essential component for cell survival as long as it is present at concentrations that remain soluble. However, when the concentration is sufficiently high for precipitation to occur, the precipitate is cytotoxic to epithelial cells, including urothelium (IARC 1999). The conditions are appropriate in rat urine for the excretion of sufficient concentrations of calcium phosphate for precipitation to occur. The precipitate also contains protein, silicates, and mucopolysaccharides. Precipitation does not occur in mice because they have urinary concentrations of calcium and phosphate approximately 10 times less than that in rats. The precipitate also does not form in primates, including humans. Conditions leading to acidification of the urine, such as co-administration with ammonium chloride, administration in AIN76A semisynthetic diet, or administration of the acid form of the corresponding sodium salt, prevents formation of the calcium phosphate precipitate and completely inhibits the urothelial cytotoxicity, proliferation, and tumorigenicity (IARC 1999). Thus, for these sodium salts, the mechanism leading to the induction of urothelial tumors in rats not only occurs only at high doses, above a threshold, but is speciesspecific. Extensive epidemiological investigations have not shown an increased risk of bladder cancer in individuals consuming saccharin (Elcock and Morgan 1993). Consumption of ascorbate (vitamin C) is associated with a decreased risk of bladder cancer in humans (Cohen 1998; Cohen et al. 2000; Johansson and Cohen 1997). Based on these mechanistic considerations, including species-specificity and a threshold dose–response, the IARC down-classified saccharin from possibly carcinogenic to humans (Group 2B) to not classifiable as to their carcinogenicity to humans (Group 3) (IARC 1999), and the United States National Toxicology Program removed saccharin from its list of carcinogens (NTP 2000).
19.7.
EPIDEMIOLOGY
Numerous studies have examined the relationship of urinary tract solids to toxicity and to bladder cancer in humans (Burin et al. 1995; Cohen et al. 2000; La Vecchia et al. 1991; RBCWG 1995). The evidence suggests that urinary amorphous precipitate and urinary crystals of any kind are not associated with cytotoxicity or deleterious effects in humans. Crystalluria in humans is not associated with any toxicological response (McPherson et al. 2006; Pearle and Lotan 2007). In some instances it can be an indication of the propensity of the individual to form calculi from these substances, such as calcium oxalate, or occasionally it can be an indication of systemic metabolic disturbances, such as gout, oxalosis, or hypercalcemia. The relationship of calculi to human bladder cancer remains unclear (Burin et al. 1995; Cohen et al. 2000; La Vecchia et al. 1991; RBCWG 1995). Several epidemiologic studies have not found any evidence for a relationship of urinary tract calculi to human bladder cancer. However, occasional studies have found a small but statistically significant increased risk of bladder cancer in association with exposure to calculi (Burin et al. 1995). In humans, urinary tract calculi are generally not present for long periods of time, in contrast to rodents (DeSesso 1995). This is because of the normal anatomy
19.8. RISK ASSESSMENT
511
of the human lower urinary tract. Formation of calculi in humans generally leads to obstruction of the lower urinary tract, either at the kidney pelvic–ureteral junction, the site at which the ureter crosses the pelvic brim, or where the ureter enters the bladder, as well as at the urethral outlet of the bladder. When calculi are large enough to lead to obstruction, they cannot be readily excreted, leading to excruciating pain. An individual promptly consults a physician on an emergency basis, resulting in either (a) spontaneous evacuation of the calculus by hydration of the patient or (b) treatment by ultrasound or surgery. There are a few situations in which lower urinary tract calculi can be retained for substantial periods of time in humans (DeSesso 1995; Pearle and Lotan 2007). One is the presence of calculi in the kidney pelvis, frequently leading to formation of what are referred to as staghorn calculi. Other situations that can result in the prolonged presence of calculi include bladder diverticuli and the neurogenic bladder associated with paraplegia. However, in circumstances with long-standing urinary tract calculi, patients also have bacterial cystitis (Schaeffer and Schaeffer 2007). Bacterial cystitis is a known risk factor for a slightly increased risk for development of bladder cancer. Thus, it is difficult to ascertain whether the cases of bladder cancer developing in patients with long-standing urinary tract calculi are related to the calculus or whether they are related to the bacterial cystitis, which is associated with the calculi. Moreover, the types of tumors associated with bacterial cystitis and calculi, as well as with other infectious inflammatory processes in the bladder, such as schistosomiasis, frequently are squamous cell carcinomas, in contrast to the usual transitional (urothelial) cell carcinomas that occur in the bladder (Oyasu 1995). In rodents, the tumors associated with urinary tract solids are for the most part transitional (urothelial) cell tumors rather than squamous cell proliferations.
19.8.
RISK ASSESSMENT
From the above review, the overall risk assessment for humans of any chemical producing bladder cancer in rodents based on a mode of action involving the formation of urinary tract solids needs to take into account qualitative and quantitative differences between rodent species and humans and also consideration of a threshold dose response (the necessity for the presence of sufficiently high concentrations of solute to form a precipitate in the urine). Without question, whether involving substances directly or indirectly leading to the formation of calculi or other urinary tract solids, there is a nonlinear, threshold dose response that is based on the physical chemical properties of the solutes and the composition of the urine. Since the composition of human and rodent urine is strikingly different, this must be taken into account in any quantitative comparison between the species. Furthermore, the formation of urinary precipitate or crystals appears to be insufficient for the production of toxicity to the human urothelium in contrast to many instances of cytotoxicity being produced in rodents by such solids, particularly in rats. In addition, the anatomical differences between rodents and humans need to be taken into account in any risk assessment. Humans generally do not retain calculi for long periods of time
512
CHAPTER 19 URINARY TRACT CALCULI AND BLADDER TUMORS
whereas the horizontal, quadruped rodent can retain calculi in the dome of the bladder essentially for its lifetime since the calculi do not completely obstruct the urethral outlet. Thus, in assessing potential bladder cancer risk for humans based on studies in rodents, consideration of a threshold dose response is the foremost consideration. The differences in composition of the urine, anatomic differences, and especially exposure differences between rodents and humans must be taken into account. Furthermore, the evidence for a relationship for urinary tract calculi to an increased risk of bladder cancer is relatively weak and is complicated by the usual association of bacterial cystitis with the presence of long-standing calculi. Urinary precipitate and crystals are not relevant to human carcinogenesis, in contrast to rodents. The potential for formation of urinary tract crystalluria and calculi in humans can be readily assessed in the clinical setting (McPherson et al. 2006; Pearle and Lotan 2007). This can be accomplished by routine collection of urine for urinalysis, including sediment analysis for crystals, cells, and casts. As indicated above, the presence of abnormal crystals or an increase in crystals is not sufficient to indicate urothelial toxicity in humans. It is only the presence of calculi that poses any potential for urinary toxicity. Since calculi will frequently be associated with obstruction and consequent pain, this will be a clinical observation that is readily made by the patient with corroboration by the clinician. Many substances that can lead to the production of calculi in rodents do not appear to do so in humans. Thus, sulfonamides can produce crystalluria and calculi frequently in rodents, whereas sulfonamide crystalluria is common in humans but calculi are rare (McPherson et al. 2006; Pearle and Lotan 2007). Another example is muraglitazar (Dominick et al. 2006), which leads to formation of a variety of urinary solids in rats, including calculi, but it is not associated with the formation of urinary crystalluria or calculi in humans. Overall, the presence of urinary tract solids can readily be assessed with appropriate studies of urine and with proper collection methodology, so that the mode of action can be established in the rodent model. It clearly represents a threshold phenomenon, and estimation of exposure levels for humans can be made. It appears doubtful, however, whether urinary tract calculi actually pose a cancer hazard to humans, and therefore they are not a cancer risk for humans.
REFERENCES Arnold, L. L., Cano, M., St John, M. K., Healy, C. E., and Cohen, S. M. (2001). Effect of sulfosulfuron on the urine and urothelium of male rats. Toxicol Pathol 29, 344–352. Bryan, G. T. (1969). Pellet implantation studies of carcinogenic compounds. J Natl Cancer Inst 43, 255–261. Burin, G. J., Gibb, H. J., and Hill, R. N. (1995). Human bladder cancer: Evidence for a potential irritationinduced mechanism. Food Chem Toxicol 33, 785–795. Clayson, D. B. (1974). Editorial: Bladder carcinogenesis in rats and mice: possibility of artifacts. J Natl Cancer Inst 52, 1685–1689. Clayson, D. B., and Cooper, E. H. (1970). Cancer of the urinary tract. In Advances in Cancer Research, Vol. 13, Klein, G., and Weinhouse, S., eds., Academic Press, New York, pp. 271–381.
REFERENCES
513
Clayson, D. B., Fishbein, L., and Cohen, S. M. (1995). Effects of stones and other physical factors on the induction of rodent bladder cancer. Food Chem Toxicol 33, 771–784. Cohen, S. M. (1995). Role of urinary physiology and chemistry in bladder carcinogenesis. Food Chem Toxicol 33, 715–730. Cohen, S. M. (1998). Urinary bladder carcinogenesis. Toxicol Pathol 26, 121–127. Cohen, S. M., Cano, M., Johnson, L. S., StJohn, M. K., Asamoto, M., Garland, E. M., Thyssen, J. H., Sangha, G. K., and van Goethem, D. L. (1994). Mitogenic effects of propoxur on male rat bladder urothelium. Carcinogenesis 15, 2593–2597. Cohen, S. M., Johansson, S. L., Arnold, L. L., and Lawson, T. A. (2002). Urinary tract calculi and thresholds in carcinogenesis. Food Chem Toxicol 40, 793–799. Cohen, S. M., Ohnishi, T., Clark, N. M., He, J., and Arnold, L. L. (2007). Investigations of rodent urinary bladder carcinogens: Collection, processing, and evaluation of urine and bladders. Toxicol Pathol 35, 337–347. Cohen, S. M., Shirai, T., and Steineck, G. (2000). Epidemiology and etiology of premalignant and malignant urothelial changes. Scand J Urol Nephrol Suppl, 105–115. DeSesso, J. M. (1989). Confounding factors in direct bladder exposure studies. Comments in Toxicol 3, 317–334. DeSesso, J. M. (1995). Anatomical relationships of urinary bladders compared: Their potential role in the development of bladder tumours in humans and rats. Food Chem Toxicol 33, 705–714. Dominick, M. A., White, M. R., Sanderson, T. P., Van Vleet, T., Cohen, S. M., Arnold, L. E., Cano, M., Tannehill-Gregg, S., Moehlenkamp, J. D., Waites, C. R., and Schilling, B. E. (2006). Urothelial carcinogenesis in the urinary bladder of male rats treated with muraglitazar, a PPAR alpha/gamma agonist: Evidence for urolithiasis as the inciting event in the mode of action. Toxicol Pathol 34, 903–920. Elcock, M., and Morgan, R. W. (1993). Update on artificial sweeteners and bladder cancer. Regul Toxicol Pharmacol 17, 35–43. Fisher, M. J., Sakata, T., Tibbels, T. S., Smith, R. A., Patil, K., Khachab, M., Johansson, S. L., and Cohen, S. M. (1989). Effect of sodium saccharin and calcium saccharin on urinary parameters in rats fed Prolab 3200 or AIN-76 diet. Food Chem Toxicol 27, 1–9. Gold, L. S., Manley, N. B., Slone, T. H., and Ward, J. M. (2001). Compendium of chemical carcinogens by target organ: Results of chronic bioassays in rats, mice, hamsters, dogs, and monkeys. Toxicol Pathol 29, 639–652. Hard, G. C. (1995). Species comparison of the content and composition of urinary proteins. Food Chem Toxicol 33, 731–746. IARC (1987). Overall Evaluations of Carcinogenicity: An Updating of IARC Monographs Volumes 1 to 42. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans Supplement 7, http:// monographs.iarc.fr/ENG/Monographs/suppl7/Suppl7.pdf. IARC (1999). Consensus report. In Species Differences in Thyroid, Kidney and Urinary Bladder Carcinogenesis, Capen, C. C., Dybing, E., Rice, J. M., and Wilbourn, J. D., eds., Vol. 147, International Agency for Research on Cancer, Lyon, France, pp. 5–9. IARC (2008). Some industrial and cosmetic dyes, and related exposures. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans 99, in press. Johansson, S. L., and Cohen, S. M. (1997). Epidemiology and etiology of bladder cancer. Semin Surg Oncol 13, 291–298. Jull, J. W. (1979). The effect of time on the incidence of carcinomas obtained by the implantation of paraffin wax pellets into mouse bladder. Cancer Lett 6, 21–25. La Vecchia, C., Negri, E., D’Avanzo, B., Savoldelli, R., and Franceschi, S. (1991). Genital and urinary tract diseases and bladder cancer. Cancer Res 51, 629–631. Marengo, S. R., Chen, D. H., Kaung, H. L., Resnick, M. I., and Yang, L. (2002). Decreased renal expression of the putative calcium oxalate inhibitor Tamm–Horsfall protein in the ethylene glycol rat model of calcium oxalate urolithiasis. J Urol 167, 2192–2197. McPherson, R. A., Ben-Ezra, J., and Zhao, S. (2006). Basic examination of urine. In Henry’s Clinical Diagnosis and Management by Laboratory Methods, 21st edition, McPherson, R. A., and Pincus, M. R., eds., Saunders, New York, pp. 393–425.
514
CHAPTER 19 URINARY TRACT CALCULI AND BLADDER TUMORS
Meek, M. E., Bucher, J. R., Cohen, S. M., Dellarco, V., Hill, R. N., Lehman-McKeeman, L. D., Longfellow, D. G., Pastoor, T., Seed, J., and Patton, D. E. (2003). A framework for human relevance analysis of information on carcinogenic modes of action. Crit Rev Toxicol 33, 591–653. NTP (2000). Appendix B. Summary—Actions on the nomination of saccharin for delisting from the Report on Carcinogens. In Report on Carcinogens, 9th edition, National Toxicology Program, Washington, D.C., pp. B3–B4. NTP (2008). Long-term study reports and abstracts. National Toxicology Program, http://ntp.niehs.nih. gov/index.cfm?objectid=D16D6C59-F1F6-975E-7D23D1519B8CD7A5. Okamura, T., Garland, E. M., Masui, T., Sakata, T., St John, M., and Cohen, S. M. (1991). Lack of bladder tumor promoting activity in rats fed sodium saccharin in AIN-76A diet. Cancer Res 51, 1778–1782. Olson, M. J., Johnson, J. T., and Reidy, C. A. (1990). A comparison of male rat and human urinary proteins: Implications for human resistance to hyaline droplet nephropathy. Toxicol Appl Pharmacol 102, 524–536. Oyasu, R. (1995). Epithelial tumours of the lower urinary tract in humans and rodents. Food Chem Toxicol 33, 747–755. Pearle, M. S., and Lotan, Y. (2007). Urinary lithiasis: Etiology, epidemiology, and pathogenesis. In Campbell–Walsh Urology, Vol. 2, 9th edition, Wein, A. J., Kavoussi, L. R., Novick, A. C., Partin, A. W., and Peters, C. A., eds., Saunders, Philadelphia, pp. 1363–1392. RBCWG (1995). Urinary bladder carcinogenesis: Implications for risk assessment. Rodent Bladder Carcinogenesis Working Group. Food Chem Toxicol 33, 797–802. Sakata, T., Masui, T., St John, M., and Cohen, S. M. (1988). Uracil-induced calculi and proliferative lesions of the mouse urinary bladder. Carcinogenesis 9, 1271–1276. Schaeffer, A. J., and Schaeffer, E. M. (2007). Infections of the urinary tract. In Campbell–Walsh Urology, Vol. 1, 9th edition, Wein, A. J., Kavoussi, L. R., Novick, A. C., Partin, A. W., and Peters, C. A., eds., Saunders, Philadelphia, pp. 223–303. Shirai, T., Fukushima, S., Tagawa, Y., Okumura, M., and Ito, N. (1989). Cell proliferation induced by uracil-calculi and subsequent development of reversible papillomatosis in the rat urinary bladder. Cancer Res 49, 378–383. Tiltman, A. J., and Friedell, G. H. (1972). Effect of feeding N-(4-(5-nitro-2-furyl)-2-thiazolyl)formamide on mitotic activity of rat urinary-bladder epithelium. J Natl Cancer Inst 48, 125–129.
PART
V
METHODS FOR INFORMING CANCER RISK QUANTIFICATION
CH A P TE R
20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS: A STATE-OFTHE-ART OVERVIEW Yin-tak Woo David Y. Lai
20.1.
INTRODUCTION
During the past decade, there has been an explosive growth in the interest of using qualitative as well as quantitative structure–activity relationships analysis or modeling—collectively known as (Q)SAR—in predicting the carcinogenic potential of chemicals. Despite tremendous advancement in predictive technology, carcinogenicity remains to be one of the most difficult toxicological endpoints to predict because of the complexity of its mechanisms of action and the difficulty of obtaining robust, well-balanced databases needed for effective (Q)SAR studies. Meanwhile, the scientific, industrial, and regulatory communities are under increasingly intense pressure to expand the use of (Q)SAR from the traditional research and development (R&D) tool to health and environmental protection and regulatory purposes. The user base has also substantially expanded from experienced research scientists to a broader, heterogeneous base that may include nonscientists with limited knowledge of chemical carcinogenesis. It is important to point out that proper use of (Q)SAR predictions requires basic understanding of (a) the complexity of the toxic endpoint of interest, (b) the validity and applicability of the specific (Q)SAR method for the query chemical, and (c) the limitations of the method, the uncertainty or degree of confidence of the prediction, and the need for supportive evidence. The purposes of this chapter are to provide a background document to address some of these issues along with descriptions of some practical guiding principles and structural alerts or factors for (Q)SAR users. The chapter aims to provide (a) an overview of (Q)SAR analysis or modeling, (b) basic knowledge of the essence of mechanism-based SAR exemplified by genotoxic carcinogens, nongenotoxic carcinogens, and fibers,
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
517
518
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
particles, and nanomaterials, (c) descriptions of possible uses of (Q)SAR and some widely used (Q)SAR methods/models, and (d) some future perspectives of (Q) SAR field.
20.2. OVERVIEW OF (Q)SAR ANALYSIS AND MODELING 20.2.1.
Types of (Q)SAR
The carcinogenic potential of chemicals may be assessed by different types of (Q) SAR analysis which include (a) qualitative or semiquantitative SAR, (b) traditional QSAR analysis, (c) formalized computerized models or softwares, and (d) other methods such as biologically based or molecular modelling methods [e.g., Rabinowitz et al. (2008)] or integrative methods. Qualitative or semiquantitative SAR may involve (a) human expert judgment or computer-assisted identification of structural features (e.g., “structural alert”) that may contribute to carcinogenicity, (b) assessment of factors that affect absorption/ distribution/metabolism/excretion (ADME), and (c) consideration of other supportive information as a basis for prediction. The predicted results may be expressed qualitatively as positive/negative or semiquantitatively in a relative scale such as low/moderate/high. For chemicals with abundant data on closely related homologues or analogues, “read across” or “trend analysis” may be conducted to project the carcinogenic potential of the query chemical by comparison to homologues or analogues. In traditional QSAR methods, the carcinogenic potential of the query chemical is predicted quantitatively (usually in TD50, the dose inducing a tumor incidence of 50% in rodents) by using mathematical equations/models that relate the carcinogenic activity of a training set of structurally related chemicals to a combination of their physicochemical properties and other molecular descriptors (e.g., topological, quantum mechanical) using various statistical methods, such as regression analysis, principal component and factor analysis, discriminant and pattern recognition analysis, and similarity analysis. A number of computerized predictive softwares or models have been developed to predict carcinogenic potential of chemicals. These include (a) knowledge rule-based expert systems that capture human expertise, (b) programs that combine human expert decisions with statistical and correlative approaches, and (c) machine learning, neuronal networking, artificial intelligence (AI), or data mining systems to identify molecular fragments of interest, discover SAR features, induce knowledge rules, and/or develop decision logic. Computational methods or approaches are also being used to develop (a) biologically based models such as 2-D or 3-D receptor modeling, docking, and ligand SAR and (b) integrative models that incorporate or combine both chemical and biological information. (Q)SAR may also be classified as (a) statistically based (relying on statistical, deterministic, or probabilistic association) or (b) mechanistically based (e.g., receptor modeling, electrophilicity-based), or they may be a combination of both. Ideally,
20.2. OVERVIEW OF (Q)SAR ANALYSIS AND MODELING
519
(Q)SAR studies should strive to achieve statistical association with mechanistic foundation. The importance of mechanistic understanding in (Q)SAR will be discussed further in the following section. For model development, (Q)SAR studies may be undertaken using structurally closely related sets of chemicals (termed congeneric or homogeneous) or structurally diverse sets of chemicals (termed global, noncongeneric, or heterogeneous). In general, predictive models using congeneric data require fewer chemicals for model development and tend to perform better (presumably because they are more likely to act by a particular mechanism) but are more limited in scope whereas global, noncongeneric models tend to be more versatile but more likely to yield false negative results if the predicted chemical is not well-represented in the training knowledge base or database. Beyond structural homogeneity, (Q)SAR studies may also be conducted on classes of chemicals with similar biological activity/function (e.g., peroxisome proliferators) to identify common factors beyond chemical structure.
20.2.2. Criteria for Assessing Validity and Scientific Soundness of (Q)SAR Most validation studies [e.g., Benigni and Bossa (2008), Benigni and Zito (2004), and Mayer et al. (2008)] tend to focus on the predictive accuracy/concordance of (Q)SAR methods. The predictive accuracy may be separated into two components: (a) sensitivity, which measures the ability to correctly detect positive chemicals (i.e., avoid false negatives), and (b) specificity, which measures the ability to correctly predict negative chemicals (i.e., avoid false positives). The receiver operating characteristic (R.O.C.) curve can graphically express both components and has often been used to compare the relative performance of different methods [e.g., Benigni and Bossa (2008)]. It should be emphasized that the predictive accuracy of any (Q) SAR method is often chemical batch-specific; that is, the accuracy shown for one batch of chemicals is not necessarily applicable to another batch of chemicals. Beyond predictive accuracy, other factors must be considered to assess the validity and scientific soundness of the (Q)SAR methods. For example, for (Q)SAR methods involving expert judgment, the factors that should be considered in assessing the scientific validity and soundness include: (a) knowledge, expertise, and predictive track record of the experts involved, (b) the scope and purpose of the (Q)SAR analysis, (c) the extent of consideration of relevant literature and supportive evidence, (d) the extent of consideration of relevant structural and mechanistic information, and (e) the articulation of the scientific basis for prediction, reasoning rationale, confidence and uncertainty, and knowledge gaps, if any. The critical requirements and pitfalls for conducting (Q)SAR studies have been the subject of many recent publications [e.g., Benigni et al. (2007), Cronin and Schultz (2003), Doull et al. (2007), Helma (2004), and Woo and Lai (2003)]. The Organization for Economic Cooperation and Development (OECD) recently conducted an international workshop at Setubal, Portugal to define the principles (now often referred to as the Setubal principles) for considering a (Q)SAR model for regulatory purposes. Essentially, the workshop panel concluded that, for regulatory
520
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
uses, (Q)SARs should (a) be associated with a defined endpoint of regulatory importance, (b) take the form of an unambiguous algorithm, (c) have a defined domain of applicability, (d) be associated with appropriate measure of goodness of fit, robustness, and predictivity, and (e) preferably have a mechanistic basis (OECD 2004). It should be emphasized that, for mechanistically complex endpoints such as carcinogenicity, a sound mechanistic basis should be strongly preferred if high confidence is needed. An in-depth discussion of the criteria needed to assess the scientific soundness of (Q)SAR was the focus of an International Life Science Institute (ILSI) and U.S. Environmental Protection Agency (EPA) workgroup that dealt with issues related to (a) selection of proper endpoint or characteristics for (Q)SAR analysis, (b) knowledge/data coverage for knowledge-based/statistically based (Q)SAR models, (c) methodology, selection, and handling of molecular descriptors, (d) validation considerations, (e) transparency and rationale, (f) confidence and uncertainty, and (g) strengths, weaknesses, and limitations. The readers are referred to the workgroup report (Doull et al. 2007) for details.
20.2.3. Difficulties of (Q)SAR Modeling/Prediction of Chemical Carcinogens (Q)SAR studies, no matter how well conducted, are always subject to the limitation of a variety of inherent endpoint-specific difficulties. This is particularly true for the carcinogenicity endpoint. Some of the major difficulties include: (a) mechanistic complexity that hampers (Q)SAR analysis (see discussion below); (b) lack of wellbalanced training database required for good (Q)SAR studies because researchers tend to favor conducting studies that lead to positive findings over those that are likely to be negative; (c) high variability of long-term studies leading to inconsistent or equivocal findings that may require further studies; (d) species, strain, and gender differences complicating interpretation of data and human relevancy; (e) the commonly used quantitative parameter, TD50, does not fully take into account other important consideration such as tumor multiplicity, malignancy, and latency period; (f) complications from the use of maximum tolerated doses often used in many cancer bioassays; and (g) high cost of prospective, external validation [e.g., Benigni and Zito (2004)]. To fully interpret the outcome of (Q)SAR predictions and evaluate reliability, it would be advisable to “test drive” the (Q)SAR models with related chemicals of known activity to ensure that the model is properly trained, to examine the rationale and analogues that led to the predictions, and to peruse the original study data of the key analogs.
20.2.4.
Importance of Mechanistic Understanding
Mechanistic understanding is especially crucial for (Q)SAR analysis/modeling of chemical carcinogens because of the complexity and multistage, multifactorial process of carcinogenesis (see discussion below). The importance of mechanistic considerations in (Q)SAR analysis of chemical carcinogens has been discussed by Woo and Lai (2003). Mechanistic considerations can improve (Q)SAR study by
20.3. MECHANISM-BASED SAR ANALYSIS
521
helping to (a) select the most appropriate molecular descriptors (e.g., electrophilicity versus receptor-based), (b) serve as a criterion to assess whether the training database is suitable for making predictions on the chemicals of interest, (c) stratify the training database into smaller but mechanistically more homogenous subsets to improve predictive capability, (d) interpret outliers, (e) guide hypothesis testing to fill data gaps, and (f) assess the human significance of predictions based on animal data.
20.3. MECHANISM-BASED SAR ANALYSIS OF CHEMICAL CARCINOGENS, FIBERS, AND PARTICLES/NANOPARTICLES 20.3.1.
Basic Principles
Chemical carcinogenesis is a multistage, multifactorial process, which conceptually consists of three operational stages: initiation, promotion, and progression. Initiation involves a mutational event that may include gene mutation, chromosome aberration, translocation, and instability. Promotion involves clonal expansion of initiated cells to reach a critical mass by a variety of means such as cell proliferation, inhibition of programmed cell death, persistent chronic inflammation, inhibition of terminal differentiation, and loss of growth control. Progression may involve a second mutational event, the loss of tumor suppressor gene, impairment of immune surveillance, and acquisition of ability to metastasize. The underlying mechanisms of these three stages differ significantly; therefore, the key molecular descriptors for (Q)SAR analysis differ accordingly. To be a complete carcinogen, a chemical must be able to trigger, either directly or indirectly, activity in all three stages of the process. The relative importance of the chemical’s contribution to each of these three stages differs from chemical to chemical. Based on the predominant mechanism of action, carcinogens may be classified as genotoxic and epigenetic/nongenotoxic. Genotoxic carcinogens, also known as DNA-reactive carcinogens, generally are chemicals that directly interact with DNA either as parent chemicals or as reactive metabolites to form DNA adducts or lesions which, if unrepaired, may initiate carcinogenesis. Epigenetic carcinogens are agents that act through secondary mechanisms that do not involve direct DNA damage. Mechanism-based SAR analysis basically involves comparison of an untested chemical with structurally related compounds for which carcinogenic activity is known. Considering the most probable mechanism(s) of action, the structural features and functional properties of the untested chemical are evaluated and compared with those of the reference compounds with focus on how the differences between the untested chemical and the reference compounds may affect the potential mechanism of action. These include consideration of (a) SAR knowledge, (b) toxicokinetic and toxicodynamic parameters that may affect the delivery of biologically active intermediate to target tissue(s) for interaction with key macromolecules that may contribute to carcinogenesis, and (c) available supportive evidence. For chemicals with limited knowledge base and information, human expert judgment with delineation of rationale and possible knowledge gaps is often needed. For chemicals with
522
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
abundant SAR information, structural alerts, knowledge rules, decision trees, expert systems, and predictive software may be identified or developed to facilitate the process. Depending on the mechanism of action, the approaches to (Q)SAR analysis may be totally different. In this section, the basic approaches to mechanism-based SAR analysis of (a) genotoxic carcinogens, (b) nongenotoxic carcinogens, and (c) fibers and particles are presented. Some of the approaches and knowledge have been captured in the U.S. EPA’s OncoLogic Cancer Expert System for predicting carcinogenic potential of chemicals (Woo and Lai 2005; Woo et al. 1995).
20.3.2.
SAR of Genotoxic Carcinogens
There are numerous examples of chemical carcinogens that act predominantly by genotoxic mechanisms [see Arcos et al. (1982), Woo and Lai (2003), and Woo et al. (1985b, 1988)]. Virtually all these carcinogens are either electrophiles per se (also called direct-acting genotoxic carcinogens) and can be metabolically activated to electrophilic intermediates. The following are a variety of functional groups that can directly bind covalently to DNA without requirement metabolic activation. The reasons for their direct-acting activity are also discussed along with some of the factors that may affect their ability to exert carcinogenic activity. 20.3.2.1.
Direct-Acting Electrophilic Functional Groups
a. Strained Ring Systems: Strained rings such as (a) epoxide, (b) aziridine (also known as ethyleneimine), (c) lactone, and (d) sultone (see structures below) can readily generate reactive electrophilic intermediates due to their propensity to open up the ring. Upon acidification, epoxides and aziridines can generate carbonium ions. Lactones and sultones can generate acylating intermediates and carbonium ions. The ability of the lactone and sultone rings to open up decreases with the increase in ring size due to reduction in ring strain. The introduction of a double bond to the ring may restore some of the activity especially if adjacent to the carbonyl group.
O
N
(a)
(b)
O
S O (c)
O
O
O (d)
b. Alkyl esters of Moderate and Strong Acids: Alkyl esters of moderate and strong acids such as (a) sulfate, (b) phosphate, (c) tosylate, and (d) methanesulfonate (see below) can serve as alkylating agents. The alkylating activity is dependent on the size of the alkyl group with the relative activity following the order methyl >> ethyl > propyl > butyl; beyond butyl, there is hardly any activity. For alkyl esters of dibasic (e.g., sulfate) and tribasic (e.g., phosphate) acids, the alkylating activity is completely eliminated if any one of the alkyl group is hydrolyzed (e.g., monoalkyl sulfate or dialkyl phosphate).
20.3. MECHANISM-BASED SAR ANALYSIS
523
[ RO]2 SO2 [ RO]3 PO ROSO2 C6 H 4 CH3 ROSO2 CH 2 − R = alkyl (c) (a ) (d) (b) c. Haloalkanes and Substituted Haloalkanes: Haloalkanes with one halogen (other than fluorine) atom at the terminal end(s) of the alkyl chain are potential alkylating agents because the halogen is a good leaving group. In general, the alkylating activity of haloalkanes decreases with (a) the decrease in the leaving tendency of the halogen in the order, I > Br >> Cl, and (b) the increase in the size of the alkyl chain. The introduction of additional halogen(s) to the terminal carbon can also decrease the alkylating activity because the electron withdrawing activity of additional halogen(s) may hinder the departure of the first halogen. In contrast, the introduction of either (a) a heteroatom (e.g., N, S, or O), (b) double bond, or (c) aryl group at the carbon bearing the halogen can significantly increase the alkylating activity by facilitating the departure of halogen. XCH 2 − XCH 2 O− XCH 2 S− XCH 2 N− XCH 2 CH=CH− XCH 2 ArX = Cl, Br, or I d. N-Mustards and S-Mustards: N-Mustards (a) and S-mustards (b) are potent alkylating agents. The nitrogen and sulfur atom may facilitate the departure of chlorine (or bromine or iodine) by providing a resonance stabilizing mechanism through cyclization of the carbonium ion to form aziridiuim or episulfonium ion.
[ XCH 2 CH 2 ]2 N− [ XCH 2 CH 2 ]2 S X = Cl, Br, or I (b)
(a )
e. N-Nitrosamides: N-Nitrosamides, which include (a) N-nitrosoureas, (b) Nnitrososguanidines, and (c) N-nitrosourethanes (see below), can generate alkylating intermediates without metabolic activation. Thiols and alkalis can catalyze the process. The alkylating activity is dependent on the size and nature of the R group attached to nitrogen bearing the nitroso group. R′
R N CO
R″
N NO
(a)
H
NH
R
R
R=Alkyl
R'-O-CO N
N
N
O2N
NO
NO (b)
(c)
f. Aldehydes and Substituted Aldehydes: Aldehydes are highly reactive electrophiles capable of crosslinking DNA as well as reacting with protein. Owing to rapid oxidation, either chemically or metabolically, to unreactive carboxylic acids, most of the target organs of aldehydes tend to be at or close to portal of entry. However, individuals deficient with aldehyde dehydrogenase may be more susceptible to aldehydes. The reactivity of aldehydes decreases with the increase in the size of the alkyl chain. The introduction of α,β-double bond may increase the reactivity of aldehydes. O H C
O H C CH CH
524
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
g. Quinones and Quinoid Compounds: Quinones and quinoid compounds (e.g., 1,4-quinone, 1,2-quinone, 1,4-quinonediminine, 1,4-quinonemethide; see structures below) are reactive electrophiles capable of reacting with –SH compounds. They may be generated during oxidation of ortho/para (but not meta) hydroquinones or aromatic diamines or hydroxytoluene compounds. The oxidation process may involve one electron oxidation that could also generate free radicals. O
O
NH
O
NH
CH2
O
O
h. Michael Addition Acceptors: The presence of a carbonyl, sulfonyl, or phosphoryl group at the α-carbon of a terminal vinyl group (see structures below) can polarize the double bond and impart partial positive charge on the terminal carbon make it electrophilic. The ability to serve as Michael Addition acceptor is highly susceptible to substitution on the vinyl group. For example, methacrylates are substantially less active than acrylates. O
O
O
CH2 CH C
CH2 CH S
CH2 CH P
O i. Arylating Agents: Aryldiazonium compounds can generate arylating agents after departure of nitrogen. Pyridine-type heteroaromatic compounds with halogen at the ortho position, as well as nitroaromatic compounds with halogen ortho/para to the nitro group(s), are also arylating agents (see below). Although the fluoro group is not a good leaving group in haloalkanes, it can be activated when ortho to ring nitrogen and aromatic nitro group(s) and can be even a better leaving group than the other halogens. + Ar N N
X N
X O2N
NO2
j. Acylating Agents and Isocyanates: Benzoyl, acyl, or carbamoyl halides (including fluoride), dihalocarbonyl compounds (e.g., phosgene), and anhydrides are potent acylating agents. Isocyanate group can react with a hydroxyl functional group to form an urethane linkage. They can all be readily hydrolyzed, are very short-lived, and are therefore mainly of portal of entry (e.g., inhalation) concern.
20.3. MECHANISM-BASED SAR ANALYSIS
Ar or R
O
O
O X
N
525
O X
X
X
O
N C O
O X = F, Cl, Br, or I; R = alkyl; Ar = aryl
All the above substructures/functional groups can be considered as structural alerts (SA) of genotoxicity. The presence/attachment of one or more of these groups in a molecule is suggestive of carcinogenic potential. Whether their presence may actually impart carcinogenic activity is dependent on a variety of factors such as (a) the nature of the SA (e.g., reactivity versus stability, hard versus soft electrophile, the size of the alkyl group for alkylating agent), (b) the physicochemical properties of the molecule to which the SA is attached (e.g., impeding versus facilitating the ability of SA to reach target tissue), (c) the microenvironment surrounding the SA (e.g., steric hindrance versus resonance stabilization), (d) the exposure scenario (particularly for highly reactive SA that can be readily detoxified), and (e) dosage and frequency of the exposure. Judicious use of SA is needed to avoid oversensitivity and ensure reasonable specificity in predicting carcinogenic potential of chemicals. Depending on the specific conditions of the SA-bearing chemical, a different concern level for carcinogenic potential may be predicted. The following are some of the “rules” that may be used to modify the prediction. Direct-Acting SA Concern-Enhancing or “Boosting” Rules a. Presence of two or more functional groups with high molecular flexibility at terminal positions 2–6 atoms apart (e.g., linear alkyl chain but not cycloalkyl ring) b. Low-molecular-weight volatile compounds c. Attachment to intercalating moiety (e.g., linear 3-ring planar molecule) d. Attachment to molecules that are or resemble normal body constituents (e.g., nucleosides, amino acids) e. Attachment to molecules that contain chemical structure capable of exercising resonance stabilization (e.g., α,β-double bond) f. Attachment of additional genotoxic functional groups g. Anticipated exposure that may lead to direct access to potential targets (e.g., inhalation, parenteral injection) Direct-Acting SA Concern-Mitigating or “Busting” Rules a. Physiochemical properties indicative of negligible bioavailability at the route of exposure of concern/evaluation b. For alkyl esters of dibasic (e.g., sulfuric) and tribasic (e.g., phosphoric) acids, absence of one ester group (e.g., monoallkyl sulfate or dialkyl phosphate)
526
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
c. Alkylating agents with alkyl group higher than butyl group d. Presence of substituent adjacent to the electrophilic group or bulky substituent(s) on the molecule of attachment e. Positioning of all electrophilic functional groups in the middle of the molecule with none at any terminal position f. Easily hydrolysable reactive compounds (e.g., α-haloethers, acylating agents, isocyanates) via oral route g. Easily oxidizable reactive compounds (e.g., aldehydes) via oral (dietary) route 20.3.2.2. SAR of Genotoxic Carcinogens that Require Metabolic Activation. The classical major structural classes of genotoxic carcinogens that require metabolic activation include homocyclic and heterocyclic polycyclic aromatic hydrocarbons, aromatic amines, N-nitrosamines, aflatoxin type furocoumarins, carbamates, benzene and alkenylbenzenes, and compounds with terminal double bonds. It is obvious that not all chemicals in these classes are carcinogenic. The key common features for most of the potent carcinogens in these classes are: (a) propensity to generate electrophilic intermediates, especially at or near their target organ; (b) availability of a stabilizing mechanism to allow transport of reactive intermediates from the site of activation to the site of interaction for DNA covalent binding; (c) favorable molecular size, shape, and planarity; (d) characteristics of persistent DNA adducts or chromosomal lesions; and (e) ability to act on various stages of carcinogenesis. Knowledge of the key metabolic activation pathway(s) of chemicals in a specific structural class can provide important clues and approaches to effective (Q)SAR analysis and identification of structural features that may contribute to or reduce carcinogenic potential. Some of these principles may be illustrated by the following examples. N-Nitrosamines. The vast majority of the more than 400 N-nitroso compounds that have been tested for carcinogenic activity are carcinogenic [see Arcos et al. (1982) and Lijinsky (1992)]. As a result, N-nitrosamines are often presumed or predicted to be carcinogenic without careful examination of structural features. However, it is well known that, with few exceptions, the predominant initial metabolic activation pathway for N-nitrosamines is α-hydroxylation. Since the presence of an α-hydrogen is needed for α-hydroxylation, it can be mechanistically predicted that substitution(s) that replace α-hydrogen in dialkylnitrosamines can lead to reduction or elimination of carcinogenic potential. This is consistent with the experimental findings that the relative carcinogenic potency of diethylnitrosamine is much greater than di-sec-propylnitrosamine (with fewer α-hydrogen), which, in turn, is greater than the inactive di-tert-butylnitrosamine (with no α-hydrogen). Some of the other substituents that are known or can be expected to reduce/eliminate carcinogenic potential of N-nitrosamines include: (a) acidic group, fluoro group, or any bulky/ unmetabolizable groups at the α-carbon, (b) branching of alkyl groups or bulky substituents in the vicinity of the α-carbon, and (c) large alkyl groups with total exceeding 12 carbons.
527
20.3. MECHANISM-BASED SAR ANALYSIS
Polycyclic Aromatic Hydrocarbons (PAHs). Although PAHs are a wellknown structural class of carcinogens, relatively few are really potent [see Richard and Woo (1990)]. Virtually all the potent PAH carcinogens contain 4–6 rings with a bay/fjord region because metabolic activation to bay/fjord ring diol epoxide has been shown to be the key pathway. Opening of the epoxide ring generates an electrophilic carbonium ion that can be stabilized by the ring system to give it time to reach and bind to DNA. bay region 10 9 8
distorted bay region 2 3 CH3 1
fjord region 14 13
4 5 L region K region Benzo[a]pyrene
4
12
7
7
8
11
CH3
Dibenzo[a,l]pyrene
7,12-Dimethylbenz[a]anthracene
O
O
HO
OH OH
OH Reactive dihydrodiol epoxides
Other structural features that may enhance carcinogenic potential include: (a) blocking of the L-region by ring fusion to prevent detoxification and (b) a methyl group at the immediate vicinity of bay region (e.g., 12-methyl of benz[a]anthracene or 5-methyl of chrysene) to slightly distort the bay region. On the other hand, the structural features that may reduce or eliminate carcinogenic potential include (a) ring substitution at each and every bay/fjord region benzo ring(s), (b) bulky substituent(s) at virtually any ring, (c) acidic group at any ring, (d) more than four linear rings, (e) PAHs with high degree of symmetry, and (f) PAHs with less than four rings and no methyl at the L-region. Aromatic Amines. The SAR of aromatic amines has been extensively studied. The predominant activation pathway is oxidation of the amino group to generate electrophilic nitrenium ion which can be resonance-stabilized by the aryl moiety to make it stable enough to travel from site of activation to reach and bind to DNA. Molecular planarity is also favorable for carcinogenicity due to ease of DNA intercalation and binding and more accessible to metabolic activation. The critical structural features can best be illustrated by the following molecule along with reasoning:
5′
6′
6
R 4
A
X
B
4′
3
2
2′
3′
5
N R′
528
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
Critical Position/Factor
Effect on Carcinogenic Potential
Brief Reasoning
a. Amino group
↑ if R/R′ = H, CH3 ↓↓ if R/R′ = t-alkyl Relative activity: 4- >> 2- ≥ 3↑ if one CH3 ↓↓ if bulky group ↑ if –, –O–, –S– ↓↓ if –CH2– CH2– ↓↓ if bulky group ↑ if –NH2 ↑ if –F ↓↓ if –COOH, –SO3H
Metabolic activation
b. Position of amino group c. 3- and 5d. Intercyclic group (X) e. 2-, 2′-, 6-, 6′f. 4′-
Resonance stabilization Flanking effect Allows conjugation Electron insulating Distorts planarity Extended conjugation Blocks detoxification ↓ absorption, ↑ excretion
The details of these SAR approaches and how these factors can be manipulated to design safer chemicals have been described (Lai et al. 1996).
20.3.3.
SAR of Nongenotoxic Carcinogens
The scientific literature on epigenetic mechanisms of chemical carcinogens has been growing at an accelerated pace in the past several years because of the increasing importance of mechanistic understanding in elucidating the molecular basis of carcinogenesis, considering human relevance of animal data, and modeling quantitative risk assessment. Epigenetic carcinogens are chemicals, which induce cancer without covalently binding to DNA or directly causing DNA damage. They may act via a variety of mechanisms such as (a) receptor-mediated cell proliferation, (b) generation of reactive oxygen species and free radicals to cause oxidative stress and secondary DNA damage, (c) perturbation of DNA methylation leading to aberrant gene expression, (d) hormonal imbalance or disturbance of homeostatic status of cells, (e) cytotoxicity with subsequent compensatory regenerative hyperplasia, (f) persistent chronic inflammation, (g) inhibition of gap junctional intercellular communication, (h) disturbance of signal transduction, (i) reduction of programmed cell death (apoptosis), (j) mitogenesis, (k) tissue/cell overload with foreign body or certain metals, (l) inhibition of microtubulin polymerization, and (m) impairment of immune surveillance (Woo and Lai 2003). As may be expected from the multiple mechanisms, (Q)SAR analysis of epigenetic carcinogens is very difficult and is dependent on the specific mechanism involved. Unlike genotoxic carcinogens that center on DNA as the common initial target molecule, the initial targets of epigenetic carcinogens may be distributed throughout the cell (e.g., nuclear receptor, cell membrane, cytoplasmic protein, organelles, etc). Some mechanisms (e.g., signal transduction) may involve complex molecular and cell biology and, therefore, may necessitate high-throughput assays or toxicogenomic studies and computational biology [e.g., Dix et al. (2007), Kavlock et al. (2008), Vinken et al. (2008)]. By far the most extensively studied mechanism of epigenetic carcinogens is receptor-mediated cell proliferation. Xenobiotic ligand-induced activation of several
20.3. MECHANISM-BASED SAR ANALYSIS
529
nuclear receptors (Pascussi et al. 2008; Safe 2001) for enzyme induction (arylhydrocarbon receptor or AhR, pregnane X receptor or PXR, constitutive androstane receptor or CAR) have been shown to contribute to increased cell proliferation as the mode of carcinogenic action of a variety of hepatocarcinogens. Peroxisome proliferation activating receptor α (or PPARα)-mediated cell proliferation and oxidative stress are believed to be the mode of carcinogenic action of peroxisome proliferators. Mechanistically, since receptor binding is noncovalent and therefore reversible, (Q)SAR analysis of receptor-mediated epigenetic carcinogens essentially involves two components: (a) ability of the chemical to fit the receptor and serve as agonist/antagonist and (b) long biological half-life to allow continuous or sustained binding/activation of the receptor. Receptor fitness can be studied by either (a) analyzing the structural features of active compounds to find the favorable molecular size, shape, and thickness/planarity or (b) using a computer-assisted 2D or 3D receptor modeling or docking study. Biological half-life can be experimentally measured by looking for structural features suggestive of resistance to metabolism—for example, fluorination (stable C–F bond), ω-1 branching of fatty acids (inhibition of β-oxidation), presence/absence of two adjacent ring positions in aromatic compounds (which allows ring oxidation). These principles can be illustrated by following examples with emphasis of identification of structural features that are predictive of epigenetic carcinogenic potential. SAR Prediction of Carcinogenic Potential of TCDD-Related Compounds. It is now well-documented that the broad spectrum of toxicity of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) is mediated by the AhR (Poland et al. 1976; Safe 2001). Earlier studies showed that TCDD binds to the AhR with extremely high affinity in the picomolar range. The ligand-binding site is hydrophobic and preferentially accommodates planar nonpolar ligands with molecular dimensions approximating a 3 × 10 Å rectangle. Subsequent studies showed a wider range of ligand types including dibenzofurans, biphenyls, naphthalenes, PAHs, and indole carbazoles. The AhR is now predicted to have either a single ligand-binding pocket of 14 × 12 × 5 Å or two ligand-binding sites with one for TCDD-like compounds and the other for the larger PAH-like compounds. A 3D QSAR study provided a detailed characterization of the molecular binding domain of the AhR (Waller and McKinney 1995). From the SAR point of view, the dibenzo-p-dioxin, dibenzofuran, and planar biphenyl rings are approximately isosteric. A comparison of the available carcinogenicity data of unsubstituted dibenzo-p-dioxin and its 2,7-dichloro-, 2,3,7,8tetrachloro-, and 1,2,3,6,7,8/1,2,3,7,8,9-hexachloro- and octachloro- congeners [see Woo et al. (1985b)] suggested that the presence of chlorine substitution at all the four lateral 2,3,7,8-positions can provide TCDD with the most optimal molecular size/shape/planarity and metabolic stability for optimal carcinogenic activity. Lower chlorinated compounds are more likely to be metabolized due to the presence of more adjacent unsubstituted ring positions that favors metabolism, whereas more highly chlorinated compounds may increase the size beyond the optimal range. For the polychlorinated biphenyls (PCBs), the most carcinogenic congener was the one with all six lateral 3,3′,4,4′,5,5′-positions chlorinated. The requirement for planar
530
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
structure can be observed by the loss of dioxin-like activities of PCBs with ring substitution(s) at one or more of the ortho (2,2′6,6′-) position(s) which can lead to distortion from planar structure. Subsequent AhR binding data indicated that polychlorinated dibenzofuran congeners have binding activity similar to that of their dibenzo-p-dioxin counterparts (Woo and Lai 2003). These SAR considerations have been captured in EPA’s OncoLogic Cancer Expert System for predicting carcinogenic potential of chemicals completed in 1999–2000 (Woo and Lai 2005; Woo et al. 1995). In 2006, the NTP completed the cancer bioassay of a number of new congeners of polychlorinated dibenzofurans and PCBs. The results of these studies are summarized in the table below and compared to the OncoLogic predictions. As can be seen from the table, the mechanism-based SAR approach of the OncoLogic system allowed the system to accurately predict the carcinogenic activities. Cl
O
Cl
Cl
O
Cl
2,3,7,8-TCDD
Cl
Cl
Cl
O
Cl Cl
Cl
Cl
# 521 # 520 # 525 # 529
Cl Cl
2,3,4,7,8-PCDB
2,2′,4,4′,5,5′-HCB
3,3′,4,4′,5′-PCB
NTP TR
Cl
Cl
Cl
Cl
Chemical Agent 2378-TCDD 33′44′5′-PCB 23478-PCDB 22′44′55′-HCB
Cl
Cl
Cl
NTP Bioassay Result in Female Rats Evidencea
Target Organ (s)
Incidenceb
OncoLogic Predictionc
CE CE SE EE
Liver/lung/oral Liver/lung/oral Liver/oral Liver
25/53 13/53 4/53 2/53
H HM M LM
a
CE, clear evidence; SE, some evidence; EE, equivocal evidence.
b
Highest incidence seen in any one specific target organ.
c
H, high; HM, high/medium; M, medium; LM, low/medium.
SAR of Hepatocarcinogenic Peroxisome Proliferators. Another welldocumented example of receptor involvement in epigenetic chemical carcinogenesis is PPARα-mediated cell proliferation and oxidative stress as the mode of hepatocarcinogenic action of peroxisome proliferators (Cattley et al. 1998; Gonzalez and Shah 2008; Klaunig et al. 2003; Yu et al. 2003). PPARα belongs to the steroid receptor superfamily that is involved in physiological functions such as lipid
20.3. MECHANISM-BASED SAR ANALYSIS
531
metabolism and energy transfer. The physiological ligands of the receptor are quite diverse and include fatty acids. The SAR of peroxisome proliferators has been reviewed ((Bentley et al. 1993; Lai 2004; Lake and Lewis 1993; Woo and Lai 2003)). Figure 20.1 shows the chemical structures for a variety of hepatocarcinogenic peroxisome proliferatiors. However, at first glance the chemical structures appear to be diverse. From the mechanistic point of view, there are two common features: (a) Virtually all the
(1) Substituted Phenoxyacid Pharmaceuticals and Pesticides Cl
O
CH3
CH3
CH3 COOC2H5
O
CH3
Cl
Clofibrate
COOH CH3
Cl
CH3 COOH CH3
O (CH2)3 H3C
Ciprofibrate
Gemfibrozil Cl
CH3
Cl
O
CH3
COOCH3
O
COOH
CH3
H3C
CH2
COOH
H COO C
O
CF3
2,4-D
SCH2COOH
WY-14.643
Cl O
CH3
Nafenopino
Cl
N N
CH3
Methylclofenapate
Cl
H N
NO
O CH3
O C2H5
Lactofen Cl O
Cl
O
CH2
COOH
Cl
CH3
NH
CH2
CH2
O
COOH CH3
Cl
Bezafibrate
2,4,5-T
(2) Alkyl Carboxylic Acids and Precursors COOH
C2H5
O CH2 CH (CH2)3 CH3 ( CH2 )4
C2H5
Cl Cl
O
CF3 (CF2)6
CH3CH2CH2CH2CHCH2OH
COOH O
Cl
O CH2 CH (CH2)3 CH 3 C2H5
TCA
2-EH
Perfluorooctanoic Acid
DEHA
(3) Phthalate Esters C2H5
O O
CH2 CH (CH2)3 CH3
O
CH2 CH (CH2)3 CH3
O
C2H5
DEHP
Figure 20.1.
CH3
O O
O O
O
CH2
CH
(CH2)5 CH3
CH2
CH
(CH2)5
CH3
DINP
CH3
O
CH2 CH2 CH2 CH 3
O
CH2
O
BBP
Chemical structures of some major classes of peroxisome proiferators.
532
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
compounds contain (either per se or after oxidation of a hydrolysis product) a hydrophilic group (in most cases, a carboxylic acid) at one end and a nonpolar moiety at the other end, and (b) most of the compounds contain structural features that are suggestive of resistance to metabolic detoxification. For example, for substituted phenoxy acid pharmaceuticals and pesticides, one of the key structural elements for peroxisome proliferative activity is chlorine substitution at the 4position of the phenyl ring. When the chlorine atom is at the 2- or 3-position of the phenyl ring, peroxisome proliferation is totally abolished. The 4-position is usually the most important position for detoxification. Chlorine substitution at the 4-position may discourage metabolic detoxification as well as exert some electronic effect. Most of the more potent compounds also have substitution at the ω-1 carbon (the carbon next to the terminal carboxylic acid). In the substituted n-alkylcarboxylic acids, most of the active compounds are also substituted at the ω-1 carbon, which renders the compound resistant to metabolism; the most effective substituent is an ethyl group. Several 2-ethylhexyl-containing esters such as di-(2-ethylhexyl) adipate (DEHA), di-(2-ethylhexyl)phthalate (DEHP), and di-(2-isononyl) phthalate (DINP) induce peroxisome proliferation because they are readily hydrolyzed to yield 2-ethylhexanol, which is further oxidized to 2-ethylhexanoic acid, an active peroxisome proliferator (Reddy and Lalwai 1983). As compared to 2-ethylhexanoic acid, the 3- and 4-isomers are virtually inactive in peroxisomal responses [cited in Lake and Lewis (1993)]. Perfluorooctanoic acid (PFOA), a rodent hepatocarcinogen with long biological half-life, is also believed to owe its activity to metabolic resistance because of the strength of the C–F bonds (Olsen et al. 2007). The human relevance of PPARα-mediated rodent hapatocarcinogenesis has been a subject of intensive debate for more than a decade. Whereas the relevancy cannot be totally disregarded, considering the totality of the evidence, most regulatory agencies are supportive of the conclusion that this mechanism is not relevant to humans (IARC 1994; Klaunig et al. 2003; Lai 2004). Nevertheless, the SAR features mentioned above can be effectively used to interpret rodent cancer data and assess the significance of human cancer risk. SAR of Rodent Thyroid Carcinogenesis. Hormonal imbalance leading to overcompensation of trophic hormone production and subsequent cell proliferation is an important mechanism of epigenetic carcinogenesis. The rodent thyroid gland is particularly susceptible to chemical carcinogens via disruption of the pituitary– thyroid axis feedback mechanism. A number of possible mechanisms (e.g., inhibition of iodide transport, perturbation of biosynthesis or catabolism or secretion of thyroid hormones) have been identified (Hill et al. 1998; Woo and Lai 2003). Of these, the most well established SAR feature is the association of thiourea/ thionamide substructure with rodent thyroid carcinogens (see structures below). In addition to these compounds, the thyroid carcinogenic activity of a few ethylenbisdithiocarbamate (e.g., mancozeb) in rats has also been attributed to the presence of ethylenethiourea as metabolite and/or contaminant. The molecular mechanism is not clearly understood but is believed to involve inhibition of thyroid hormone biosynthetic enzyme by thiourea moiety via formation of disulfide bridge. It should
20.3. MECHANISM-BASED SAR ANALYSIS
533
be noted that the presence of thiourea/thioamide moiety alone may not be sufficient evidence of an epigenetic mechanism of thyroid carcinogenesis. Negative genotoxicity data or lack of genotoxic structural alerts may be needed to ascertain epigenetic mechanism. On the other hand, finding of goitrogenic/antithyroid activity may provide further support. The criteria and science policy for determining whether a thyroid carcinogen should be considered genotoxic or epigenetic and the subsequent use of appropriate quantitative risk assessment methodology have been discussed by the U.S. EPA (EPA 1998).
R
R″ N
R N
R H
O
HN
NH
S
S
N,N′-Dicyclohexylthiourea (R=H; R′=R″=C6H11) N,N'-Diethylthiourea (R=H; R′=R″=C2H5) Trimethylthiourea (R=R′=R″=CH3)
2-Thiouracil (R=H) 6-Methylthiouracil (R=CH3) 6-n-Propylthiouracil (R=C3H7)
HN
NH S
Ethylenethiourea
NH-C(S)-SH NH-C(S)-SH Ethylenebisdithiocarbamate
SAR of Alpha-2u-Nephropathy and Male Rat Kidney Tumor. A growing list of nonmutagenic chemicals and chemical mixtures of diverse structures has been found to induce kidney tubule cell tumors in male rats but not in female rats or other animal species. Mechanistic studies showed that these chemicals or their metabolites/components bind reversibly to a male rat-specific α2u-globlin (an 18,700-dalton protein synthesized under androgenic control in hepatic parenchymal cells of mature male rats of various strains [except inbred NBR strain], secreted into the blood, and excreted in urine) to form a protein complex that is resistant to hydrolytic degradation. An excessive accumulation of α2u-globulin-containing hyaline droplets in renal proximal tubules ensues. The protein overload causes cytotoxicity leading to necrosis of the tubule epithelial cells, followed by sustained regenerative cell proliferation (in the P2 segment of proximal tubules), hyperplasia, and eventual induction of renal tubule tumors. Considering the fact that α2u-nephropathy is mainly a male rat-specific mechanism, both the U.S. EPA (1991) and the International Agency for Research on Cancer (IARC) concluded that chemicals that induce renal tumors solely as a result of α2u-nephropathy should not be of human significance in risk assessment (EPA 1991; IARC 1999a). A possible useful structural alert of α2u-nephropathy-inducing carcinogen may be a small to medium-sized t-alkyl alcohol group. A variety of chemicals (isooctane in unleaded gasoline, methyl isobutyl ketone, methyl t-butyl ether, t-butyl alcohol, propylene glycol t-butyl ether) that have been shown to be α2u-nephropathy-inducing carcinogens by the U.S. NTP either are or can be metabolized to t-alkyl alcohols (see below).
534
CHAPTER 20
CH3
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
CH3
CH3
O
CH3
HOH3C
CH3
CH3 CH CH2 C CH3 CH3 CH CH2 C CH3 CH3 C CH3 CH3 CH O C CH3 CH3
O CH3
MIBK
Isooctane (TMP)
CH3
CH3
CH3 C CH2 C CH3 OH
CH3
2-OH-TMP
CH3
O
CH3 C CH2 C CH3
CH3 CH3 C CH3
OH Major metabolite of MIBK
CH3 PG t-Butyl ether
MTBE
OH t-Butyl alcohol
Further studies have shown that 2,2,4-trimethyl 2-pentanol (2-OH-TMP), a metabolite of 2,2,4-trimethylpentane (TMP), is capable of binding to α2u-globulin (Lock et al. 1987). The ability of 2-OH-TMP and other compounds to bind α2uglobulin has been investigated using molecular modeling. The results showed that 2-OH-TMP has the highest binding affinity to α2u globulin and that a number of structurally related compounds also have comparable binding affinity to α2u-globulin (Borghoff et al. 1991).
20.3.4.
SAR of Fibers, Particles, and Nanomaterials
Fibers, particles, and nanoparticles are physical agents that can cause so-called “foreign-body” or “solid-state carcinogenesis” (IARC 1999b; Woo et al. 1988)). A number of mineral fibers and particles including asbestos and crystalline silica have been shown to be carcinogenic or possibly carcinogenic to humans (Table 20.1). Although the mechanisms of fiber/particle carcinogenesis are still not clearly understood, studies of asbestos and other fibrous and nonfibrous particles for the last few decades have revealed a number of physical and chemical characteristics that are related to their biological activities. The toxic and carcinogenic potential of other untested fibers, particles, and nanoparticles may also be predicted using mechanismbased SAR analysis with the understanding of their mechanisms of carcinogenic/ toxic action. 20.3.4.1.
Fibers and Particles
20.3.4.1.1. Mechanisms of Fiber and Particle Carcinogenesis/Toxicity. The mechanisms leading to the development of lung cancers and/or malignant mesothelioma by exposure to fibers and particles are not clearly understood. Asbestos fibers may act as direct or indirect carcinogens. Current hypotheses of fiber-induced tumorigenesis based on cellular and molecular mechanisms include (i) the generation of free radicals that damage DNA, (ii) the physical interference of fibers with cell division, (iii) the fiber-stimulated cellular proliferative response of target cells, (iv) the fiber-induced chronic inflammatory response with release of reactive oxygen
20.3. MECHANISM-BASED SAR ANALYSIS
TABLE 20.1.
535
Carcinogenicity of Fibers and Particlesa
Fibers Group 1. Carcinogenic to Humans Asbestos Erionite
Group 2B. Possibly Carcinogenic to Humans Refractory ceramic fibers Special purpose glass fibers (e.g., E-glass microfibers) Palygorskite (attapulgite), long fibers (length >5 μm)
Group 3. Not Classifiable as to their Carcinogenicity Glass wool Continuous glass filament Rock (stone) wool Slag wool Para-Aramid Wollanstonite Sepiolite Palygorskite (attapulgite), short fibers (length <5 μm)
Particles Group 1. Carcinogenic to Humans Crystalline silica
Group 2B. Possibly Carcinogenic to Humans Titanium dioxideb Carbon blackb
Group 3. Not Classifiable as to their Carcinogenicity Amorphous Silica a
IARC (1987, 1997, 2002).
b
IARC (2006).
species, cytokines, and growth factors, and (v) fibers acting as co-carcinogens or carriers of chemical carcinogens to target cells (Bernstein et al. 2005; IARC 1996). Direct genotoxic and mitogenic effects of asbestos, as well as of some manmade fiber s, have been detected in in vitro assays [e.g., (Dopp et al. 1997; Osgood 1994; Wang et al. 1999)]. The physical presence of phagocytosized fibers can interfere with chromosome segregation during mitosis and result in aneuploidy and
536
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
other chromosomal abnormalities observed in cultured cell systems (Barrett et al. 1989; Jaurand 1997). In vitro data only provide weak evidence for a direct genotoxic action of crystalline silica, and there is no convincing evidence for a direct genotoxic action of crystalline silica in vivo (IARC 1997b). Although direct genotoxicity and other hypothesized mechanisms may play a role and cannot be ruled out in fiber/particle carcinogenesis, inflammation appears to be a key element since it includes cell proliferative effects, oxidant-induced damage via inflammatory cells, and pathogenesis of fibrosis. Recent experimental evidence based on animal models has shown a correlation among inflammation, fibrogenesis, and tumorigenesis of the pulmonary system exposed to inhaled fibers/ particles (IARC 1997a; Woo et al. 1988). A sequential pattern of cellular responses to inhaled fibers/particles is induced in the lung: (a) aggregation of alveolar macrophages, (b) phagocytosis of the particles by macrophages, (c) necrosis/lysis of macrophages with the release of cellular content, including the ingested particles, (d) accumulation of other macrophages and fibroblasts, (e) persistent inflammation of epithelial cells and production of collagen, (f) fibrosis, and (g) tumors (Figure 20.2). The recruitment and activation of alveolar macrophages and inflammatory cells in response to persistent fibers in the lungs are accompanied by release of cytokines, reactive oxygen species that could induce oxidative stress, cell proliferation, fibrosis, and ultimately tumors. Fibers may produce DNA damages through oxidoreductive processes during phagocytosis. Among the various cellular mediators released from alveolar macrophages and epithelial cells, reactive oxygen species (e.g., superoxide anion, hydrogen peroxide) are radiomimetic. It is likely that fibers/ particles are genotoxic via the production of these reactive oxygen species (IARC 1997b; Topinka et al. 2006). A number of in vitro assays have demonstrated the production of reactive oxygen species in rodent alveolar marcrophages and human polymorphonuclear leukocytes exposed to asbestos (Blake et al. 2007) and different
Recruitment of alveolar macrophages ↓ Necrosis of macrophage with the release of active factors, including the ingested particles/fibers ↓ Activation of epithelial cells; release of reactive oxygen/ nitrogen species; causing DNA damages/genetic alteration ↓ Release of cytokines and growth factors by inflammatory cells leading to cell proliferation and fibrosis ↓ Tumors Figure 20.2. Sequential pattern of cellular responses to inhaled particles/fibers.
20.3. MECHANISM-BASED SAR ANALYSIS
537
samples of refractory ceramic fibers (IARC 2002). In vivo and in vitro studies of crystalline silica on alveolar epithelial cells also provide support for an inflammation-dependent mechanism for crystalline silica-induced genotoxicity (IARC 1997b). Up-regulation of growth factor expression has been observed at sites of asbestos fiber deposition in rat lungs: Platelet-derived growth factor (PDGF) and transforming growth factor (TGF)-β are hypothesized to trigger fibroblast proliferation and collagen synthesis, respectively, while TGF-α is mitogenic for alveolar epithelial cells (Brody et al. 1997; Brown et al. 2004). Increased cell proliferation resulting from persistent inflammation and the release of growth factors can enhance the rate of genetic alterations associated with neoplastic transformation. Meanwhile, poor vascularization, impaired intercellular communication, and disturbance of cellular growth regulation, resulting from chronic collagenous fibrosis, can promote tumorigenesis.
20.3.4.1.2. Key Characteristics Related to Fiber and Particle Carcinogenicity/Toxicity. Fibers are generally defined as inorganic materials in the form of elongated particles at least three times longer than wide (aspect ratio ≥3). Both physical and chemical parameters of fibers are related to their biologic activity: fiber geometry and dimensions, biopersistence in the lungs, chemical composition, and surface reactivity (IARC 1996). Dose, dimension, and durability (the three D’s) of fibrous particles are key parameters with respect to their pathogenicity. Although the mechanisms involved in fiber carcinogenesis are not clearly understood, there appears to be a general belief that fiber dimension and tissue burden, which is determined by the doses and rates of deposition and clearance, are of primary importance. Experimental evidence accumulated over the last three decades has shown that long, thin fibers are more carcinogenic than short and thick fibers. The early pioneering work of Stanton and Wrench (1972) and Pott et al. (1974) demonstrated that dimension, in particular the length, of fibers is a most important determinant of fiber pathogenicity (Pott et al. 1974; Stanton and Wrench 1972). A number of studies since then have been published to demonstrate that longer fibers consistently induce greater effects than short fibers (Davis and Jones 1988; Davis et al. 1996; Dogra and Donaldson 1995; Goodglick and Kane 1990; Mossman 1990). It appears that fibers longer than 20 μm have the greatest carcinogenic potency, given other parameters (dose, durability) being equal. This conclusion was also reached more recently from an evaluation of results of a number of in vivo and in vitro studies which affirmed that long, thin fibers (longer than 20 μm, thinner than 1 μm) are the most pathogenic (Miller et al. 1999). The findings that long, thin fibers are more carcinogenic than short, thick fibers can be explained by the aerodynamics of fiber deposition and the clearance mechanism of fibers in the respiratory tract. The deposition of fibers in the lungs is determined primarily by the diameter. In humans, only fibers having a diameter of about 3.5 μm or less are respirable and are readily deposited in the lung by sedimentation. Thin fibers with a length up to 200 μm, on the other hand, may be able to travel to distal segment of the lung and deposit in the alveoli. After deposition, some short
538
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
fibers (usually less than 5 μm) are phagocytized by alveolar macrophages while some are transported to the gastrointestinal tract by the mucociliary system. Some short fibers may also enter the pleural and peritoneal cavities via lymphatic drainage. Since long fibers are only partially engulfed by macrophages, they are less likely to be cleared out of the alveolar compartment and may remain in the lungs for a longer period of time to interact with epithelial cells. Hence, both the aerodynamic consideration of fiber deposition and the clearance mechanism of the fibers in the respiratory tract are consistent with the Stanton hypothesis that long, thin fibers are more carcinogenic than short, thick fibers (Stanton et al. 1977). A limiting factor for alveolar macrophage phagocytosis appears to be the diameter of these cells. The respective values for alveolar macrophage diameters have been reported to range between 10.5 and 13 μm for the rat and between 14 and 21 μm for humans (Crapo et al. 1983; Krombach et al. 1997). There is evidence that the carcinogenic effects of inhaled fibers are most strongly associated with the lung burden (internal dose/biopersistence) of long fibers. In general, biopersistence can be viewed as the sum of the effects of biodurability (e.g., dissolution, leaching) and the result of physiological mechanical clearance mechanisms. The chemical composition of the fiber is an important determinant for the biodurability/biopersistence of the fiber. Some long fibers in the lung can disintegrate (disintegration rate is dependent on their chemical properties), leading to shorter fibers that can be removed by the macrophages. Therefore, fiber chemistry influences fiber carcinogenesis primarily through its role in determining biodurability/biopersistence. In general, fibers with high alkali (Na, K) or alkali earth oxide (e.g., Na2O, K2O) contents and low contents of Al2O3, Fe2O3, TiO2 tend to have high solubility and low durability (Searl 1994). Durability has been measured in vitro using simulated lung fluid on refractory ceramic fibers (RCF) and several other fibers. The in vitro dissolution rate constants, Kdis, in units of ng/cm2/h, were: crocidolite asbestos, 0.15; amosite asbestos, 1.3; E glass wool (MMVF 32), 6.5; RCF (RCF1), 7.6; rock wool (MMVF 21), 17; glass wool (MMVF 10), 195; and slag wool (MMVF 22), 220 (Maxim et al. 1999). Since crocidolite asbestos, amosite asbestos, E glass wool (MMVF 32), and RCF (RCF1) are carcinogenic, and rock wool, glass wool, and slag wool are not (see Table 20.1), the cutoff value of Kdis for fiber carcinogenicity appears to lie between 7.6 and 17. The in vitro dissolution rates have been shown to correlate well with the in vivo biopersistence and pathogenicity of some asbestos and man-made vitreous fibers (Bernstein et al. 2005; Hesterberg et al. 1998; IARC 2002). Biopersistence of various fibers >20 μm expressed as weighted clearance half-time (WT1/2) and 90% clearance time (T-90) was measured in the rat short-term inhalation studies. Fiber types with a WT1/2 >50 days and a T-90 >200 days induced some degree of pulmonary pathology, while fiber types with biopersistence times below these levels induced no, or only transient, inflammation (Hesterberg et al. 1998). However, for fibers with the same dimension, those with a higher surface charge density have been shown to be more carcinogenic, suggesting that the carcinogenicity of fibers may also be related to surface properties such as surface chemistry, reactivity, and surface area of the fiber (Bonneau et al. 1986a,1986b). It has been suggested that the carcinogenicity of some fibers may be a function of the
20.3. MECHANISM-BASED SAR ANALYSIS
539
aspect ratio, the dose, and other chemical properties (e.g., surface charge density) of the fiber/particle; a sufficient quantity of short, thin fibers/particles (e.g., crystalline silica) may also be carcinogenic. For particles, surface activity is the fundamental aspect of their toxicity. Surface reactivity has been shown to be most important for the toxicity/carcinogenicity of some particles such as crystalline silica and TiO2 (Clouter et al. 2001; Fubini 1997; Warheit et al. 2007; Warheit et al. 2006). Particle surface characteristics have been considered to be key factors in free radicals and reactive oxygen species (ROS) formation and in the development of fibrosis and cancer by quartz (crystallized silica). It has been shown that surface modification of silica affects its cytotoxicity, inflammation responses, and fibrogenicity (Duffin et al. 2002; Fubini 1997). Other investigators (Duffin et al. 2002; Oberdörster et al. 1994) have noted that biological effects of particles correlate better with their specific surface area than with their mass. In addition, it has been shown that lung overload of lowsurface-reactivity poorly soluble particles (PSP)(e.g., carbon black, some TiO2) at high lung burden dose expressed as surface area can also elicit inflammatory responses (Duffin et al. 2002; Oberdörster et al. 1994). The relevance of the rat lung response to particle overload for human assessment has been the subject of an international workshop. Because in some human cohorts (e.g., coal workers), very high lung burdens of poorly soluble particles have been observed, the consensus view of the workshop was that there are insufficient data to conclude that the PSP-induced tumor response in the rat model is not relevant for human hazard identification (ILSI 2000).
20.3.4.1.3. Mechanism-Based SAR Analysis of Fibers and Particles. Studies of asbestos and other fibers have shown that the dimension, durability, and dose (the three D’s) of fibrous particles are the key parameters with respect to achieving a sufficient lung burden/biopersistence and their pathogenicity. In general, fibers with a smaller diameter will penetrate deeper in the lungs while long fibers (longer than the diameter of alveolar macrophages) will only be cleared slowly. In addition to fiber length, chemical factors play an important role in fiber durability and biopersistence; fibers with high alkali or alkali earth oxide contents and low contents of Al2O3, Fe2O3, and TiO2 tend to have low durability and hence low biopersistence (Searl 1994). Based on what we know about the mechanisms and key characteristics related to fiber carcinogenicity/toxicity, it appears that those fibers that are durable (e.g., with high contents of Al2O3, Fe2O3, TiO2, or a dissolution rate constant, Kdis <10) and have a diameter <3.5 μm and a length >20 μm are expected to be carcinogenic. For durable and respirable fibers (diameter <3.5 μm) with lengths <20 μm, they may also be carcinogenic if they have high surface reactivity. Positive mechanistic data (e.g., genotoxicity, ROS release, cytotoxicity, inflammation) from in vitro or in vivo studies may lend support to their carcinogenicity/toxicity concern. On the other hand, negative mechanistic data from in vitro studies will not negate their concern since in vitro assays have limitations and often produce conflicting results depending on the cell type, species, and conditions of exposure (Bernstein et al. 2005).
540
CHAPTER 20
TABLE 20.2.
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
Mechanism-based SAR Analysis of Fibers and Particles
Fibers/Particles with Carcinogenicity Potentiala Long fibers Short fibers Particles, crystalline Particles, crystalline
Dimensions
Solubility/Durability/ Biopersistence
Diameter <3.5 μm, length >20 μm Diameter <3.5 μm, length <20 μm Diameter <10 μm Diameter <10 μm
Durable/biopersistent (e.g., Kdis<10) Durable/biopersistent (e.g., Kdis<10) Poorly soluble Poorly soluble
Surface Properties High/low surface reactivity High surface reactivity High surface reactivity Low surface reactivity; high surface area/ lung burden
a
Carcinogenicity potential increases with the increase of aspect ratio (length/diameter), durability/biopersistence, and/or surface properties (reactivity, charge density, area, etc.).
Studies of mineral particles have demonstrated that the toxic and carcinogenic effects are related to the surface area and surface activity of inhaled particles (Duffin et al. 2002; Fubini 1997; Oberdörster et al. 1994). Particle surface characteristics are considered to be key factors in free radicals and reactive oxygen species formation and in the development of fibrosis and cancer by quartz (crystallized silica) (Fubini 1997). Based on what we know about the mechanisms and key characteristics related to the carcinogenicity/toxicity of crystalline silica, it appears that any PSPs that are crystalline and respirable (<10 μm) and have high surface reactivity are expected to have carcinogenic potential. Low-surface-activity PSPs at high lung burden doses expressed as surface area may also be of toxicity/carcinogenicity concern (Table 20.2). 20.3.4.2.
Nanomaterials
20.3.4.2.1. Possible Mechanisms and Key Characteristics of Nanomaterials. A nanoparticle/nanomaterial is generally defined as a particle/ material having a physicochemical structure greater than typical atomic/molecular dimensions but at least one dimension smaller than 100 nm. It includes particles/ materials engineered or manufactured by humans on the nanoscale with specific physicochemical composition and structure to exploit properties and functions associated with its dimensions. Some of the common nanoparticle types are (1) carbon-based materials (e.g., nanotubes, fullerenes), (2) metal-based materials (e.g., nanogold, nanosilver, quantum dots, metal oxides), and (3) dendrimers (e.g., dendritic forms of ceramics). The toxic effects of nanoparticles have not been clearly characterized. Based on analogy to fibers and particles and what we know about their toxicity and mechanisms, it seems possible that some nanomaterials may act similarly as those carcinogenic fibers and particles. It has been suggested that carbon nanotubes could have features of both nanoparticles and fibers and may exhibit some of their effects through oxidative stress and inflammation (Donaldson et al. 2006). Nanoparticles of
20.3. MECHANISM-BASED SAR ANALYSIS
541
various chemical compositions have been shown to generate ROS in both in vivo and in vitro studies (Brown et al. 2001; Wilson et al. 2002). It has been demonstrated that nanoparticles preferentially mobilize to mitochondria and since mitochondria are redox active organelles, nanoparticles may alter ROS production and interfere with anti-oxidant defenses (Li et al. 2003; Oberdörster et al. 2005b). Nanoparticles have also been implicated in interfering with cell signaling via ROS-mediated activation of cytokine gene expression (Brown et al. 2004). Therefore, like fibers and particles, ROS release following exposure to nanoparticles may contribute to oxidative stress, DNA damage, and proliferation of preneoplastic cells leading to cancer. Indeed, a 2-year inhalation study has shown a statistically significant increase in lung cancer in rats exposed to nano-size (15–40 nm) TiO2 at an average concentration of 10 mg/m3 (Heinrich et al. 1995). Exposing the mesothelial lining of the body cavity of mice, as a surrogate for the mesothelial lining of the chest cavity, to long multiwalled carbon nanotubes has also resulted in asbestos-like, length-dependent, pathogenic behavior (Brown et al. 2004; Poland et al. 2008). Because of their nanoscale and unique physicochemical properties, it is believed that some nanoparticles can have toxicological properties that differ from their bulk materials. Many of the special properties of nanoparticles are due to the nano-size and an extremely large surface-to-volume ratio relative to bulk materials. As a particle decreases in size, the surface area increases and a greater proportion of atoms/molecules is found at the surface compared to those inside. Thus, nanoparticles have a much larger surface area per unit mass and a higher potential for biological interaction compared with larger particles. The increase in the surface-to-volume ratio results in the increase of the particle surface energy, which may become reactive. Therefore, as materials reach the nanoscale, they often display different chemical and electronic properties/reactivity and can have toxicological properties that differ from their bulk material. For example, even a typically inert bulk compound such as gold can elicit a biological response when it is introduced as a nanomaterial (Goodman et al. 2004). Shvedova et al. (2005) reported unusual inflammatory and fibrogenic pulmonary responses to specific nanomaterials, suggesting that they may injure the lung by new mechanisms. Similarly, studies conducted by Lam et al. (2004) and Warheit et al. (2004) examining the pulmonary toxicity of carbon nanotubes, have provided evidence that manufactured nanomaterials can display unique toxicity (Lam et al. 2004; Warheit et al. 2004). As particle size decreases, the toxicity of a particle generally increases. Data from some pulmonary studies in rats have demonstrated that exposures to some metal/ metal oxide nanoparticles produced enhanced toxicity responses when compared with larger-sized particles of similar chemical composition. For instance, in a subchronic inhalation study in rats, ultrafine TiO2 particles (20 nm) have been shown to elicit a persistently higher inflammatory reaction in the lungs compared to the largersized (250 nm) TiO2 (Oberdörster 2000; Oberdörster et al. 1994). As with other larger toxic particles, a correlation between particle surface area and toxic effects was observed, supporting the notion for inhaled particles that the particle surface area rather than the mass of the retained particles is the most relevant dose metric. However, the size of nanoparticles alone may not be the critical factor in the toxicity of nanoparticles. Chemical composition is another important parameter for the characterization of nanomaterials, which comprise nearly all substance classes—
542
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
for example, metal/metal oxides, organic/inorganic compounds, polymers, and biomolecules. Some nanomaterials can also be a combination of the above components in core-shell or other complex structures. Dependent on the particle surface chemistry, reactive groups on a particle surface will certainly modify the biological effects. For example, pulmonary exposure to nanoscale TiO2 particles did not produce more cytotoxic or inflammatory effects to the lungs of rats compared to the larger size (in micron) crystalline silica, which has much higher surface reactivity (Warheit et al. 2006). Exposure to nanoparticles of different compositions or crystal structures can produce differential pulmonary effects. The cytotoxicity, cell proliferation, inflammation, and histopathological responses in the lungs were compared in rats instilled intratracheally with P25 ultrafine TiO2 nanoparticles (80/20 anatase/ rutile) and two ultrafine rutile TiO2 nanoparticles of similar sizes and surface areas. Exposure to P25 ultrafine TiO2 nanoparticles (80/20 anatase/rutile) produced marked pulmonary inflammation, cytotoxicity, and adverse lung tissue effects. In contrast, only transient inflammation was produced following exposures to the two ultrafine rutile TiO2 nanoparticles (Warheit et al. 2007). In an in vitro study using human lung epithelial cells or human dermal fibroblasts, anatase TiO2 nanoparticles have been shown to be more chemically reactive and 100 times more cytotoxic than rutile TiO2 nanoparticles of similar particle sizes (Sayes et al. 2006). The larger cytotoxic responses of anatase TiO2 nanoparticles were attributed to a larger production of reactive oxygen species due to a superior photocatalyst of the anatase crystal phase and differences inherent in the crystal structures of the two crystal phases. Under ambient conditions, some nanoparticles can form aggregates or agglomerates. These aggregates/agglomerates have various forms, from dendritic structure to chain or spherical structures. Different aggregate/agglomerate structures/states of single-walled carbon nanotubes (SWCNT) have been associated with distinct regional responses of mice lungs (Shvedova et al. 2005). To maintain the characteristics of nanoparticles, they are often stabilized with coatings or molecular adducts to prevent aggregation/agglomeration. The properties of nanoparticles can be significantly altered by surface modification and the distribution of nanoparticles in the body strongly depends on the surface characteristics. For instance, hydroxylated fullerene (C60) has been demonstrated to be much less toxic than unsubstituted fullerene and induce distinct types of cell death by different mechanisms (Isakovic et al. 2006). Changes of surface properties by coating of various nanoparticles (e.g., nano-CdSe/ZnS) with different types and concentrations of surfactants have also been shown to change their body distribution and the effects on the biological systems significantly (Araujo et al. 1999; Kirchner et al. 2005). Nanoscale materials are known to have various shapes and structures such as spherical, needle-like, tubes, platelets, and so on. The effects of the shape on the toxicity of nanomaterials are unclear. The shape of nanomaterials may have effects on the kinetics of deposition and absorption to the body. Inhaled particles in the nanosize range can certainly deposit in all parts of the respiratory tract including the alveolar region of the lungs. Dependent upon the specific application, oral, dermal, and other routes of exposure are also possible for nanoparticles. Because of their small size, they may pass into cells directly through the cell membrane or penetrate the skin and distribute throughout the body once translocated to the blood circula-
20.3. MECHANISM-BASED SAR ANALYSIS
543
TABLE 20.3. Some Physicochemical Properties that can Influent Toxicity/Carcinogenicity Potential of Nanomaterials
Physicochemical Properties Size/size distribution Shape Agglomeration/ aggregation state Crystal structure
Chemical composition
Surface area
Surface chemistry Surface charge and density
Examples At least one dimension <100 nm, e.g., diameter <20 nm, length >10 μm Spherical, needle-like, tubes, platelets Dendritic to chain or spherical structures Crystalline (e.g., anatase TiO2, rutile TiO2), amorphous Metal/metal oxides, organic/inorganic compounds, polymers as well as biomolecules
Major Effects Surface area, particle surface energy and biological reactivity Biopersistence and the kinetics of deposition and absorption to the body Size/size distribution, and kinetics of deposition and absorption to the body Photocatalytic and surface reactivity/ biological effects Particle solubility/durability, surface chemistry, reactive groups, and particle surface reactivity/biological effects As surface area increases, a greater proportion of atoms/molecules are at the surface compared to those inside, thus having a much larger surface area per unit mass and a higher potential for biological interaction; increase in surface-to-volume ratio also results in the increase of particle surface energy Particle solubility/durability, surface reactivity/biological effects Surface reactivity/biological effects
tion. There is evidence that nanoparticles can translocate from the portal of entry, the respiratory tract, via different pathways to other organs/tissues and makes them uniquely different from larger-sized particles in that they may induce direct adverse responses in remote organs. For example, there are data from animal studies showing possible translocation of inhaled nanoparticles to the nervous system and other organs/tissues (Oberdörster 2001; Oberdörster et al. 2002). The toxic/carcinogenic effects of extra-pulmonary organs/tissues remain to be investigated. Some physicochemical properties that can influence the toxic/carcinogenic potential of nanomaterials are shown in Table 20.3. 20.3.4.2.2. Mechanism-Based SAR Analysis of Nanomaterials. A number of studies have demonstrated that nanoparticle toxicity is extremely complex and there is a strong likelihood that biological activity of nanoparticles will depend
544
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
on a variety of physicochemical properties such as particle size/size distribution, shape, agglomeration state, crystal structure, chemical composition, surface area, and surface properties (chemistry, surface charge, etc.) (Derfus et al. 2004; Nemmar et al. 2003; Oberdörster et al. 2005a,b; Sayes et al. 2006). At present, the information on the toxicity/carcinogenicity of nanomaterials is too limited to allow SAR analysis for predicting their toxic/carcinogenic potential, especially on extrapulmonary organs/tissues. Based on the possible mechanisms of action and limited toxicity database of some nanomaterials, it appears that surface reactivity may be the critical factor in the toxicity of nanomaterials. While the nano-size (and thus increased surface area) alone may contribute to increased surface reactivity for some nanomaterials (e.g., nano-gold, nano-silver), other physicochemical properties (e.g., dimension and shape, agglomeration state, chemical composition, crystal structure, surface coating) may be modifiers and important determinants to their surface reactivity and thus their cytotoxic/carcinogenic potential (Figure 20.3). In order to identify and better understand the key element(s) of engineered nanoparticles that may contribute to the mechanisms of nanoparticle toxicity for risk assessment, it is important to develop standard toxicity test protocols and reference materials representing each nanoparticle type with various standardized properties. A number of national and international projects have been initiated to develop standard toxicity test protocols, database collection, and computerized SAR models development [e.g., Aitken et al. (2007), NSTC (2007)].
20.4. USES OF (Q)SAR IN CANCER HAZARD/RISK ASSESSMENT AND BRIEF OVERVIEW OF PREDICTIVE SYSTEMS/SOFTWARES 20.4.1. Evolving Uses of (Q)SAR in Cancer Hazard Identification and Risk Assessment (Q)SAR analysis was originally developed mainly as a research tool for laboratory scientists. It has now evolved into a multipurpose screening or even regulatory tool for a wide spectrum of users (Doull et al. 2007). The possible uses may be loosely classified into the following categories: (a) Product Development: (Q)SAR is a crucial tool in the research and development of new products in the pharmaceutical and chemical industries. In addition to contributing to the discoveries and development of products with desired applications, (Q)SAR has also been used to design environmentally safer chemicals, as well as to search for safer substitutes to existing chemicals. (b) Hazard Identification and Risk Assessment: For untested chemicals, (Q)SAR has been used to identify potential hazards, design experimental studies, and select surrogate/analogue chemicals that may be used for interim risk assessments. For tested chemicals, (Q)SAR may be used to (a) provide input into weight-of-evidence assessments, (b) elucidate potential mechanisms of action,
Nanomaterials Chemical Factors
Physical Factors
Agglomerate/ Aggregate
Shape
Size
Surface Coating
Crystallinity
Chemical Composition
Surface Area
Deposition Kinetics
Surface Chemistry Surface Energy
Tissue Deposition & Clearance
Surface Reactivity
Biopersistence
Biological Effects (e.g., cytotoxicity, cancer)
545
Figure 20.3.
Physicochemical properties that can modify the biological effects of nanomaterials.
Solubulity & Durability
546
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
(c) assess human relevance of animal data, and (d) contribute to selection of quantitative risk assessment methodology. (c) Health and Environmental Protection: (Q)SAR has been widely used to assess the human health hazard of chemicals and to protect workers and general public from exposure to a variety of chemicals from various media [e.g., food additives (Mayer et al. 2008)]. It also has been used as a tool for prioritization of environmentally occurring chemicals (e.g., air pollutants or disinfection byproducts in drinking water) for testing and monitoring so that limited resources can be used effectively [e.g., Woo et al. 2002)]. (d) Regulation: (Q)SAR has been used extensively for regulating new chemicals for which test data are not available or are inadequate [e.g., Lai and Woo (2001)]. There also is increasing worldwide interest [e.g., Cronin and Schultz (2003)] in expanding the use for other regulatory purposes such as grouping chemicals into categories, filling data gaps for classification and labeling, listing in or delisting from hazardous substances lists, or even outright banning. Depending on the impact of such regulation, (Q)SAR should be used in conjunction with biological predictive tests along with articulation of limitations and uncertainties.
20.4.2. Brief Overview of (Q)SAR Systems/Softwares for Predicting Carcinogenic Potential of Chemicals Depending on the specific screening needs and time limitation, a variety of (Q)SAR predictive methods may be used. Among these, OncoLogic, MultiCase, TOPKAT, and DEREK have been used for years, whereas the newer additions include Lazar, and ToxTree. OncoLogic, Lazar, and ToxTree are now all publicly accessible or downloadable, whereas MultiCase, TOPKAT, and DEREK are commercial programs. These programs are briefly discussed below along with their links if available. OncoLogic is a mechanism-based, human expert/knowledge-rule-based expert system program. It was jointly developed by the U.S. EPA and LogicChem, Inc. and is now freely downloadable from the Agency (http://www.epa.gov/oppt/ newchems/tools/oncologic.htm). The current version 6.0 is capable of making prediction of carcinogenic potential of over 50 organic chemical classes, fibers, polymers, and some metals. The input may be chemical structure, name, or CAS number. The output consists of prediction of semiquantitative ranking of carcinogenic potential along with reasoning and rationale. Lazar is a relatively new, freely accessible (http://lazar.in-silico.de/) tool for predicting carcinogenic activity of chemicals based on statistical/correlative approach. The system was trained using the approximately 1500 compounds in Carcinogenic Potency Database (CPDB or Gold Database). Input is chemical structure by drawing package or SMILES formula. Output includes prediction of “+” or “−” activity along with rationales (structure features and similar compounds) and reliability index that may provide some indication of whether the query chemical is within the predictive domain of the program.
20.4. BRIEF OVERVIEW OF PREDICTIVE SYSTEMS/SOFTWARES
547
ToxTree is the newest cancer hazard estimation tool downloadable at the European Chemical Bureau website (http://ecb.jrc.it/qsar/qsar-tools/index.php? c=TOXTREE). It is a structural alert-based program with a decision tree approach. MultiCASE MC4PC is a statistical/correlative program that contains a predefined predictive module for carcinogenicity, which was designed by capturing SAR information from approximately 1200 compounds with well-defined chemical structures and carcinogenicity data. The program uses an algorithm that reduces each of these compounds into all possible 2–10 atom fragments. It creates a training set (library) relating these fragments to their carcinogenic potential, and other modulators. The fragments of a compound of unknown carcinogenicity can then be generated and compared to the fragment library to arrive at computational scores. Results of analysis are subject to a set of human expert rules to arrive at a prediction as to whether the test molecule is active and to determine the level of confidence with prediction (www.multicase.com/products/prod01.htm). TOPKAT (Toxicity Prediction by Komputer-Assisted Technology) is a correlative program using Kier & Hall electrotopological states (E-states) as well as shape, symmetry, molecular weight, and log P as descriptors to build statistically robust Quantitative Structure Toxicity Relationship (QSTR) models for a variety of endpoints including carcinogenicity. It validates its assessments via a univariate analysis of the descriptors, a patented multivariate analysis of the fit of the query structure in Optimum Prediction Space, and by similarity searching in descriptor space (Enslein et al. 1994). TOPKAT has been used to predict carcinogenic activity both qualitatively and quantitatively. DEREK for Windows is a computer-based application that uses a knowledge base approach to predict toxic hazards (including carcinogenicity) of chemicals. It is developed by Lhasa Limited (http://www.lhasalimited.org/index.php? cat=221&owner=220&sub_cat=221), with SAR contribution from members. The cancer module contains alerts describing structural features associated with carcinogenicity. Each alert is supported by comment, references, and examples describing the evidence on which it has been based. A number of comparative/validation studies have been conducted using external data or prospective bioassay studies from the US NTP [e.g., Benigni and Zito (2004), Mayer et al. (2008)]. Essentially, the predictive performance may differ from different batches of chemicals. Each of these software packages has its own unique strengths and weaknesses for which the user should be aware. It is advisable to “test drive” the software using tested chemicals that are structurally related to your query chemical to ascertain that the software is properly trained. Use of more than one software is also advisable to compare and contrast in light of the rationales given. In addition to the above software packages, QSAR analysis of chemical carcinogens may also be conducted using QSAR equations [e.g., Benigni (2003, 2005)]. Most of these studies used the TD50 values in the Gold Database. In general, QSAR equations developed from using structurally homogeneous series of chemical are much more reliable than those using heterogeneous series, particularly if the statistical association can also be backed by mechanistic understanding (Doull et al. 2007).
548
20.5.
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
FUTURE PERSPECTIVES
(Q)SAR analysis is an indispensable screening tool in product development, hazard identification and risk assessment. However, for the cancer endpoint, much more work is still needed before (Q)SAR can be used with confidence for regulatory purposes. Several future directions should be explored and include the following: Critical Evaluation and Improvement of Current (Q)SAR. There are now many (Q)SAR predictive software programs for cancer, and more are being developed. Although internally cross-validated, their predictive capabilities for chemicals outside training sets remain to be affirmed. More critical evaluations such as the US NTP predictive exercises [e.g., Benigni and Zito (2004)] should be conducted to gain valuable feedback to accentuate strengths and reduce weaknesses of each program. Combining several different methods with complementing strengths may improve the predictive capability. Expressions of confidence in the results and the attendant uncertainty, avoidance of false negatives due to lack of information and false positive due to overconservatism, and defining of predictive domains are important issues to consider. With the increasing regulatory attention to mode of action and human relevancy [e.g., Cohen et al. (2004); Meek et al. (2003)], (Q)SAR programs trained using animal data alone without mechanistic input should be improved. Ideally, (Q)SAR programs should be able to predict not only carcinogenic potential but also possible target organ(s), likely mechanism of action, human relevancy, and susceptible exposure scenarios, along with articulation of rationale and data gaps, if any. Expansion of Database/Knowledge Base. The predictive capability of (Q) SAR methods is limited by the input of training knowledge/data/information. In the past several years, a variety of new toxicology databases/knowledge bases have been constructed or initiated. These include the US National Institute of Enviromental Health Sciences’ Genetic Alterations in Cancer (GAC) database and the Chemical Effects in Biological Systems (CEBS) knowledge base [see Waters and Jackson (2008)], as well as the U.S. EPA’s Distributed Structure-Searchable Toxicity (DSSTox) database (http://www.epa.gov/NCCT/dsstox/) and the High Production Volume Information System (HPVIS) (http://www.epa.gov/hpvis/). The most recent EPA additions include (a) Medium Production Volume (MPV) chemical cluster analysis under the Chemical Assessment and Management Program (ChAMP) (http://www.epa.gov/champ/pubs/hbpdocs.htm), (b) Toxicology Reference Database (ToxRefDB) program of rodent bioassay data of over 300 pesticides [see Martin et al. (2009)], and (c) Aggregated Chemical Toxicity Resource (ACToR) (http:// actor.epa.gov/actor/) (Judson et al. 2008). These databases/knowledge bases are invaluable resources for data mining to find structural alerts, structural features or boundaries, biomarkers, and toxicological/genomic signatures useful for predicting the carcinogenic potential of chemicals, as well as better mechanistic understanding for rational risk assessment. Utilization of Input from Emerging Technologies/Computational Biology/ Bioinformatics. Functional criteria or activity–activity relationships analysis of
20.5. FUTURE PERSPECTIVES
549
using short-term tests or biomarkers to predict carcinogenicity has long been used as an alternative predictive tool. The powerful emerging technologies—“omics” technology (genomics, proteomics, metabonomics, systemomics, etc.), and highthroughput screening (HTS) assays—have now been increasingly utilized along with computational biology and bioinformatics tools to serve as an alternative predictive tool, as well as to provide support or supplement (Q)SAR for the most effective predictions. Toxicogenomics studies have shown great success in helping to identify mechanisms of action of chemical carcinogens [e.g., Ellinger-Ziegelbauer et al. (2009), Hester et al. (2006), Ward et al. (2006)]. The Office of Pesticide Programs of the EPA has been collaborating with the EPA’s research laboratories to explore the use of toxicogenomic input as a part of weight-of-evidence for determining mode of action and human relevancy of carcinogenic pesticides (Hester et al. 2006; Martin et al. 2007; Ward et al. 2006). Toxicogenomic analysis of gene expression profiles has also shown great promise as a prioritizing or predictive tool for identifying potential carcinogens [e.g., Ellinger-Ziegelbauer et al. (2008)]. A pan-European carcinoGENOMICS project for the development of omics-based in vitro carcinogenicity screening assays is currently under way (Vinken et al. 2008). In 2007, the EPA launched a large-scale ToxCast project (Dix et al. 2007) to develop predictive HTS and genomic bioactivity signatures useful for screening toxicants (including carcinogens), characterizing toxicity pathways, and prioritizing further testing. The readers are referred to the ToxCast website (http://www.epa.gov/ncct/ toxcast/) for details and progress reports. Expansion of Integrative Approaches. It was long recognized that (Q)SAR analysis based on structural/chemoinformatic input alone is not sufficient to predict all chemical carcinogens; functional/bioinformatic input is often needed to optimize predictive capability (Woo et al. 1985a, 1988). An integrative approach of combining mechanistically complementary short-term biological predictive tests as a basis to predict carcinogenic potential of chemicals was developed and used in the EPA’s OncoLogic expert system. (Woo et al. 1998). With the increasing use of toxicogenomics and large-scale HTS testing, expansion of high-quality databases, new approaches to data integration and relational exploration [e.g., Richard (2006)], and better understanding of molecular and systems biology, more effective integrative approaches to predict/screen carcinogens are being developed. In addition to the ToxCast project mentioned above, several recent publications— (a) use of cell viability HTS to improve QSAR (Zhu et al. 2008), (b) use of integrative approach of considering multiple mechanisms and toxicogenomics in carcinogen prediction (Guyton et al. 2009), (c) advances in database aggregation toward development of a systems biology-based predictive system (Waters and Jackson 2008)—all offer some insight of integrative approaches for the most effective strategy for predicting and screening potential carcinogens for health and environmental protection. Disclaimer. The scientific views expressed in this chapter are solely those of the authors and do not necessary reflect the views and policies of the U.S. Environmental Protection Agency.
550
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
REFERENCES Aitken, R. J., Hankin, S. M., Tran, C. L., Donaldson, K., Stone, V., Cumpson, P., Johnstone, J., Chaundhry, Q., and Cash, S. (2007). REFNANO: Reference materials for engineered nanoparticle toxicology and metrology. Institute of Occupational Medicine (IOM) United Kingdom. Final report on Project CB01099, http://www.iom-world.org/pubs/REFNANOReport.pdf, 1–30. Araujo, L., Lobenberg, R., and Kreuter, J. (1999). Influence of the surfactant concentration on the body distribution of nanoparticles. J Drug Target 6, 373–385. Arcos, J. C., Woo, Y. T., and Argus, M. F. (1982). Chemical Induction of Cancer, Vol. IIIA: Aliphatic Carcinogens, Academic Press, New York. Barrett, J. C., Lamb, P. W., and Wiseman, R. W. (1989). Multiple mechanisms for the carcinogenic effects of asbestos and other mineral fibers. Environ Health Perspect 81, 81–89. Benigni, R., ed. (2003). Quantitative Structure–Activity Relationship (QSAR) Models of Mutagens and Carcinogens, CRC Press, Boca Raton, FL. Benigni, R. (2005). Structure–activity relationship studies of chemical mutagens and carcinogens: mechanistic investigations and prediction approaches. Chem Rev 105, 1767–1800. Benigni, R., and Bossa, C. (2008). Structure alerts for carcinogenicity, and the Salmonella assay system: A novel insight through the chemical relational databases technology. Mutat Res 659, 248–261. Benigni, R., Netzeva, T. I., Benfenati, E., Bossa, C., Franke, R., Helma, C., Hulzebos, E., Marchant, C., Richard, A., Woo, Y. T., and Yang, C. (2007). The expanding role of predictive toxicology: An update on the (Q)SAR models for mutagens and carcinogens. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev 25, 53–97. Benigni, R., and Zito, R. (2004). The second National Toxicology Program comparative exercise on the prediction of rodent carcinogenicity: Definitive results. Mutat Res 566, 49–63. Bentley, P., Calder, I., Elcombe, C., Grasso, P., Stringer, D., and Wiegand, H. J. (1993). Hepatic peroxisome proliferation in rodents and its significance for humans. Food Chem Toxicol 31, 857–907. Bernstein, D., Castranova, V., Donaldson, K., Fubini, B., Hadley, J., Hesterberg, T., Kane, A., Lai, D., McConnell, E. E., Muhle, H., Oberdorster, G., Olin, S., and Warheit, D. B. (2005). Testing of fibrous particles: Short-term assays and strategies. Inhal Toxicol 17, 497–537. Blake, D. J., Bolin, C. M., Cox, D. P., Cardozo-Pelaez, F., and Pfau, J. C. (2007). Internalization of Libby amphibole asbestos and induction of oxidative stress in murine macrophages. Toxicol Sci 99, 277–288. Bonneau, L., Malard, C., and Pezerat, H. (1986a). Studies on surface properties of asbestos. II. Role of dimensional characteristics and surface properties of mineral fibers in the induction of pleural tumors. Environ Res 41, 268–275. Bonneau, L., Suquet, H., Malard, C., and Pezerat, H. (1986b). Studies on surface properties of asbestos. I. Active sites on surface of chrysotile and amphiboles. Environ Res 41, 251–267. Borghoff, S. J., Miller, A. B., Bowen, J. P., and Swenberg, J. A. (1991). Characteristics of chemical binding to alpha 2u-globulin in vitro—Evaluating structure-activity–relationships. Toxicol Appl Pharmacol 107, 228–238. Brody, A. R., Liu, J. Y., Brass, D., and Corti, M. (1997). Analyzing the genes and peptide growth factors expressed in lung cells in vivo consequent to asbestos exposure and in vitro. Environ Health Perspect 105(Suppl 5), 1165–1171. Brown, D. M., Donaldson, K., Borm, P. J., Schins, R. P., Dehnhardt, M., Gilmour, P., Jimenez, L. A., and Stone, V. (2004). Calcium and ROS-mediated activation of transcription factors and TNF-alpha cytokine gene expression in macrophages exposed to ultrafine particles. Am J Physiol Lung Cell Mol Physiol 286, L344–L353. Brown, D. M., Wilson, M. R., MacNee, W., Stone, V., and Donaldson, K. (2001). Size-dependent proinflammatory effects of ultrafine polystyrene particles: A role for surface area and oxidative stress in the enhanced activity of ultrafines. Toxicol Appl Pharmacol 175, 191–199. Cattley, R. C., DeLuca, J., Elcombe, C., Fenner-Crisp, P., Lake, B. G., Marsman, D. S., Pastoor, T. A., Popp, J. A., Robinson, D. E., Schwetz, B., Tugwood, J., and Wahli, W. (1998). Do peroxisome proliferating compounds pose a hepatocarcinogenic hazard to humans? Regul Toxicol Pharmacol 27, 47–60.
REFERENCES
551
Clouter, A., Brown, D., Hohr, D., Borm, P., and Donaldson, K. (2001). Inflammatory effects of respirable quartz collected in workplaces versus standard DQ12 quartz: Particle surface correlates. Toxicol Sci 63, 90–98. Cohen, S. M., Klaunig, J., Meek, M. E., Hill, R. N., Pastoor, T., Lehman-McKeeman, L., Bucher, J., Longfellow, D. G., Seed, J., Dellarco, V., Fenner-Crisp, P., and Patton, D. (2004). Evaluating the human relevance of chemically induced animal tumors. Toxicol Sci 78, 181–186. Crapo, J. D., Young, S. L., Fram, E. K., Pinkerton, K. E., Barry, B. E., and Crapo, R. O. (1983). Morphometric characteristics of cells in the alveolar region of mammalian lungs. Am Rev Respir Dis 128, S42–S46. Cronin, M. T. D., and Schultz, T. W. (2003). Pitfalls in QSAR. J Mol Struct: THEOCHEM 622, 39–51. Davis, J. M., and Jones, A. D. (1988). Comparisons of the pathogenicity of long and short fibres of chrysotile asbestos in rats. Br J Exp Pathol 69, 717–737. Davis, J. M. G., Brown, D., Cullen, R. T., Donaldson, K., Jones, A. D., Miller, B. C., McIntosh, C., and Searl, A. (1996). A comparison of methods of determining and predicting the pathogenicity of mineral fibers. Inhal Toxicol 8, 747–770. Derfus, A. M., Chan, W. C. W., and Bhatia, S. N. (2004). Probing the cytotoxicity of semiconductor quantum dots. Nano Lett 4, 11–18. Dix, D. J., Houck, K. A., Martin, M. T., Richard, A. M., Setzer, R. W., and Kavlock, R. J. (2007). The ToxCast program for prioritizing toxicity testing of environmental chemicals. Toxicol Sci 95, 5–12. Dogra, S., and Donaldson, K. (1995). Effect of long and short fibre amosite asbestos on in vitro TNF production by rat alveolar macrophages: The modifying effect of lipopolysaccharide. Ind Health 33, 131–141. Donaldson, K., Aitken, R., Tran, L., Stone, V., Duffin, R., Forrest, G., and Alexander, A. (2006). Carbon nanotubes: A review of their properties in relation to pulmonary toxicology and workplace safety. Toxicol Sci 92, 5–22. Dopp, E., Schuler, M., Schiffmann, D., and Eastmond, D. A. (1997). Induction of micronuclei, hyperdiploidy and chromosomal breakage affecting the centric/pericentric regions of chromosomes 1 and 9 in human amniotic fluid cells after treatment with asbestos and ceramic fibers. Mutat Res 377, 77–87. Doull, J., Borzelleca, J. F., Becker, R., Daston, G., DeSesso, J., Fan, A., Fenner-Crisp, P., Holsapple, M., Holson, J., Craig Llewellyn, G., MacGregor, J., Seed, J., Walls, I., Woo, Y. T., and Olin, S. (2007). Framework for use of toxicity screening tools in context-based decision-making. Food Chem Toxicol 45, 759–796. Duffin, R., Tran, C. L., Clouter, A., Brown, D. M., MacNee, W., Stone, V., and Donaldson, K. (2002). The importance of surface area and specific reactivity in the acute pulmonary inflammatory response to particles. Ann Occup Hyg 46, 242–245. Ellinger-Ziegelbauer, H., Aubrecht, J., Kleinjans, J. C., and Ahr, H. J. (2009). Application of toxicogenomics to study mechanisms of genotoxicity and carcinogenicity. Toxicol Lett 186, 36–44. Ellinger-Ziegelbauer, H., Gmuender, H., Bandenburg, A., and Ahr, H. J. (2008). Prediction of a carcinogenic potential of rat hepatocarcinogens using toxicogenomics analysis of short-term in vivo studies. Mutat Res 637, 23–39. Enslein, K., Gombar, V. K., and Blake, B. W. (1994). International Commission for Protection Against Environmental Mutagens and Carcinogens. Use of SAR in computer-assisted prediction of carcinogenicity and mutagenicity of chemicals by the TOPKAT program. Mutat Res 305, 47–61. EPA (1991). Alpha2u-globulin: Association with chemically induced renal toxicity and neoplasia in the male rat. EPA/625/3-91/091F, 1–118. EPA (1998). Assessment of thyroid follicular cell tumors. EPA/630/R-97/002, 1–42. Fubini, B. (1997). Surface reactivity in the pathogenic response to particulates. Environ Health Perspect 105(Suppl 5), 1013–1020. Gonzalez, F. J., and Shah, Y. M. (2008). PPARalpha: Mechanism of species differences and hepatocarcinogenesis of peroxisome proliferators. Toxicology 246, 2–8. Goodglick, L. A., and Kane, A. B. (1990). Cytotoxicity of long and short crocidolite asbestos fibers in vitro and in vivo. Cancer Res 50, 5153–5163.
552
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
Goodman, C. M., McCusker, C. D., Yilmaz, T., and Rotello, V. M. (2004). Toxicity of gold nanoparticles functionalized with cationic and anionic side chains. Bioconjug Chem 15, 897–900. Guyton, K. Z., Kyle, A. D., Aubrecht, J., Cogliano, V. J., Eastmond, D. A., Jackson, M., Keshava, N., Sandy, M. S., Sonawane, B., Zhang, L., Waters, M. D., and Smith, M. T. (2009). Improving prediction of chemical carcinogenicity by considering multiple mechanisms and applying toxicogenomic approaches. Mutat Res 681, 230–240. Heinrich, U., Fuhst, R., Rittinghausen, S., Creutzenberg, O., Bellmann, B., Koch, W., and Levsen, K. (1995). Chronic inhalation exposure of Wistar rats and two different strains of mice to diesel engine exhaust, carbon black, and titanium dioxide. Inhal Toxicol 7, 533–556. Helma, C. (2004). Data mining and knowledge discovery in predictive toxicology. SAR QSAR Environ Res 15, 367–383. Hester, S. D., Wolf, D. C., Nesnow, S., and Thai, S. F. (2006). Transcriptional profiles in liver from rats treated with tumorigenic and non-tumorigenic triazole conazole fungicides: Propiconazole, triadimefon, and myclobutanil. Toxicol Pathol 34, 879–894. Hesterberg, T. W., Hart, G. A., Chevalier, J., Miiller, W. C., Hamilton, R. D., Bauer, J., and Thevenaz, P. (1998). The importance of fiber biopersistence and lung dose in determining the chronic inhalation effects of X607, RCF1, and chrysotile asbestos in rats. Toxicol Appl Pharmacol 153, 68–82. Hill, R. N., Crisp, T. M., Hurley, P. M., Rosenthal, S. L., and Singh, D. V. (1998). Risk assessment of thyroid follicular cell tumors. Environ Health Perspect 106, 447–457. IARC (1987). Overall evaluation of carcinogenicity. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. International Agency for Research on Cancer, Suppl. 7. Lyon, France. IARC (1994). Peroxisome proliferation and its role in carcinogenesis: Views and expert opinions of an IARC working group, Lyon, 7–11 December 1994. IARC Technical Publication No. 24, http:// monographs.iarc.fr/ENG/Publications/techrep24/IARCrep24.pdf. IARC (1996). Mechanisms of fibre carcinogenesis. IARC Scientific Publications 140, 1–135. IARC (1997). Silica, Some Silicates, Coal Dust and para-Aramid Fibrils. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans 68, http://monographs.iarc.fr/ENG/Monographs/vol68/ index.php. IARC (1999a). Species differences in thyroid, kidney and urinary bladder carcinogenesis. IARC Sci Pub 147, 1–225. IARC (1999b). Surgical implants and other foreign bodies. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans 74, http://monographs.iarc.fr/ENG/Monographs/vol74/index.php. IARC (2002). Man-made vitreous fibers. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. International Agency for Research on Cancer 81, Lyon, France. IARC (2006). Carbon black. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. International Agency for Research on Cancer 93, Lyon, France. http://monographs.iarc.fr/ENG/ Meetings/93-carbonblack.pdf. ILSI (2000). The relevance of the rat lung response to particle overload for human risk assessment: A workshop consensus report. ILSI Risk Science Institute Workshop Participants. Inhal Toxicol 12, 1–17. Isakovic, A., Markovic, Z., Todorovic-Markovic, B., Nikolic, N., Vranjes-Djuric, S., Mirkovic, M., Dramicanin, M., Harhaji, L., Raicevic, N., Nikolic, Z., and Trajkovic, V. (2006). Distinct cytotoxic mechanisms of pristine versus hydroxylated fullerene. Toxicol Sci 91, 173–183. Jaurand, M. C. (1997). Mechanisms of fiber-induced genotoxicity. Environ Health Perspect 105 (Suppl 5), 1073–1084. Judson, R., Richard, A., Dix, D., Houck, K., Elloumi, F., Martin, M., Cathey, T., Transue, T. R., Spencer, R., and Wolf, M. (2008). ACToR—Aggregated computational toxicology resource. Toxicol Appl Pharmacol 233, 7–13. Kavlock, R. J., Ankley, G., Blancato, J., Breen, M., Conolly, R., Dix, D., Houck, K., Hubal, E., Judson, R., Rabinowitz, J., Richard, A., Setzer, R. W., Shah, I., Villeneuve, D., and Weber, E. (2008). Computational toxicology—A state of the science mini review. Toxicol Sci 103, 14–27. Kirchner, C., Liedl, T., Kudera, S., Pellegrino, T., Munoz Javier, A., Gaub, H. E., Stolzle, S., Fertig, N., and Parak, W. J. (2005). Cytotoxicity of colloidal CdSe and CdSe/ZnS nanoparticles. Nano Lett 5, 331–338. Klaunig, J. E., Babich, M. A., Baetcke, K. P., Cook, J. C., Corton, J. C., David, R. M., DeLuca, J. G., Lai, D. Y., McKee, R. H., Peters, J. M., Roberts, R. A., and Fenner-Crisp, P. A. (2003). PPARalpha
REFERENCES
553
agonist-induced rodent tumors: modes of action and human relevance. Crit Rev Toxicol 33, 655–780. Krombach, F., Munzing, S., Allmeling, A. M., Gerlach, J. T., Behr, J., and Dorger, M. (1997). Cell size of alveolar macrophages: an interspecies comparison. Environ Health Perspect 105(Suppl 5), 1261–1263. Lai, D. Y. (2004). Rodent carcinogenicity of peroxisome proliferators and issues on human relevance. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev 22, 37–55. Lai, D. Y., and Woo, Y. T. (2001). Structure–activity relationship (SAR) and its roles in cancer risk assessment. In Genetic Toxicology and Cancer Risk Assessment, Choy, W. N., ed., Marcel Dekker, New York, pp. 1–406. Lai, D. Y., Woo, Y. T., Argus, M. F., and Arcos, J. C. (1996). Cancer risk reduction through mechanismbased molecular design of chemicals. In Designing Safer Chemicals: Green Chemistry for Pollution Prevention, DeVito, S. C., and Garrett, R. L., eds., American Chemical Society, Washington, D.C., pp. 62–73. Lake, B. G., and Lewis, D. F. V. (1993). Structure–activity relationships for chemically induced peroxisome proliferation in mammalian liver. In Peroxisomes: Biology and Importance in Toxicology and Medicine, Gibson, G. G., and Lake, B. G., eds., Taylor & Francis, London, pp. 313–342. Lam, C. W., James, J. T., McCluskey, R., and Hunter, R. L. (2004). Pulmonary toxicity of singlewall carbon nanotubes in mice 7 and 90 days after intratracheal instillation. Toxicol Sci 77, 126–134. Li, N., Sioutas, C., Cho, A., Schmitz, D., Misra, C., Sempf, J., Wang, M., Oberley, T., Froines, J., and Nel, A. (2003). Ultrafine particulate pollutants induce oxidative stress and mitochondrial damage. Environ Health Perspect 111, 455–460. Lijinsky, W. (1992). Chemistry and Biology of N-Nitroso Compounds, Cambridge University Press, Cambridge. Lock, E. A., Charbonneau, M., Strasser, J., Swenberg, J. A., and Bus, J. S. (1987). 2,2,4-Trimethylpentaneinduced nephrotoxicity. II. The reversible binding of a TMP metabolite to a renal protein fraction containing alpha 2u-globulin. Toxicol Appl Pharmacol 91, 182–192. Martin, M. T., Brennan, R. J., Hu, W., Ayanoglu, E., Lau, C., Ren, H., Wood, C. R., Corton, J. C., Kavlock, R. J., and Dix, D. J. (2007). Toxicogenomic study of triazole fungicides and perfluoroalkyl acids in rat livers predicts toxicity and categorizes chemicals based on mechanisms of toxicity. Toxicol Sci 97, 595–613. Martin, M. T., Judson, R. S., Reif, D. M., Kavlock, R. J., and Dix, D. J. (2009). Profiling chemicals based on chronic toxicity results from the U.S. EPA ToxRef Database. Environ Health Perspect 117, 392–399. Maxim, L. D., Mast, R. W., Utell, M. J., Yu, C. P., Boymel, P. M., Zoitos, B. K., and Cason, J. E. (1999). Hazard assessment and risk analysis of two new synthetic vitreous fibers. Regul Toxicol Pharmacol 30, 54–74. Mayer, J., Cheeseman, M. A., and Twaroski, M. L. (2008). Structure–activity relationship analysis tools: Validation and applicability in predicting carcinogens. Regul Toxicol Pharmacol 50, 50–58. Meek, M. E., Bucher, J. R., Cohen, S. M., Dellarco, V., Hill, R. N., Lehman-McKeeman, L. D., Longfellow, D. G., Pastoor, T., Seed, J., and Patton, D. E. (2003). A framework for human relevance analysis of information on carcinogenic modes of action. Crit Rev Toxicol 33, 591–653. Miller, B. G., Jones, A. D., Searl, A., Buchanan, D., Cullen, R. T., Soutar, C. A., Davis, J. M., and Donaldson, K. (1999). Influence of characteristics of inhaled fibres on development of tumours in the rat lung. Ann Occup Hyg 43, 167–179. Mossman, B. T. (1990). In vitro studies on the biologic effects of fibers: Correlation with in vivo bioassays. Environ Health Perspect 88, 319–322. Nemmar, A., Hoylaerts, M. F., Hoet, P. H., Vermylen, J., and Nemery, B. (2003). Size effect of intratracheally instilled particles on pulmonary inflammation and vascular thrombosis. Toxicol Appl Pharmacol 186, 38–45. NSTC (2007). Prioritization of environmental, health, and safety research needs for engineered nanoscale materials. Nanotechnology Environmental and Health Implications Working Group, National Science and Technology Council (NSTC), http://www.nano.gov/Prioritization_EHS_Research_Needs_ Engineered_Nanoscale_Materials.pdf, 1–8.
554
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
Oberdörster, G. (2000). Toxicology of ultrafine particles: In vivo studies. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 358, 2719–2740. Oberdörster, G. (2001). Pulmonary effects of inhaled ultrafine particles. Int Arch Occup Environ Health 74, 1–8. Oberdörster, G., Ferin, J., and Lehnert, B. E. (1994). Correlation between particle size, in vivo particle persistence, and lung injury. Environ Health Perspect 102(Suppl 5), 173–179. Oberdörster, G., Maynard, A., Donaldson, K., Castranova, V., Fitzpatrick, J., Ausman, K., Carter, J., Karn, B., Kreyling, W., Lai, D., Olin, S., Monteiro-Riviere, N., Warheit, D., and Yang, H. (2005a). Principles for characterizing the potential human health effects from exposure to nanomaterials: Elements of a screening strategy. Part Fibre Toxicol 2, 8. Oberdörster, G., Oberdörster, E., and Oberdörster, J. (2005b). Nanotoxicology: An emerging discipline evolving from studies of ultrafine particles. Environ Health Perspect 113, 823–839. Oberdörster, G., Sharp, Z., Atudorei, V., Elder, A., Gelein, R., Lunts, A., Kreyling, W., and Cox, C. (2002). Extrapulmonary translocation of ultrafine carbon particles following whole-body inhalation exposure of rats. J Toxicol Environ Health A 65, 1531–1543. OECD (2004). OECD principles for the validation, for regulatory purposes, of (quantitative) structure– activity relationship models. 37th Joint Meeting of the Chemicals Committee and Working Party on Chemicals, Pesticides and Biotechnology in November 2004, http://www.oecd.org/dataoecd/33/37/37849783.pdf, 1–2. Olsen, G. W., Burris, J. M., Ehresman, D. J., Froehlich, J. W., Seacat, A. M., Butenhoff, J. L., and Zobel, L. R. (2007). Half-life of serum elimination of perfluorooctanesulfonate, perfluorohexanesulfonate, and perfluorooctanoate in retired fluorochemical production workers. Environ Health Perspect 115, 1298–1305. Osgood, C. J. (1994). Refractory ceramic fibers (RCFs) induce germline aneuploidy in Drosophila oocytes. Mutat Res 324, 23–27. Pascussi, J. M., Gerbal-Chaloin, S., Duret, C., Daujat-Chavanieu, M., Vilarem, M. J., and Maurel, P. (2008). The tangle of nuclear receptors that controls xenobiotic metabolism and transport: Crosstalk and consequences. Annu Rev Pharmacol Toxicol 48, 1–32. Poland, A., Glover, E., and Kende, A. S. (1976). Stereospecific, high affinity binding of 2,3,7,8-tetrachlorodibenzo-p-dioxin by hepatic cytosol. Evidence that the binding species is receptor for induction of aryl hydrocarbon hydroxylase. J Biol Chem 251, 4936–4946. Poland, C. A., Duffin, R., Kinloch, I., Maynard, A., Wallace, W. A., Seaton, A., Stone, V., Brown, S., Macnee, W., and Donaldson, K. (2008). Carbon nanotubes introduced into the abdominal cavity of mice show asbestos-like pathogenicity in a pilot study. Nat Nanotechnol 3, 423–428. Pott, F., Huth, F., and Friedrichs, K. H. (1974). Tumorigenic effect of fibrous dusts in experimental animals. Environ Health Perspect 9, 313–315. Rabinowitz, J. R., Goldsmith, M. R., Little, S. B., and Pasquinelli, M. A. (2008). Computational molecular modeling for evaluating the toxicity of environmental chemicals: Prioritizing bioassay requirements. Environ Health Perspect 116, 573–577. Reddy, J. K., and Lalwai, N. D. (1983). Carcinogenesis by hepatic peroxisome proliferators: Evaluation of the risk of hypolipidemic drugs and industrial plasticizers to humans. Crit Rev Toxicol 12, 1–58. Richard, A. M. (2006). Future of toxicology—Predictive toxicology: An expanded view of “chemical toxicity”. Chem Res Toxicol 19, 1257–1262. Richard, A. M., and Woo, Y. T. (1990). A CASE-SAR analysis of polycyclic aromatic hydrocarbon carcinogenicity. Mutat Res 242, 285–303. Safe, S. (2001). Molecular biology of the Ah receptor and its role in carcinogenesis. Toxicol Lett 120, 1–7. Sayes, C. M., Wahi, R., Kurian, P. A., Liu, Y., West, J. L., Ausman, K. D., Warheit, D. B., and Colvin, V. L. (2006). Correlating nanoscale titania structure with toxicity: A cytotoxicity and inflammatory response study with human dermal fibroblasts and human lung epithelial cells. Toxicol Sci 92, 174–185. Searl, A. (1994). A review of the durability of inhaled fibres and options for the design of safer fibres. Ann Occup Hyg 38, 839–855. Shvedova, A. A., Kisin, E. R., Mercer, R., Murray, A. R., Johnson, V. J., Potapovich, A. I., Tyurina, Y. Y., Gorelik, O., Arepalli, S., Schwegler-Berry, D., Hubbs, A. F., Antonini, J., Evans, D. E., Ku, B.
REFERENCES
555
K., Ramsey, D., Maynard, A., Kagan, V. E., Castranova, V., and Baron, P. (2005). Unusual inflammatory and fibrogenic pulmonary responses to single-walled carbon nanotubes in mice. Am J Physiol Lung Cell Mol Physiol 289, L698–L708. Stanton, M. F., Laynard, M., Tegeris, A., Miller, E., May, M., and Kent, E. (1977). Carcinogenicity of fibrous glass: Pleural response in the rat in relation to fiber dimension. J Natl Cancer Inst 58, 587–603. Stanton, M. F., and Wrench, C. (1972). Mechanisms of mesothelioma induction with asbestos and fibrous glass. J Natl Cancer Inst 48, 797–821. Topinka, J., Loli, P., Dusinska, M., Hurbankova, M., Kovacikova, Z., Volkovova, K., Kazimirova, A., Barancokova, M., Tatrai, E., Wolff, T., Oesterle, D., Kyrtopoulos, S. A., and Georgiadis, P. (2006). Mutagenesis by man-made mineral fibres in the lung of rats. Mutat Res 595, 174–183. Vinken, M., Doktorova, T., Ellinger-Ziegelbauer, H., Ahr, H. J., Lock, E., Carmichael, P., Roggen, E., van Delft, J., Kleinjans, J., Castell, J., Bort, R., Donato, T., Ryan, M., Corvi, R., Keun, H., Ebbels, T., Athersuch, T., Sansone, S. A., Rocca-Serra, P., Stierum, R., Jennings, P., Pfaller, W., Gmuender, H., Vanhaecke, T., and Rogiers, V. (2008). The carcinoGENOMICS project: Critical selection of model compounds for the development of omics-based in vitro carcinogenicity screening assays. Mutat Res 659, 202–210. Waller, C. L., and McKinney, J. D. (1995). Three-dimensional quantitative structure–activity relationships of dioxins and dioxin-like compounds: Model validation and Ah receptor characterization. Chem Res Toxicol 8, 847–858. Wang, Q. E., Han, C. H., Yang, Y. P., Wang, H. B., Wu, W. D., Liu, S. J., and Kohyama, N. (1999). Biological effects of man-made mineral fibers (II)—Their genetic damages examined by in vitro assay. Ind Health 37, 342–347. Ward, W. O., Delker, D. A., Hester, S. D., Thai, S. F., Wolf, D. C., Allen, J. W., and Nesnow, S. (2006). Transcriptional profiles in liver from mice treated with hepatotumorigenic and nonhepatotumorigenic triazole conazole fungicides: Propiconazole, triadimefon, and myclobutanil. Toxicol Pathol 34, 863–878. Warheit, D. B., Laurence, B. R., Reed, K. L., Roach, D. H., Reynolds, G. A., and Webb, T. R. (2004). Comparative pulmonary toxicity assessment of single-wall carbon nanotubes in rats. Toxicol Sci 77, 117–125. Warheit, D. B., Webb, T. R., Reed, K. L., Frerichs, S., and Sayes, C. M. (2007). Pulmonary toxicity study in rats with three forms of ultrafine-TiO2 particles: Differential responses related to surface properties. Toxicology 230, 90–104. Warheit, D. B., Webb, T. R., Sayes, C. M., Colvin, V. L., and Reed, K. L. (2006). Pulmonary instillation studies with nanoscale TiO2 rods and dots in rats: Toxicity is not dependent upon particle size and surface area. Toxicol Sci 91, 227–236. Waters, M., and Jackson, M. (2008). Databases applicable to quantitative hazard/risk assessment— Towards a predictive systems toxicology. Toxicol Appl Pharmacol 233, 34–44. Wilson, M. R., Lightbody, J. H., Donaldson, K., Sales, J., and Stone, V. (2002). Interactions between ultrafine particles and transition metals in vivo and in vitro. Toxicol Appl Pharmacol 184, 172–179. Woo, Y. T., Arcos, J. C., and Lai, D. Y. (1985a). Structural and functional criteria for suspecting chemical compounds of carcinogenic activity: State-of-the-art of predictive formalism. In Handbook of Carcinogen Testing, Noyes Publications, Park Ridge, IL, pp. 1–637. Woo, Y. T., and Lai, D. (2003). Mechanisms of action of chemical carcinogens and their role in structure– activity relationships (SAR) analysis and risk assessment. In Quantitative Structure–Activity Relationship (QSAR) Models of Chemical Mutagens and Carcinogens, Benigni, R., ed., CRC Press, Boca Raton, FL, pp. 41–80. Woo, Y. T., and Lai, D. (2005). OncoLogic: A mechanism-based expert system for predicting the carcinogenic potential of chemicals. In Predictive Toxicology, Helma, C., ed., Taylor and Francis, Boca Raton, FL, pp. 385–413. Woo, Y. T., Lai, D., Arcos, J. C., and Argus, M. F. (1988). Chemical Induction of Cancer, Vol. IIIC: Natural, Metal, Fiber, and Macromolecular Carcinogens, Academic Press, New York. Woo, Y. T., Lai, D., McLain, J. L., Manibusan, M. K., and Dellarco, V. (2002). Use of mechanism-based structure–activity relationships analysis in carcinogenic potential ranking for drinking water disinfection by-products. Environ Health Perspect 110(Suppl 1), 75–87.
556
CHAPTER 20
(Q)SAR ANALYSIS OF GENOTOXIC AND NONGENOTOXIC CARCINOGENS
Woo, Y. T., Lai, D. Y., Arcos, J. C., and Arcos, J. C. (1985b). Chemical Induction of Cancer, Vol. IIIB: Aliphatic and polyhalogenated carcinogens, Academic Press, Orlando, FL. Woo, Y. T., Lai, D. Y., Argus, M. F., and Arcos, J. C. (1995). Development of structure–activity relationship rules for predicting carcinogenic potential of chemicals. Toxicol Lett 79, 219–228. Woo, Y. T., Lai, D. Y., Argus, M. F., and Arcos, J. C. (1998). An integrative approach of combining mechanistically complementary short-term predictive tests as a basis for assessing the carcinogenic potential of chemicals. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev C16, 101–122. Yu, S., Rao, S., and Reddy, J. K. (2003). Peroxisome proliferator-activated receptors, fatty acid oxidation, steatohepatitis and hepatocarcinogenesis. Curr Mol Med 3, 561–572. Zhu, H., Rusyn, I., Richard, A., and Tropsha, A. (2008). Use of cell viability assay data improves the prediction accuracy of conventional quantitative structure–activity relationship models of animal carcinogenicity. Environ Health Perspect 116, 506–513.
CH A P TE R
21
PHYSIOLOGICALLY BASED PHARMACOKINETIC (PBPK) MODELS IN CANCER RISK ASSESSMENT Mathieu Valcke Kannan Krishnan
21.1.
INTRODUCTION
The process of risk assessment for carcinogens, depending upon the jurisdictions, has involved one of the following approaches (Younes et al. 1998): 1. Estimation of risk at low doses based on linear extrapolation of response observed at an effective dose 2. Expression of the dose–response relationship near the experimental range in terms of potency estimates 3. Comparison of the effective dose with the exposure dose to compute a margin of exposure 4. Advice to control the exposure levels to the maximum extent possible so as to reduce risk to humans Frequently, however, a quantitative approach is applied on the basis of (i) problem formulation and hazard identification, (ii) dose–response assessment, (iii) exposure assessment, and (iv) risk characterization (EPA 2005a; NRC 1983; WHO 1999). In this regard, modeling of the dose–response data using a linearized multistage model has been applied to chemical carcinogens, particularly for those chemicals that are thought to act by nonthreshold mechanisms. Accordingly, a unit cancer risk—that is, probability of developing cancer over the life time per unit of exposure—is developed. The unit risk values associated with the oral and inhalation routes are referred to as the cancer slope factor (CSF) and the inhalation unit risk (IUR). Following the publication of United States Environmental Protection
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
557
558
CHAPTER 21 PBPK MODELS IN CANCER RISK ASSESSMENT
Agency’s (EPA) revised cancer risk guidelines (EPA 2005a), the emphasis has shifted toward the characterization of the mode of action for choosing the appropriate extrapolation approach—that is, linear versus nonlinear model. Accordingly, the CSF for genotoxicants or nonthreshold carcinogens is derived on the basis of linear extrapolation from the point of departure (POD) or from the region of observations, whereas a nonlinear threshold analysis [e.g., no observed adverse effect level (NOAEL) divided by uncertainty factors (UFs)] is applied in other instances to develop a margin of exposure (MOE) (EPA 2005a). Either approach may additionally require interspecies and/or route-to-route extrapolations of the POD. The extrapolation of cancer response across doses, species, routes, lifestages, and mixture complexity continues to represent a challenge. The scientific basis of these various extrapolations can be enhanced by conducting them with the use of an appropriate dose metric—that is, a measure of internal dose consistent with the mode of action of the carcinogen. The “dose metric” then provides a better measure of the target organ exposure to toxic moiety than does the potential dose or administered dose (EPA 2006; Clewell et al. 2002a). It is, however, impossible to obtain the measures of the dose metric in the target organ of people or animals exposed to carcingens at various doses by routes and scenarios of relevance to risk assessment. Therefore, there is increasing emphasis on the development of mechanism-based models to simulate the dose metric of carcinogens for various species, routes, doses, mixtures and lifestages. In this regard, physiologically based pharmacokinetic (PBPK) models are uniquely useful. This chapter presents the characteristics of PBPK models, as well as their application in cancer risk assessment of environmental and occupational contaminants.
21.2. PBPK MODELING: CHARACTERISTICS AND APPROACHES PBPK models represent the organism as a set of tissue compartments interconnected by systemic circulation and facilitate the simulation of pharmacokinetics in the organism (Figure 21.1) (Fiserova-Bergerova 1975; Haggard 1924; Kety 1951; Mapleson 1963). Typically, the development of PBPK models intended for use in cancer risk assessment begins with a literature search focused on assembly of information on the nature and dose–response relationship of the critical effect(s), the mode of carcinogenic action, the pathways and rates of absorption, and the distribution, metabolism, and excretion (ADME) of the carcinogen, as well as the physiology of the species of interest (i.e., tissue weights and blood flow rates). The development and evaluation of PBPK models are based on six general principles as described in detail by several authors (Andersen et al. 1995; Barton et al. 2007; Chiu et al. 2007; Clark et al. 2004; Clewell and Clewell 2008; Gentry et al. 2004; Kohn 1995; Loizou et al. 2008). First, the scope for the use of PBPK models in a particular risk assessment has to be clearly identified since it essentially determines the intended model complexity, model capability, and the extent of model evaluation. Here, the goal is really to specify the purpose for which the PBPK model would be used in a cancer risk
21.2. PBPK MODELING: CHARACTERISTICS AND APPROACHES
Inhalation
Dermal contact
A r t e r i a l b l o o d
Exhalation
Lungs Skin Richely perfused Poorly perfused Fat
V e n o u s b l o o d
Kidneys Liver
Metabolism Excretion
Figure 21.1. Conceptual representation of a PBPK model permitting the simulation of dermal, pulmonary, and oral exposures.
559
560
CHAPTER 21 PBPK MODELS IN CANCER RISK ASSESSMENT
assessment and outline how it would be used to address a specific source of uncertainty in such an assessment, based on current understanding of the mode of action. For example, if the uncertainty in a cancer risk assessment relates to lack of knowledge regarding rat-to-human differences in tissue dosimetry, then the resulting PBPK model should be able to provide reliable estimates of the tissue dose of the toxic moiety in rats and humans exposed by the relevant route(s) of exposure. Since a single PBPK model may not be sufficient to address all areas of uncertainty, it is essential, at the outset, to match the capability of the model with the needs of a risk assessment in terms of the species, lifestage, exposure routes/windows, and dose metrics (Clark et al. 2004). Second, the model structure should reflect a balance between plausibility (i.e., reflective of physiological reality and consistency with current state of knowledge) and parsimony (i.e., minimal but essential elements characterizing a system and for which required data are available). As shown in Figure 21.1, typically the PBPK model represents a series of anatomically relevant tissue compartments that receive the carcinogenic chemical or its metabolite via the arterial blood and lose via the effluent venous blood. The need for representing a particular organ or tissue as a separate compartment or for lumping a number of tissues into a single compartment in a PBPK model should be based the consideration of the tumor site(s), mode of action, toxicokinetic mechanisms, and portals of entry (exposure routes) of the carcinogen under study (Clewell and Clewell 2008; Krishnan and Andersen 2007). For example, a model intended to simulate dose to specific cell types in an organ (e.g., sinusoidal cells in the liver) will be more complex than the one for hydrophilic chemicals whose concentrations are fairly similar across tissues. Third, the equations employed in a PBPK model should be consistent with the state of knowledge or biologically plausible hypotheses of the mechanisms of ADME for the particular chemical. In this regard, the uptake of chemicals in systemic circulation is described as either a diffusion-limited or perfusion-limited process (Gerlowski and Jain 1983), and metabolic clearance in individual tissues or tissue groups is described using a maximal velocity and Michaelis constant, intrinsic clearance, or hepatic extraction ratio (Krishnan and Andersen 2007). The mass balance differential equations accounting for uptake clearance, efflux clearance, and metabolic clearance are formulated as a function of identifiable input parameters (Table 21.1). Fourth, simulation of system behavior (i.e., pharmacokinetic profiles of chemicals and/or their metabolites in blood and tissues) should be conducted using numerical algorithms that are proven to be adequate for solving differential equations of the type used in PBPK models and should specifically be capable of dealing with stiff systems. In this regard, a number of algorithms (Runge Kutta, Euler, etc.) and software are adequate for this purpose (Chiu et al. 2007). It is important to verify the accuracy of the mathematical and computational implementations of PBPK models systematically. Thus, the model codes should be free of syntax/mathematical errors, the units of input parameters and variables should be compatible, and the chemical mass balance and physiological mass balance in the model should be respected at all times (Balci 1997).
21.2. PBPK MODELING: CHARACTERISTICS AND APPROACHES
561
TABLE 21.1. Examples of Equations Used in PBPK Models of Lipophilic Volatile Organic Chemicals Metabolized Principally in Liver
Equationsa
Compartment Blood
Tissue
Qp ∗ Ci + Qc ∗ Cv Qalv + ( Qc Pba ) ⎛ ⎞ Cv = ⎜ ∑ (Qt ∗Cvt )⎟ Qc ⎝ t ⎠ dAt = Qt ∗ ( Ca − Cvt ) dt
Ca =
Cvt = At/(Vt * Pt : b) Ct = At/Vt Skin Liver
dAsk = K p ∗ S ∗ (Cm − Csk Psk:m ) dt dAl ⎡ V ∗ Cvt ⎤ = Ql ∗ ( Ca − CVl ) − ⎢ max + Dor dt ⎣ K m + Cvt ⎥⎦ Dor = Ko * Fb * Aing
a Notation: Qalv, Qc, Qt, Vt, Pba, Ptb, Vmax, Km, Ko, and Kp refer to alveolar ventilation rate, cardiac output, tissue blood flow rate, tissue volume, blood : air partition coefficient, tissue : blood partition coefficient, maximal velocity for metabolism, Michaelis–Menten constant, oral absorption constant, and dermal permeability constant; Dor, Aing, Fb, and S refers to oral dose, amount ingested, fraction of bioavailability, and dermal surface in contact with chemical; C refers to concentration and subscripts i, v, a, t, vt, sk, and m refer to inhaled, venous, arterial, tissue, venous blood leaving tissue, skin, and chemical in the media in contact with skin and liver.
Fifth, the PBPK model parameters should be estimated using valid in vivo, in vitro, or in silico methods. The current methods for estimation and analysis of chemical-specific parameters as well as biological input data for PBPK models can be found in Krishnan and Andersen (2007) as well as in Lipscomb and Ohanian (2007). Particular emphasis should be placed on the biological and mechanistic relevance of the input parameters with regard to the mode of action and ADME of the chemical as well as the species and lifestage bing modeled [e.g., Arms and Travis (1988), Davies and Morris (1993), Gentry et al. (2004), ICRP (1975), Price et al. (2003b)]. Additionally, the allometric scaling and in vitro to in vivo extrapolation of model parameters, if performed, should be consistent with the current state of knowledge on physiology and biochemical reactions (Krishnan and Andersen 1991a; Lipscomb and Poet 2008). Sixth, and finally, the adequacy of model structure as well as parameter values should be evaluated based on comparison of mode predictions with experimental data that had not been used for calibration purpose. This process essentially evaluates whether the PBPK model is capable of providing reliable predictions of the various dose metrics of potential use in a cancer risk assessement. The model should not only reproduce consistently the shape of the pharmacokinetic time-course curve (i.e., including bumps and valleys) and not just provide satisfactory fit only to a portion of the curve. Evaluation or validation of PBPK models should be regarded
562
CHAPTER 21 PBPK MODELS IN CANCER RISK ASSESSMENT
in the context of specific applications (i.e., interspecies extrapolation, intraspecies extrapolation, route to route extrapolation), since no single model will be universally valid and applicable for all purposes. The quantitative fit of PBPK model simulations to experimental data can be evaluated using one of several methods (Krishnan and Andersen 2007). In this regard, the most common method involves visual inspection of the plots of simulated values (or the residuals) against a common independent variable, usually time. Second, formal statistical tests such as leave-one-out crossvalidation, multivariate analysis of variance, and lack-of-fit F test may be applied in specific instances (Krishnan and Andersen 2007). Finally, to quantitate the difference between simulations and experimental data, the discrepancy indices that involve calculation of the root mean square of error reflecting the difference between the simulated and experimental data along the time-course curve may be applied (Krishnan et al. 1995). The PBPK model evaluation should involve more than just a comparison of simulations with experimental data; importantly, it should consist of systematic analyses of the impact of parameter sensitivity, uncertainty, and variability (Table 21.2). During the performance of these analyses, however, it is critical to ensure that the physiological and chemical mass balance of the model is respected and that the plausible range and covariance of parameters is not violated.
TABLE 21.2.
Uncertainty, Sensitivity, and Variability Analyses in PBPK Modeling
Analysis Features
Uncertainty
Sensitivity
Variability
Objective
To evaluate the impact of the lack of precise knowledge of parameter values or structure on the tissue dose. • Monte Carlo methods • Stochastic surface analysis • Fuzzy simulation approach Predictions of pharmacokinetic behavior or tissue dose associated with given set of parameters or structure.
To evaluate the magnitude of change in dose metric for a unit change in the value(s) of iput parameters. • Monte Carlo methods • Local and global analyses
To evaluate the magnitude of change in dose metric due to population variability of population input parameters. • Subject-specific population modeling • Monte Carlo methods • Markov Chain Monte Carlo simulations • Population distributions (probability) of dose metric for an exposure situation • Percentage of variance in the output (e.g., dose metric) attributed to each parameter ’s variance
Methods
Outputs
Sensitivity coefficients
21.3. PBPK MODELS IN CANCER RISK ASSESSMENT
21.3.
563
PBPK MODELS IN CANCER RISK ASSESSMENT
The use of internal dose or tissue dose, instead of the administered dose, facilitates a better characterization of the dose–response relationship for carcinogens (EPA 2006; Clewell et al. 2002a; Slikker et al. 2004) because the carcinogenic response(s) are more closely and directly related to the concentration of the putative toxic chemical in the target tissue (i.e., dose metric) (Andersen and Dennison 2001). Since the tumor response in laboratory animals is often observed at dose levels and scenarios not directly relevant to human exposures, it becomes essential to conduct analyses and extrapolations based on appropriate dose measures. The measure of dose to target, often referred to as the “dose metric,” reflects the biologically active form of chemical (parent chemical, metabolite, or adducts), its level (concentration or amount), its duration (instantaneous, daily, lifetime, or a specific developmental period), and its intensity (peak, average, or integral) as well as the biological matrix (e.g., blood, target tissue) that is consistent with the mode of action of the carcinogen (Figure 21.2) (Andersen et al. 1987; Clewell et al. 2002a; Voisin et al. 1990). For example, in the case of carcinogens producing reactive intermediates, the amount of metabolite produced per unit time or the amount of metabolite in target tissue over a period of time (e.g., 24 hr) have been used as dose metrics (Andersen et al. 1987; Andersen and Dennison 2001). By facilitating the simulation of the dose metrics for use in cancer dose– response analysis, the PBPK models address the uncertainty associated with interspecies, route-to-route, and high-dose to low-dose extrapolations (Andersen et al. 1993; Andersen and Krishnan 1994; Clewell et al. 2002a; Clewell and Andersen 1987; Melnick and Kohn 2000). Since the first demonstration of the application of PBPK models in cancer risk assessment by Andersen and co-workers in 1987, there have been substantial efforts to evaluate the appropriate dose metrics and cancer risk associated with a number of other volatile organic chemicals using the PBPK modeling approach (Table 21.3). These risk assessments have been based on the PBPK model simulations of a variety of dose metrics that reflect the current state
Exposure /Administered Dose PBPK model ''Dose metric'' Initiation Process
Promotion/Repair Process
Carcinogenic Response
Figure 21.2. assessment.
Illustration of the rationale guiding the use of PBPK models in cancer risk
564
CHAPTER 21 PBPK MODELS IN CANCER RISK ASSESSMENT
TABLE 21.3. Examples of PBPK Models Developed for Assessing Dose Metrics and Cancer Risks
Chemical
Extrapolation Performed
Most Recent Human Risk Estimation Approach
1,4-Dioxane
HL, IS
Unit risk
Methylene chloride
HL, IS
Unit risk
Trichloroethylene
HL, IS
Threshold-MOE
Tetrachloroethylene Acrylamide Chloroprène Dioxin Vinyl chloride
InS HL, IS HL, IS HL, IS HL, IS, RR
Unit risk Unit risk Threshold-MOE Unit risk Unit risk
Acrylonitrile 2-Butoxyethanol Chloroform
HL, IS RR HL, IS, RR
Threshold-MOE – Threshold-RfD
Formaldehyde
HL, IS
Unit risk
Methyl methacrylate Styrene
IS
Threshold-RfC
IS
–
Vinyl acetate
HL, IS
Threshold-RfC
Reference Reitz et al. (1990b), Leung and Paustenbach (1990), Sweeney et al. (2008) Andersen et al. (1987), Marino et al. (2006), David et al. (2006), Marino and Starr (2007) Fisher and Allen (1993), Clewell et al. (2000), Clewell and Andersen (2004) Byczkowski and Fisher (1995) Doerge et al. (2008) Himmelstein et al. (2004) Maruyama and Aoki (2006) Reitz et al. (1996), Clewell et al. (2001) Kirman et al. (2000, 2005) Poet et al. (2003) Reitz et al. (1990a), Meek et al. (2002), Levesque et al. (2002), Liao et al. (2007b) Casanova et al. (1996), Schlosser et al. (2003) Andersen et al. (2000, 2002) Cruzen et al. (2002), Sarangapani et al. (2003) Bogdanffy et al. (1999, 2001), Andersen et al. (2002)
Abbreviations: HL, High dose-to-low dose; IS, interspecies; InS, intraspecies; RR, route-to-route. RfD, reference dose; RfC, reference concentration.
of knowledge on the mode of action of chemical carcinogens (e.g., production of metabolite (rate or amount), area under the tissue concentration versus time curve, intercellular change in pH, receptor occupancy, DNA–protein crosslinks, glutathione depletion, amount bound to biological macromolecules). The PBPK-based cancer risk assessment process has generally involved the following steps (Krishnan and Andersen 1991b): 1. Development and validation of a PBPK model for the test species, in order to evaluate the relationship between the administered dose and potential dose
21.3. PBPK MODELS IN CANCER RISK ASSESSMENT
565
metrics (e.g., concentration of parent chemical or metabolite in the target tissue). 2. Selection of the appropriate dose metric for the risk assessment, based on the results of the preceding step and mode of action of the chemical. 3. Characterization of the quantitative relationship between the dose metric and the cancer incidence observed in bioassay(s) to estimate the dose metric-based slope factor or threshold level. 4. Back-calculation of the potential dose or exposure concentration associated with the dose metric for a predetermined (i.e., acceptable) risk level (e.g., 1 × 10−6) or threshold level, using the human PBPK model, based on the assumption of equivalent tissue responses for equivalent dose metrics regardless of the species (Andersen et al. 1987; Krishnan and Andersen 1991b). Whereas some cancer assessments may require the use of PBPK models for multiple purposes, others might focus on their specific application with regard to high-dose to low-dose, interspecies, intraspecies, or route-to-route extrapolations, as discussed below.
21.3.1. High-Dose to Low-Dose and Interspecies Extrapolation 21.3.1.1. Issue. Since high doses of chemicals are often administered in animal cancer bioassays, the tumor incidence observed in such studies is not directly proportional or directly related to the administered dose (or exposure concentration); and this, in part, is due to the lack of linear relationship between the dose to target tissue and dose administered in the bioassays (Slikker et al. 2004). In other words, the target tissue dose of the toxic moiety may be disproportional to the administered doses used in animal bioassays, if nonlinearity in absorption, distribution, metabolism and/or excretion processes occurs in the dose range employed in the cancer bioassay. Table 21.4 summarizes the exposure concentration versus tumor response data for vinyl chloride. These data indicate that the incidence of angiosarcoma in male and female rats does not increase proportionally between 500 ppm and 6000 ppm (despite the 12-fold increase in the exposure concentration); rather, it attains a plateau consistent with the pattern expected when a reactive metabolite produced by a saturable process is the carcinogenic moiety. Aspects such as this need to be given due consideration before extrapolating to humans; otherwise the human-equivalent dose determined for a specific level of cancer risk (e.g., 1 in a million) may be overly conservative or not sufficiently health-protective. In this regard, the conventional default approaches have involved the use of linearized multistage model (for high-dose to low-dose extrapolation) along with an interspecies scaling of oral doses according to body surface differences (EPA 2005a). The body surface scaling has often been implemented by converting oral doses from one species (e.g., rats) to another (e.g., humans) on the basis of body weight raised to a fractional power ranging from 0.67 to 0.75 (Krishnan and Andersen 1991b). This procedure presumes that equal doses expressed on the basis of a fractional power of body weight, when administered daily over a lifetime, will result in equal
566
CHAPTER 21 PBPK MODELS IN CANCER RISK ASSESSMENT
TABLE 21.4.
Dose–Response Assessment for Angiosarcoma Induced by Vinyl Chloride
Inhalation Exposurea (ppm)
Lifetime Average Dose Metricsb (mg metabolized per liter of liver)
Tumor Probability
0 0.6 3.0 6.0 15.0 32.5 130.3 163.4 221.0 250.7
0 0 0 0.02 0.1 0.03 0.08 0.21 0.29 0.40
0 0.6 3.0 5.90 14.6 31.3 103.4 116.9 134.4 143.7
0 0 0 0 0.02 0 0.04 0 0.23 0.18
Male Rats 0 1 5 10 25 50 250 500 2500 6000 Female Rats 0 1 5 10 25 50 250 500 2500 6000 a
4 hr/day, 5 days/week, 52 of 147 weeks (0–25 ppm) or 52 of 135 weeks (50–6000 ppm); data from Gehring et al. (1978, 1979). b
Determined by PBPK modeling (Clewell et al. 2001).
internal dose and lifetime cancer risks across mammalian species. This three-fourths power scaling relationship is generally considered to reflect differences in rates of basal metabolism. This approach appears relevant when (i) the toxic moiety is the parent compound (which is not always the case for chemical carcinogens), (ii) the major clearance is via hepatic metabolism, and (iii) the intrinsic hepatic clearance is proportional to the body surface area (Krishnan and Andersen 1991a). Therefore, for cancer dose–response analysis, it is more relevant to use an appropriate dose metric based on an understanding of the MOA for the conduct of high-dose to lowdose and interspecies extrapolation, as done with the use pf PBPK models (Andersen et al. 1987; Clewell et al. 2002a), and the use of internal dose or delivered dose at target tissue in such analysis has been encouraged. These models, in effect, account for the difference and interplay among mechanistic determinants of uptake and disposition, which are typically physiological, biochemical, or physicochemical in nature, in simulating the extent of target organ exposure to the toxic moiety of a chemical carcinogen.
567
21.3. PBPK MODELS IN CANCER RISK ASSESSMENT
21.3.1.2. Role of PBPK Models. PBPK models facilitate high-dose to lowdose extrapolation of tissue dosimetry by accounting for the dose-dependency of capacity-limited processes (e.g., saturable metabolism). Nonlinearity arising from mechanisms other than saturable metabolism, such as enzyme induction, enzyme inactivation, depletion of glutathione reserves, and binding to macromolecules, has also been described with PBPK models (Clewell and Andersen 1987; Krishnan and Andersen 2001). For conducting high-dose to low-dose extrapolation with PBPK models, there is no need to change the numerical values of input parameters with the exception of exposure concentration or administered dose. However, for purposes of interspecies extrapolation (e.g., rat to human) using PBPK models, the species-specific input parameters should be used. Qualitative and/or quantitative differences are factored into the existing structure of PBPK models, whereas the existing equations, parameter definitions and structure of the PBPK model are kept the same during this process. Figure 21.3 illustrates this approach for 1,4-dioxane. The PBPK model, written and solved in Microsoft EXCEL® (the modeling template can be obtained by contacting the chapter authors), was used to simulate the amount metabolized, area under the concentration–time curve (AUC), and maximum or peak concentration (Cmax) in rats and humans. The interspecies and high-dose to low-dose extrapolations have been accomplished for a number of chemical carcinogens: dichloromethane (DCM) (Andersen et al. 1987; David et al. 2006; Marino et al.
Venous blood [ ] (mg/L)
Parameters: Plood flows Volumes Partition coefficient Metabolic constants Tissues % of CO Q(L/hr) % of BW Volume (L) tissue:air tissue:blood Vmax (mg/h) Km (mg/L) Rat 0.25 100 4.66 Body 100 4.66 Lung 100 0.46 851 0.0175 0.33 7 Fat 7 0.72 1348 0.1485 1.57 59.4 SPT 33.6 0.30 560 0.0345 1.92 13.8 RPT 41.1 1.00 1862 4.81 7.96 0.0085 0.85 3.4 Liver 18.3 1861 0.0185 7.4 Blood 0.2275 4.66 Total Human 70 100 100 301.52 Body 100 301.52 Lung 0.52 865 14.98 21.4 5.2 15.68 Fat 0.90 1503 17.43 24.9 24.9 75.08 SPT 0.34 560 23.45 33.5 RPT 47.2 142.32 0.90 1500 3757.25 626.21 2.31 3.3 Liver 22.7 68.45 1666 5.53 7.9 Blood 63.7 301.52 Total 0.1 0.08 0.06
Human Rat
0.04 0.02 0 0
5
10
Time (hours)
15
Rat results AUC (mg. h/L) Cmax (mg/L) Amnt. metab. (mg/L of liver) Human results AUC (mg. h/L) Cmax (mg/L) Amnt. metab. (mg/L of liver)
0.6 0.03 1.465 0.182 0.083 8.166
Figure 21.3. Illustration of the use of PBPK models in rat-human extrapolation using Microsoft Excel spreadsheet, for an inhalation exposure to 0,7 ppm of 1,4-dioxane [physiological data taken from Sweeney et al. (2008)].
568
CHAPTER 21 PBPK MODELS IN CANCER RISK ASSESSMENT
2006), dioxin (Maruyama and Aoki 2006), 1,4-dioxane (Leung and Paustenbach 1990; Reitz et al. 1990b; Sweeney et al. 2008); chloroprene (Himmelstein et al. 2004); acrylonitrile (Kirman et al. 2005); tricholoroethylene (Clewell et al. 2004), and acrylamide (Doerge et al. 2008).
21.3.2.
Intraspecies Extrapolation
21.3.2.1. Issue. There has been a concern about the adequacy of cancer risk assessment approaches, particularly with respect to the susceptibility of children of various age groups, as compared to adults. Indeed, there is some evidence that earlylife exposures can lead to higher cancer risks compared to exposure of the same duration occurring later in life (Ginsberg 2003). Exposure to carcinogens present in breast milk and associated risks of cancer later in life have been modeled [e.g., Byczkowski and Fisher (1995)]. The adult–child differences in susceptibility, in part, are due to differences in internal dose resulting from identical exposures. The concentration of the toxic chemical reaching the target tissue could be different between children and adults since a number of factors determining absorption, distribution, metabolism, and excretion of chemicals change with age (Alcorn and McNamara 2003; Clewell et al. 2002b; Ginsberg et al. 2002; Gladtke 1973; Haddad et al. 2001b; Price et al. 2003a; Strolin and Baltes 2003). In spite of that, cancer risk assessments have not generally considered intraspecies variability in pharmacokinetics or pharmacodynamics; the use of the upper bounds on maximum likelihood estimate (MLE) in cancer assessments has generally been thought to be sufficiently conservative to protect susceptible populations. Recently, however, EPA has emphasized the importance of early-life exposures associated with childhood cancer as well as early-life exposures that may lead to cancers later in life (EPA 2005b). In this regard, EPA suggests that an additional adjustment factor (age-dependent adjustment factor, ADAF) to the cancer slope or unit risk value be considered for application, to account for enhanced susceptibility in early life (i.e., neonates and young children) from exposure to carcinogens exhibiting a mutagenic MOA (EPA 2005b). This factor is suggested to be 3 for the children aged 2–16, and it is thought to be 10 for the children aged 0–2. Thus, the cancer risk (R) for given lifetime average daily exposure (E) would be calculated using the slope factor q as follows: R = E × [(q × 10 × 2 70 ) + (q × 3 × 14 70 ) + (q × 54 70 )]
It should be noted that ADAFs generally account for increased susceptibility related to adult–child differences in pharmacodynamic processes such as cell proliferation rates and numbers of cells with proliferative potential (EPA 2005b); thus even if pharmacokinetic differences are accounted for when extrapolating from a POD, these adjustment factors would still likely be applied for chemicals with a mutagenic MOA unless the carcinogenicity data were derived from young animals or humans (EPA 2005b). A distinct advantage of using PBPK models is that this assumption can be tested, because these models allow a systematic evaluation of the adult–child differences in physiological determinants and their impact on the dose
21.3. PBPK MODELS IN CANCER RISK ASSESSMENT
569
metrics of carcinogens (Clewell et al. 2002b, 2004; Gentry et al. 2003; Ginsberg et al. 2004a, 2004b; Nong et al. 2006; Price et al. 2003a). PBPK models also facilitate the dose-construction over life-time by accounting for temporal changes in physiological flow rates, metabolic clearance and body composition parameters [e.g., David et al. (2006), Haddad et al. (2006), Liao et al. (2007a), Pelekis et al. (2003), Sarangapani et al. (2003), Tan et al. (2006, 2007), Verner et al. (2008)], which in turn can be used in cancer risk assessment. 21.3.2.2. Role of PBPK Models. PBPK models can be used to estimate the dose metrics in different subgroups of a given population, animal or human, by adjusting the values of input parameters reflective of ADME determinants for each subgroup. The subgroups may be defined according to their age (e.g., neonates/pups, versus adults/mature) or according to their ethnicity (e.g., caucasian/asiatics) and genotype (e.g., wild-type versus polymorphic individuals related to metabolic enzymes). In doing so, values assigned to the parameters of the model can be deterministic, usually representing the average individual belonging to a particular group. However, defining each parameter with a probability density function (PDF) can facilitate probabilistic PBPK modeling, which can in turn permit evaluation of the impact of uncertainty and inter-individual variability within each subgroup on the overall internal dose metric values and corresponding risk. These PDFs can either be developed on the basis of in vitro data, population biomonitoring data, or limited in vivo human data analyzed with Markov Chain Monte Carlo (MCMC) simulation [e.g., El-Masri et al. (1999), Jonsson and Johanson (2001, 2002)]. The resulting a posteriori distributions of input parameters can then be used in a PBPK model, and the distributions of dose metrics and risk estimates can be obtained with Monte Carlo simulations (Figure 21.4). The quantitative nature of the change in various physiological, biochemical, and physicochemical parameters together determine the outcome (i.e., tissue dosimetry) of relevance to risk assessment. Several authors used PBPK models to evaluate the impact of the age-dependent change in pharmacokinetic determinants on the blood concentration profile of carcinogenic chemicals following inhalation or oral exposures (Price et al. 2003a) and volatile organic compounds (VOCs) (Clewell and Andersen 2004; Pelekis et al. 2001; Sarangapani et al. 2003). These simulations indicated that the blood concentrations of parent chemicals are generally greater in neonates because the net amount metabolized by cytochrome P450 2E1 (CYP2E1) was computed to be greater in adults than in neonates. While the adult–children differences in the amount metabolized during the first year of life might in part be related to differences in hepatic enzyme content, these differences in the following years are also influenced by the differences in liver blood flow rate (e.g., ∼96 L/hr in adults versus 16 L/hr in 6-year-old). Age-related differences in dosemetric may be simulated not only with lifestage-specific PBPK models, but also using lifetime PBPK models. In the later case, the time-weighted exposure is taken into consideration as well, in the overall evaluation of the contribution of the various lifestages to the exposure and dose metrics of relevance to the cancer mode of action of the chemicals evaluated (Clewell et al. 2004).
570 Lungs
Expérimental data
Richely perfused
+
Poorly perfused
Physiological
Adipose tissues
Physiological
Physicochemical
Lungs
Lungs Liver Liver Rest of body Rest of body
Liver
Biochemical
Physicochemical
Biochemical
PBPK model
MCMC simulations
Optimized distributions of model parameters
MC simulations
A priori distributions of model parameters
F r e q u e n c y Dose
Prediction of dose metrics Figure 21.4. Schematic of probabilistic PBPK modeling using Markov Chain Monte Carlo simulations approach.
21.3. PBPK MODELS IN CANCER RISK ASSESSMENT
21.3.3.
571
Route-to-Route Extrapolation
21.3.3.1. Issue. Route-to-route extrapolation is necessary if (i) the cancer dose– response data were obtained for an exposure route that is different from the anticipated human exposure route, (ii) the chemical induces tumors at a site different from the portal of entry, and (iii) the chemical is systemically absorbed, resulting in an effective internal dose (ECETOC 2003; IGHRC 2005). Conventionally, route-toroute extrapolations for systemic toxicants and carcinogens have been performed on the basis of administered dose and 100% absorption for each of the exposure routes. These assumptions are questionable since the tissue dose is not always linearly related to the administered dose, for the various exposure routes. For a scientifically sound dose extrapolation across routes, the exposure route-specific first pass effect should be accounted for. Further, multi route exposure consideration is also important while deriving the guideline values for carcinogens, particularly those found in drinking water. In this case, the dose received via inhalation and dermal contact might be as importrant as the oral route, resulting in greater tissue dose of the toxic moiety than that for the single, assumed principal exposure route (i.e., oral) [e.g., Krishnan and Carrier (2008)]. 21.3.3.2. Role of PBPK Models. PBPK models facilitate the conduct of routeto-route extrapolation of the CSF or the risk-specific dose of carcinogens by accounting for the route-specific rate and magnitude of absorption, first-pass effect, and metabolism (Clewell and Andersen 2004). Following the addition of equations specific to the absorption and clearance of carcinogens for each route of exposure, the PBPK model facilitates the simulation of the dose metric associated with the applied doses given by different routes (Blancato and Chiu 1993; Clewell and Andersen 1987; EPA 2006; Gerrity et al. 1990; Krishnan and Andersen 2001) (see Figure 21.1). The PBPK-based route-to-route extrapolation in cancer risk assessment frequently begins with the determination of a slope factor, associated with the response data for one exposure route, on the basis of the appropriate dose metrics (Dt), e.g., 2 × 10−4 per milligram metabolized per day per g of tissue. Then, the PBPK model, parameterized for other exposure route(s) of interest, is used to determine the exposure dose that generates the same Dt—that is, that corresponding to a predetermined risk level (e.g., 1 × 10−6) (Clewell et al. 2001). When linear extrapolation is appropriate, the following equation is used: P ( d ) = URt × Dt where P(d) is the excess cancer risk and URt is the CSF based on tissue dose estimated with the PBPK model.
21.3.4. Extrapolation from Individual Carcinogens to Mixtures 21.3.4.1. Issue. The assessment of cancer risk associated with simultaneous or sequential exposure to initiators and/or promoters presents a challenge. The cancer risk associated with human exposure to chemical mixtures is frequently assessed
572
CHAPTER 21 PBPK MODELS IN CANCER RISK ASSESSMENT
either by considering the mixture as a single entity or by summing the response associated with the components (EPA 2000). Woo et al. (1994) developed an approach to compute the consequence of the interaction among carcinogens (Woo et al. 1994). This approach integrates available information on interactions among the carcinogenic components of a mixture to determine an interaction weighting ratio (IWR), as follows: IWR =
1 + ( p ⋅ Hsyn ) + ( q ⋅ H pro ) 1 + ( r ⋅ Hant ) + ( s ⋅ H inh )
where H terms represent the hazard-modification scores for interactions (synergism (syn), promotion (pro), antagonism (ant), and inhibition (inh). The terms p, q, r, and s are simply weighting factors reflecting the relative importance of each form of interaction. The IWR in turn is used to calculate an exponent index, which corresponds to a scale of the cancer risk associated with the mixture, based on the assumption of independent joint action. The product of IWR and exponent index is then used to classify a carcinogenic chemical mixture according to its level of concern as low, marginal, moderate, high-moderate, or high. This approach is appropriate for use as a qualitative aid in the cancer dose–response assessment of chemical mixtures. For addressing more effectively the issues related to extrapolations for dose–response assessment of carcinogenic chemical mixtures, mechanistic modeling frameworks are more useful. Biologically based dose–response models allow the evaluation of the cumulative impact of multiple carcinogens by accounting for the interactions that can occur at the pharmacokinetic or pharmacodynamic level, the magnitude of which may be a function of dose, route, and species. Indeed, the pharmacokinetic interactions in general, and metabolic interactions more specifically, can lead to changes in the tissue dose of the toxic moiety of one or more mixture components (Haddad et al. 2001b). If the tissue dose of a mixture component is altered during mixed exposures, a corresponding change in the risk level is likely, depending upon the shape of the dose–response curve (Krishnan et al. 2002). 21.3.4.2. Role of PBPK Models. PBPK models facilitate the prediction of the change in tissue dose of chemicals during exposures to mixtures of substances interacting at the ADME level (Haddad et al. 2001a). By accounting for the exposure concentration of mixture components and the interaction mechanisms at binary level, these models uniquely allow the prediction of the kinetics of chemicals in increasingly complex mixtures (Dobrev et al. 2001; Haddad et al. 2001a; Haddad and Krishnan 1998; Krishnan et al. 1994; Simmons 1997; Yang et al. 1995). Accordingly, PBPK models developed for individual chemicals can simulate the consequence of metabolic interactions at multiple levels, once the individual chemical descriptions are linked at the binary level (Figure 21.5). In this modeling approach, the kinetics of mixture components is simulated using a single set of physiological characteristics and multiple sets of chemicalspecific parameters. Binary connections between mixture components are established within the model on the basis of interaction mechanism(s). Thus, for example, once
21.3. PBPK MODELS IN CANCER RISK ASSESSMENT
573
Lungs Fat Lungs Fat Richly perfused tissues
Richly perfused tissues Slowly perfused tissues Liver
Slowly perfused tissues Liver
B
Lungs Fat Richly perfused tissues Slowly perfused tissues Liver
C
A D Liver Fat Richly perfused tissues Slowly perfused tissues Lungs
Figure 21.5. Conceptual representation of a PBPK model for a hypothetical mixture of four chemicals (ABCD) competing for hepatic metabolism.
the inhibitory effect of A on B metabolism via its Michaelis–Menten constant is described, the reduced rate of B metabolism and the corresponding increase in its blood concentration are simulated. The increase in the blood concentration of B during mixed exposures (compared to individual exposures) will likely lead to greater inhibition of the metabolism of other mixture components. Thus, during mixed exposures, the binary inhibition constants are not modified but the concentration of the inhibitors is altered as a function of the number and nature of the chemicals in the mixture (Krishnan et al. 2002). The data on altered blood kinetics and dose metrics simulated by mixture PBPK models can then be used in the cancer risk assessment process. Accordingly, for mixtures of carcinogens exhibiting threshold mechanism of action, the cancer risk assessment process would involve (Haddad et al. 2001a): • Characterizing the dose metrics associated with the guideline value of each of the mixture components • Obtaining predictions of dose metrics for each mixture component in humans, based on information on exposure condition defined as input to the mixture PBPK model
574
CHAPTER 21 PBPK MODELS IN CANCER RISK ASSESSMENT
• Determining the sum total of the ratios of the results of the above two steps for each component during mixed exposures Similarly, for carcinogens with slope factor, risk assessment of chemicals in mixtures would involve (Haddad et al. 2001a): • Establishment of a tissue dose metric-based slope factor (q*tissue ) using the animal PBPK model for each of the mixture constituents • Computation of the dose metric for each mixture constituent (Dtissue) associated with human exposure conditions using mixture PBPK models • Integration of the results from the above two steps to determine the cancer * response (P(d)) during mixed exposures, as follows: P ( d ) = ∑ q*tissue Dtissue In essence then, the PBPK modeling approach improves upon the dose addition and response-addition methods applied to carcinogenic chemical mixtures on the basis of data on the tissue dose of chemicals in mixtures, as well as the magnitude of pharmacokinetic interactions.
21.4. PBPK MODELS IN CANCER RISK ASSESSMENT: CASE STUDIES 21.4.1.
Dichloromethane (Methylene Chloride)
Andersen et al. (1987) developed a PBPK model for dichloromethane (DCM) that was capable of simulating the flux through the glutathione (GSH) conjugation pathway as well as the CYP-mediated oxidative pathway. In order to identify the appropriate dose metric for the cancer risk assessment of DCM, the predicted flux through each metabolic pathway as well as the concentration of the parent compound associated with the exposure concentrations used in the mouse bioassay (1–4000 ppm) were compared with the observed tumor incidence (Table 21.5). DCM being nonreactive, Andersen et al. (1987) considered it to be an unlikely candidate responsible for the tumorigenicity. Hence, the relationship between the tissue exposure to its metabolites and tumor incidence was examined. Whereas the simulated dose metric based on CYP-mediated oxidative pathway varied very little between 2000 and 4000 ppm, the flux through the GSH pathway increased with increasing exposure
TABLE 21.5. Improvement of Cancer Dose–Response Assessment for Dichloromethane Using PBPK Models to Compute Relevant Dose Metrics
Exposure (ppm) 0 2000 4000 a
CYP Dose Metricsa (mg metabolized/volume liver/day)
GST Dose Metricsa (mg metabolized/volume liver/day)
Tumor Prevalence
0 3575 3701
0 851 1811
0.06 0.33 0.83
Determined by PBPK modeling.
21.4. PBPK MODELS IN CANCER RISK ASSESSMENT: CASE STUDIES
575
concentrations of DCM and corresponded with the degree of DCM-induced liver tumors at the high exposure concentrations (Table 21.5). In this case, the low-dose extrapolation of the target tissue dose was performed on the basis of flux through the GSH conjugation pathway, which gave rise to a 21-fold lower risk estimate compared to the conventional linear extrapolation approach (Andersen et al. 1987). This discrepancy arises from the consideration of the saturation of CYP-mediated DCM metabolism at high-exposure concentrations in the PBPK models and the accurate prediction of the disproportionate increase in the flux through glutathione conjugation pathway at high exposure concentrations. By accounting for the speciesspecific differences in the rates of metabolism via CYP and GSH pathways as well as the physiology, the PBPK model simulated target tissue dose and human risk estimates that were about 100- to 200-fold lower than those predicted initially by the conventional approaches relying upon body surface scaling of exposure doses (Andersen et al. 1987). Following the data published by NTP (1986) and Andersen et al. (1987), the EPA (1987) proposed an IUR for DCM of 4.7 × 10−7 per ug/m3, based on deterministic mouse PBPK model applied to the data on lung and liver cancer combined with a surface area correction to estimate dose in humans (Andersen et al. 1987; EPA 1987; NTP 1986). More recently, Marino et al. (2006) refined the original PBPK model for DCM by introducing a probabilistic component in the physiological parameters of both the mouse and human models. Using Markov Chain Monte Carlo simulations, these authors first optimized the prior distributions of the parameter values in the mouse model using kinetic data from close-chamber, inhalation, and intravenous studies. The PBPK model with posterior distributions of input parameters was then applied to simulate the NTP bioassay experiments. The simulated dose metric of interest, namely the amount of DCM metabolized by the GSH pathway per day per volume of tissue, and consequently the cancer potency factors, were 3–4 times higher than those simulated by Andersen et al. (1987) and the EPA (1987) for lung and liver. The MCMC approach applied to the human DCM model by David et al. (2006) yielded risk estimates that were severalfold lower than the EPA’s estimate obtained using the previous versions of the deterministic PBPK model. More recently, to estimate the impact of using probabilistic instead of deterministic PBPK approach on the determination of the unit risk factor, Marino and Starr (2007) examined four hypothetical combinations of probabilistic distributions of parameters while modeling the exposure of the two mouse groups involved in the NTP study (Marino and Starr 2007). These combinations represented increasing levels of uncertainty and variability. The variability resulting from the dose metric estimation was quite similar in each of the four scenarios for both lung and liver. Combining the dosimetry data with the binomial distribution data of the NTP assay showed that regardless of the scenario examined, there was less than a 10% difference between central tendency and upper percentile value of the unit risk factor.
21.4.2.
Vinyl Chloride
Vinyl chloride (VC) has been reported to induce angiosarcoma of the liver in exposed animals and workers [e.g., Gehring et al. (1978, 1979)]. The dose–response
576
CHAPTER 21 PBPK MODELS IN CANCER RISK ASSESSMENT
for this chemical shows saturation at high exposure concentrations, consistent with the toxic moiety being the reactive metabolite, cyanoethylene oxide (CEO). Clewell et al. (1995) developed a PBPK model for vinyl chloride to predict the dose metrics associated with CEO (Clewell et al. 1995). The results of this modeling effort were then combined with a linearized multistage (LMS) model to estimate risk for human populations after a lifetime exposure to 1 ppb in air. The PBPK-based risk analysis, conducted on the basis of the amount of VC metabolized per unit time normalized to the volume of tissue, yielded 95% upper confidence limit (UCL) varying between 0.4 and 4.2 × 10−6 per ppb in human epidemiological studies. Applying this analysis to rodent bioassay data, similar risk estimates were obtained (i.e., varying between 1.1 and 3.3 × 10−6 per ppb in female rats and 1.5 and 5.2 × 10−6 per ppb in male rats). Subsequently, Clewell et al. (2001) extended the model to include the initial metabolism of VC occurring via two saturable pathways, one representing low capacity-high affinity oxidation by CYP2E1 and the other (in the rodent) representing higher capacity-lower affinity oxidation by other isozymes of P450 (Clewell et al. 2001). The authors simulated (a) alternative dose metrics (total amount of metabolite generated divided by the volume of liver (i.e., the tissue into which it is produced), (b) total amount conjugated with GSH, and (c) total amount of metabolite produced but not detoxified by reaction with GSH per L liver) for evaluation of their association with the tumor outcome. Even though roughly similar results were obtained for the various dose metrics, based on biological plausibility considerations, Clewell et al. (2001) used the dose metric related to metabolite production (milligrams formed per day per liter of liver) as the basis for the conduct interspecies, high-dose to low-dose and route-to-route extrapolations. Contrary to the risk assessment conventional approaches, the risk estimates based on dose metrics of relevance to mode of action, obtained with the PBPK models, were coherent with epidemiological risk estimates and were lower by a factor of 80 than those obtained in risk assessments based on default approaches (Clewell et al. 2001).
21.4.3.
Chloroform
Reitz et al. (1990a) evaluated the hepatic cancer risk for humans based on data for chloroform obtained in mice and rats. The amount of metabolite covalently bound to macromolecules, the rate of hepatocyte death, and subsequent overproliferation were used as dose metrics in the dose–response assessment. Thus, a threshold dose– response relationship based on relevance to the MOE was developed. In this regard, a virtually safe dose (VSD) in humans was calculated by intially dividing the PBPKderived no-observed effect level in rodent liver by uncertainty factor (totaling 1000) and then by deriving the human-equivalent exposure dose with the human PBPK model. The resulting acceptable chloroform concentrations were 2840 ppb in air and 13,900 ppb in water. The latter assessment has been further improved with the use of an integrated PBPK-BBDR (biologically based dose–response) model for chloroform by Liao et al. (2007b). These authors conducted MCMC simulations to optimize mouse and rat metabolic parameters as well as pharmacodynamic parameters. This model was then used to (a) simulate renal dosimetry resulting from an exposure associated with renal and liver cytolethality and (b) develop a human
21.4. PBPK MODELS IN CANCER RISK ASSESSMENT: CASE STUDIES
577
lifetime PBPK model by scaling up the rodent pharmacokinetic determinants. Agedependant variations in physiological and metabolic parameters for six age groups were considered for both sexes. On the basis of these groups, continuous oral and inhalation lifetime exposures associated with a threshold in liver and renal cytolethality were predicted. The values obtained were 0.4 mg/kg-day and 3 mg/kg-day for oral exposures and 0.09 ppm and 0.9 ppm for inhalation exposure, respectively. The cancer risk assessments for chemicals such as chloroform, which occur as drinking water contaminants, should take into account the additional exposures via noningestion routes as well. The contribution of dermal and inhalation routes to the target tissue dose could be as important as the principal exposure route (i.e., oral ingestion) [e.g., Krishnan and Carrier (2008)]. Levesque et al. (2002) applied a multi-route PBPK model to simulate human exposures during showering and to compute the amount of chloroform metabolized and bound to renal and hepatic macromolecules over a 24-hr period (Levesque et al. 2002). This study indicated that, following a 10-min shower, the metabolite concentration would be about 0.01 μg chloroform-equivalent/kg tissue in kidneys and liver, whereas multi-route exposures (shower plus drinking water ingestion plus indoor air inhalation) would result in 5–10 times greater exposure of the target tissues. More recent modeling studies indicate that the sum of inhalation and dermal exposures resulting from a 15-min shower correspond to about 70% and 50% of the chloroform dose resulting from oral ingestion of 1.5 L of water for a 70-kg adult, as calculated based on absorbed dose and amount metabolized, respectively (Haddad et al. 2006). In a recent health risk assessment for chloroform, the contribution of multi-route exposures was evaluated and used in setting the drinking water goals (HC 2004a). Multi-route PBPK modeling was also used to identify exposures consistent with human biomonitoring data in order to assess the cancer risk associated with the latter (Tan et al. 2006).
21.4.4.
Dioxane
Leung and Paustenbach (1990) applied a PBPK model to predict the liver concentration of 1,4-dioxane in rats following oral doses of 14.3 and 1184 mg/kg-day given via drinking water. These doses corresponded, respectively, to NOAEL and the dose at which significant incidence of liver tumors was observed. Application of the conventional approaches of interspecies and high-dose to low-dose extrapolation (i.e., administered dose and body surface area scaling combined with the linearized multistage modeling) led to a risk-specific dose of 0.055 mg/kg-day in rats and 0.011 mg/kg-day in humans (for an excess cancer risk level of one in a million). Using the time-weighted average lifetime liver concentration of dioxane as the dose metric, the PBPK modeling approach indicated that dose associated with a liver cancer risk of 10−5 in humans might be as much as 80 times greater than that obtained using the conventional approaches. The authors attribute this discrepancy to the nonlinearity of the dose–response relationship, due to saturable metabolism of dioxane, which is adequately taken into account by the PBPK model. In order to reduce the uncertainty associated with the dioxane model of Reitz et al. (1990b), Sweeney et al. (2008) focused on the consideration of the production of a specific
578
CHAPTER 21 PBPK MODELS IN CANCER RISK ASSESSMENT
metabolite β-hydroxyacetic acid (HEAA), as well as better estimates of input parameters (e.g., blood:air and tissue:air partition coefficients as well as metabolic rate constants) for the PBPK model of dioxane in mice, rats, and humans.
21.4.5.
Trichloroethylene
The assessment of liver cancer risks associated with human exposure to trichloroethylene (TCE) was initially conducted by Fisher and Allen (1993) using a PBPKmodeling approach. The use of the amount of TCE metabolized per day as dose metric used in the linearized multistage model led to 10 ppb in air and 7 μg/L in water as acceptable concentrations—that is, environmental levels corresponding to a population cancer risk of 1 in 106 (Fisher and Allen 1993). Corresponding values based on circulating levels of the metabolite, trichloroacetic acid, were 10 times and twice lower than those based on the amount of TCE metabolized per unit time, whereas the acceptable TCE concentration in air as defined by the EPA at the time was 90 times lower. A number of authors subsequently investigated the dose metrics and cancer risks associated with TCE [e.g., Bois (2000), Clewell and Andersen (2004), Clewell et al. (1995, 2000), Cronin et al. (1995)]. In particular, Clewell et al. (1995) expanded the PBPK model by including descriptions for metabolites (dichloroacetic acid, chloral and dichlorovinylcysteine) and three target tissues (liver, lung, and kidney). In concordance with the cytotoxicity-mediated carcinogenicity of TCE (Bogen and Gold 1997), the results of the dosimety extrapolation performed with the PBPK model in humans and rodents were used to derive a human POD. Accordingly, virtually safe concentrations for lung, kidney, and liver cancer of 800, 8000, and 66 μg TCE/L air and 15,000, 30,000, and 265 μg TCE/L water, respectively, were obtained (Clewell and Andersen 2004). In deriving the guideline values for TCE in water, however, it is important to consider the additional dose received via dermal and inhalation exposures. In this regard, the relative importance of exposure routes has been evaluated by various agencies and researchers for TCE in drinking water. Assuming a water ingestion rate of 1.5 L/day, the L-equivalent values of 2.35 for inhalation and 1.1 for dermal route, based on absorbed dose, and 1.65 and 0.68 based on amount metabolized, were obtained with the PBPK modeling approach by Haddad et al. (2006). These values are consistent with relevant data available in the published literature (Lindstrom and Pleil 1996; Weisel and Jo 1996). Health Canada, in deriving its drinking water guidelines for TCE, used the results of a PBPK modeling effort in order to determine the systemic exposures resulting from dermal dose and inhalation of TCE associated with 30-min bath (HC 2004b).
21.4.6.
Volatile Organic Chemical Mixtures
Haddad et al. (2001a) conducted a cancer risk assessment for inhaled volatile organic chemical mixtures containing DCM and benzene. These authors used PBPK models to calculate the change in the tissue dose of the putative toxic moiety as well as the cancer risk during mixed exposures as compared to single chemical exposures. In the case of DCM, the GSH conjugate is the relevant dose surrogate (Andersen et al.
21.5. CONCLUDING REMARKS
Change in cancer risk
3.5
579
3.21
3 2.5 2
1.8
Dichloromethane Benzene
1.5 1
1
1
0.93
0.82
0.5 0 Single
Mixture A
Mixture B
Figure 21.6. PBPK model-based cancer risk assessment for 10 ppm of dichloromethane or 0.5 ppm of benzene alone or in mixture with 10 ppm each of toluene, m-xylene, and ethylbenzene (Mixture A); 5 ppm of toluene, 20 ppm of m-xylene and 40 ppm of ethylbenzene (Mixture B). Based on data from Haddad et al. (2001a).
1987) and the flux through this activation pathway increased disproportionately during mixed exposures to toluene, benzene, m-xylene and ethylbenzene all which compete with DCM for hepatic CYP2E1 metabolism. The competitive interaction among these substrates for oxidative metabolism leads to an increase in the flux of DCM metabolized through the GST pathway, thus resulting in an increase, by up to a factor of approximately 3, of its cancer risk (Figure 21.6). The PBPK modeling approach, however, suggested that cancer risk associated with benzene would decrease during mixed exposures compared to single chemical exposures, since the rate of formation of oxidative metabolites from benzene would be reduced during concurrent exposure to other P450 2E1 substrates (toluene, m-xylene, and ethylbenzene) (Haddad et al. 2001a).
21.5.
CONCLUDING REMARKS
PBPK modeling involves the development of mathematical descriptions of the interrelationships among critical parameters that determine the ADME of chemical carcinogens in biota. These models are part of a systems approach to the study of how chemicals gain entry into, distribute within, and are eliminated from the body. The biological and mechanistic basis of these models enables them to be used for the conduct of various extrapolations essential for the risk assessment of individual and mixtures of chemical carcinogens. The PBPK modeling efforts should ensure that (i) the assumptions on which the model is based are appropriate, (ii) the coding of model equations is errorless, (iii) the model parameter values are accurate, and (iv) the model is adequately evaluated/validated (Krishnan and Andersen 2007). Needless to say, the model is as good as the input parameters. Therefore, accurate parameterization is fundamentally important for constructing useful PBPK models. A number of prototypical descriptions developed and parameterized for chemical carcinogens, described in this chapter, can serve as examples for developing and
580
CHAPTER 21 PBPK MODELS IN CANCER RISK ASSESSMENT
applying PBPK models in the case of other chemicals of interest. Since each chemical carcinogen may possess some unique properties and challenges related to its MOA, novel model structures and methods of parameter estimation might continue to evolve in this field. However, unlike the default approaches used in conventional risk assessment, the PBPK models are versatile and are useful not only for generating a risk number, but also to identify critical data gaps and sources of significant uncertainty in an assessment.
REFERENCES Alcorn, J., and McNamara, P. J. (2003). Pharmacokinetics in the newborn. Adv Drug Deliv Rev 55, 667–686. Andersen, M. E., Clewell, H. J., III, and Frederick, C. B. (1995). Applying simulation modeling to problems in toxicology and risk assessment—A short perspective. Toxicol Appl Pharmacol 133, 181–187. Andersen, M. E., Clewell, H. J. I., Gargas, M. L., Smith, F. A., and Reitz, R. H. (1987). Physiologicallybased pharmacokinetics and risk assessment process for methylene chloride. Toxicol Appl Pharmacol 87, 185–205. Andersen, M. E., and Dennison, J. E. (2001). Mode of action and tissue dosimetry in current and future risk assessments. Sci Total Environ 274, 3–14. Andersen, M. E., Krewski, D., and Withey, J. R. (1993). Physiological pharmacokinetics and cancer risk assessment. Cancer Lett 69, 1–14. Andersen, M. E., and Krishnan, K. (1994). Physiologically based pharmacokinetics and cancer risk assessment. Environ Health Perspect 102(Suppl 1), 103–108. Andersen, M. E., Green, T., Frederick, C. B., and Bogdanffy, M. S. (2002). Physiologically based pharmacokinetic (PBPK) models for nasal tissue dosimetry of organic esters: assessing the state-ofknowledge and risk assessment applications with methyl methacrylate and vinyl acetate. Regul Toxicol Pharmacol 36, 234–245. Andersen, M. E., Sarangapani, R., Gentry, P. R., Clewell, H. J. III., Covington, T.R., and Frederick, C. B. (2000). Application of a hybrid CFD-PBPK nasal dosimetry model in an inhalation risk assessment: an example with acrylic acid. Toxicol Sci 57, 312–325. Arms, A. D., and Travis, C. C. (1988). Reference Physiological Parameters in Pharmacokinetic Modeling, Vol. EPA/600/6-88/004, U.S. EPA, Washington, D.C. Balci, O. (1997). Verification, validation and accreditation of simulation models. In Proceedings of the 1997 Winter Simulation Conference, Andradottir, S., Healy, K. J., Withers, D. H., and Nelson, B. L., eds., Society for Computer Simulation, San Diego, CA, pp. 135–141. Barton, H. A., Chiu, W. A., Woodrow, S. R., Andersen, M. E., Bailer, A. J., Bois, F. Y., Dewoskin, R. S., Hays, S., Johanson, G., Jones, N., Loizou, G., Macphail, R. C., Portier, C. J., Spendiff, M., and Tan, Y. M. (2007). Characterizing uncertainty and variability in physiologically based pharmacokinetic models: state of the science and needs for research and implementation. Toxicol Sci 99, 395–402. Blancato, J. N., and Chiu, N. (1993). Predictive modeling for uptake and tissue distribution from human exposures. In Craun, G., ed., Safety of Water Disinfection: Balancing Chemical and Microbial Risks. ILSI Press, Washington, DC, pp. 303–316. Bogdanffy, M. S., Plowchalk, D. R., Sarangapani, R., Starr, T. B., and Andersen, M. E. (2001). Modeof-action-based dosimeters for interspecies extrapolation on vinyl acetate inhalation risk. Inhal Toxicol 13, 377–396. Bogdanffy, M. S., Sarangapani, R., Plowchalk, D. R., Jarabek, A. M., and Andersen, M. E. (1999). A biologically-based risk assessment for vinyl acetate-induced cancer and noncancer inhalation toxicity. Toxicol Sci 51, 19–35. Bogen, K. T., and Gold, L. S. (1997). Trichloroethylene cancer risk: Simplified calculation of PBPKbased MCLs for cytotoxic end points. Regul Toxicol Pharmacol 25, 26–42.
REFERENCES
581
Bois, F. Y. (2000). Statistical analysis of Clewell et al. PBPK model of trichloroethylene kinetics. Environ Health Perspect 108(Suppl 2), 307–316. Byczkowski, J. Z., and Fisher, J. W. (1995). A computer program linking physiologically based pharmacokinetic model with cancer risk assessment for breast-fed infants. Comput Methods Programs Biomed 46, 155–163. Casanova, M., Conolly, R. B., and Heck, H. A. (1996). DNA-protein cross-links (DPX) and cell proliferation in B6C3F1 mice but not Syrian Golden Hamsters exposed to dichloromethane: pharmacokinetics and risk assessment with DPX as dosimeter. Fund Appl Toxicol 31, 103–116. Chiu, W. A., Barton, H. A., Dewoskin, R. S., Schlosser, P., Thompson, C. M., Sonawane, B., Lipscomb, J. C., and Krishnan, K. (2007). Evaluation of physiologically based pharmacokinetic models for use in risk assessment. J Appl Toxicol 27, 218–237. Clark, L. H., Woodrow Setzer, R., and Barton, H. A. (2004). Framework for evaluation of physiologically-based pharmacokinetic models for use in safety or risk assessment. Risk Anal 24, 1697–1717. Clewell, H. J., and Andersen, M. E. (2004). Applying mode-of-action and pharmacokinetic considerations in contemporary cancer risk assessments: An example with trichloroethylene. Crit Rev Toxicol 34, 385–445. Clewell, H. J., Gentry, P. R., Covington, T. R., Sarangapani, R., and Teeguarden, J. G. (2004). Evaluation of the potential impact of age- and gender-specific pharmacokinetic differences on tissue dosimetry. Toxicol Sci 79, 381–393. Clewell, H. J., Gentry, P. R., Gearhart, J. M., Allen, B. C., and Andersen, M. E. (1995). Considering pharmacokinetic and mechanistic information in cancer risk assessments for environmental contaminants: Examples with vinyl chloride and trichloroethylene. Chemosphere 31, 2561–2578. Clewell, H. J., Gentry, P. R., Gearhart, J. M., Allen, B. C., and Andersen, M. E. (2001). Comparison of cancer risk estimates for vinyl chloride using animal and human data with a PBPK model. Sci Total Environ 274, 37–66. Clewell, H. J., III, Andersen, M. E., and Barton, H. A. (2002a). A consistent approach for the application of pharmacokinetic modeling in cancer and noncancer risk assessment. Environ Health Perspect 110, 85–93. Clewell, H. J., III, Gentry, P. R., Covington, T. R., and Gearhart, J. M. (2000). Development of a physiologically based pharmacokinetic model of trichloroethylene and its metabolites for use in risk assessment. Environ Health Perspect 108(Suppl 2), 283–305. Clewell, H. J., Teeguarden, J., McDonald, T., Sarangapani, R., Lawrence, G., Covington, T., Gentry, R., and Shipp, A. (2002b). Review and evaluation of the potential impact of age- and gender-specific pharmacokinetic differences on tissue dosimetry. Crit Rev Toxicol 32, 329–389. Clewell, H. J. I., and Andersen, M. E. (1987). Dose, species and route extrapolation using physiologicallybased pharmacokinetic models. Drinking Water and Health 8, 159–182. Clewell, R. A., and Clewell, H. J., III (2008). Development and specification of physiologically based pharmacokinetic models for use in risk assessment. Regul Toxicol Pharmacol 50, 129–143. Cronin, W. J., Oswald, E. J., Shelley, M. L., Fisher, J. W., and Flemming, C. D. (1995). A trichloroethylene risk assessment using a Monte Carlo analysis of parameter uncertainty in conjunction with physiologically-based pharmacokinetic modeling. Risk Anal 15, 555–565. Cruzen, G., Carlson, G. P., Johnson, K. A., Andrews, L. S., Banton, M. I., Bevan, C., and Cushman, J. R. (2002). Styrene respiratory tract toxicity and mouse lung tumors are mediated by CYP2Fgenerated metabolites. Reg Toxicol Pharmacol 35, 308–319. David, R. M., Clewell, H. J., Gentry, P. R., Covington, T. R., Morgott, D. A., and Marino, D. J. (2006). Revised assessment of cancer risk to dichloromethane II. Application of probabilistic methods to cancer risk determinations. Regul Toxicol Pharmacol 45, 55–65. Davies, B., and Morris, T. (1993). Physiological parameters in laboratory animals and humans. Pharm Res 10, 1093–1095. Dobrev, I. D., Andersen, M. E., and Yang, R. S. (2001). Assessing interaction thresholds for trichloroethylene in combination with tetrachloroethylene and 1,1,1-trichloroethane using gas uptake studies and PBPK modeling. Arch Toxicol 75, 134–144. Doerge, D. R., Young, J. F., Chen, J. J., Dinovi, M. J., and Henry, S. H. (2008). Using dietary exposure and physiologically based pharmacokinetic/pharmacodynamic modeling in human risk extrapolations for acrylamide toxicity. J Agric Food Chem 56, 6031–6038.
582
CHAPTER 21 PBPK MODELS IN CANCER RISK ASSESSMENT
ECETOC (2003). Derivation of Assessment Factors for Human Health Risk Assessment, European Centre for Ecotoxicology and Toxicology of Chemicals (Technical Report No. 86). Brussels, Belgium. El-Masri, H. A., Bell, D. A., and Portier, C. J. (1999). Effects of glutathione transferase theta polymorphism on the risk estimates of dichloromethane to humans. Toxicol Appl Pharmacol 158, 221–230. EPA (1987). Technical Analysis of New Methods and Data Regarding Dichloromethane Hazard Assessments, Vol. EPA/600/8-87/029A, U.S. EPA Office of Health and Environmental Assessment, Washington, D.C. EPA (2000). Supplemental Guidance for Conducting Health Risk Assessment of Chemical Mixtures, U.S. EPA Risk Assessment Forum, Washington, D.C. EPA (2005a). Guidelines for Carcinogen Risk Assessment, Vol. EPA/630/P-03/001B, U.S. EPA Office of Research and Development, Washington, D.C. EPA (2005b). Supplemental Guidance for Assessing Susceptibility from Early-Life Exposure to Carcinogen, Vol. EPA630/R-03/003F, U.S. EPA Office of Research and Development, Washington, D.C. EPA (2006). Approaches for the Application of Physiologically Based Pharmacokinetic (pbpk) Models and Supporting Data in Risk Assessment (Final Report), U.S. EPA, Vol. EPA/600/r-05/043f, Washington, D.C. Fiserova-Bergerova, V. (1975). Mathematical modeling of inhalation exposure. J Combust Toxicol 32, 201–210. Fisher, J. W., and Allen, B. C. (1993). Evaluating the risk of liver cancer in humans exposed to trichloroethylene using physiological models. Risk Anal 13, 87–95. Gehring, P. J., Watanabe, P. G., and Park, C. N. (1978). Resolution of dose–response toxicity data for chemicals requiring metabolic activation: Example—vinyl chloride. Toxicol Appl Pharmacol 44, 581–591. Gehring, P. J., Watanabe, P. G., and Park, C. N. (1979). Risk of angiosarcoma in workers exposed to vinyl chloride as predicted from studies in rats. Toxicol Appl Pharmacol 49, 15–21. Gentry, P. R., Clewell, H. J., and Andersen, M. E. (2004). Good Modeling Practices for Pharmacokinetic Models in Chemical Risk Assessment. Contract report submitted to Health Canada. Gentry, P. R., Covington, T. R., and Clewell, H. J. III. (2003). Evaluation of the potential impact of pharmacokinetic differences on tissue dosimetry in offspring during pregnancy and lactation. Regul Toxicol Pharmacol 38, 1–16. Gerlowski, L. E., and Jain, R. K. (1983). Physiologtically based pharmacokinetic modeling principles and applications. J Pharm Sci 72, 1103–1127. Gerrity, T. R., Henry, C. J., and Birnbaum, L. (1990). Principles of Route-to-Route Extrapolation for Risk Assessment. Elsevier, New York. Ginsberg, G., Hattis, D., Russ, A., and Sonawane, B. (2004a). Physiologically based pharmacokinetic (PBPK) modeling of caffeine and theophylline in neonates and adults: Implications for assessing children’s risks from environmental agents. J Toxicol Environ Health A 67, 297–329. Ginsberg, G., Hattis, D., and Sonawane, B. (2004b). Incorporating pharmacokinetic differences between children and adults in assessing children’s risks to environmental toxicants. Toxicol Appl Pharmacol 198, 164–183. Ginsberg, G., Hattis, D., Sonawane, B., Russ, A., Banati, P., Kozlak, M., Smolenski, S., and Goble, R. (2002). Evaluation of child/adult pharmacokinetic differences from a database derived from the therapeutic drug literature. Toxicol Sci 66, 185–200. Ginsberg, G. L. (2003). Assessing cancer risks from short-term exposures in children. Risk Anal 23, 19–34. Gladtke, E. (1973). Pharmacokinetics in relation to age. Boll Chim Farm 112, 333–341. Haddad, S., Beliveau, M., Tardif, R., and Krishnan, K. (2001a). A PBPK modeling-based approach to account for interactions in the health risk assessment of chemical mixtures. Toxicol Sci 63, 125–131. Haddad, S., and Krishnan, K. (1998). Physiological modeling of toxicokinetic interactions: Implications for mixture risk assessment. Environ Health Perspec 106, 1377–1384. Haddad, S., Restieri, C., and Krishnan, K. (2001b). Characterization of age-related changes in body weight and organ weights from birth to adolescence in humans. J Toxicol Environ Health A 64, 453–464.
REFERENCES
583
Haddad, S., Tardif, G. C., and Tardif, R. (2006). Development of physiologically based toxicokinetic models for improving the human indoor exposure assessment to water contaminants: Trichloroethylene and trihalomethanes. J Toxicol Environ Health A 69, 2095–2136. Haggard, H. W. (1924). The absorption, distribution and elimination of ethyl ether. Analysis of the mechanism of the absorption and elimination of such a gas or vapor as ethyl ether. J Biol Chem 59, 753–770. HC (2004a). Chloroform in drinking water. H. C. Federal-Provincial-Territorial Committee on Drinking Water, Ottawa, Ontario. HC (2004b). Trichloroethylene in drinking water. H. C. Federal-Provincial-Territorial Committee on Drinking Water, Ottawa, Ontario. Himmelstein, M. W., Carpenter, S. C., Evans, M. V., Hinderliter, P. M., and Kenyon, E. M. (2004). Kinetic modeling of beta-chloroprene metabolism: II. The application of physiologically based modeling for cancer dose response analysis. Toxicol Sci 79, 28–37. ICRP (1975). International Commission on Radiation Protection: Report of the Task Group on Reference Man, Pergamon Press, New York. IGHRC (2005). Guidelines on route-to-route extrapolation of toxicity data when assessing health risks of chemicals. Research Council Institute for Environment and Health, Draft v6.0, University of Leicester, Leicester, UK. Jonsson, F., and Johanson, G. (2001). A Bayesian analysis of the influence of GSTT1 polymorphism on the cancer risk estimate for dichloromethane. Toxicol Appl Pharmacol 174, 99–112. Jonsson, F., and Johanson, G. (2002). Physiologically based modeling of the inhalation kinetics of styrene in humans using a bayesian population approach. Toxicol Appl Pharmacol 179, 35–49. Kety, S. S. (1951). The theory and application of the exchange of inert gas at the lungs. Pharmacol Rev 3, 1–41. Kirman, C. R., Hays, S. M., Kedderis, G. L., Gargas, M. L., and Strother, D. E. (2000). Improving cancer dose-response characterization by using physiologically based pharmacokinetic modeling: an analysis of pooled data for acrylonitrile-induced brain tumors to assess cancer potency in the rat. Risk Anal 20, 135–151. Kirman, C. R., Gargas, M. L., Marsh, G. M., Strother, D. E., Klaunig, J. E., Collins, J. J., and Deskin, R. (2005). Cancer dose–response assessment for acrylonitrile based upon rodent brain tumor incidence: Use of epidemiologic, mechanistic, and pharmacokinetic support for nonlinearity. Regul Toxicol Pharmacol 43, 85–103. Kohn, M. C. (1995). Achieving credibility in risk assessment models. Toxicol Lett 79, 107–114. Krishnan, K., and Andersen, M. E. (1991a). Interspecies scaling in pharmacokinetics. In Rescigno, A., and Thakkur, A. K., eds., New Trends in Pharmacokinetics. Plenum Press, New York, pp. 203–226. Krishnan, K., and Andersen, M. E. (1991b). Physiological modeling and cancer risk assessmrent. In Rescigno, A., and Thakur, A. K., eds., New Trends in Pharmacokinetics. Plenum Press, New York, pp. 335–354. Krishnan, K., and Andersen, M. E. (2001). Physiologically based pharmacokinetic modeling in toxicology. In Hayes, A. W., ed., Principles and Methods of Toxicology. Vol. 4. Taylor & Francis, Philadelphia, pp. 193–241. Krishnan, K., and Andersen, M. E. (2007). Physiologically based pharmacokinetic and toxicokinetic models. In Hayes, A. W., ed., Principles and Methods of Toxicology. Vol. 5. CRC Press, Boca Raton, FL, pp. 231–292. Krishnan, K., and Carrier, R. (2008). Approaches for evaluating the relevance of multiroute exposures in establishing guideline values for drinking water contaminants. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev 26, 300–316. Krishnan, K., Clewell, H. J., III, and Andersen, M. E. (1994). Physiologically based pharmacokinetic analyses of simple mixtures. Environ Health Perspect 102(Suppl 9), 151–155. Krishnan, K., Haddad, S., Beliveau, M., and Tardif, R. (2002). Physiological modeling and extrapolation of pharmacokinetic interactions from binary to more complex chemical mixtures. Environ Health Perspect 110(Suppl 6), 989–994. Krishnan, K., Haddad, S., and Pelekis, M. (1995). A simple index for representing the discrepancy between simulations of physiological pharmacokinetic models and experimental data. Toxicol Ind Health 11, 413–422.
584
CHAPTER 21 PBPK MODELS IN CANCER RISK ASSESSMENT
Leung, H. W., and Paustenbach, D. J. (1990). Cancer risk assessment for dioxane based upon a physiologically-based pharmacokinetic modeling approach. Toxicol Lett 51, 147–162. Levesque, B., Ayotte, P., Tardif, R., Ferron, L., Gingras, S., Schlouch, E., Gingras, G., Levallois, P., and Dewailly, E. (2002). Cancer risk associated with household exposure to chloroform. J Toxicol Environ Health A 65, 489–502. Liao, K. H., Tan, Y. M., and Clewell, H. J., III (2007a). Development of a screening approach to interpret human biomonitoring data on volatile organic compounds: Reverse dosimetry on biomonitoring data for trichloroethylene. Risk Anal 27, 1223–1236. Liao, K. H., Tan, Y. M., Conolly, R. B., Borghoff, S. J., Gargas, M. L., Andersen, M. E., and Clewell, H. J. III. (2007b). Bayesian estimation of pharmacokinetic and pharmacodynamic parameters in a mode-of-action-based cancer risk assessment for chloroform. Risk Anal 27, 1535–1551. Lindstrom, A. B., and Pleil, J. D. (1996). A methodological approach for exposure assessment studies in residences using volatile organic compound-contaminated water. J Air Waste Manag Assoc 46, 1058–1066. Lipscomb, J. C., and Ohanian, E. V. eds. (2007). Toxicokinetics and Risk Assessment. Informa Healthcare, New York. Lipscomb, J. C., and Poet, T. S. (2008). In vitro measurements of metabolism for application in pharmacokinetic modeling. Pharmacol Ther 118, 82–103. Loizou, G., Spendiff, M., Barton, H. A., Bessems, J., Bois, F. Y., d’Yvoire, M. B., Buist, H., Clewell, H. J., III, Meek, B., Gundert-Remy, U., Goerlitz, G., and Schmitt, W. (2008). Development of good modelling practice for physiologically based pharmacokinetic models for use in risk assessment: The first steps. Regul Toxicol Pharmacol 50, 400–411. Mapleson, W. W. (1963). An electric analog for uptake and elimination in man. J Appl Physiol 18, 197–204. Marino, D. J., Clewell, H. J., Gentry, P. R., Covington, T. R., Hack, C. E., David, R. M., and Morgott, D. A. (2006). Revised assessment of cancer risk to dichloromethane: Part I Bayesian PBPK and dose– response modeling in mice. Regul Toxicol Pharmacol 45, 44–54. Marino, D. J., and Starr, T. B. (2007). Probabilistic dose–response modeling: Case study using dichloromethane PBPK model results. Regul Toxicol Pharmacol 49, 285–300. Maruyama, W., and Aoki, Y. (2006). Estimated cancer risk of dioxins to humans using a bioassay and physiologically based pharmacokinetic model. Toxicol Appl Pharmacol 214, 188–198. Meek, M. E., Beauchamp, R., Long, G., Turner, L., and Walker, M. (2002). Chloroform: exposure estimation, hazard characterization, and exposure-response analysis. J Toxicol Environ Health 5 (B), 283–334. Melnick, R. L., and Kohn, M. C. (2000). Dose–response analyses of experimental cancer data. Drug Metab Rev 32, 193–209. Nong, A., McCarver, D. G., Hines, R. N., and Krishnan, K. (2006). Modeling interchild differences in pharmacokinetics on the basis of subject-specific data on physiology and hepatic CYP2E1 levels: A case study with toluene. Toxicol Appl Pharmacol 214, 78–87. NRC (1983). Risk Assessment in the Federal Government: Managing the Process, National Academy Press, Washington, D.C. NTP (1986). Toxicology and carcinogenesis studies of dichloromethane (methylene chloride) (CAS No. 75-09-2) in F344/N rats and B6C3F1 mice (inhalation studies). National Toxicology Program, NTP-TR-1986. Pelekis, M., Gephart, L. A., and Lerman, S. E. (2001). Physiological-model-based derivation of the adult and child pharmacokinetic intraspecies uncertainty factors for volatile organic compounds. Regul Toxicol Pharmacol 33, 12–20. Pelekis, M., Nicolich, M. J., and Gauthier, J. S. (2003). Probabilistic framework for the estimation of the adult and child toxicokinetic intraspecies uncertainty factors. Risk Anal 23, 1239–1255. Poet, T. S., Soelberg, J. J., Weitz, K. K., Mast, T. J., Miller, A. A., Thrall, B. D., and Corley, R. A. (2003). Mode of action and pharmacokinetic studies of 2-butoxyethanol in the mouse with an emphasis on forestomach dosimetry. Toxicol Sci 71, 176–189. Price, K., Haddad, S., and Krishnan, K. (2003a). Physiological modeling of age-specific changes in the pharmacokinetics of organic chemicals in children. J Toxicol Environ Health A 66, 417–433. Price, P. S., Conolly, R. B., Chaisson, C. F., Gross, E. A., Young, J. S., Mathis, E. T., and Tedder, D. R. (2003b). Modeling interindividual variation in physiological factors used in PBPK models of humans. Crit Rev Toxicol 33, 469–503.
REFERENCES
585
Reitz, R. H., Gargas, M. L., Andersen, M. E., Provan, W. M., and Green, T. L. (1996). Predicting cancer risk from vinyl chloride exposure with a physiologically based pharmacokinetic model. Toxicol Appl Pharmacol 137, 253–267. Reitz, R. H., Mandrela, A. L., Corley, R. A., Quast, J. F., Gargas, M. L., Andersen, M. E., Staats, D. A., and Conolly, R. B. (1990a). Estimating the risk of liver cancer associated with human exposures to chloroform using physiologically-based pharmacokinetic modeling. Toxicol Appl Pharmacol 105, 443–459. Reitz, R. H., McCroskey, P. S., Park, C. N., Andersen, M. E., and Gargas, M. L. (1990b). Development of a physiologically-based pharmacokinetic model for risk assessment with 1,4-dioxane. Toxicol Appl Pharmacol 105, 37–54. Sarangapani, R., Gentry, P. R., Covington, T. R., Teeguarden, J. G., and Clewell, H. J., III (2003). Evaluation of the potential impact of age- and gender-specific lung morphology and ventilation rate on the dosimetry of vapors. Inhal Toxicol 15, 987–1016. Schlosser, P. M., Patrick, D. L., Conolly, R. B., Janszen, D. B., and Kimbell, J. S. (2003). Benchmark dose risk assessment for formaldehyde using airflow modeling and a single-compartment, DNA-protein cross-link dosimetry model to estimate human equivalent doses. Risk Analysis 23, 473–487. Simmons, J. E. (1997). Application of physiologically based pharmacokinetic modelling to combination toxicology. Food Chem Toxicol 34, 1067–1073. Slikker, W., Jr., Andersen, M. E., Bogdanffy, M. S., Bus, J. S., Cohen, S. D., Conolly, R. B., David, R. M., Doerrer, N. G., Dorman, D. C., Gaylor, D. W., Hattis, D., Rogers, J. M., Woodrow, S. R., Swenberg, J. A., and Wallace, K. (2004). Dose-dependent transitions in mechanisms of toxicity. Toxicol Appl Pharmacol 201, 203–225. Strolin, B. M., and Baltes, E. L. (2003). Drug metabolism and disposition in children. Fundam Clin Pharmacol 17, 281–299. Sweeney, L. M., Thrall, K. D., Poet, T. S., Corley, R. A., Weber, T. J., Locey, B. J., Clarkson, J., Sager, S., and Gargas, M. L. (2008). Physiologically based pharmacokinetic modeling of 1,4-dioxane in rats, mice, and humans. Toxicol Sci 101, 32–50. Tan, Y. M., Liao, K. H., and Clewell, H. J., III (2007). Reverse dosimetry: Interpreting trihalomethanes biomonitoring data using physiologically based pharmacokinetic modeling. J Expo Sci Environ Epidemiol 17, 591–603. Tan, Y. M., Liao, K. H., Conolly, R. B., Blount, B. C., Mason, A. M., and Clewell, H. J. (2006a). Use of a physiologically based pharmacokinetic model to identify exposures consistent with human biomonitoring data for chloroform. J Toxicol Environ Health A 69, 1727–1756. Verner, M. A., Charbonneau, M., Lopez-Carrillo, L., and Haddad, S. (2008). Physiologically based pharmacokinetic modeling of persistent organic pollutants for lifetime exposure assessment: A new tool in breast cancer epidemiologic studies. Environ Health Perspect 116, 886–892. Voisin, E. M., Ruthsatz, M., Collins, J. M., and Hoyle, P. C. (1990). Extrapolation of animal toxicity to humans: Interspecies comparisons in drug development. Regul Toxicol Pharmacol 12, 107–116. Weisel, C. P., and Jo, W. K. (1996). Ingestion, inhalation, and dermal exposures to chloroform and trichloroethene from tap water. Environ Health Perspect 104, 48–51. WHO (1999). Principles for the Assessment of Risks to Human Health from Exposure to 4 Chemicals, Vol. EHC 210, World Health Organisation, Geneva, Switzerland. Woo, Y. T., Di Carlo, F. J., Arcos, J. C., Argus, M. F., Polansky, G., and DuBose, J. (1994). Assessment of carcinogenic hazard of chemical mixtures through analysis of binary chemical interaction data. Environ Health Perspect 102(Suppl 9), 113–118. Yang, R. S. H., El-Masri, H. A., Thomas, R. S., and Constan, A. A. (1995). The use of physiologicallybased pharmacokinetic pharmacodynamic dosimetry models for chemical mixtures. Toxicol Lett 82– 83, 497–504. Younes, M., Sonich-Mullin, C., and Meek, M. E. (1998). Risk: Assessment and management. In Herzstein, J. A., Bunn, W. B., and Fleming, L. E., eds., International Occupational and Environmental Medicine. Mosby, St Louis, MO.
CH A P TE R
22
GENOMICS AND ITS ROLE IN CANCER RISK ASSESSMENT Banalata Sen Douglas C. Wolf Vicki Dellarco
22.1.
INTRODUCTION
The traditional risk assessment paradigm is based on exposure–dose–response (see Chapter 1).* The individual is exposed to a chemical or other stressor at some dose, and a response in the organism or tissue is elicited. Though precursor events such as target cell proliferation may be used as the response in place of a frank effect such as tumors, the rodent cancer bioassay has formed the basis for health risk assessment and regulatory decisions for several decades. Several extrapolations or inference methods are necessary when using the results from experimental animal models to predict human health consequences (see Chapter 14). Unless there is evidence to the contrary, it is typically assumed that rodent data predict responses in humans and that findings at high experimental doses predict effects at environmental exposure levels. It is also assumed, when there are no data to the contrary, that effects in young adult animals predict responses for other life stages. These assumptions have been the center of intense discussion and debate. Conventional rodent toxicity studies characterize adverse effects of a chemical primarily on apical endpoints such as clinical signs or pathological states. Evidence of organ toxicity in the form of an apical endpoint does not always provide mechanistic understanding of the toxicity involved (see Chapter 13). The exposure of rodents in a cancer bioassay model can result in species-specific responses that are not relevant to humans (e.g., alpha2u-globulin-induced rat renal tumors) (see Chapter 18) (EPA 1991). Rodents may also have increased sensitivity to a particular toxicity pathway relative to humans (e.g., disruption of thyroid homeostasis and thyroid follicular tumors in rodents) (EPA 1998; IARC 2001). There are rodent responses to chemical treatment in tissues where there is a high spontaneous incidence to develop *The opinions expressed in this chapter are those of the authors and are not meant to represent the opinions or policies of the U.S. Environmental Protection Agency.
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
586
22.1. INTRODUCTION
587
tumors. For example, the rodent liver is a common target (irrespective of the database analyzed) (see Chapter 16); but unlike the rodent, the incidence of liver cancer in humans occurs at a much lower frequency compared to the much higher frequency of prostate and breast cancer (Ries et al. 2008). Prostate tumors are rarely diagnosed in rodents (Thayer and Foster 2007). The traditional linear default procedure for extrapolating risk to environmental exposures has been a longstanding and often controversial issue. Conventional rodent studies may result in an inability to discern dose–response relationships below the observed range of empirical data partly due to species differences in background tumor rates. Also the consideration of the mode of carcinogenic action is extremely important in helping interpret the relevancy of the laboratory animal data and guide appropriate extrapolation to humans (including the use of no effect levels and margins of safety). An organized approach to risk assessment and research in support of risk assessment was presented by the U.S. National Research Council (NRC), which describes how the biologically effective dose is related to the precursor biological response that is ultimately related to the adverse health consequence of exposure to the stressor of concern (NRC 1994). Data gaps are identified that feed back into the design of supportive research that further informs the risk assessment. Though this approach represents advancement in the risk assessment process, it continues to be based on the traditional paradigm. More recently, cancer risk assessments have been developed through the process of describing a mode of action (see Chapter 13). The U.S. Environmental Protection Agency (EPA) defines a mode of action as “… a sequence of key events and processes, starting with interaction of an agent with a cell, proceeding through operational and anatomical changes, and resulting in cancer formation” (EPA 2005). It is unlikely that complete knowledge of how an agent causes cancer will exist, certainly for the near term. Thus, “mode” of action is contrasted in the U.S. EPA’s Guidelines for Carcinogenic Risk Assessment with “mechanism” of action, “which implies a more detailed understanding and description of events, often at the molecular level, than is meant by mode of action” (EPA 2005). In 1996, the U.S. EPA formally proposed to make information on mode of carcinogenic action a pivotal component of the cancer risk assessment process. The most frequent comment about this proposal was that more guidance was needed on how to evaluate an agent’s mode of carcinogenic action. In response to this comment, the U.S. EPA finalized an analytical framework for judging whether available evidence supports a mode of carcinogenic action postulated for an animal response to chemical treatment (EPA 2005). This framework was developed in conjunction with work performed by the International Programme for Chemical Safety (IPCS) (Sonich-Mullin et al. 2001). This mode of action framework is based on considerations for causality in epidemiologic investigations originally articulated by Sir Austin Bradford Hill (Hill 1965), and it includes considerations of dose–response and temporal concordance, consistency, specificity, biological plausibility, and coherence. Later, the International Life Sciences Institute (ILSI) further developed the mode of action framework by incorporating a human relevance component (Meek et al. 2003). Once an animal mode of action is established, qualitative and quantitative comparisons of each key event between the experimental animal and humans (e.g., consideration of comparative biology, kinetics/metabolism,
588
CHAPTER 22 GENOMICS AND ITS ROLE IN CANCER RISK ASSESSMENT
anatomical variations, and relevant human disease states) enable a conclusion as to likely relevance of the mode of action for human risk. There is a great deal of interest in improving the assessment of potential human cancer risk by incorporating information on mode of action; however, the current regulatory testing paradigm, which is based on internationally validated test guidelines and performance of studies under Good Laboratory Practice (GLP) standards, is both expensive and time-consuming (see Chapter 14). For example, a guideline compliant, GLP 28-day repeated dose study costs in excess of $150,000.00 and takes up to a year to complete, from developing the study protocol to finalizing the study report. Though guideline/GLP studies are considered the “gold standard” for evaluating the toxicity of chemical entities, they provide little information on mode of action. Molecular tools may, however, provide a more cost- and time-effective approach to evaluating the potential toxicity of a chemical and have the added advantage of providing information about a chemical’s mode of action. The U.S. NRC’s report on Toxicity Testing in the 21st Century: A Vision and a Strategy emphasizes the importance of (1) moving from apical endpoints in evaluating safety to understanding how a chemical may perturb a normal cellular pathway (i.e., mode of action) and (2) making maximum use of modern tools of biology (NRC 2007). The discussion that follows provides an overview of “-omics” technologies and their use in the risk assessment paradigm. Case studies are presented where genomic data were used for elucidating modes of action, followed by a discussion of the use of genomics for predicting toxicity.
22.2.
“-OMICS” TECHNOLOGIES
Standard molecular biology techniques such as enzyme-linked immunosorbant assays (ELISA) and Western immunoblotting have furthered the understanding of specific proteins involved in mechanisms of toxicity, and quantitative polymerase chain reaction (qPCR) has helped identify the involvement of specific genes. These methods were designed to investigate a particular biochemical pathway by evaluating one protein or one gene at a time. However, with the new advances in highdensity molecular technologies (such as “-omics”) described briefly below, one can take a more comprehensive approach and obtain a “global” view of the cellular machinery involved in “toxicity pathways” following a toxic insult. Trancriptomics is a semiquantitative evaluation of the expression of genes through analysis of mRNA (Figure 22.1). The commonly used methods allow investigation of the entire genome of the target species of interest. Though there are many platforms and evaluation methods, the ultimate goal of all the scientific and research approaches is to determine which genes have transcribed a message and which have not. Proteomics can be a semiquantitative evaluation of proteins that have been translated from expressed genes to specific analysis and quantitation of individual proteins. Proteomics is a more complex analysis than transcriptional profiling because analysis of functionally expressed proteins includes quantifying not only the ones that are present, but also their three-dimensional structure and the presence of functional groups (Figure 22.1).
22.3. GENOMICS AND THE NEW RISK ASSESSMENT PARADIGM
589
DNA Gene mRNA
Transcriptomics
Amino Acid Chain Protein structure Post-translational modification
Proteomics
Protein function Metabolites
Metabolomics
Figure 22.1. Genomics is a science that evaluates the expression of transcribed genes (transcriptomics) from the inherited genome, translated proteins (proteomics), functionally active proteins, and the end products of metabolism, which are controlled by the functional proteins (metabolomics). See insert for color representation of this figure.
Metabolomics is a quantitative measure of the end products of metabolism in biological tissues and liquid products such as serum and urine (Figure 22.1). Metabolites, in this context, are not the degradation products of the metabolic and elimination processes associated with exposure to a xenobiotic. They are the normal end products of cellular metabolism of the natural constituents of the tissues and organism such as nitrogenous wastes, energy metabolism, and nucleic acid synthesis and degradation. Toxicogenomics is an approach that examines how the entire genome is involved in biological responses to environmental toxicants and stressors. It combines information from mRNA profiling (i.e., transcriptomics), cell or tissue protein profiling (i.e., proteomics), and metabolite profiling (i.e., metabolomics) to improve our understanding of how environmental stressors and toxicants may induce disease in the general population or susceptible individuals based on genetics (e.g., mutation, polymorphism, etc.) or lifestage (e.g., children). One potential product of this approach is to produce system-based computational models that can be used to predict potential adverse health effects and prioritize additional toxicity testing.
22.3. GENOMICS AND THE NEW RISK ASSESSMENT PARADIGM The likelihood of an adverse health effect is dependent on the interplay of several factors, including the target tissue dose of the toxicant, the genetics of the individual
590
CHAPTER 22 GENOMICS AND ITS ROLE IN CANCER RISK ASSESSMENT
and species, and phenotypic variability. The genetic background of the group or individual will determine what genes are available for transcription in response to the exposure. Proteins encoded by the transcribed genes get expressed in response to the stressor and elicit the response in the target tissue. Some of these proteins will be involved in cellular metabolism and determine the metabolic profile from the stressed tissue and individual. Thus, the genetic makeup of the individual determines the probability of the response (i.e., which genes are transcribed and which proteins are present) and whether the response progresses to an adverse effect. An improved description of the precursor biological response of concern and the mode of action can be developed through genomic-based identification of key events. This is the basis of the new paradigm for risk assessment (NRC 2007). All of these methodologies for examination of the processes from gene expression to protein function provide quantitative data, which can be organized within spreadsheets and examined using computer software-based statistical algorithms. These approaches allow one to interpret the data, to identify collections of expressed genes, and to determine functional or nonfunctional biological pathways. Examining these data in an organized way may result in the description of toxicity pathways that are specifically related to exposure to a stressor of concern and related to the development of an adverse health effect. Ultimately, this approach will provide a mechanistic description that can be used to identify key events that describe a mode of action resulting in the lesion of concern after exposure to a stressor or xenobiotic. The application of genomic technologies allows a significant refinement of the traditional risk assessment paradigm (Figure 22.2). With major advancements in the field of microarrays and bioinformatics, it is now possible to mine gene expression changes across thousands of genes to derive expression signatures that can identify relevant biological pathways and elucidate mechanisms of toxicity that lead to overt phenotypic changes. Gene expression profiling has the potential to inform possible key events in the modes of carcinogenic action, such as mutagenicity, inhibition of cell death, regenerative cell-proliferationinduced cytotoxicity, and immune suppression. The case studies discussed in the next section illustrate how gene expression profiles, after a chemical exposure, and in association with conventional toxicological endpoints, add valuable information to the overall cancer risk assessment process. These examples highlight the utility of toxicogenomic approaches for informing hazard identification and mode of action analysis.
22.4. 22.4.1.
CASE STUDIES Perfluorooctanoic Acid (PFOA)
Perfluorooctanoic acid (PFOA), a perfluoroalkyl acid (PFAA), is a peroxisome proliferator activator receptor-alpha (PPARα) activator. Its hepatic carcinogenicity and toxicity is mediated by activation of PPARα in the liver (see Chapter 17). Gene expression studies revealed that the majority of the genes induced by PFOA are involved in the transport and metabolism of fatty acids (Guruge et al. 2006; Rosen et al. 2008a,b). Up-regulation of multiple genes involved in the peroxisomal fatty
22.4. CASE STUDIES
591
Exposure Genome Dose Transcriptome Response
Adverse Health Effect
Proteome Metabolome
Figure 22.2. New Risk Assessment Paradigm. The goal in this new paradigm is to characterize the likelihood of the occurrence for an adverse health effect. The probability of this occurring is dependent of several factors. Characterizing the target tissue dose of the toxicant is still a critical feature. The genetic background of the group or individual will determine what genes are available to be transcribed in response to the exposure. The transcribed genes code for the translated proteins that are expressed in response to the stressor and that elicit the response in the target tissue. Some of these proteins will be involved in cellular metabolism and will determine the metabolic profile from the stressed tissue and from the individual. Thus, the interplay between the stressor at the tissue level, the genetic makeup of the individual, the genes that are transcribed, and the proteins that are present determine the probability of a response and whether it progresses to an adverse effect.
acid beta-oxidation pathway by PFOA may be a consequence of its structural similarity to endogenous fatty acids. The oxidative damage caused by PFOA to DNA and proteins is enhanced by its inability to induce the expression of anti-oxidative genes. PFOA alters cholesterol biosynthesis in the liver by affecting the expression of the rate-limiting enzyme hydroxymethylglutaryl-CoA reductase (Hmgcr); however, this does not alter cell membrane fluidity, an effect that has been observed with the structurally similar PFAA, perfluorooctanesulfonic acid (PFOS), suggesting alternative mechanisms for the two chemicals. Mechanistic data from transcriptional profile analysis that identified genes that control critical pathways for structurally related chemicals, like PFOA and PFOS, can be used to inform the development of relative potency factors.
22.4.2.
Formaldehyde and Glutaraldehyde
Formaldehyde and glutaraldehyde have been evaluated using the human relevance framework for determining their carcinogenic and toxic modes of action. Both aldehydes are toxic to the nasal epithelium in rats and mice, but only formaldehyde is carcinogenic at high doses (McGregor et al. 2006). Formaldehyde induces sustained increases in reparative cell proliferation and DNA–protein crosslinks. It is mutagenic in the nasal respiratory tissue lining of the rodents. The postulated mode of action for chronic exposure to formaldehyde includes sustained cytotoxicity and
592
CHAPTER 22 GENOMICS AND ITS ROLE IN CANCER RISK ASSESSMENT
cell proliferation as key events that lead to neoplasia, especially in obligatory nose breathers like rats and mice. The airway lesions caused by the chemicals, though qualitatively similar, were more severe in glutaraldehyde- than in formaldehydeexposed animals. Gene expression changes observed following exposure to aldehydes were consistent with the phenotypic changes, namely epithelial hyperplasia, metaplasia, and associated inflammatory cell infiltration and increased cell proliferation, that occur in the nasal epithelium in response to a persistent cytotoxic injury (Hester et al. 2003, 2005). Transcriptional changes associated with increased cell proliferation include increased expression of proliferating cell nuclear antigen (PCNA), cell division protein 25B, DNA topoisomerase IIB, and cyclin-dependent kinases. Comparative analyses by Hester et al. of gene expression profiles of the aldehydes have provided valuable insights into their different toxic response. The idea that formaldehyde either causes greater DNA damage or has a greater ability to induce DNA damage repair has been suggested by the increased induction of DNA repair genes by formaldehyde compared to glutaraldehyde. Though both aldehydes induce apoptosis, transcriptional profiling revealed that the two chemicals affect apoptosis by different mechanisms. Differential apoptosis-related gene expression patterns suggest that formaldehyde induces apoptosis through induction of the Fas-ligand and TNF receptor with subsequent induction of downstream caspases. In contrast, glutaraldehyde induces genes that are associated with the mitochondrial bcl-2 family, BAD and Bax. Transcriptional profiling of glutaraldehyde-exposed nasal tissue suggests that it causes a diminished DNA repair response due to the repressed induction of DNA repair genes and stimulated cell death as an alternative protective mechanism. In contrast, the transcriptional profile of formaldehydeexposed nasal tissue suggests that there is incomplete repair of cells experiencing growth arrest. These cells may persist, accumulate more defects, and eventually develop into tumors. This comparative analysis of gene expression profiles of a nasal carcinogen with a noncarcinogen provided an example of how genomics coupled with phenotypic indicators can be used to inform the mode of action by which a chemical may cause cancer (Hester et al. 2005).
22.4.3.
Conazoles
No unique patterns of tissue-specific responses have been identified for the induction of liver tumors by specific tumorigenic conazoles (i.e., propiconazoleand triadimefon) compared to a nontumorigenic conazole (i.e., myclobutanil) (Allen et al. 2006; Wolf et al. 2006). However, hepatic gene expression profiles in rats and mice have been identified for potential molecular pathways that may be involved in the induction of liver tumors by these chemicals (Chen et al. 2009; Hester and Nesnow 2008; Hester et al. 2006; Ward et al. 2006). Treatment with each of the three conazoles resulted in similar histopathological and clinical chemistry results. A common nongenotoxic mechanism of rodent hepatocarcinogenesis involves the activation of the constitutive androstane receptor (CAR) (see Chapter 16), hepatocyte hypertrophy, the induction of CYP2B, the induction of cell proliferation, and the inhibition of apoptosis has been proposed for conazoles (Chen et al. 2009; Peffer et al. 2007). These alterations were produced by
22.5. USE OF GENOMICS IN PREDICTIVE TOXICOLOGY
593
all three conazoles. Studies with rats and mice found similar results when perturbations of different cellular processes resulted in patterns that distinguished triadimefon from propiconazole. In mice, activation of nuclear receptors leading to overexpression of CYPs and associated induction of oxidative stress was the initial response to propiconazole and triadimefon as inferred from their gene expression profiles (Ward et al. 2006). These expression changes suggest genomic damage resulting from reactive oxygen and aldehyde species. Up-regulation of cholesterol biosynthesis, influencing the retinoic acid catabolism pathway and stimulating cell proliferation through key signaling pathways, has been hypothesized as a plausible mode of action for triadimefon-induced mouse liver tumors. A hypothesized mode of action for propiconazole-induced mouse liver tumors includes the following key events: (a) increased cell proliferation through down-regulation of the PTEN tumor suppressor pathway and (b) up-regulation of the WNT-β-catenin signaling pathway for cell growth, differentiation, and tumorigenesis (see Chapter 5). In rats, triadimefon effected molecular pathways associated with M-Ras regulation, cytoskeleton remodeling, and apoptosis by mitochondrial proteins, whereas propiconazole uniquely altered transcription of amino acid metabolism, ubiquitin metabolism, inflammation signaling, and glutathione metabolism (Hester et al. 2006). Pathway level analysis of the gene expression data in rats suggests that the tumor promotion potential of propiconazole occurs by the perturbation of pathways involved in inflammation response, growth factor receptor, and intracellular signaling pathways. Triademifon uniquely affects pathways involved in metabolic processes affecting amino acids, prostaglandins, fatty acids and leukotrienes, apoptosis, growth factors, cytoskeletal modeling, or G-protein-linked signaling. Dysregulation of cell cycle and metabolic growth processes represent key events for hepatotoxicity in rats treated with triadimefon. In contrast, stress response characterized by acute inflammation and protein metabolism represent key events that are characteristic of propiconazole-induced hepatotoxicity in rats.
22.5. USE OF GENOMICS IN PREDICTIVE TOXICOLOGY Toxicogenomics provides the opportunity for a comprehensive view of biological mechanisms and pathways that may be perturbed after a toxic insult. It may also afford the capability to predict toxicity and classify chemicals based on specific patterns of altered gene expression. An early study of toxicogenomics was able to classify chemicals based on their gene signatures (Ellinger-Ziegelbauer et al. 2005). The authors observed similar gene expression patterns corresponding to the liver tissue from rats exposed to peroxisome-proliferating agents (i.e., clofibrate, Wyeth 14,643, and gemfibrozil) compared to the expression profile for an enzyme-inducing agent (i.e., phenobarbital). Since 2002, extensive generation of microarray data has led to the creation of many reference databases. A comparison of gene signatures of five hepatotoxicants (i.e., myclobutanil, propiconazole, triadimefon, PFOA, and PFOS) to a genomic signature reference database showed matches to hepatotoxicity related genomics signatures for all five compounds (Martin et al. 2007). The authors
594
CHAPTER 22 GENOMICS AND ITS ROLE IN CANCER RISK ASSESSMENT
cross-referenced the gene expression profiles for the five chemical treatment groups with 128 literature-curated pathways relevant to pharmacological and toxicological processes. Although all the chemicals were found to modulate the regulation of cytochrome P450 genes, the three triazoles perturbed CYP genes that are known to be regulated by CAR whereas PFOA and PFOS induced PPARα-regulated genes. When gene expression profiles of nongenotoxic hepatocarcinogens (methapyrilene, diethylstilbestrol, WY-14,643, and phenobarbital) were compared with profiles of genotoxic carcinogens (2-nitrofluorene, dimethylnitrosamine, 4[methylnitrosamino]-1-[3-pyridyl]-1-butanone, and aflatoxin B1), distinct cellular pathways were affected by both classes of compounds (Hamadeh et al. 2002). Signatures typical of DNA damage response and activation of proliferative survival signaling signatures are characteristic of the genotoxic carcinogens. The nongenotoxic carcinogens show responses to oxidative DNA or protein damage as well as cell cycle progression and signs of regeneration. As mentioned earlier, the resources and time required for testing the carcinogenic potential of chemicals using rodent bioassays represents a significant barrier to obtaining data on the thousands of chemicals used in commerce. To overcome these limitations, identification of early biomarkers of carcinogenicity using highdensity data will be highly desirable because these data can be obtained with fewer resources and less time and can provide valuable information about mode of action. For example, gene expression profiles from livers of rats treated up to 14 days have been successfully used to extract biomarkers discriminating groups of nongenotoxic and genotoxic hepatocarcinogens to calculate classifier profiles with up to 88% accuracy (Ellinger-Ziegelbauer et al. 2005, 2008). Although fine-tuning and validation of the approach are needed, toxicogenomics data suggest that it might be possible to predict long-term toxicity based on short-term acute exposure data for carcinogens.
22.6.
CONCLUSIONS
Beyond providing valuable mechanistic information for addressing mode(s) of action, gene expression data can also be used to describe quantitatively the degree to which the gene expression changes occur across dose and time. Tools are being developed that address the quantitative aspects of risk assessment using genomics data. Thomas et al. (2007) developed a method to integrate benchmark dose estimates with genomic data to assess the functional effects of chemical exposure. This method allows for a comprehensive survey of molecular changes associated with chemical exposure and the identification of reference doses based on cellular processes affected as determined by changes in gene expression (Thomas et al. 2007). Yu et al. (2006) developed a method to interpret dose and time dependency of microarray data using a systems-based approach. Instead of basing their estimates on the expression of single genes, the authors performed a global measurement based on pathway response and gene ontology to generate an unbiased view of biological processes and cellular components that are affected by toxicant exposure at the transcriptional level (Yu et al. 2006).
REFERENCES
595
Carcinogenesis is a complex process involving multiple genetic alterations (see Chapter 5). Molecular investigations at the whole-genome level allow for a comprehensive investigation of the process. The power of genomics to understand the cancer process is amplified when combined with conventional analytic techniques such as biochemical analysis and histopathology. A combination of new and standard practices in toxicology will enhance our ability to inform mode of action analysis of chemical-induced toxicity and cancer. Toxicogenomic analysis shows that it requires more than a single mutated gene or altered pathway to sufficiently inform the mode of action and to discriminate between different classes of chemicals. Combinations of pathway-associated gene networks derived from gene expression profiles will prove to be a robust approach in predicting mechanisms of toxicity and classifying chemicals. This type of detailed mechanistic information fits into the mode of action framework and provides an improved scientific basis for determining human relevancy in cancer risk assessment.
REFERENCES Allen, J. W., Wolf, D. C., George, M. H., Hester, S. D., Sun, G., Thai, S. F., Delker, D. A., Moore, T., Jones, C., Nelson, G., Roop, B. C., Leavitt, S., Winkfield, E., Ward, W. O., and Nesnow, S. (2006). Toxicity profiles in mice treated with hepatotumorigenic and nonhepatotumorigenic triazole conazole fungicides: Propiconazole, triadimefon, and myclobutanil. Toxicol Pathol 34, 853–862. Chen, P. J., Padgett, W. T., Moore, T., Winnik, W., Lambert, G. R., Thai, S. F., Hester, S. D., and Nesnow, S. (2009). Three conazoles increase hepatic microsomal retinoic acid metabolism and decrease mouse hepatic retinoic acid levels in vivo. Toxicol Appl Pharmacol 234, 143–155. Ellinger-Ziegelbauer, H., Gmuender, H., Bandenburg, A., and Ahr, H. J. (2008). Prediction of a carcinogenic potential of rat hepatocarcinogens using toxicogenomics analysis of short-term in vivo studies. Mutat Res 637, 23–39. Ellinger-Ziegelbauer, H., Stuart, B., Wahle, B., Bomann, W., and Ahr, H. J. (2005). Comparison of the expression profiles induced by genotoxic and nongenotoxic carcinogens in rat liver. Mutat Res 575, 61–84. EPA (1991). Alpha2u-globulin: Association with chemically induced renal toxicity and neoplasia in the male rat. EPA/625/3-91/091F, 1–118. EPA (1998). Assessment of thyroid follicular cell tumors. EPA/630/R-97/002, 1–52. EPA (2005). Guidelines for carcinogen risk assessment. EPA/630/P-03/001F, 1–166, http://oaspub.epa. gov/eims/eimscomm.getfile?p_download_id=439797. Guruge, K. S., Yeung, L. W., Yamanaka, N., Miyazaki, S., Lam, P. K., Giesy, J. P., Jones, P. D., and Yamashita, N. (2006). Gene expression profiles in rat liver treated with perfluorooctanoic acid (PFOA). Toxicol Sci 89, 93–107. Hamadeh, H. K., Bushel, P. R., Jayadev, S., Martin, K., DiSorbo, O., Sieber, S., Bennett, L., Tennant, R., Stoll, R., Barrett, J. C., Blanchard, K., Paules, R. S., and Afshari, C. A. (2002). Gene expression analysis reveals chemical-specific profiles. Toxicol Sci 67, 219–231. Hester, S. D., Barry, W. T., Zou, F., and Wolf, D. C. (2005). Transcriptomic analysis of F344 rat nasal epithelium suggests that the lack of carcinogenic response to glutaraldehyde is due to its greater toxicity compared to formaldehyde. Toxicol Pathol 33, 415–424. Hester, S. D., Benavides, G. B., Yoon, L., Morgan, K. T., Zou, F., Barry, W., and Wolf, D. C. (2003). Formaldehyde-induced gene expression in F344 rat nasal respiratory epithelium. Toxicology 187, 13–24. Hester, S. D., and Nesnow, S. (2008). Transcriptional responses in thyroid tissues from rats treated with a tumorigenic and a non-tumorigenic triazole conazole fungicide. Toxicol Appl Pharmacol 227, 357–369.
596
CHAPTER 22 GENOMICS AND ITS ROLE IN CANCER RISK ASSESSMENT
Hester, S. D., Wolf, D. C., Nesnow, S., and Thai, S. F. (2006). Transcriptional profiles in liver from rats treated with tumorigenic and non-tumorigenic triazole conazole fungicides: Propiconazole, triadimefon, and myclobutanil. Toxicol Pathol 34, 879–894. Hill, A. B. (1965). The Environment and disease: Association or causation? Proc R Soc Med 58, 295–300. IARC (2001). Some Thyrotropic Agents. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans 79, 10–17. Martin, M. T., Brennan, R. J., Hu, W., Ayanoglu, E., Lau, C., Ren, H., Wood, C. R., Corton, J. C., Kavlock, R. J., and Dix, D. J. (2007). Toxicogenomic study of triazole fungicides and perfluoroalkyl acids in rat livers predicts toxicity and categorizes chemicals based on mechanisms of toxicity. Toxicol Sci 97, 595–613. McGregor, D., Bolt, H., Cogliano, V., and Richter-Reichhelm, H. B. (2006). Formaldehyde and glutaraldehyde and nasal cytotoxicity: Case study within the context of the 2006 IPCS Human Framework for the Analysis of a cancer mode of action for humans. Crit Rev Toxicol 36, 821–835. Meek, M. E., Bucher, J. R., Cohen, S. M., Dellarco, V., Hill, R. N., Lehman-McKeeman, L. D., Longfellow, D. G., Pastoor, T., Seed, J., and Patton, D. E. (2003). A framework for human relevance analysis of information on carcinogenic modes of action. Crit Rev Toxicol 33, 591–653. NRC (1994). Science and Judgment in Risk Assessment, National Academies Press, Washington, D.C., pp. 1–652. NRC (2007). Toxicity Testing in the 21st Century: A Vision and a Stategy, National Academies Press, Washington, D.C., 1–196. Peffer, R. C., Moggs, J. G., Pastoor, T., Currie, R. A., Wright, J., Milburn, G., Waechter, F., and Rusyn, I. (2007). Mouse liver effects of cyproconazole, a triazole fungicide: Role of the constitutive androstane receptor. Toxicol Sci 99, 315–325. Ries, L. A. G., Melbert, D., Krapcho, M., Stinchcomb, D. G., Howlader, N., Horner, M. J., Mariotto, A., Miller, B. A., Feuer, E. J., Altekruse, S. F., Lewis, D. R., Clegg, L., Eisner, M. P., Reichman, M., and Edwards, B. K. (2008). Annual SEER incidence and US death rats 1975–2005, National Cancer Institute, Bethesda MD, based on November 2007 SEER data submission, posted to the SEER website, 2008. http://seer.cancer.gov/csr/1975_2005/index.html. Rosen, M. B., Abbott, B. D., Wolf, D. C., Corton, J. C., Wood, C. R., Schmid, J. E., Das, K. P., Zehr, R. D., Blair, E. T., and Lau, C. (2008a). Gene profiling in the livers of wild-type and PPARalpha-null mice exposed to perfluorooctanoic acid. Toxicol Pathol 36, 592–607. Rosen, M. B., Lee, J. S., Ren, H., Vallanat, B., Liu, J., Waalkes, M. P., Abbott, B. D., Lau, C., and Corton, J. C. (2008b). Toxicogenomic dissection of the perfluorooctanoic acid transcript profile in mouse liver: Evidence for the involvement of nuclear receptors PPAR alpha and CAR. Toxicol Sci 103, 46–56. Sonich-Mullin, C., Fielder, R., Wiltse, J., Baetcke, K., Dempsey, J., Fenner-Crisp, P., Grant, D., Hartley, M., Knaap, A., Kroese, D., Mangelsdorf, I., Meek, E., Rice, J. M., Younes, M., and International Programme on Chemical Safety (2001). IPCS conceptual framework for evaluating a mode of action for chemical carcinogenesis. Regul Toxicol Pharmacol 34, 146–152. Thayer, K. A., and Foster, P. M. (2007). Workgroup report: National Toxicology Program Workshop on Hormonally Induced Reproductive Tumors—Relevance of Rodent Bioassays. Environ Health Perspect 115, 1351–1356. Thomas, R. S., Allen, B. C., Nong, A., Yang, L., Bermudez, E., Clewell, H. J., 3rd, and Andersen, M. E. (2007). A method to integrate benchmark dose estimates with genomic data to assess the functional effects of chemical exposure. Toxicol Sci 98, 240–248. Ward, W. O., Delker, D. A., Hester, S. D., Thai, S. F., Wolf, D. C., Allen, J. W., and Nesnow, S. (2006). Transcriptional profiles in liver from mice treated with hepatotumorigenic and nonhepatotumorigenic triazole conazole fungicides: Propiconazole, triadimefon, and myclobutanil. Toxicol Pathol 34, 863–878. Wolf, D. C., Allen, J. W., George, M. H., Hester, S. D., Sun, G., Moore, T., Thai, S. F., Delker, D., Winkfield, E., Leavitt, S., Nelson, G., Roop, B. C., Jones, C., Thibodeaux, J., and Nesnow, S. (2006). Toxicity profiles in rats treated with tumorigenic and nontumorigenic triazole conazole fungicides: Propiconazole, triadimefon, and myclobutanil. Toxicol Pathol 34, 895–902. Yu, X., Griffith, W. C., Hanspers, K., Dillman, J. F., 3rd, Ong, H., Vredevoogd, M. A., and Faustman, E. M. (2006). A system-based approach to interpret dose- and time-dependent microarray data: Quantitative integration of gene ontology analysis for risk assessment. Toxicol Sci 92, 560–577.
CH A P TE R
23
COMPUTATIONAL TOXICOLOGY IN CANCER RISK ASSESSMENT Jerry N. Blancato
23.1.
INTRODUCTION
A new era in toxicology testing has arrived (Committee on Toxicity Testing and Assessment of Environmental Agents: National Research Council 2007).* Suddenly toxicologists can study hundreds of biologic pathways at a myriad of levels within cells, tissues, organs, and organisms. The advent of high-throughput screening (HTS) studies and high content analysis (HCA) allows for studying multiple phenomena simultaneously. Furthermore, these complex phenomena can be now related to one another in ways that could previously only be speculated (Kavlock et al. 2007; Kavlock 2007). Thus the complex nature of interacting biologic actions can be better described, and the effect from exposure to exogenous stresses can be more accurately predicted. Likewise we can expect that this new advancing frontier of toxicology will change the basic approach to risk assessment. It is reasonable that these new methods in toxicology will be widely applied and only go from promise to acceptance as a result of rigorous comparisons to known biology. We should thus strive for acceptance and use of such findings in the risk assessment process as well. Risk assessment over the last half-century has for many individual cases served us well, but has proceeded on an extremely slow pace and has left us with considerable uncertainty. There are certainly thousands of compounds and thousands of exposure scenarios that remain untested and thus contribute to the great uncertainty surrounding risk from environmental exposures. This is further complicated when we consider that exposure to environmental chemicals and stressors does not occur simplistically. There is a myriad of co-exposures for each individual exposure. Time and frequency of exposure, concomitant ingestion of pharmaceuticals and dietary supplements, lifestyle behaviors, and underlying chronic diseases are all factors that may greatly change the expected impact of environmental exposure. Susceptibility and vulnerability are critical areas of concern that can be better addressed with newer and more rapid methods. *This document has been reviewed and approved for publication by the U.S. EPA but does not necessarily reflect Agency policy.
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
597
598
23.2.
CHAPTER 23 COMPUTATIONAL TOXICOLOGY IN CANCER RISK ASSESSMENT
RISK ASSESSMENT: HISTORICAL PERSPECTIVE
While risk assessment in the context of protecting public health has been performed for many years, it is the 1983 U.S. National Academy of Sciences Report (Committee on the Institutional Means for Assessment of Risks to Public Health: Commission on Life Sciences: National Research Council 1983) that has served as the tenet for practicing risk assessors (see Chapter 1). Risk assessment was defined as “… the characterization of the potential adverse health effects of human exposures to environmental hazards.” The predictive aspect of risk assessment was set by the use of the word potential. A fundamental expectation of the risk assessment process was that it should attempt to accurately predict adverse effects before there is evidence of disease in the population. Thus, risk assessment goes beyond the mere description of epidemiological and clinical case–control studies. In that report, the committee defined logical components of a risk assessment which still serve as guiding principles today. They were and are (a) hazard assessment or the qualitative determination that a stressor poses a hazard as evidence by causal evidence of an ill effect, (b) the dose–response assessment or a quantitative relationship between the magnitude of exposure and the probability of adverse health effects, (c) the exposure assessment or the extent of human exposure to the stressor in question from the environments of the population of concern, and (d) the risk characterization or the description of the nature and magnitude of risk and associated uncertainties under relevant exposure conditions for the population of concern. The hazard identification and dose–response assessments have required many data often derived from in vivo animal studies (see Chapter 14). As would be expected, the burden of proof from these studies has been rightfully set very high. Despite this high burden, numerous questions arose and many still remain. Many are associated with the meaning and concordance of results in animal studies with ill effects in humans. As a result, many toxicity studies are conducted in more than one species. Animal studies, especially those for carcinogenicity, for many reasons are conducted at very high doses. These high doses do not often mirror environmental concentrations to which the general population is exposed. They also leave considerable uncertainty regarding operational modes of action at true exposure levels (see Chapter 13). While such studies may imply or even establish a causal relationship between exposure and disease, they often shed little light upon exact mechanisms of this disease. As a result, little is known how different exposure scenarios, concomitant exposures, windows of vulnerability, and low-level exposures impact the occurrence of disease or the magnitude of the exposure–effect response. Furthering the conundrum are the expense and time required for such studies. Risk assessment has evolved somewhat since 1983. Let us consider the early risk assessment process, called here the classical era. Hazard was considered as a simple yes or no question. Exposure was used instead of the dose normally used in classic dose–response models in pharmacology studies. The mechanisms between exposure and the observed effects were usually unknown or a virtual “black box.” Next comes “neoclassical era” or where we are today. Here there is a foray into a better understanding and accounting for mechanisms of action. Time and resource requirements continue to be high and even increased in many cases.
23.3. ENHANCEMENTS IN QUANTITATIVE RISK ASSESSMENT
599
The evolution into this era began in the 1980s and still continues today. This evolution, if it continues along proper channels, will take us to the “modern era” or where we should be going. As more and more is learned about internal events, the better we can quantify and predict disease outcomes, characterize the uncertainty, and replace default black boxes with scientific knowledge.
23.3. ENHANCEMENTS IN QUANTITATIVE RISK ASSESSMENT 23.3.1. Physiologically Based Pharmacokinetic (PBPK) Modeling The move from what here is called the classical era to the neoclassical era centered on the describing the magnitude of an internal dose metric. The dose–response function has more meaning and considerably less uncertainty if the metric of dose is toxicologically relevant and biologically and mathematically related to site of toxicity. Many compounds of environmental relevance are transformed once entering a living system. This biotransformation sometimes exhibits nonlinear behavior across the different exposure levels of interest. Furthermore, the nonlinear behaviors are not always exactly mirrored across the different species. Thus, as has been obviously recognized for some time (Andersen 1981), if the nonlinear relation is different in test species versus the species of interest (e.g., human), using exposure or absorbed dose in the dose–response function could result in considerable error when performing extrapolations. PBPK models have been developed for a number of applications for several decades (Bischoff et al. 1970; Gibaldi and Perrier 1975; Jain et al. 1981; King and Dedrick 1981) and essentially are a series of mass balance equations describing the transport, transformation, and distribution of a drug or toxin within defined compartments (see Chapter 21). In physiologic models those compartments are typically anatomic organs, although typically some organs may be “lumped” rather than explicitly described. Lumping is a fine art but generally is done for compartments that are important for pharmacokinetic reasons but have no toxicological relevance. Also, lumped compartments should have similar pharmacokinetic properties. Techniques for lumping have been described (Bernareggi and Rowland 1991). Considerations for the formulation of the structure of PBPK models are often casespecific but some generic principles have been described (Blancato and Bischoff 1993; Gerlowski and Jain 1983). The application of PBPK modeling to better understand the disposition of potentially toxic compounds soon followed. Styrene and dichloromethane (DCM, or methylene chloride) are notable examples (Angelo et al. 1984; Angelo and Pritchard 1984; Ramsey and Andersen 1984; Andersen et al. 1987) and applications continue well into the current era (Blancato et al. 2007). While there are many uses for computational methods such as PBPK models in risk assessment, some knowledge of the mode of action is crucial for using them to enhance the characterization and quantification of the dose–response function. All
600
CHAPTER 23 COMPUTATIONAL TOXICOLOGY IN CANCER RISK ASSESSMENT
of the exact mechanisms that lead to the mode of action are almost never completely known. However, the more that are known, the lower the uncertainty in the assessment. Here we have used the term “toxicologically relevant dose.” In order to judge some metric of dose to be relevant, knowledge of mode and the concomitant pharmacokinetics is necessary. The selection of a suitable dose metric requires understanding of the biologic processes acting on the chemical and the chemical pharmacodynamic actions within the organism as well as knowledge about the pharmacokinetics of the compound. There is, however, a fine balance between necessary descriptive detail and parsimony in modeling, especially when the cost and time required for data gathering can be great. Let us look at two somewhat different cases to illustrate the points of model structure and selection of dose metric. The first case is the well-studied and often cited case of DCM (Andersen et al. 1987; Blancato et al. 1987; Angelo et al. 1984; Angelo and Pritchard 1984). Earlier assessments had been based on exposure assuming total absorption and that mode was linearly related to the amount of parent DCM entering the organism. However, careful study of the toxicology findings in light of the pharmacokinetics revealed a set of assumptions that were plausible and far more reasonable. The details of these carefully studied assumptions can be found elsewhere (Andersen et al. 1987; Blancato et al. 1987). A brief summary follows. DCM can be metabolized by two different pathways: (a) an oxidative pathway operative at low doses and saturable and (b) a glutathione-S-transferase (GST)-mediated pathway more active at higher doses and apparently linear within normal exposure and dosing levels in both animals and humans. Inspection of the tumor data in animals showed tumors to occur only at high doses, and the dose–response function followed the linear pathway. Furthermore, the quantitative metabolic profile for the GST-mediated pathway is different in animals when compared to humans. After extensive study and numerous experiments, it was accepted that the proper dose metric for DCM risk assessment should be the total amount of metabolite formed from the GST-mediated pathway in the tissues of concern, either the liver or the lung. For several reasons the lung was chosen as the location of the metric for risk assessment purposes. This was a great step forward for risk assessment. First, a reasonable dose metric was chosen for which there was evidence of a close mechanistic relationship with the potential adverse outcome. Second, because of this understanding and the knowledge of the metabolism and pharmacokinetics, a PBPK model was constructed. This model enabled a quantification of the dose metric in different species and for different dosing and exposure regimens and routes. Thus interspecies and interdose extrapolations could be done more rationally and with less uncertainty. Third, because of the first two developments the risk assessment was based on mode of action hence the relevance of dose, exposure levels, and test species relevance was better understood and considered. As a result, the cancer potency for DCM was changed by close to an order of magnitude from the previous value. In contrast, the second case illustrates use under different conditions. In this case a PBPK model was developed for methyl tertiary butyl ether (MTBE) (Blancato et al. 2007). There was more uncertainty on the exact mode of action and even on the details of metabolism beyond two pathways. Several models had been published,
23.3. ENHANCEMENTS IN QUANTITATIVE RISK ASSESSMENT
601
and models and data were examined (Licata et al. 2001; Prah et al. 2004; Rao and Ginsberg 1997). Toxicologists that reviewed the toxicity data determined that purposes of the assessment the parent (MTBE itself) compound was the critical species and that the area under the curve of the parent in the blood was the best metric for the risk assessment. The formulated model was then tested against available data and found to be in acceptable agreement with the data for the parent compound. Given this agreement the model was applied, but simulation results only for the parent compound were of interest and used. Furthermore, the application of the model was for equating internal doses [parent AUC (area under the curve) in blood] across species and across routes of administration or exposure (rodent versus human and inhalation versus oral). After the initial assessment the model was further applied to assess the impact or range of possible outcomes from variation in key and sensitive processes that affected the simulated dose metric. More fully discussed in the referenced paper (Blancato et al. 2007) the range of possible metabolic activity was observed in the data of several biochemical studies and applied by using the model. The resultant simulations showed the plausible predicted risks within that range. Illustrated here was a different use of a computational model. It should be noted that even a model using somewhat more limited data (no adequate time course concentration disposition in humans, especially for metabolic species) was still a valuable tool for predicting a dose metric of choice for which there were more data (parent compound) and for predicting across different routes of exposure and between different species. Furthermore, the model was used to assess potential impact on that risk from variability in a human population of interest. Thus it served as an in silico laboratory. In summary, PBPK modeling advanced risk assessment by better defining and quantifying the dose which would then be used in formulating and applying a dose– response function. In this way, a more relevant measure of dose, closer to the actual mechanism of action and more closely related to the toxicity, was employed. However, several barriers still existed. Variability of pharmacokinetic processes and their uncertainty would be addressed later. Uncertainty in mechanisms of action and differences in sensitivity between species and within a species were not addressed by PBPK modeling and had to be tackled elsewhere.
23.3.2.
Pharmacokinetic Variability and Uncertainty
On first construct, PBPK models are formulated for one average or ideal individual. Anatomic and physiologic parameters are rarely determined for each individual animal in a study or each individual in a clinical study. Certainly for risk assessment across a population it would be impossible to determine individual values for the myriad of anatomic and physiologic parameters. Thus, average values are often selected. These have usually been determined over many decades and are generally accepted as accurate for the average individual in a population. Such values were species- and strain-specific but did not, at least early on, indicate gender differences. Furthermore, values are generally for adults in the species and for humans they were generally for adult, healthy males. Typically, diurnal variation and other withinindividual variation were not considered. All of these limitations could be—and, in
602
CHAPTER 23 COMPUTATIONAL TOXICOLOGY IN CANCER RISK ASSESSMENT
time, were—addressed with more advanced modeling approaches. There are many papers and articles showing how this variability is handled (Cox 1996; Clewell et al. 1994, 1999, 2004). There are also several databases published with physiologic parameters for various species, including the human (US EPA 2007; Santolucito and Blancato 1991; International Life Sciences Institute 1994).
23.3.3.
Pharmacodynamic and Dose–Response Modeling
Pharmacodynamics can be considered to be the description as to how a substance impacts a living system. Dose–response modeling is a variant on that concept that quantifies the relationship between dose or exposure and the phenotypical, often clinical, response observed. Under a broad construct, dose–response models can be empirical descriptions of the relationship. These use any of several different classic mathematical functions to describe the data. Alternately, they may be pharmacodynamic models that are mechanistically based descriptions accounting for the mechanisms that lead to the toxicity (Zhang et al. 2007). The pharmacodynamically based dose–response models are obviously desired because of their advantages. Because of their description of mechanisms, more rational interspecies and dose-to-dose extrapolations can be made. Furthermore, when exposure is considered for the at-risk populations the impact of actual exposure conditions, including patterns and timing, can be better considered in the risk assessment. In actuality, many dose–response functions resemble empirical approaches but include some biologic mechanistic information. The best examples in cancer assessment include the famous work of Moogolvkar from the 1980s (Moolgavkar et al. 1980; Moolgavkar and Knudson 1981; Moolgavkar and Luebeck 1990).
23.4. COMPUTATIONAL TOXICOLOGY AND FUTURE RISK ASSESSMENTS It is clear that pharmacokinetically and pharmacodynamically based risk assessments have several clear advantages over those based on classical empirical functions. Obviously, risk results from far more than exposure to a chemical. Different biochemical reactions, physiologic processes, and individual and species susceptibility all contribute to the exact outcome. Understanding and then accounting for these mechanisms identifies the uncertainty and reduces some of that uncertainty that often accompanies risk assessments. This is particularly true for interspecies and dose-todose extrapolation and for accounting for actual exposure conditions and variability. Using such models, however, results in numerous parameters that are not easy or possible to measure. Many have held that such models merely increase uncertainty and make little or no improvement to risk assessment (Crump et al. 2007). Here it is maintained that identifying the uncertainties and quantifying some of the biologic process is a distinct improvement over classical methods. Let us consider the analogy of walking through a dark room. If no source of lighting is available, one has to walk through the room hoping not to be stopped by numerous obstacles that
23.4. COMPUTATIONAL TOXICOLOGY AND FUTURE RISK ASSESSMENTS
603
may be present. The walker may successfully negotiate through the room with no problems or catastrophes to the structures in the room. However, it cannot be predicted if the walker will be successful, and in fact no predetermined path can be determined without any knowledge of possible obstacles. If the walker instead has even dim light, some of the obstacles could be seen and some better path could be traced and followed. Better judgments are made as result of the light, even a dim one. Thus, is it not better to have some dim light rather than none at all? As better light sources are developed and used, more and more of the obstacles in the room are identified and can be accounted for and even avoided. But if instead we choose not to develop better light sources and turn them on, we will never be sure whether we can successfully traverse the room and we shall as well always be uncertain as to what damage may result to structures within the room. However, as noted, the number of assessments using these approaches have been few. What exactly has limited their use? The necessary mathematical descriptions have been rather easily formulated. But as discussed, to use these quantitative models, several pieces of information are needed. Many biological and physiological parameters are generic with respect to the chemical of interest. Biochemical information related to metabolism and mechanism of action are more chemical specific, although commonalities often exist across various chemicals. Information on metabolic and mechanistic pathways has been lacking and resource-intensive to measure. Very costly and time-consuming in vivo studies on an almost chemical-by-chemical basis have been the typical approach. Also, extrapolating human toxicity from animal studies has been problematic at best.
23.4.1.
21st-Century Toxicology
Over the last one or two decades, there have been significant advances in new and exciting fields such as genomics, proteomics, and systems biology which have resulted in new tools. These have the have the potential of generating a tremendous amount of relevant toxicity and metabolic data. This large amount of data can then result in more rapid development of models that can be used to test a myriad of relevant exposure conditions and, by identifying key pathways, could significantly reduce uncertainty. In 2007 the NRC published a report of its review and recommendation on how to move toxicity testing beyond the long used classical methods that have served us for years but are fraught with limitations already discussed (Committee on Toxicity Testing and Assessment of Environmental Agents: National Research Council 2007). The NRC states that new testing methods should be used that focus on toxicity pathways. Such testing relies on the rapidly evolving information from these new and advanced techniques. This will lead to the better understanding on how genes, proteins, receptors, regulatory networks, and other molecules help maintain the normal and complex biological functions of living organisms. This understanding is then combined with exposure and dose information to better elucidate how environmental stressors perturb the biologic pathways and lead to frank disease. Furthermore, these methods offer the promise to help us identify and characterize vulnerabilities based on inherent molecular and genetic properties, not just preconceived notions on age, gender, and ethnicity.
604
CHAPTER 23 COMPUTATIONAL TOXICOLOGY IN CANCER RISK ASSESSMENT
23.4.1.1. High-Throughput Testing. High-throughput screening (HTS) assays have come about as a result of the revolution in “omic” technologies and their automation (see Chapter 22). These allow for rapid in vitro testing of numerous biological pathways. Tests can range from genomic arrays to testing on whole organisms in vitro such as the zebra fish. With automation a wide array of biologic targets can be tested for changes resulting from exposure to test substances. These approaches developed after the “genomic revolution” and have been utilized by the pharmaceutical industry to identify target candidates as therapeutic agents. This rapid screening process avoids expensive and long-term testing of candidates that ultimately show little promise of success as pharmacological agents. Only those testing “positive” and identifying therapeutic pathways in these tests are then tested further for development. Obviously, such tests need to be designed to avoid false positives; otherwise the value of the screening is diminished. Readily the question arises if such techniques can be used to screen for or identify potential toxicity? The pharmaceutical industry has less interest in this approach. Once candidates are identified as positive there are then further tested for therapeutic potential and eventually tested for toxicity, often in vivo. However, the HTS studies have greatly narrowed down the array of target chemicals. Regulatory requirements for drugs require extensive toxicity testing and clinical trials before the drug is approved. Again, it should be emphasized the goal of screening here is to narrow the field of candidates to those most likely to have a therapeutic effect. As discussed in the introduction, there are an innumerable number of environmental chemicals for which little or no toxicity information is available and for many others for which it is inadequate. HTS studies could be invaluable at three related areas. First, they could identify those chemicals with the greatest potential toxicity and hazard which would then be prioritized for further testing. Second, they may be able to identify pathways of toxicity. Third, they may be able to identify and quantify metabolic pathways. A variation on this approach is to expand the HTS studies to include what are called high-content assays (HCA). These are not necessarily positioned for screening but because of automated means and informatic-based analysis methods (discussed later) are well-positioned to identify toxicity pathways (Tencza and Sipe 2004; Vogt and Lazo 2007). Again, they are in vitro-based, but they target several known or suspected pathways simultaneously to determine activity in those pathways. The key to both HTS and HCA is that numerous and different tests can be performed across many chemicals simultaneously. So instead of extensive in vivo studies at high dose for one chemical to simply identify toxicity such as carcinogenicity, these studies can study numerous chemicals at various doses and dosing regimens in a very short time. In addition, HTS are already often designed to use human proteins as targets. Also, HCA can be designed to use human cell lines that might be more relevant to actual human exposure conditions. At the very least, cells from multiple species could be tested in short order. The array of tests performed can also include purely computational methods to identify, for example, potential for receptor interaction or metabolic pathway. Both structure–activity-based studies and quantum-based computational techniques are among the methods employed (see Chapter 20). The advent of relational
23.4. COMPUTATIONAL TOXICOLOGY AND FUTURE RISK ASSESSMENTS
605
databases that allow for relating chemical structure with toxicity has been a major advancement (Martin et al. 2009; Richard 2006; Richard et al. 2008). A current resource is the Distributed Structure-Searchable Toxicity (DSSTox) Database Network (US EPA 2008). Furthermore, in silico metabolic simulators are being developed to predict and quantify metabolism (Mekenyan et al. 2007). So then a combination of HTS, HCA, and in silico methods can help identify potentially toxic or carcinogenic potential, identify pathways leading to perturbations and possible disease, and identify and possibly quantify metabolic pathways. This potentially addresses many of the shortfalls of classic toxicity testing. The US EPA is pursing such a path with its ToxCast™ (Dix et al. 2007). Using data from HTS bioassays, ToxCast™ is building computational models to forecast the potential human toxicity of chemicals. The endpoints being studied include biochemical assays of protein function, cell-based transcriptional reporter assays, multicell interaction assays, transcriptomics on primary cell cultures, and developmental assays in zebra fish embryos. In the first phase, all the chemicals being tested have been tested in traditional toxicity tests. This allows for comparison and evaluation of the rapid assay approaches used. Subsequent phases will expand the universe of chemicals to include more toxicity pathways and more testing. Eventually and after evaluation, the system will be used to test chemicals without the benefit of extensive classic toxicity information. Such approaches, if available at the time, could have helped provide key information for the cancer risk assessments identified earlier, DCM for example. In the case of DCM, there was considerable uncertainty regarding which metabolic pathway was responsible for the tumors. HTS and HCA approaches may have identified or provided further pathway evidence that might have resulted in a shorter timeframe to reach a decision and increased the confidence in the decision. Even at the time of the risk assessment, it was unclear whether pancreatic or other organs were vulnerable to cancer resulting from DCM exposure. Epidemiologic studies were incomplete at the time. Results from these studies were followed for several years before a decision on the pancreatic site was reached. Possibly, HCA and HTS testing could have helped resolve the issue sooner. More clear, however, is that HCA and HTS testing would have also helped reduce uncertainty with interspecies interpretation. There was clear evidence of tumors in experimental rodents. Using HCA and HTS testing in both rodent and human cells lines, for example, may have helped assess the relevance of those rodent tumors to human risk. While extremely promising, this approach—especially for identifying and quantifying toxic and metabolic pathways—is not yet proven and does raise some questions for application. Its use in pharmaceutical screening is more targeted. Often the desired target system is known and selected for the testing, resulting in a limited field of targets needing evaluation. For toxicity testing, there are any of a large number of toxic endpoints needing testing. False negatives are of serious concern, as would be excessive false positives. For the former, potentially toxic chemicals could be misclassified as “safe” or of little concern. Too many false positives would result in a diminished value of this approach. Furthermore, how to interpret in vitro and in silico testing relative to in vivo and actual exposure conditions is a key issue that has plagued such testing methods in the past. However, given the power of HTS
606
CHAPTER 23 COMPUTATIONAL TOXICOLOGY IN CANCER RISK ASSESSMENT
and HCA assays, it is extremely important to solve at least part of this conundrum. Computational models to make the connection between exposures, in vitro conditions, and effect will have to be developed. 23.4.1.2. Analysis, Interpretation, and Informatics. “Informatics develops new uses for information technology … to solve specific problems in areas as diverse as biology, …” (UNLV 2008). Bioinformatics can be thought of bringing together advanced information technology and management with biologic information and advanced computational techniques. The approach involves the formation of relational databases to house complex and varied data on biologic data. Some have proposed that the tremendous amount of biologic data such as generated through the new molecular techniques correlations are sufficient and can replace hypothesis testing. “We can throw the numbers into the biggest computing clusters the world has ever seen and let statistical algorithms find patterns where science cannot” (Anderson 2008). In effect, this approach is actually advanced correlation testing. With enough data and analysis, the line between correlation and causality is blurred. It is not apparent if such “nonhypotheses” approaches will suffice, at least in the near term, for quantitative cancer risk assessment or risk assessment in general. We should realize however, that such approaches allow for rapid and advanced pattern searching that can help form new and more rational hypotheses. As with the dim light in a dark room, we must accept here the widely held premise that more knowledge and information on mechanisms leading to carcinogenicity and knowledge of biotransformation pathways are crucial for accurate assessments whose uncertainty is better characterized and reduced. It clearly follows then that informatics, especially using the extensive data from HTS and HCA, while not the sole solution, can only help us more quickly obtain such information and reduce uncertainties. Programs such as the U.S. EPA’s ToxCast™ and the associated development of ACToR database are the most extensive endeavor in this regard (Judson et al. 2008; Richard et al. 2008). For the first time, hundreds of chemicals and thousands upon thousands of data can be organized, examined, and analyzed in a structured manner. Combined with machine learning techniques that can take advantage of the immense text data on the internet from many scientific and publicly available databases, we can now postulate and test pathways of toxicity and biotransformation within minutes or hours rather than weeks and years. Significant gains were made in the 20th century with the advent and development of advanced analytical techniques in chemistry and the electron microscope in biology and medicine. Examples are far too numerous to mention here, but suffice it to say that they produced quantum leaps in our ability to measure and observe changes in living systems. With these advances, we have begun to see the “molecules of life.” The early part of this century now brings us the technology to study these data and information to determine how those molecules of life operate and maintain living systems. 23.4.1.3. Systems Biology. This then brings us back to modeling and to the next generation of modeling. As discussed, modeling in risk assessment has been of
23.4. COMPUTATIONAL TOXICOLOGY AND FUTURE RISK ASSESSMENTS
607
two genres. Either data were analyzed by statistical means looking for correlations or hypotheses on mechanisms of action were mathematically described. Such models describe how one chemical, or a few chemicals at best, can impact the living system, in this case leading to carcinogenicity. Few would deny that actual living systems are far more complicated and have extensive networks of redundancy to sustain the steady-states necessary for and maintaining a healthy existence. When those steadystate conditions are altered by disease, we often use pharmacologic means to return them back to those necessary for healthy conditions. Chemical risk assessment then can be thought of as the identification, characterization, and even quantification of how chemicals cause an alteration of the normal steady-state conditions. The next generation of models will hopefully take us to this next level. We must better understand the operation of complex networks of processes within living systems. Modeling a single pathway or a few biological pathways for a single chemical or a few chemicals will not suffice. The promise though is that we now have data and information and computational means to analyze and describe those data. Furthermore, we can now organize them into models that are beginning to characterize the complexities of the living system. We must think differently from our traditional ways. It would perhaps be an insurmountable task to use only sequential observational studies with expert knowledge to hope to understand such complexities. While knowledge and expertise must never be abandoned and cannot be replaced, we must use new technologies and paradigms to augment and expand that expert knowledge. Given the amount and complexity of available data, we need to look to more advance modeling of systems biology. Spivey (2004) states “… systems biology attempts to harness the power of mathematics, engineering, and computer science to analyze and integrate data from all the ‘omics’ and ultimately create working models of entire biological systems.” “Omics” here can be expanded to include HTS, HCA, and the extensive network of data and knowledge of biological systems. Thus systems biology and modeling uses modern computational techniques and power to harness that knowledge and data for understanding how exposure to chemicals might or might not lead to perturbations and frank clinical disease. Systems modeling can be simply described as “the application of mathematical models and reasoning to the understanding of biological systems and the explanation of biological phenomena” (US EPA 2006). While the term systems biology may be new, the concept of understanding and describing biological networks and interactions is not. Harvey first described the circulation and cardiac activity in the 1600s (Encyclopaedia Britannica Online 2008). Even before this, going back hundreds of years to Galen in the classical era, circulation was the object of interest and description. A more “modern” example would be the Frank–Starling Law of the Heart first published in the early 1900s (Klabunde 2007). Concentrating again on circulation within the living mammalian system, this description is far more extensive and is an accurate description of what we know even today. This elegant presentation describes the rigorous relationship between cardiac output and venous return which is necessary for a properly functioning circulatory system. Beyond that it describes the interrelationships between the various organs and systems that work to maintain that rigorous relationship.
608
CHAPTER 23 COMPUTATIONAL TOXICOLOGY IN CANCER RISK ASSESSMENT
Let us here define or describe a system as an organized entity that survives the demands of a changing environment by being adaptable and at the same time maintaining a status quo—or maintain life. The system is defined by the level of organization being described. It could be the whole organism, a system within the organism, or a system within a tissue or even within a cell. The key is that the components of the system are interrelated or networked: Each component is influenced by other components, and each component influences other components; these influences are both direct and indirect. So the system then represents a network of interconnected and interlaced regulatory and communication pathways. These pathways are interconnected to regulate and maintain the normal function of the system. Systems models are developed at various levels of complexity and at various scales. They can be a serially connected group of submodels, often mechanistic in nature, that describe fate and transport in the environment, exposure by the organisms of interest, pharmacokinetics, and the pharmacodynamics. To this we can add the knowledge-based descriptions of the HTS and HCA data as well as the knowledge gained from the machine learning scouring of the available literature. Alternately, they can be described in a parallel way at the various scales. Large-scale datasets are modeled statistically to establish an understanding of biological or clinical outcomes including the impact of exposure, genes, regulatory proteins, and protective mechanisms. Mathematical descriptions can be combined with clinical and histopathologic descriptions as well. Such mathematical approaches have recently been used, for example, with the visualization from colonoscopy to determine the optimal screening times and to simulate the progression of colorectal cancer (Jeon et al. 2008). These approaches expand the empirical modeling to which we are accustomed. Because of the large amount of data and the pattern recognition capabilities, far more can be accomplished with such modeling than was ever possible with the empirical curve-fitting routines of the past. As mentioned in the section on informatics, pathway analysis can also be accomplished in using these advanced techniques. This can then help explain the mechanistic basis of perturbed pathways and thus augment the statistically based modeling. It can also serve as the basis for formulating the mechanistically based pharmacodynamic models. However, these models must, like the aforementioned Frank–Starling Law, take into account all interrelated and interacting regulatory pathways. HCA and HTS can identify those regulatory relationships within the system. Progress has been made at using such mechanistic models in quantitative assessments of carcinogenicity (Little et al. 2008). As described, PBPK models are systems of mass balance equations describing the mass or concentration of chemicals at various loci within the tissues. Most often the loci have been at the level of the whole organ. However, the mass can be described at as fine a resolution as is necessary and as knowledge allows. For example, a PBPK model can describe the concentration at the cellular or subcellular level (Blancato and Bischoff 1993). Ultimately, the chemical input into pharmacodynamic and systems level models could easily be the output of PBPK models at whatever the necessary resolution. For example, if a perturbation in a specific cell is directly related to a concentration or mass of chemical at that cell, the PBPK model would be expanded to describe that mass or concentration. This though raises
23.5. CONCLUSION
609
some interesting and challenging issues. PBPK models describe mass transfer. By nature, such descriptions are continuous functions often requiring considerable parameters and knowledge to estimate those parameters. The greater the resolution of a PBPK model, usually the more resource-intensive and difficult it is to make such estimates. This may where other computational and in vitro models will be needed to accurately and reliably predict those parameters. Developing PBPK models by a combination of in vitro and in silico experimental methods will be key to creating “high-throughput PBPK models.” This will help alleviate the problem of differing scales between system models and practical PBPK models. By such means it will be easier to develop high-resolution PBPK models. In addition, measured data can be used with statistical means to estimate such parameters. Such datasets then are used in Bayesian approaches to continuously improve the reliability of the estimates. As such, measured data serve to both evaluate model output and estimate the key parameters that allow a model to function. Ultimately, this cadre of models at various scales from molecular to tissue levels is the basis of the U.S. EPA’s virtual tissue models now under way (Shah 2008; Knudsen 2008). Using data from the ToxCast™, the statistical inference models are being developed. As more of these data are obtained, the model can be expanded to include mechanistic knowledge of key pathways and biotransformation. Recall that ToxCast™ data come from HCA, HTS, and in silico computational methods so that large amounts of data can be quickly generated. An iterative approach is very possible and utilized. In addition, with knowledge gained from structured and rational mining of toxicology (largely in vivo) information already published, a complex and informative library of knowledge is built. This library can be expanded and reanalyzed as the need arises. These virtual tissue models are designed to first describe relevant biology of the organ through both advanced empirical methods and mechanistically based models. Then they are exercised to study and include knowledge of perturbations caused by exposure to environmental insults. Another advantage of such modeling systems is that they can be parsimonious or as complex as the risk assessment need requires and the data and knowledge allow. The parsimony is helpful in avoiding overparameterized models.
23.5.
CONCLUSION
Quantitative risk assessment and cancer risk assessment in particular have long relied on modeling to describe and predict perturbations that then lead to frank appearance of cancer in a population. Society on the whole and regulatory bodies must be able to predict which exposure conditions might lead to catastrophic results. Although incidences of disease have occurred and no doubt will continue to occur, the desire is to avoid such incidences. A fundamental goal of risk assessment is prediction. Thus, the goal of regulation is to avoid undesirable incidences. Risk assessment informs the regulators on the magnitude and impact of potential incidences. The risk manager then takes action to avoid the actual fruition of those predictions. Toxicology studies, data analysis, and predictive modeling have been and will continue to be the cornerstone for such risk assessments. Assessments based
610
CHAPTER 23 COMPUTATIONAL TOXICOLOGY IN CANCER RISK ASSESSMENT
on all the available data analyzed and described in a rational fashion accounting for as much mechanistic biology as possible should always be the desired goal. Such knowledge-based assessments avoid over- and underestimates of risk. While underestimation of risk is always a most undesirable outcome, it must also be remembered that, given societal and resource pressures and limitations, overestimation of risk is not necessarily protective of public health. Overestimatation of risk can lead to substitutions that are, in fact, more dangerous and can rob valuable resources from studying and acting upon other, more real dangers. These new approaches outlined here give great promise for achieving the goals of more accurate risk predictions. The fact that great challenges continue cannot and should not be obscured. For example, the use of in vitro and in silico data is not without shortcomings. But with the advent of these new techniques for assaying, computation, and data analysis, none of these types of data have to be used to the exclusion of the others. Ultimately, the only hope of reducing the cost and time required by traditional toxicology is to develop and apply combinations of HTS, HCA, in silico, in vitro, in vivo, and advanced modeling approaches to the problem. We clearly see how HTS and HCA studies combined with advanced informatics can reveal a great deal about modes of action and pathways to toxicity. Likewise, these approaches will gain an understanding of the biological basis for susceptibility. These data will also “feed” the models that will describe and ultimately predict phenotypical changes and outcomes in organisms. While true for all toxicological concerns, this is especially so for cancer because traditional testing by bioassay often requires years of costly experiments. Clearly, this new approach gives the opportunity to shed more light into the dark room.
REFERENCES Andersen, M. E. (1981). Saturable metabolism and its relationship to toxicity. Crit Rev Toxicol 9, 105–150. Andersen, M. E., Clewell, H. J., 3rd, Gargas, M. L., Smith, F. A., and Reitz, R. H. (1987). Physiologically based pharmacokinetics and the risk assessment process for methylene chloride. Toxicol Appl Pharmacol 87, 185–205. Anderson, C. (2008). The end of theory: The data deluge makes the scientific method obsolete. Wired 16. Angelo, M. J., Bischoff, K. B., Pritchard, A. B., and Presser, M. A. (1984). A physiological model for the pharmacokinetics of methylene chloride in B6C3F1 mice following i.v. administrations. J Pharmacokinet Biopharm 12, 413–436. Angelo, M. J., and Pritchard, A. B. (1984). Simulations of methylene chloride pharmacokinetics using a physiologically based model. Regul Toxicol Pharmacol 4, 329–339. Bernareggi, A. M., and Rowland (1991). Physiologic modeling of cyclosporin kinetics in rat and man. J Pharmacokinet Biopharm 19, 21–50. Bischoff, K. B., Dedrick, R. L., and Zarharko, D. D. (1970). Preliminary model for methotrexate pharmacokinetics. J Pharm Sci 59, 149–154. Blancato, J. N., and Bischoff, K. B. (1993). The application of pharmacokinetic models to predict target dose. In Health Risk Assessment: Dermal Inhalation Exposure and Absorption of Toxicants, Wang, R. G. M., Knnak, J., and Maibach, H., eds., CRC Press, Inc., Boca Raton, FL, pp. 31–46. Blancato, J. N., Evans, M. V., Power, F. W., and Caldwell, J. C. (2007). Development and use of PBPK modeling and the impact of metabolism on variability in dose metrics for the risk assessment of methyl tertiary butyl ether (MTBE). J Environ Protection Sci 1, 29–51.
REFERENCES
611
Blancato, J. N., Hopkins, J., and Rhomberg, L. R. (1987). Update to the health assessment document and addendum for dichloromethane (methylene chloride): Pharmacokinetics, mechanism of action and epidemiology. United States Environmental Protection Agency, Washington, D.C. Clewell, H. J., 3rd, Lee, T. S., and Carpenter, R. L. (1994). Sensitivity of physiologically based pharmacokinetic models to variation in model parameters: Methylene chloride. Risk Anal 14, 521–531. Clewell, H. J., Gearhart, J. M., Gentry, P. R., Covington, T. R., VanLandingham, C. B., Crump, K. S., and Shipp, A. M. (1999). Evaluation of the uncertainty in an oral reference dose for methylmercury due to interindividual variability in pharmacokinetics. Risk Anal 19, 547–558. Clewell, H. J., Gentry, P. R., Covington, T. R., Sarangapani, R., and Teeguarden, J. G. (2004). Evaluation of the potential impact of age- and gender-specific pharmacokinetic differences on tissue dosimetry. Toxicol Sci 79, 381–393. Committee on the Institutional Means for Assessment of Risks to Public Health: Commission on Life Sciences: National Research Council (1983). Risk Assessment in the Federal Government: Managing the Process, National Academies Press, Washington, D.C. Committee on Toxicity Testing and Assessment of Environmental Agents: National Research Council (2007). Toxicity Testing in the 21st Century: A Vision and A Strategy, National Academies Press, Washington D.C. Cox, L. A., Jr. (1996). Reassessing benzene risks using internal doses and Monte-Carlo uncertainty analysis. Environ Health Perspect 104(Suppl 6), 1413–1429. Crump, K., Subrmaniam, R., and Chen, C. (2007). Uncertainty inherent in biologically-based models. In Resources for the Future, October 22–23, 2007, Washington, D.C. Dix, D. J., Houck, K. A., Martin, M., Richard, A. M., Setzer, R. W., and Kavlock, R. J. (2007). The Toxcast program for prioritizing toxicity testing of environmental chemicals. Toxicol Sci 95, 5–12. Encyclopaedia Britannica Online (2008). An Anatomical Disquisition on the Motion of the Heart and Blood in Animals. Gerlowski, L. K., and Jain, R. (1983). Physiologically based pharmacokinetic modeling: Principles and applications. J Pharm Sci 72, 1103–1127. Gibaldi, M., and Perrier, D. (1975). Pharmacokinetics, Marcel Dekker, New York. International Life Sciences Institute (1994). Physiological Parameter Values for PBPK Models, Risk Science Institute, Washington, D.C. Jain, R. K., Gerlowski, L. K., Weisbrod, J. M., Wang, J., and Perrson, R. N. (1981). Kinetics of uptake, distribution, and excretion of zinc in rats. Ann Biomed Eng 9, 347–361. Jeon, J., Meza, R., Moolgavkar, S. H., and Luebeck, E. G. (2008). Evaluation of screening strategies for pre-malignant lesions using a biomathematical approach. Math Biosci 213, 56–70. Judson, R. F., Elloumi, F., Setzer, R. W., Li, Z., and Shah, I. A. (2008). A comparison of machine learning algorithms for chemical toxicity classification using a simulated multi-scale data model. BMC Bioinform 9, 241–256. Kavlock R. J. (2007). Computational toxicology. In AltTox.org. Kavlock R. J., Ankley, G., Blancato, J., Breen, M., Conolly, R., Dix, D., Houck, K., Hubal, E., Judson, R., Rabinowitz, J., Richard, A., Setzer, R. W., Shah, I., Villeneuve, D., and Weber, A. E. (2007). Computational toxicology—A state of the science mini review. Toxicol Sci http://toxsci.oxfordjournals. org/cgi/content/abstract/kfm297v1?ck=nck. King, F. G., and Dedrick, R. L. (1981). Physiologic model for the pharmacokinetics of 2’deoxycoformycin in normal and leukemic mice. J Pharm Biopharm 9, 519–534. Klabunde, R. E. (2007). Cardiovascular physiology concepts: Frank–Starling mechanism. In CV Pharmacology.Com. Knudsen, T. B. (2008). The Virtual Embryo Project. In Environmental Bioinformatics & Computational Toxicology Center Symposium, Piscataway, NJ. Licata, A. C., Dekant, W., Smith, C. E., and Borghoff, S. J. (2001). A physiologically based pharmacokinetic model for methyl tert-butyl ether in humans: implementing sensitivity and variability analyses. Toxicol Sci 62, 191–204. Little, M. P., Heidenreich, W. F., Moolgavkar, S. H., Schollnberger, H., and Thomas, D. C. (2008). Systems biological and mechanistic modelling of radiation-induced cancer. Radiat Environ Biophys 47, 39–47.
612
CHAPTER 23 COMPUTATIONAL TOXICOLOGY IN CANCER RISK ASSESSMENT
Martin, M. T., Judson, R. S., Reif, D. M., Kavlock, R. J., and Dix, D. J. (2009). Profiling chemicals based on chronic toxicity results from the U.S. EPA ToxRef Database. Environ Health Perspect 117, 392–399. Mekenyan, O., Jones, W. J., Kolanczyk, R. C., and Schmieder, P. K. (2007). Design and performance of a xenobiotic metabolism database manager for metabolic simulator enhancement and chemical risk analysis. In SETAC Europe 17th Annual Meeting, Porto, Portugal. Moolgavkar, S. H., Day, N. E., and Stevens, R. G. (1980). Two-stage model for carcinogenesis: Epidemiology of breast cancer in females. J Natl Cancer Inst 65, 559–569. Moolgavkar, S. H., and Knudson, A. G., Jr. (1981). Mutation and cancer: A model for human carcinogenesis. J Natl Cancer Inst 66, 1037–1052. Moolgavkar, S. H., and Luebeck, G. (1990). Two-event model for carcinogenesis: Biological, mathematical, and statistical considerations. Risk Anal 10, 323–341. Prah, J., Ashley, D., Blount, B., Case, M., Leavens, T., Pleil, J., and Cardinali, F. (2004). Dermal, oral, and inhalation pharmacokinetics of methyl tertiary butyl ether (MTBE) in human volunteers. Toxicol Sci 77, 195–205. Ramsey, J. C., and Andersen, M. E. (1984). A physiologically based description of the inhalation pharmacokinetics of styrene in rats and humans. Toxicol Appl Pharmacol 73, 159–175. Rao, H. V., and Ginsberg, G. L. (1997). A physiologically-based pharmacokinetic model assessment of methyl t-butyl ether in groundwater for a bathing and showering determination. Risk Anal 17, 583–598. Richard, A. M. (2006). Future of toxicology—Predictive toxicology: An expanded view of “chemical toxicity”. Chem Res Toxicol 19, 1257–1262. Richard, A. M., Yang, C., and Judson, R. (2008). Toxicity data informatics: Supporting a new paradigm for toxicity prediction. Toxicol Mechanisms Methods 8, 103–118. Santolucito, J., and Blancato, J. (1991). Physiological Parameters for Pharmacokinetic Modeling, US EPA, Las Vegas, NV. Shah, I. A. (2008). Representing chemical-induced liver Injury for multiscale tissue modeling. In Conference of Semantics in Healthcare and Life Sciences, Boston, MA. Spivey, A. (2004). Systems biology: the big picture. Environ Health Perspect 112, A938–A943. Tencza, S. B., and Sipe, M. A. (2004). Detection and classification of threat agents via high-content assays of mammalian cells. J Appl Toxicol 24, 371–377. UNLV (2008). University of Nevada—Las Vegas, Las Vegas. US EPA (2006). Computational Toxicology, Basic Information. US EPA (2007). Database of Physiological Parameters for Early Life Rats and Mice. US EPA (2008). Distributed Structure-Searchable Toxicity (DSSTox) Database Network, Vol. 2008. Vogt, A., and Lazo, J. S. (2007). Implementation of high-content assay for inhibitors of mitogen-activated protein kinase phosphatases. Methods 42, 268–277. Zhang, X., Tsang, A. M., Okino, M. S., Power, F. W., Knaak, J. B., Harrison, L. S., and Dary, C. C. (2007). A physiologically based pharmacokinetic/pharmacodynamic model for carbofuran in Sprague– Dawley rats using the exposure-related dose estimating model. Toxicol Sci 100, 345–359.
PART
VI
GENERAL APPROACHES FOR QUANTIFYING CANCER RISKS
CH A P TE R
24
LINEAR LOW-DOSE EXTRAPOLATIONS Michael Dourson Lynne Haber
24.1.
INTRODUCTION
As discussed in previous chapters, cancer assessment begins by evaluating the weight of evidence (WOE) regarding a chemical’s carcinogenicity, based on all of the relevant data, including human and animal carcinogenicity data, information on toxicokinetics and toxicodynamics, genotoxicity, mechanistic data, genomics, and structure–activity relationships. Based on all of this information, a determination is made regarding (a) a chemical’s carcinogenic potential and (b) its mode of action (MOA) via a key event(s). Information on both the carcinogenic potential and MOA is used to inform the appropriate approach for dose–response assessment, which is typically based on experimental animal or human studies of cancer incidence as a function of dose rate. In brief, several MOAs are possible for tumor development, all of which ultimately lead to loss of cell growth control. These MOAs include, for example, mutation of DNA leading to loss of cell growth control; oxidative stress leading to mutations or chromosomal disruption, leading to loss of cell growth control; hormonal excess leading to inappropriate cellular growth; and cytotoxicity leading to cell regeneration and the replication of cells that have undergone mutations from endogenous exposures, leading to loss of cell growth control. Several MOAs may be simultaneously occurring, or they may be occurring at different parts of the dose– response curve. Despite the variety of these, and other, differing MOAs, however, dose–response assessment options have been limited to just a few approaches: • Linear extrapolation of the cancer incidence with dose rate without a threshold, leading to the determination of a risk per unit dose, and a corresponding dose corresponding to a specified risk level [risk-specific dose (RSD), e.g., the EPA’s N-nitrosodimethylamine, (EPA 2009a)]. This approach is used by the EPA for chemicals that are known to act via a mutagenic MOA, and for chemicals for which the MOA is not known, as a health-protective default.
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
615
616
CHAPTER 24 LINEAR LOW-DOSE EXTRAPOLATIONS
• Extrapolation to a practical threshold or subthreshold dose, such as the determination of a reference dose [RfD, e.g., EPA’s perchlorate (EPA 2009b)] or tolerable daily intake [TDI, e.g., The Netherlands’s National Institute for Public Health and the Environment or RIVM’s trichloroethylene (RIVM 2009)], or development of health-based occupational exposure limits by many groups [e.g., American Conference of Governmental Industrial Hygienists (ACGIH 2006)]. • Use of a biologically based dose–response (BBDR) model that can capture multiple MOAs, which may dominate in different portions of the dose– response curve [e.g., the formaldehyde assessment by The Hamner Institutes for Health Sciences, formerly Chemical Institute of Industrial Toxicology (CIIT 2009)], which has been used in regulatory settings (Conolly et al. 2004). • Both a linear extrapolation in the low part of the dose–response curve and a steeper curve reflecting a different MOA in the higher part of the dose– response curve—TERA’s assessment for thyroid tumors resulting from acrylamide exposure (Dourson et al. 2008). • A hybrid approach where a point of departure is identified in the range of the data or near the data. A risk management determination is made based on this estimate, the cancer classification, and estimated exposures, such as for Health Canada’s 1,3-butadiene assessment (Health Canada 2009). The purpose of this chapter is to present the rationale for linear low-dose extrapolation and to critically consider the rationale in light of recent research. We begin by presenting the historic basis for use of linear extrapolation in the regulatory context. This approach has been based on the combination of presumed health protectiveness in the absence of adequate extrapolation, considerations based on first principles regarding the shape of dose–response curves, and consideration of the mathematical implications of background exposure and population variability. We then discuss research on the shape of the dose–response curve for early events in the chemical carcinogenic process, and we consider the implications of these data for the shape of the dose–response curve. Finally, we discuss how recent considerations for evaluating mutagenicity as an MOA can help in determining the MOA, and we also discuss the approach for low-dose extrapolation. In light of the biology presented in this chapter, and elsewhere in this book, the general use of linear extrapolation provides reasonable upper-bound risk estimates that are very likely to be health-protective (or even very conservative) and also provide consistency across dose–response assessments among disparate chemicals.
24.2.
HISTORICAL
Cancer is a self-replicating disease, resembling pathogen infections in this aspect. Unlike many such infections, however, cancer often does not evoke the full-range immune response that is seen with infectious agents, in part because the tumor cell is inherently “self” and in part because the tumor cell is mammalian and thus better able to respond to the unique milieu of the host. Because of these reasons and the
24.2. HISTORICAL
Extrapolation Range
617
Observed Range
10%
ecte Proj
(Ce Lim ntr al E stim ate )
r
nea
d Li
0%
Point of Departure
RfD
(C on fid en ce
Response
) ose nD o it
UFs
LED10 ED10 Dose
Figure 24.1. Estimated hypothetical cancer risks from chemical exposures by way of a linear dose response curve. [Adapted from 3-1 of EPA (1996).]
one-hit hypothesis described below, a major focus in the cancer area is on prevention, and a key aspect of preventative measures is the minimization of exposures to chemicals known to cause cancer either in humans or in experimental animals. Such preventive measures lead to several public health strategies, such as restricting the legal smoking of cigarettes to only adults, admonishing the populace to reduce dietary intake of fats, and regulations or voluntary measures to limit exposure to known or suspected carcinogens, whether they be (a) naturally occurring, such as aflatoxin, (b) anthropogenic, such as cigarette smoke, or (c) both, such as acrylamide, arsenic, and dioxin. These latter limits typically depend on methods that estimate upper-bound hypothetical cancer risks from chemical exposures by way of a linear dose–response curve, as shown in Figure 24.1, although other methods are also used, as noted above and in Figure 24.1. Low-dose estimation of chemical cancer risks has developed because outright bans on all chemicals that cause cancer are problematic. This is due to the social utility of some carcinogenic chemicals (e.g., found in gasoline, plastics) and because increasingly more sensitive chemical detection methods are able to find chemicals in virtually all environmental media and organisms. Thus, for example, perchlorate, a known carcinogen in experimental animals, can be found in everyone, because it is naturally occurring and is used as a drug and also because certain waters and soils are contaminated with industrial and military wastes. This leads to either direct exposures or to indirect exposures through foods grown with these contaminated waters and soils. So if outright bans are problematic, what is it about the linear dose–response curve that makes it a preferred approach for cancer prevention? First, a theoretical basis for a linear response in the low-dose region exists for radiation carcinogenesis. This theoretical basis has some biological support, but this
618
CHAPTER 24 LINEAR LOW-DOSE EXTRAPOLATIONS
1000
800
700
600
500
400 150 INCIDENCE
Incidence of Leukemia / 106 Population / Year
900
300
200
100
THRESHOLD HYPOTHESIS
50 0
100
LINEAR HYPOTHESIS
20
40
60
80
100
See Inset Note: Dose is not corrected for attenuation by shielding
30 1200 18001600 1400 Distance From Hypocenter In Meters 200
400
600
800
1000
1000
Estimated Dose of Radiation (RAD) Figure 24.2. Rate of leukemia deaths in survivors of Hiroshima and Nagasaki. From Heyssel et al. (1960), with permission.
support is open to a number of interpretations, for example, as shown in Figure 24.2. In brief, this basis is supported by the fact that (1) several forms of radiation can cross biological membranes without being affected by the toxicokinetic variability among humans that are evident for perhaps all chemicals and (2) radiation can damage parts of DNA responsible for development of cancer without regard to toxicodynamic differences among humans, which is also evident for perhaps all chemicals. Humans may respond differently after this damage has occurred, most likely by differences in toxicodynamics (e.g., DNA repair capabilities), but this initial damage is thought to be without individual responder influences.
24.2. HISTORICAL
619
These two facts related to effects of radiation lead directly to the stochastic assumption. This assumption states that the initial damage of critical DNA by radiation is a matter of chance, that this chance is a matter of dose, and that the two are directly related, such that doubling the radiation dose will result in a directly proportional increase in the chance of critical DNA damage. Furthermore, the individual’s innate capabilities to either repair this critical DNA damage or otherwise mitigate the progression of cancer, for example, by way of an enhanced immune response, does not depend on the initial chance of radiation damage. In other words, the relationship between initial damage (and thus dose) and cancer risk for an individual does not vary with increasing dose; the risk per unit dose is constant at low doses. A 1977 National Academy of Sciences report (NAS 1977) was highly influential in the decision by regulatory agencies such as the EPA to adopt a default of linear low-dose (or no threshold) extrapolation for carcinogens. This report presented three key lines of reasoning: • If a single change to the DNA of a cell can transform it, and a transformed cell can be irreversibly propagated leading to cancer, then there can be no threshold, although the risk may be low. This is often referred to as the “onehit” model. • Human population heterogeneity leads some people to be at greater risk, and it reduces the likely population-based threshold toward zero. • Exposure to the chemical of interest may add directly to background disease; and if the exposure acts via the same mechanism, it would lead to linearity in the low-dose region. While the key lines of reasoning can be simplistically presented as in the above bullets, it is important to recognize that the NAS (1977) report did consider additional levels of subtleties. For example, the potential for multistage carcinogenesis was recognized in that report, although the biological understanding of these stages has increased tremendously since then, with the identification of multiple cell-cycle control points and regulatory controls that must be overcome for cancer to result. The NAS (1977) report reviewed a number of multievent (multihit or multistage) models, and joined Crump et al. (1976) in concluding that most models become linear at low doses. However, the Crump paper noted an important caveat: “It can be shown that the response will be linear in d for small dose rate d unless all paths depending upon d contain at least 2 events that occur only in the presence of the specific primary carcinogen” (emphasis added). Since cancer is now understood to be a multievent process, this modifies the question to what the rate-limiting step is, and whether later stages are dependent on the presence of the carcinogen, as discussed further below. As another example of additional subtleties in the NAS (1977) report, the report addresses population heterogeneity by noting that such heterogeneity would result in multiple thresholds (for different people or populations) and that it would be difficult to establish a single threshold. The report also notes the difficulty in distinguishing statistically between models that hypothesize multiple thresholds and
620
CHAPTER 24 LINEAR LOW-DOSE EXTRAPOLATIONS
multievent models that hypothesize no thresholds. Thus, it appears that the report authors took a health-protective approach in this line of reasoning, based on methodological issues. Based on modern risk assessment thinking, the implication of population heterogeneity is a quantitative issue of characterizing the variability and extrapolating from the general population to sensitive populations, rather than a qualitative issue of the shape of the dose–response curve. Notwithstanding these arguments for linearity and ones from radiation exposure, chemical exposures are known to differ from radiation exposures in ways that allow individual variability in toxicokinetics and toxicodynamics to affect the outcome. This evidence for interstrain, interspecies, or interindividual differences is overwhelmingly found in experimental animal studies and in human epidemiology studies, suggesting that the real dose–response curve for chemical exposure may be a combination of the stochastic curve overlaid by the impact of several sources of variability, all of which may modify the shape of the dose–response curve, or may be due to just individual variability, or may be due to dose dependent transitions in response (Slikker et al. 2004a,b). This raises the question, then, of whether the stochastic assumption, one of the theoretical bases of the linear dose response extrapolation, should be applied to chemical dose response assessment. If so, should its use be mollified? As noted below, many of the toxicokinetic and toxicodynamic considerations affecting the shape of the curve are linear at very low doses. But, this linearity at doses close to zero in toxicokinetics or toxicodynamics does not mean that the entire shape of the dose–response curve for tumor effects is linear up to experimental animal doses. It is important to distinguish between factors leading to low-dose tumor response linearity and considerations of linearity or nonlinearity of the overall dose response curve. Low-dose linearity may result in very shallow dose–response curves followed by upward curvature at higher doses; this results in risk estimates that are very different from those obtained by linear extrapolation from experimental animal data on tumors. In addition, several scientists, such as Calabrese (2009), argue that several of the supporting lines of evidence espoused by the NAS (1977) are testable and that each of the testable lines of evidence has been generally discredited and/or seriously weakened. Doull (2003) agrees, arguing that earlier statements that thresholds for the cancer response could not be proved are no longer true. Presumably this conclusion is due to a fuller understanding of carcinogenic modes of action and hormesis, as well as an understanding of the potential for nonlinearities and saturation or induction of numerous processes, such as absorption, detoxification, and repair. As discussed by Slikker et al. (2004b), these processes lead to dose-dependent transitions in toxic response, which might be more the norm than not. As a result, several judgments for thresholds or dose-dependent transitions for chemicals that cause carcinogenic responses predominantly or exclusively via a nonmutagenic MOA are now available (e.g., EPA 2009b; RIVM 2009; Dourson et al. 2008). Moreover, several scientists unequivocally reject a linear low-dose extrapolation because it ignores the impact of dose-related differences in toxicokinetics and toxicodynamics, as mentioned above, as well as other biological principles as stated in other chapters in this book. For example, Rozman et al. (1996) argue that rather than use an arithmetic scale for dose, which compresses the dose scale in the
24.2. HISTORICAL
621
very low dose area of interest into an undecipherable blip, one should evaluate dose–response based on a logarithmic scale, which expands the low-dose area of interest, since the response can be viewed near the reference point of one molecule. Rozman et al. (1996) argue that the dose–response function exists on a logarithmic scale because its underlying principle is described by a Malthusian-type differential equation. Earlier publications attesting to the use of the lognormal distribution for biological phenomenon are also found (e.g., Gaddum 1945). Other scientists, for example Waddell (2003a,b), argue that thresholds are provable when the extant cancer dose–response information is plotted on a logarithmic dose curve in such a fashion. The analyses presented by Rozman et al. (1996) and Waddell (2003a), however, are themselves controversial and have stimulated considerable discussion in the toxicological risk assessment community (e.g., Haseman 2003; Waddell 2003b; Andersen et al. 2003). Arguments against the Rozman et al. (1996) and Waddell (2003a) approach include, but are not limited to, the inappropriateness of generic judgments based on logarithmic scale plotting, the inability of any study to know true background response, and the lack of power for most studies to adequately determine a threshold dose. Andersen et al. (2003) also argue that the focus instead should be on a case-by-case judgment about the shape of the dose–response curve based on the biology and/or mode of action of individual chemicals. All of us are likely to agree with this statement. Finally, the linear low-dose extrapolation obviates de facto the concept of a threshold in the dose–response curve. The threshold concept is universally accepted for most other, noncancer toxicity and for many other biological phenomena (e.g., taste thresholds), and many scientists would argue that such thresholds (or “practical thresholds”) should apply to the cancer or mutation response due to the underlying complexities in toxicokinetics and in the dynamics of DNA damage and repair, immune surveillance, and the variation in the progression of clinical diseases.* The second underlying basis for a linear response in the low-dose region for chemical carcinogenicity is the concept of background response. This concept is exemplified by Crump et al. (1976) and Lutz (1990), for direct-acting carcinogens.† Crump et al. (1976) states that “if carcinogenesis by an external agent acts additively with any already ongoing process, then under almost any model the response will be linear at low dose.” In other words, the acceleration of an endogenous process by an exogenous agent can result in a nonthreshold, and perhaps linear, *We note here that the argument as to whether a threshold for an adverse effect exists for any chemical is really an argument as to whether a single molecule may cause an adverse effect, since by definition of the word “threshold” (a dose at which effects begin to occur), all chemicals must have a threshold for adverse effect greater than zero dose. †As described by Crump et al. (1976): “We note again here that this analysis is appropriate only for those agents that affect cancer incidence through the alteration of single cells in an irreversible and hereditable manner (e.g., chronic exposure to low-level ionizing radiation). Those agents that increase cancer by anatomical and/or physiological alteration of whole tissues and organs (e.g., dietary modification of gut flora) may or may not be described by these models. Since we do not know the relative proportion of these 2 types of carcinogens and often do not know into which category a particular agent falls, we must stress the importance of understanding basic carcinogenic mechanisms.”
622
CHAPTER 24 LINEAR LOW-DOSE EXTRAPOLATIONS
dose–response relationship for tumors in the range of environmental exposures. Note that additivity to background in this situation refers to additivity to background response, not additivity to background exposure. In addition, the consideration of the site of chemical action is at the process level, not the effect level. Thus, this concept of additivity to a background process resulting in linearity would apply only after consideration of how the chemical is acting, not simply on the basis of the same type of tumors or preneoplastic lesions. For example, this principle has been invoked as being consistent with many cases of carcinogenesis in rodents, in which chemical exposure leads to an increased tumor incidence above control only in the tissues having a high background incidence of tumors or preneoplastic lesions (e.g., NAS 2008). The presumption for this observation is that background cancer responses are being evoked by external agents or the animal’s physiology acting either above their threshold of response or by way of a nonthreshold process; furthermore, if the chemical in question causes cancer in a similar manner (e.g., by mutagenic damage to critical DNA), then linearity in the low-dose range is expected. However, in order to invoke the additivity to background consideration in these cases of high background tumors, one needs to consider the underlying biological reason for the high background level of the tumors in that particular species/ strain/sex. If the high background is unique to a specific species/strain/sex, it is likely that the background reflects physiological differences (e.g., hormonal, genetic, or other differences) related to that group and that the chemical is causing cancer by acting on a target other than those related to these specific physiological differences. Even if the chemical is acting on the same process that results in the background response, additional consideration is needed. Linear low-dose response from additivity to background would be expected in such situations only if the chemical interaction with the underlying biological process is itself a nonthreshold process. It is possible to envision situations where the underlying biological processes result in an elevated background, but there is still sufficient resiliency in the system that one molecule of a chemical will not result in increased risk (Rhomberg 2008, personal communication). It is also important to note that, as first proposed, this issue of additivity to background refers to background physiology of the organism, not a cancer response from “background” exposure to other chemicals. The potential for “background” exposures to bring a cancer dose–response that is biologically nonlinear into the linear range for a human population was noted in the EPA (2005) cancer guidelines, and again by NAS (2008). Such exposures are most properly addressed as part of a mixtures assessment. Some, or perhaps even many, of the chemicals in these mixtures have similar modes of toxic action and/or act through similar pathways. This raises the issue of how one goes about assessing risk to such mixtures. Mixtures risk assessment is routinely done by scientists based on guidelines from a number of organizations (e.g., ACGIH 2006; EPA 1986, 2000, 2009a–d). EPA (1986, 2000d), for example, suggests first developing a dose–response assessment for the mixture of concern, such as what EPA (2009c) has done for PCBs. If sufficient information is not available to conduct such a dose–response assessment, then EPA recommends conducting a dose–response assessment on the basis of a sufficiently similar mixture (EPA 2009d). If sufficient information is not available
24.2. HISTORICAL
623
for either of these two approaches, then EPA (1986, 2000) recommends conducting the dose–response assessment on the basis of individual components of the mixture. Because toxicity information is seldom available on the mixture of concern, or on a sufficiently similar mixture, the mixtures dose–response assessment is most often conducted on the basis of individual components. However, in order to conduct a credible dose–response assessment for mixtures based on the toxicity of components as suggested by EPA (1986, 2000) guidelines, the dose–response assessment of individual components cannot incorporate considerations of background exposures to other, similarly acting chemicals. Thus, this second basis for the linear dose–response assessment for individual chemicals, additivity to background, is not tenable. Rather, a mixtures dose–response assessment should be conducted on the mixture of concern, or on a sufficiently similar mixture. Absent such information, then the mixture dose–response assessment should be based on the dose–response assessments of individual chemicals, and these assessments on individual chemicals cannot in themselves incorporate background exposures to other chemicals. This thinking raises the question of how to address risk from a chemical for which background exposure (to the chemical of interest, and/or other chemicals) is already above the threshold for a key event in the carcinogenic pathway, such as sufficient receptor occupancy to turn on a key regulatory protein. In such cases, linear extrapolation from tumor data is not scientifically correct, in light of the threshold, but a simple RfD approach would not be useful, in light of the need to estimate risk above the threshold. However, it is necessary to first estimate the threshold (possibly the threshold for the key event), in order to determine that exposures are above the threshold. The need to estimate the risk from the chemical of interest at human exposure levels remains, but the observation that human exposure exceeds a threshold suggests the need to use data to directly calculate human risk (e.g., by linking receptor binding to the tumor response), rather than defaulting to linear or nonlinear methods for risk estimation. Furthermore, the statement that background exposure is above a threshold implies a “toxic equivalency” approach, and mixtures issues such as the potential for interactions should also be considered. In many situations, it can be a challenge to conduct a complete mixtures assessment based on the dose–response of individual chemicals, because insufficient individual chemical data are available to calculate risk for each chemical (NAS 2008). However, in such situations, a more scientifically accurate approach is to use methods such as risk based on surrogates or “threshold of toxicological concern” (TTC) to estimate the risk for chemicals lacking sufficient data, rather than modifying the approach to risk estimation for individual chemicals to account for other chemical exposures [e.g., see reviews by Kroes et al. (2005) and Dolan et al. (2005)]. A third basis for the linear low-dose extrapolation is often stated as an expected broadening of the dose–response curve in the human population in the low-dose region when compared with the corresponding experimental animal curve, on which many of these linear extrapolations are based (Lutz 2001). This broadening is likely due to greater variations among humans in both toxicokinetics and toxicodynamics when compared with the experimental animal, thus leading to greater variations in
624
CHAPTER 24 LINEAR LOW-DOSE EXTRAPOLATIONS
tumor response. Although greater variations in response would not strictly lead to a linear slope, it might support the argument against the estimation of a practical population threshold, especially if the background carcinogen exposure were already in excess of a hypothetical population threshold for sensitive populations (e.g., skin cancer in populations with xeroderma pigmentosa, a genetic deficiency in DNA repair capacity). Thus, the historical context of the linear-dose response assessment for carcinogens is based on three principles: First, an assumption that the stochastic process likely to occur in radiation carcinogenesis applies, along with the one-hit model, to chemical carcinogenicity. Second, that if a direct-acting carcinogen acts in a manner consistent with existing background exposures, then the additive response will be linear at low dose with almost any model. Third, that dose–response curves in humans are expected to be broader than corresponding curves in experimental animals, and the extrapolation should account for this likely occurrence. These principles have both supporting and detracting data, as well as associated supporters and detractors. We find that the second principle, that of additivity to background, is not tenable in light of EPA’s guidelines of chemical mixtures risk assessment, which directs risk scientists to develop dose–response assessments for individual chemicals prior to any additive response. Lutz (1998) argued that for DNA-reactive carcinogens, linearity of the dose– response at the low-dose end is expected, but then he went on to state that deviation from linearity in dose–response relationship for tumors may be appropriate if mechanistic considerations can explain the threshold-like shape of the dose–response curve. For example, with increasing dose, saturation of DNA repair can introduce a sublinearity, as can stimulation of cell division from high-dose toxicity. Moreover, if a carcinogen increases the rate of cell division at high dose but has an antimitogenic effect at low dose, then a J-shaped, or hormetic, curve might result. Thus, for a mechanism-based assessment of a low-dose cancer risk, information on the various modes of action and modulations should be available over the full dose range, and models should be refined to incorporate the respective information. In considering the shape of the tumor dose–response curve, it is important to recognize that the mere observation of a tumor response that appears different from linearity is not in itself an adequate justification for the determination that nonlinear low-dose extrapolation should be used for that chemical. For example, Lutz et al. (2005) conducted a simulation study, in which they binomially sampled from a study of 50 animals per group, assuming a true linear dose response from 5% to 25% tumor incidence at arbitrary dose levels of 0, 0.5, 1, 2, and 4. They found linearity in the dose–response curve of only 8 of the 20 samples (only 40%), and 4 of the simulations did not reveal the carcinogenicity at all. Three exhibited thresholds, and two showed a nonmonotonic behavior with a decrease at low dose, followed by a significant increase at high dose (“hormesis”). Thus, for a study design that included a standard number of animals/group, but more dose groups than the typical NTP bioassay, a chemical for which the true dose–response is linear can result in linear, nonlinear, threshold, or “hormetic” dose–responses, or no response at all, simply based on sampling variability. Lutz et al. (2005) went on to state that nonlinear shapes of dose cancer incidence curves are rarely seen with epidemiological data in
24.3. ISSUES RELATED TO EXTRAPOLATION FROM EXPERIMENTAL DATA
625
humans and that this discrepancy from data in rodents may in part be explained by the greater degree of variability in individual susceptibilities for tumor induction in humans, as noted in the third consideration above. Thus, they suggested that linear extrapolation of a human cancer risk could therefore be appropriate even if animal bioassays show nonlinearity. These simulation studies of Lutz et al. (2005) illustrate the difficulty and uncertainty in characterizing the shape of dose–response curve based on the standard animal study design, due to sampling uncertainties, and thus again emphasize the importance of incorporating a mechanistic understanding of the carcinogenic response.
24.3. ISSUES RELATED TO EXTRAPOLATION FROM EXPERIMENTAL DATA It is important to recognize that there is considerable uncertainty involved in extrapolating dose–response data from the high doses at which animals are tested to environmentally relevant doses. The existing testing approach reflects an effort to maximize sensitivity while controlling cost and considering animal welfare, as well as the belief that any chemicals missed would be of such low potency that they would not have a meaningful impact on the cancer incidence at environmental exposure levels. Furthermore, although the typical study design (50 animals/sex/ dose) is used for both hazard identification and quantitative dose–response assessment, the aforementioned considerations mean that the study design is driven largely by hazard identification needs, and it is recognized that there are numerous limitations in quantitatively and accurately extrapolating to low doses. Due to these limitations, low-dose extrapolation for regulatory applications has generally focused on ensuring health-protectiveness, rather than accuracy in the extrapolation. A key implication of this typical experimental animal study design is that information is available on the shape of the dose–response curve at high doses (although sampling uncertainties apply even there, as noted above and by Lutz et al. 2005), but typically little or no information exists on the shape of the curve at low (environmentally relevant) doses. Extrapolation from the animal data to low doses is necessary, but it is possible to identify many different dose–response curves that fit a given data set, resulting in very different implications in the low-dose region. Regulatory agencies have adopted approaches that avoid this model-dependence of curves at low doses. The EPA, for example, identifies a point of departure (POD) in the range of the animal data, and then it chooses a linear or nonlinear approach to extrapolate (EPA 2005). The linear extrapolation approach avoids model dependency in risk estimation, but is recognized as conservative, because it does not capture nonlinearities that may occur (or are likely to occur) at lower doses. Health Canada avoids even estimating the risk at low doses due to such uncertainties, and instead it compares the exposure and the point of departure to calculate an exposure potency index (EPI) and then determines whether the EPI is of sufficient magnitude (Meek et al. 1994), based on how likely a particular chemical is a human carcinogen. As noted above, it is important to distinguish issues related to the shape of the dose–response curve at low (environmentally relevant) doses from issues related to
626
CHAPTER 24 LINEAR LOW-DOSE EXTRAPOLATIONS
the shape of the dose–response curve when (a) extrapolating from the experimental data range to low doses and (b) the shape of the curve in the experimental range. Many tumor dose–response curves are concave upward (i.e., nonlinear) in the dose range of the animal bioassay. However, the terms “linear” and “nonlinear” extrapolation refer to the approach used to extrapolate to below the range of the bioassay, and therefore they reflect only the shape of the dose–response curve at low doses.* Nonlinearities in the animal bioassay data do not necessarily mean that nonlinear extrapolation to low doses is appropriate. Even if there is no response above controls at low experimental doses, this could be due to the bioassay having insufficient statistical power to detect a small incidence of chemical-induced tumors at the lower doses, rather than reflecting a nonlinear dose–response. The linearity in low-dose extrapolation relates to the shape of typical concave upward dose–response curves, which have minimal curvature near the origin and have higher curvature at high doses. This means that the lower the point of departure from the experimental data, the better the approximation of the true low-dose slope. Drawing a straight line from a high-response to the origin is likely to overestimate the low-dose slope. However, using a mathematical model to extrapolate the curve below the range of the data introduces additional uncertainties. The current EPA cancer guidelines (EPA 2005) recommend that the point of departure should be the lowest dose that is adequately supported by the data, an approach that balances these two considerations. These two considerations of the shape of the curve in the range of the animal data, as well as the shape of the curve used for low-dose extrapolation, are further distinguished from the shape of the curve at low doses near zero (which may apply to environmentally-relevant exposures, or still lower exposure levels). Several considerations from first principles support linearity at these very low doses approaching zero [reviewed by Slikker et al. (2004a)]. We will discuss these considerations one by one; but throughout this discussion, recall that the principles apply to doses approaching zero, and the implications are different as doses increase to experimental animal doses. The slope of the low-dose linear curves based on the following considerations may be very different from the slope between the point of departure and the origin (and from the slope in the range of the animal data). It is also noteworthy that many of these considerations are cited as contributors to nonlinearity of the dose–response curve. These factors do indeed contribute to nonlinearities in the dose–response curve (as further noted by Slikker et al. 2004a), but at very low doses the reactions can be approximated by linear relationships. • Metabolism. The Michaelis–Menten equation becomes a linear function of the concentration of the parent chemical as dose approaches zero. Low-dose linearity also results if a chemical is metabolized via competing pathways: *Quantitatively, the EPA (2005) estimates cancer risk by fitting a curve to the animal bioassay data, and identifying a POD (typically a 10% response) in the range of the data. A linear extrapolation is conducted by drawing a straight line to the origin and calculating the unit risk at any given dose. In contrast, a nonlinear extrapolation considers the overall database to determine either (a) an uncertainty factor for developing a RfD or (b) determine a safe margin of exposure between the estimated exposure level and the point of departure.
24.3. ISSUES RELATED TO EXTRAPOLATION FROM EXPERIMENTAL DATA
627
(a) one that is capacity-limited and produces a harmless metabolite and (b) one that has high capacity and produces a DNA-reactive metabolite. In this latter case, the ratio at low doses of the rate of production of the two metabolites becomes a constant. Nonlinearities in metabolism may occur at higher doses, again following Michaelis–Menten kinetics (Lehninger 1975). • Other Aspects of Toxicokinetics. Other aspects of toxicokinetics also have the potential for nonlinearities, but are linear at doses approaching zero. Nonlinearities may occur in the transport of a chemical into a cell, by either diffusion or active transport. Diffusion is a linear function of chemical concentration, and active transport at doses approaching zero is a linear function of chemical concentration, unless the expression of the transport molecule is induced. Nonlinearities may occur at higher doses, if the transport mechanism is saturated. • DNA Repair. The argument is often made that it is necessary for the rate of production of lesions to exceed the maximum rate of repair in order for tumors to result, due to the potential for DNA repair. However, this argument assumes perfect efficiency, while DNA repair, like other enzymatic processes, would follow Michaelis–Menten kinetics. Thus, the rate of repair at low doses is a linear function of the concentration of lesions, and some lesions (including mutations) may persist when the cell completes its next division. Although risk assessors often classify chemicals as genotoxic or nongenotoxic, recent research has begun to consider the potential for thresholds for genotoxic agents that indirectly interact with DNA, and even for agents that directly interact with DNA. Indeed, the journal Mutation Research has dedicated an entire issue to this topic (Vol. 464, Issue 1; 2000). Many of the ideas presented in the following paragraphs are provocative and controversial. More research is clearly needed in this area, but the ideas are presented here to encourage discussion on these issues and their implications for assessments of chemicals such as acrylonitrile, trichloroethylene, and 1,3-dichloropropene (1,3-DCP), which have mixed or weak genotoxicity data or have positive genotoxicity data coupled with evidence for other modes of action. Part of this controversy relates to different definitions of thresholds (see footnote * on page 621 and also see Table 24.1). Regulators often focus on the presence or absence of a biological threshold, or the absence of any damage, as a conservative health-protective approach. Determining whether a biological threshold exists, or whether there is some small but indistinguishable effect, is a difficult problem that depends on issues of biological understanding, study sensitivity, and statistical power. The biological threshold can be distinguished from a pragmatic threshold or statistical threshold. Biologically adverse changes may be occurring at low doses, but the slope of the curve may be so shallow that the changes are not statistically detectable, and/or they may result in risk below de minimis levels. Nonlinearity in the mutagenesis dose–response curve can occur even for chemicals that interact directly with DNA. Sofuni et al. (2000) evaluated the mutagenicity of the alkylating agents N-ethyl-N′-nitro-N-nitrosoguanidine (ENNG), methyl methanesulfonate (MMS), dimethylnitrosamine (DMN), and ethylnitrosourea
628
CHAPTER 24 LINEAR LOW-DOSE EXTRAPOLATIONS
TABLE 24.1.
Definitions of Thresholdsa
Absolute threshold Apparent threshold Biological threshold
Pragmatic threshold Statistical threshold
A concentration of a chemical below which a cell would not “notice” the presence of this agent. An immeasurable concentration of a chemical below which no damage is produced. A measureable concentration of chemical below which no damage is produced because of the chemical’s inability to undergo the necessary biochemical reactions. The lowest concentration of a chemical upon which regulatory decisions are made. The lowest concentration of a chemical that induces a statistically significant increase in the endpoint being measured.
a
Adapted from Kirsch-Volders et al. (2000). See text and footnote * on page •• for further discussion.
(ENU), in Salmonella strain TA1535 and TA1535 derivatives lacking the inducible adaptive repair gene ada, which removes alkyl groups from O6-methylguanine. They found that the mutagenic response in repair-deficient strains was linear down to the lowest dose tested, with dose-related increases in revertants in the repair-deficient strains at doses that did not increase the revertant count in the parental TA1535 strain. By contrast, in repair-competent strains, there was no apparent mutagenic response at the lowest doses. The dose–response for several of the chemicals in the parental strain was in the shape of a hockey stick; five or more doses over a dose range of more than an order of magnitude exhibited no increase over background, followed by a sharp increase in response at higher doses. While the absence of an increase in revertants in the parental strain at the lower doses could reflect the sensitivity of the assay and the difficulty in detecting a small increase over background, the consistent background response and nonlinear increase at higher doses suggests the presence of a biological threshold. In contrast to the results with the parental strain, linear or nearly linear dose–responses down to zero dose were observed in the methyltransferase-deficient strains. This study illustrates the importance of highfidelity DNA repair processes in determining the shape of the dose–response for mutagenesis at low doses, and it suggests that a threshold for mutation induction (below which all damage is repaired) may exist in the strain having normal DNA repair capacity. It is plausible to consider that DNA repair would have similar effects on the tumor dose–response. Koana et al. (2004) also found differences between repair-proficient and repair-deficient strains, using a model system of Drosophila exposed to X-irradiation and evaluated for somatic mutations. They found clear nonlinearities resembling thresholds in both repair-proficient and repair-deficient strains, with the response occurring at lower doses in the repair-deficient strain. Inducible repair mechanisms would also affect the shape of the dose–response curve for mutagenicity and tumorigenicity. For example, the response to a given (short-term) dose may vary, depending on whether the cell was previously exposed to an inducing dose. In light of these complexities, Lovell (2000) noted that inducible repair mechanisms are not easily
24.3. ISSUES RELATED TO EXTRAPOLATION FROM EXPERIMENTAL DATA
629
incorporated into existing models, one of the reasons that in vivo data are highly weighted in evaluating genotoxicity data. Results from mammalian systems illustrate the difficulty in having sufficient analytical sensitivity to detect and/or evaluate the shape of the dose–response curve. Hoshi et al. (2004) evaluated the mutagenicity of 2-amino-3,8-dimethylimidazo[4,5f]quinoxaline (MeIQx) in the livers of Big Blue transgenic rats fed a diet containing 0.001, 0.01, 0.1, 1, 10, or 100 ppm of MeIQx for 16 weeks. MeIQx is a heterocyclic amine that is a potent genotoxic carcinogen and strong bacterial mutagen. The authors found that the frequency of lac I mutants was significantly increased only at 10 or 100 ppm and that glutathione S-transferase-P (GST-P) positive foci (a preneoplastic lesion) were significantly increased only at 100 ppm, and they interpreted these data as supporting a threshold. However, while the gene mutation data are clearly nonlinear, there is a clear dose-related trend, possibly extending to the lowest dose. In contrast, the data for GST-P-positive foci are more supportive of a threshold, with the response at all doses up to 1 ppm being below background, although the variability at all doses might obscure a very shallow positive slope. Hengstler et al. (2003) reviewed data on the shape of the dose–response curves for a number of chemicals. In contrast with the examples described above, aflatoxin B1 shows a linear dose–response for tumors and DNA adducts down to exposures of 1 ppb or less in drinking water. Low-dose linearity was observed down to the ppm range for the incidence of liver tumors in Wistar rats and DNA adducts in Fisher rats exposed to N,N-diethylnitrosamine (DEN) in drinking water. Nonlinearity was observed at higher doses. These findings contrast with the results described above for bacterial mutagenicity of the closely related chemical DMN in bacteria, but it is not known whether a threshold-like nonlinearity would occur for DNA adducts and tumors at much lower in vivo doses. Interestingly, a threshold-like nonlinearity is seen for DEN induction of liver tumors in fish, and nonlinearity with very shallow initial slope is seen for liver tumors induced by weekly i.p. injection of Fisher 344 rats. Perera (1988) reviewed the data on the shape of the dose–response curves for adducts and mutations in a number of systems. He reported linearity in a variety of systems, including epidemiology studies, for the formation of DNA and protein adducts, both for chemicals that must be metabolically activated [(e.g., polycyclic aromatic hydrocarbons (PAHs) and 4-aminobiphenyl (4-ABP)], and for direct-acting chemicals, such as ethylene oxide and propylene oxide. He also noted that in vitro and in vivo studies have found a good correlation between levels of the critical DNA adduct (i.e., the one responsible for most of the chemical-induced mutations) and the frequency of induced mutations, indicating a linear dose–response for mutations. Swenberg et al. (2008) came to different conclusions in a review of the dose– response literature for mutations and DNA adducts, which are biomarkers of exposure. They concluded that biomarkers of exposure (DNA adducts) are usually linear at low doses down to zero (except when the chemical-induced adduct is also formed endogenously, in which case there is a plateau at the endogenous exposure level, and the adduct level of de minimus exposures below endogenous amounts will be lost in the noise of the background). In contrast, they found that biomarkers of effect
630
CHAPTER 24 LINEAR LOW-DOSE EXTRAPOLATIONS
(mutations) do not extrapolate to zero, but extrapolate back to the spontaneous or background number of mutations, and that the shape of the dose–response curves tended to be nonlinear, with flat curves at low doses, followed by regions of higher slope. The authors explained the shape of the dose–response curve as being a combination of curves driven by different aspects in different regions of the dose– response. At low doses they suggested that the dose–response curve is driven by the dose–response curve for mutagenesis resulting from endogenous DNA damage, whereas at higher exposures they suggested that the mutagenesis is driven by DNA damage resulting from the chemical exposure. Because the mutations (biomarkers of effect as described by Swenberg and colleagues) are further on the progression pathway and closer to cancer, the dose-response curve for cancer is better informed by the curve for mutations than that for DNA adducts. As noted by Swenberg and colleagues, part of the reason that mutations are better predictors of cancer risk than are DNA adducts is that there are a number of events between adduct formation and the development of mutations. The DNA adduct must remain in the DNA during its replication (i.e., be unrepaired), and the cell must mishandle the adduct (e.g., cause a mispairing) for a mutation to occur. Thus, results in the literature are mixed: Some reported a linear dose–response for mutations in multiple systems, whereas others reported consistent nonlinearities in the mutation dose–response. The reason for the difference between the results reported by Perera (1988) and those reported by Sofuni et al. (2000) and reviewed by Swenberg et al. (2008) is not clear, but may be related to differences in the region of the dose–response curve evaluated (e.g., lower doses tested in the latter studies, with linearity occurring at higher doses). Reaction kinetics may also argue against an absolute threshold. Since metabolism is not instantaneous, a reaction has some probability of proceeding via a less favored path. For example, a number of chemicals are detoxified by a saturable high-efficiency glutathione (GSH) pathway and are activated to a reactive epoxide (which can then react with DNA) by a high-capacity, low-efficiency pathway dependent on cytochrome P450. The presence of these competing pathways means that GSH conjugation predominates at low doses, but there is a nonzero (although low) probability for the production of the epoxide, even at low doses. This means that merely showing that one pathway dominates metabolism at low doses is not sufficient. It would be necessary to quantitatively determine the relative amount of the chemical metabolized via the GSH pathway and the P450 pathway to determine the relative contribution of activating and detoxification pathways. In such cases, physiologically based pharmacokinetic (PBPK) modeling could aid in evaluating whether mutagenic metabolites are formed in sufficient quantities to add significantly to background DNA damage rates and thus exceed a pragmatic threshold. In summary, the evaluation of the shape of the low-dose dose–response curve for tumorigenicity is not simply a matter of determining whether or not the chemical is genotoxic. The mechanism of genotoxicity and metabolic pathway need to be considered for a complete understanding of the dose–response. Finally, scientists often talk about genotoxicity as if there is a dichotomy between genotoxic and nongenotoxic chemicals, but chemicals actually fall on a
24.4. CONCLUSION
631
spectrum of genotoxicity/ nongenotoxicity (Butterworth and Bogdanffy 1999). For example, formaldehyde is a weak mutagen, but tumors are induced in the rat nasal passage only at doses producing high levels of necrosis and regenerative cell proliferation. BBDR modeling has shown that the tumorigenicity of formaldehyde is driven by cytotoxicity resulting from DNA–protein crosslinks, not from the genotoxic activity (CIIT 2009), resulting in an overall J-shaped dose–response curve (Conolly et al. 2004). Similarly, Melnick et al. (1996) noted that many of the chemicals considered to be nongenotoxic carcinogens (e.g., tamoxifen, cyproterone acetate) do have some degree of genotoxic activity, and some nongenotoxic effects can cause oxidative DNA damage and thereby initiate carcinogenesis. At the other end of the spectrum, cytotoxicity and necrosis can enhance the carcinogenicity of strong mutagens. In order to provide a structure for evaluating whether carcinogens are acting via a mutagenic MOA, EPA has published a draft Framework for Evaluating a Mutagenic MOA (EPA 2007). The draft framework explicitly recognizes that mutagenicity is not directly analogous to carcinogenicity, and just because a carcinogenic chemical is mutagenic does not mean that it is acting via a mutagenic MOA. The draft framework further describes an approach using the modified Hill criteria (EPA 2005) to evaluate whether a chemical is acting via a mutagenic MOA. This framework is important both for affirmatively evaluating whether a chemical acts via a mutagenic MOA (rather than using low-dose extrapolation as a default) and because the EPA uses an additional age-dependent adjustment factor for calculating the cancer risk from chemicals that act via a mutagenic MOA. As noted by Guyton et al. (2009), chemicals often act by multiple MOAs and it can be difficult to distinguish causative from associated key events. Genetically engineered animals lacking genes associated with key events (“knockout” animals) can be very useful for determining causation of hypothesized key events. Moore et al. (2008) described an approach to (a) aid in the evaluation of whether chemicals that are mutagenic and carcinogenic are acting via a mutagenic MOA and (b) directly test two aspects of the modified Hill criteria, temporality and dose–response. They described an assay approach using transgenic animals such as Big Blue® mice or rats to evaluate the mutation frequency in the same tissue, species, strain, and sex as the tumor target(s). The mutation frequency data are then compared with data on the tumor dose–response and the time to effect for tumors or, if available, precarcinogenic lesions, to see if the mutation data are consistent with the hypothesis that mutations are causing cancer. If the mutations are causing the tumors, they should be increased at the same or earlier times as for tumors (and any identified tumor precursors) and at the same or earlier doses.
24.4.
CONCLUSION
Linear low-dose extrapolation has been chosen by many regulatory agencies as a health-protective approach to cancer risk assessment. While this is a conservative approach, research in recent years has found many exceptions to this approach.
632
CHAPTER 24 LINEAR LOW-DOSE EXTRAPOLATIONS
Key aspects of the scientific basis for linear extrapolation are (1) extrapolation from radiation carcinogenesis, (2) consideration of additivity to background, and (3) higher variability of the human population compared to the experimental animal population. However, none of these considerations automatically result in a linear dose–response; and one of these, additivity to background, is operationally inconsistent with the risk assessment guidelines of several organizations for chemical mixtures. Chemicals differ from radiation in that, even for direct-acting carcinogens, toxicokinetic processes modulate the tissue dose of the DNA-reactive form, and these processes can affect the shape of the dose–response curve. Additivity to background would result in a linear dose–response only if (a) the chemical is acting on the same process that results in the background response and (b) the chemical interaction with the underlying biological process is itself a nonthreshold process. Greater population variability in the human population in the low-dose region when compared with the corresponding experimental animal population results in broadening of the dose–response curve. This broadening would not strictly lead to a linear slope, but might support the argument against the estimation of a practical population threshold, especially if the background carcinogen exposure were in already in excess of a hypothetical population threshold for sensitive populations. Of course, in order for one to know whether a population is in excess of such a hypothetical threshold, one has to estimate it or have measured that response using sensitive endpoints, such as biomarkers. It is important to distinguish between factors leading to linearity as dose approaches zero (low-dose linearity) and considerations of linearity or nonlinearity of the overall dose–response curve. Thus, while consideration of the kinetics of fundamental biological processes (e.g., absorption, DNA reactivity, DNA repair) indicates that these processes tend to be linear as the dose approaches zero, that linear slope may be very different from the slope derived by extrapolation from the animal tumor data. Several authors have also reported nonlinearities in DNA mutation, even for direct-acting mutagens, although issues of assay sensitivity make it difficult to distinguish nonlinearity from a true threshold. All of these considerations indicate that the biology behind the shape of the tumor dose–response curve is much more complex than a simple conclusion that mutagenic activity = linear dose-response. Ultimately, biologically based dose– response models and use of biomarker data may make it possible to extend the tumor dose–response curve to low doses based on biological data, rather than presumptions about the shape of the dose–response curve. In the shorter term, it is important to recognize that the biology is complex, and linear extrapolation from tumor data is a health-protective science policy decision.
ACKNOWLEDGMENTS The authors wish to thank Rick Hertzberg for stimulating discussions, and they wish to acknowledge Alison Willis and Valerie Ayers for technical editorial assistance.
REFERENCES
633
REFERENCES American Conference of Governmental Industrial Hygienists (ACGIH). (2006). TLVs and BEIs: Based on the documentation of the threshold limit values for chemical substances and physical agents & bioassays exposures indices. Appendix E. Threshold limit values for mixtures. Cincinnati, OH. Andersen, M. E., Conolly, R. B., and Gaylor, D. W. (2003). Thresholds of carcinogenicity in the ED01 study. Toxicol Sci 74(2), 486–487. Butterworth, B. E., and Bogdanffy, M. S. (1999). A comprehensive approach for integration of toxicity and cancer risk assessments. Regul Toxicol Pharmacol 29, 23–36. CIIT (2009). International Toxicity Estimates for Risk (ITER) IPRV entry for CIIT’s formaldehyde. Accessed at http://toxnet.nlm.nih.gov. Calabrese, E. J. (2009). The road to linearity: why linearity at low doses became the basis for carcinogen risk assessment. Arch Toxicol 83, 203–225. Conolly, R. B., Kimbell, J. S., Janszen, D., Schlosser, P. M., Kalisak, D., Preston, J., and Miller, F. J. (2004). Human respiratory tract cancer risks of inhaled formaldehyde: Dose–response predictions derived from biologically-motivated computational modeling of a combined rodent and human dataset. Toxicol Sci 82, 279–296. Crump, K. S., Hoel, D. G., Langley, C. H., and Peto, R. (1976). Fundamental carcinogenic processes and their implications for low dose risk assessment. Cancer Res 36, 2973–2979. Dolan, D., Naumann, B., Sargent, E., Maier, A., and Dourson, M. (2005). Application of the threshold of toxicological concern concept to pharmaceutical manufacturing operations. Regul Toxicol Pharmacol 43, 1–9. Doull, J. (2003). The “Red Book” and other risk assessment milestones. Hum Ecol Risk Assess 9, 1229–1238. Dourson, M. L., Hertzberg, R., Allen, B., Haber, L., Parker, A., Kroner, O., Maier, A., and Kohrman, M. (2008). Evidence-based dose response assessment for thyroid tumorigenesis from acrylamide. Regul Toxicol Pharmacol 52(3), 264–289. EPA (1986). Guidelines for the health risk assessment of chemical mixtures. EPA/630/R-98/002. EPA (1996). Proposed guidelines for carcinogen risk assessment. Office of Research and Development, Washington, D.C. EPA/600/P-92/003C. EPA (2000). Supplementary guidance for conducting health risk assessment of chemical mixtures. EPA/630/R-00/002. EPA (2005). Guidelines for carcinogen risk assessment. EPA/630/P-03/001B. EPA (2007). Framework for determining a mutagenic mode of action for carcinogenicity (external review draft). EPA 120/R-07/002-A. EPA (2009a). IRIS entry for N-nitrosodimethylamine. Accessed at http://www.epa.gov/NCEA/iris/ subst/0045.htm. EPA (2009b). IRIS entry for perchlorate. Accessed at http://www.epa.gov/NCEA/iris/subst/1007.htm. EPA (2009c). IRIS entry for PCBs. Accessed at http://www.epa.gov/NCEA/iris/subst/0294.htm. EPA (2009d). Overview summary of toxicological, chemical, and statistical criteria for defining sufficient similarity for use in the risk assessment of chemical mixtures. Published as a series of papers in J Toxicol Environ Health A 72, 429–504. Gaddum, J. H. (1945). Lognormal distributions. Nature 156(3964), 463–466. Guyton, K. Z., Kyle, A. D., Aubrecht, J., Cogliano, V. J., Eastmond, D. A., Jackson, M., Keshava, N., Sandy, M. S., Sonawane, B., Zhang, L., Waters, M. D., and Smith, M. T. (2009). Improving prediction of chemical carcinogenicity by considering multiple mechanisms and applying toxicogenomic approaches. Mutat Res 681(2–3), 230–240. Haseman, J. K. (2003). An alternative perspective: A critical evaluation of the Waddell threshold extrapolation model in chemical carcinogenesis. Toxicol Pathol 31(5), 468–470. Health Canada (2009). International toxicity estimates for risk (ITER) entry for Health Canada’s 1,3butadiene. Accessed at http://toxnet.nlm.nih.gov/. Hengstler, J. G., Bogdanffy, M. S., Bolt, H. M., and Oesch, F. (2003). Challenging dogma: Thresholds for genotoxic carcinogens? The case of vinyl acetate. Annu Rev Pharmacol Toxicol 43, 485–520.
634
CHAPTER 24 LINEAR LOW-DOSE EXTRAPOLATIONS
Heyssel, R., Brill, A. B., Woodbury, L. A., Nishimura, E. T., Ghose, T., Hoshino, T., and Yamasaki, M. (1960). Leukemia in Hiroshima atomic bomb survivors. Blood 15(3), 313–331. Hoshi, M., Morimura, K., Wanibuchi, H., Wei, M., Okochi, E., Ushijima, T., Takaoka, K., and Fukushima, S. (2004). No-observed effect levels for carcinogenicity and for in vivo mutagenicity of a genotoxic carcinogen. Toxicol Sci 81, 273–279. Kirsch-Volders, M., Aardema, M., and Elhajouji, A. (2000). Concepts of threshold in mutagenesis and carcinogenesis. Mutat Res 464, 3–11. Koana, T., Takashima, Y., Okada, M. O., Ikehata, M., Miyakoshi, J., and Sakai, K. (2004). A threshold exists in the dose–response relationship for somatic mutation frequency induced by X irradiation of Drosophila. Radiat Res 161, 391–396. Kroes, R., Kleiner, J., and Renwick, A. (2005). The threshold of toxicological concern concept in risk assessment. Toxicol Sci 86(2), 226–230. Lehninger, A. L. (1975). Biochemistry, 2nd edition, Worth Publishers, New York. Lovell, D. P. (2000). Dose–response and threshold-mediated mechanisms in mutagenesis: Statistical models and study design. Mutat Res 464, 87–95. Lutz, W. K. (1990). Dose–reponse relationship and low dose extrapolation in chemical carcinogenesis. Carcinogenesis 11(8), 1243–1247. Lutz, W. K. (1998). Dose–response relationships in chemical carcinogenesis: superposition of different mechanisms of action, resulting in linear–nonlinear curves, practical thresholds, J-shapes. Mutat Res 405(2), 117–124. Lutz, W. K. (2001). Susceptibility differences in chemical carcinogenesis linearize the dose-response relationship: Threshold doses can be defined only for individuals. Mutat Res 482, 71–76. Lutz, W. K., Gaylor, D. W., Conolly, R. B., and Lutz, R. W. (2005). Nonlinearity and thresholds in dose–response relationships for carcinogenicity due to sampling variation, logarithmic dose scaling, or small differences in individual susceptibility. Toxicol Appl Pharmacol 207, S565–S569. Meek, M., Newhook, R., Liteplo, R., and Armstrong, V. C. (1994). Approach to assessment of risk to human health for priority substances under the Canadian environmental protection act. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev 12(2), 105–134. Melnick, R. L., Kohn, M. C., and Portier, C. J. (1996). Implications for risk assessment of suggested nongenotoxic mechanisms of chemical carcinogenesis. Environ Health Perspect 104(Suppl 1), 123–134. Moore, M. M., Heflich, R. H., Haber, L. T., Allen, B.C., Shipp, A. M., and Kodell, R. L. (2008). Analysis of in vivo mutation data can inform cancer risk assessment. Regul Toxicol Pharmacol 51, 151–161. NAS (1977). Drinking Water and Health, National Academies Press, Washington, D.C. NAS (2008). Science and Decisions: Advancing Risk Assessment, National Academies Press, Washington, D.C. Perera, F. P. (1988). The significance of DNA and protein adducts in human biomonitoring studies. Mutat Res 205, 255–269. Rhomberg, L. (2008). Personal communication with L. Haber of Toxicology Excellence for Risk Assessment (TERA). RIVM (2009). ITER entry for RIVM’s trichloroethylene. Accessed at http://toxnet.nlm.nih.gov/cgi-bin/ sis/htmlgen?iter. Rozman, K. K., Kerecsen, L., Viluksela, M. K., Osterle, D., Deml, E., Viluksela, M., Stahl, B. U., Greim, H., and Doull, J. (1996). A toxicologist’s view of cancer risk assessment. Drug Metab Rev 28(1–2), 29–52. Slikker, W., Jr, Andersen, M. E., Bogdanffy, M. S., Bus, J. S., Cohen, S. D., Conolly, R. B., David, R. M., Doerrer, N. G., Dorman, D. C., Gaylor, D. W., Hattis, D., Rogers, J. M., Woodrow Setzer, R., Swenberg, J. A., and Wallace, K. (2004a). Dose-dependent transitions in mechanisms of toxicity. Toxicol Appl Pharmacol 201(3), 203–225. Slikker, W., Jr, Andersen, M. E., Bogdanffy, M. S., Bus, J. S., Cohen, S. D., Conolly, R. B., David, R. M., Doerrer, N. G., Dorman, D. C., Gaylor, D. W., Hattis, D., Rogers, J. M., Woodrow Setzer, R., Swenberg, J. A., and Wallace, K. (2004b). Dose-dependent transitions in mechanisms of toxicity: Case studies. Toxicol Appl Pharmacol 201(3), 226–294. Sofuni, T., Hayashi, M., Nohmi, T., Matsuoka, A., Yamada, M., and Kamata E. (2000). Semiquantitative evaluation of genotoxic activity of chemical substances and evidence for a biological threshold of genotoxic activity. Mutat Res 464, 97–104.
REFERENCES
635
Swenberg, J. A., Fryar-Tita, E., Jeong, Y. C., Boysen, G., Starr, T., Walker, V. E., and Albertini, R. J. (2008). Biomarkers in toxicology and risk assessment: Informing critical dose response assessment relationships. Chem Res Toxicol 21, 253–265. Waddell, W. J. (2003a). Thresholds in Chemical Carcinogenesis: What are animal experiments telling us? Toxicol Pathol 31(3), 260–262. Waddell, W. J. (2003b). Reply. Toxicol Pathol 31(6), 712–713.
CH A P TE R
25
QUANTITATIVE CANCER RISK ASSESSMENT OF NONGENOTOXIC CARCINOGENS Rafael Meza Jihyoun Jeon Suresh H. Moolgavkar
25.1.
INTRODUCTION
In the current paradigm, carcinogenesis is a process involving mutation accumulation and clonal expansion of partially altered populations of cells on the pathway to cancer. The idea that accumulation of mutations could lead to abrogation of normal control of growth and differentiation originated with Muller (1951) and Nordling (1952) and was formalized in mathematical models by Nordling (1953) and Armitage and Doll (1954). In the 1940s, Berenblum and Shubik’s seminal experiments on the mouse skin led to the operational ideas of initiation and promotion in carcinogenesis (Berenblum and Shubik 1947). The process of initiation, brought about by one or more mutations, gives rise to an initiated cell, the hallmark of which is that it and its progeny have a growth advantage over their normal neighbors. This advantage is amplified by the presence of endogenous or exogenous promoters, agents that accelerate the clonal expansion of initiated cells. Thus, the current paradigm for carcinogenesis is essentially an amalgam of the idea, going back to Muller, that multiple mutations are involved in the process and the idea derived from the concepts of initiation and promotion that, once some of these mutations have occurred, the cell could already have a growth advantage over its neighbors. It follows that carcinogens, broadly speaking, could be classified into initiators, agents that interact with the DNA to bring about mutations, and promoters, agents that do not necessarily interact with the DNA, that accelerate clonal expansion of initiated cells. More detailed description and concerpts of cancer formation can be found in the other chapters of this book (Chapters 5 and 6). Regulatory agencies such as the United States Environmental Protection Agency (EPA) have shown interest in developing distinct protocols for
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
636
25.1. INTRODUCTION
637
the risk assessment of genotoxic and nongenotoxic carcinogens. Models that explicitly incorporate both mutations and clonal expansion of partially altered cells can be used for risk assessment for both classes of carcinogen and were first proposed by Armitage and Doll (1957) and by Moolgavkar and Venzon (1979) and Moolgavkar and Knudson (1981). The main focus of this chapter is on the quantitative risk assessment of nongenotoxic carcinogens. In our experience, however, there are few agents that act exclusively as initiators or promoters. Rather, most agents appear to have both activities. Recent epidemiological evidence suggests that agents, such as ionizing radiation, that were considered quintessential mutagens, may also be promoters. We therefore focus on the interaction of initiation and promotion in the carcinogenic process. We begin with a discussion of the two-stage clonal expansion (TSCE) model and then go on to discuss its multistage extension, the multistage clonal expansion (MSCE) model. After discussing these models briefly, we give examples of analyses of epidemiological and toxicological data using the models and discuss features that are particularly relevant to promotion. The models can also be used to investigate temporal patterns of risk when agents may have opposing actions on initiation and promotion. For example, an agent may decrease rates of mutation while increasing promotion of initiated cells. We discuss briefly a recent paper analyzing such a situation.
25.1.1.
The Hazard or Incidence Function
The standard measure for cancer risk is the hazard or age-specific incidence function of cancer. Let T be the time to malignant transformation of a particular tissue. Let P(t) = P[T ≤ t] be the probability that cancer occurs before age t. Then, the hazard or incidence function is defined as h ( t ) = lim
Δt →0
P [ t < T ≤ t + Δt T > t ] Δt
(25.1)
So, the hazard measures the instantaneous rate at which cancer occurs in a previously cancer free tissue. The hazard can be expressed also in terms of the survival function, S(t) ≡ P[T > t] = 1 − P(t), h ( t ) = lim
Δt →0
1 P [T > t ] − P [T > t + Δt ] − S ′ ( t ) = P [T > t ] S (t ) Δt
(25.2)
where S′(t) denotes the derivative of the survival function with respect to t. The statistical estimator for the hazard is the incidence or mortality rate of cancer in a population (number of cases/population at risk).
25.1.2.
Two-Stage Clonal Expansion (TSCE) Model
The two-stage clonal expansion model (TSCE) was originally developed by Moolgavkar and Venzon (1979) and Moolgavkar and Knudson (1981). This model assumes that malignant transformation of susceptible stem cells occurs as a result
638
CHAPTER 25
QUANTITATIVE CANCER RISK ASSESSMENT OF NONGENOTOXIC CARCINOGENS
Figure 25.1. Two-stage clonal expansion model.
of two specific, irreversible and hereditary events. In its most general form, the number of normal (susceptible) stem cells as a function of age, X(t), is modeled as a stochastic process. However, for simplicity we concentrate only in the case when the normal stem cells are assumed to follow a deterministic process (Moolgavkar et al. 1988; Moolgavkar and Knudson 1981; Moolgavkar and Luebeck 1979). The first event in the TSCE model occurs when a normal stem cell becomes “initiated.” Initiation is modeled as a nonhomogeneous Poisson process of intensity μ0(t)X(t), where X(t) is the number of susceptible stem cells at age t. Initiated cells, also called intermediate cells, are assumed to expand clonally; this expansion is modeled as a birth–death and mutation process. This means that initiated cells can divide into two initiated cells with rate α(t), die or differentiate with rate β(t), and divide into one initiated and one malignant cell with rate μ1(t). Figure 25.1 gives a schematic representation of this model. 25.1.2.1. Constant Model Parameters. If we assume that the number of susceptible stem cells is constant during life and that the model parameters are constant, then we can find a closed-form expression for the hazard and survival function under the TSCE model (Luebeck and Moolgavkar 2002): q− p ⎛ ⎞ S ( t ) = ⎜ − pt ⎝ qe − pe − qt ⎟⎠
μ0 X α
(25.3)
and h (t ) =
μ0 X pq (e − qt − e − pt ) α qe − pt − pe − qt
(25.4)
where p=
1⎡ 2 − (α − β − μ1 ) − (α − β − μ1 ) + 4αμ1 ⎤ ⎦ 2⎣
(25.5)
25.1. INTRODUCTION
q=
1⎡ 2 − (α − β − μ1 ) + (α − β − μ1 ) + 4αμ1 ⎤ ⎣ ⎦ 2
639 (25.6)
Note that pq = −αμ1 and (p + q) = −(α − β − μ1). Also notice that if μ1 << 1, then p ≈ −(α − β − μ1) and q ≈ αμ1/(α − β − μ1). 25.1.2.2. Age-Dependent Model Parameters. Multistage models can be used to evaluate the effects of carcinogens on cancer risk, by assuming that the model parameters are functions of the exposure dose at any particular age. Similarly, multistage models can also be used to evaluate the effects of preventive measures on cancer risk—for example, the use of nonsteroidal anti-inflammatory drugs as a preventive measure to reduce the risk of polyps in the colon. To evaluate the effects of agents with age-dependent exposure, we need expressions of the hazard and survival functions when the model parameters depend on age. Closed-form expressions for the hazard and survival functions of the TSCE model are not available in the case of general age-dependent parameters, so numerical methods are necessary (Crump et al. 2005; Little 1995; Little and Wright 2003; Meza 2006). However, closed-form expressions for the hazard and survival function of the TSCE model in the case of piecewise constant parameters were derived by Heidenreich et al. (1997): ⎧ n μ0, j X ⎛ q j − p j ⎞ ⎫ S ( t ) = exp ⎨∑ ln ⎜ ⎬ ⎝ f j ( t j −1, t n ) ⎟⎠ ⎭ ⎩ j =1 α j n μ X n μ X ∂ 1 ∂ 0, j 0, j h (t ) = ∑ ln ( f j ( t j −1, t n )) = ∑ f j ( t j −1, t n ) ∂t n f j ( t j −1, t n ) ∂t n j =1 α j j =1 α j
(25.7) (25.8)
where n is the number of age–periods with different parameter values before age tn ≡ t; [tj−1,tj], j = 1, … , n, denote the endpoints of the jth age–period; t0 = 0, and μ0,j, αj, pj, qj, μ1,j denote the parameter values during the jth age–period, and y n = 0,
y j −1 =
α j −1 ( y j − p j ) q j eq j (t j −1 −t j ) + ( q j − y j ) p j e p j (t j −1 −t j ) αj f j ( t j −1, t n )
f j ( t j −1, t n ) = ( y j − p j ) exp {q j ( t j −1 − t j )} + ( q j − y j ) exp { p j ( t j −1 − t j )}
∂ fn ( t n−1, t n ) = [ exp {qn ( t n−1 − t n )} − exp { pn ( t n−1 − t n )}] pn qn ∂t n ∂ ∂ f j ( t j −1, t n ) = [ exp {q j ( t j −1 − t j )} − exp { p j ( t j −1 − t j )}] y j ∂t n ∂t n
(25.9) (25.10) (25.11) (25.12)
∂ α ( q − pn ) e( pn + qn )(tn−1 −tn ) y n−1 = n−1 n pn qn ∂t n αn ( fn (tn−1, tn ))2
(25.13)
α j −1 ( q j − p j ) e( p j + q j )(t j −1 −t j ) ∂ ∂ y j y j −1 = ∂t n ∂t n αj ( f j (t j −1, tn ))2
(25.14)
2
2
640
CHAPTER 25
QUANTITATIVE CANCER RISK ASSESSMENT OF NONGENOTOXIC CARCINOGENS
25.1.2.3. Time Between Malignant Conversion and Clinical Detection or Death. In its simplest form, the TSCE model does not account for the time between the appearance of the first malignant cell and cancer incidence or mortality (usually termed “lag time”). Several approaches can be taken to address this issue. The simplest one is to assume that the lag time is constant. Although somewhat unrealistic, this approach adds only one more parameter to the model, and the computation of the hazard or survival function at age t from the corresponding TSCE model functions is straightforward. A more realistic approach is to assume a distribution for the lag time, so interindividual variability can be taken into consideration. A gamma distribution has often been used. This adds two more parameters to the model, the mean and the standard deviation of the gamma distribution, and calculation of the hazard and survival functions involves the convolution of the gamma distribution with the TSCE survival and hazard functions. Specifically, S2 ( t − t lag ) ⎧⎪ S (t ) = ⎨ t ⎩⎪1 − ∫0 (1 − S2 (u )) f ( t − u ) du
h2 ( t − t lag ) ⎧⎪ h (t ) = ⎨ t ⎩⎪ ∫0h2 (u ) S2 (u ) f ( t − u ) du S ( t )
(
)
if lag time is constant if lag time is gamma − distributed
(25.15)
if lag time is constant if lag time is gamma − distributed
(25.16)
where h2(t) and S2(t) represent the TSCE model hazard and survival, respectively, and f(·) is the gamma density. In general, the integrals cannot be evaluated in closedform and numerical methods for integration are required. Another approach is to incorporate the dynamics of malignant cells explicitly into the model and assume that the malignant compartment follows also a birth– death–mutation process. The “mutation rate” of the malignant compartment is then interpreted as the rate of cancer detection or death (per malignant cell). Although this approach can be thought as more realistic and complete, it has some disadvantages. Having models with more than one clonal expansion stage makes impossible the calculation of closed-form expressions for the hazard and survival function, so numerical methods are required (Crump et al. 2005; Little 1995; Little and Wright 2003; Meza 2006). In addition, the parameters of the malignant compartment may not be identifiable from incidence or mortality data alone, so explicit modeling of the malignant cell dynamics may just overparameterize the model unnecessarily. Nevertheless, models with more than one clonal expansion have been implemented successfully and used to analyze the incidence and mortality of several cancers (Little 1995; Little and Wright 2003). However, no significant improvement in fit has been obtained when compared with models with a single clonal expansion stage, suggesting that more data on intermediate lesions is required (Little and Li 2007; Luebeck and Moolgavkar 2002; Meza et al. 2008b).
25.1.3.
Multistage Clonal Expansion (MSCE) Model
There is clear biological evidence that more than two mutational events are required for malignant transformation in most tissues (Bodmer et al. 1987; Jones et al. 2008; Luebeck and Moolgavkar 2002). The TSCE model is the simplest model that
25.1. INTRODUCTION
Figure 25.2.
641
MSCE model.
captures the initiation–promotion–conversion paradigm in carcinogenesis. However, this model can be extended to allow for multiple mutational events on the pathway to carcinogenesis. Here we discuss extensions of the TSCE model that allow for multiple mutational events prior to initiation. These extensions have been used in the past to analyze the incidence and mortality of several cancers (Hazelton et al. 2006; Jeon et al. 2006, 2008; Luebeck and Moolgavkar 2002; Luebeck et al. 2008; Meza et al. 2005, 2008b). A graphical representation of a general k-stage model with multiple pre-initiation stages is shown in Figure 25.2. In this model, it is assumed that normal susceptible stem cells (stage 0) have to go through k − 2 pre-initiation stages, before being able to expand clonally (initiated stage). Normal cells become pre-initiated (stage 1) according to a Poisson process with intensity μ0(t)X(t), where X(t) is the number of normal stem cells at age t. Each cell in the pre-initiation stage i can divide (with rate μi(t)) into one stage i and one stage i + 1 cell, i = 1, … , k − 2. Once a cell reaches the initiation stage (k − 1), it expands clonally via a birth–death– mutation process with rates α(t), β(t), μk−1(t), respectively. Figure 25.2 shows a schematic representation of the k-stage MSCE model. 25.1.3.1. Constant Model Parameters. The hazard and survival functions for the MSCE model with constant parameters have been derived previously (Luebeck and Moolgavkar 2002; Meza et al. 2005). In particular, the survival function for the k-stage model, Sk(t) (k ≥ 3), can be calculated iteratively by
(
Sk , j ( t ) = exp − μ k − j ∫0 (1 − Sk , j −1 ( t − u )) du t
)
(25.17)
for j = 3, … , k, where Sk(t) = (Sk,k(t))X and q− p ⎛ ⎞ Sk ,2 ( t ) = ⎜ − pt ⎝ qe − pe − qt ⎟⎠
μ k −2 α
(25.18)
Here, p=
(
1 2 − (α − β − μk −1 ) − (α − β − μ k −1 ) + 4αμ k −1 2
)
(25.19)
642
CHAPTER 25
QUANTITATIVE CANCER RISK ASSESSMENT OF NONGENOTOXIC CARCINOGENS
q=
(
1 2 − (α − β − μk −1 ) + (α − β − μ k −1 ) + 4αμk −1 2
)
(25.20)
Note that pq = −αμk−1 and p + q = −(α − β − μk−1). The hazard of the k-stage model can be obtained from the survival function using Eq. (25.2). In particular,
μ0 X ⎛ qp (e − qt − e − pt ) ⎞ α ⎜⎝ qe − pt − pe − qt ⎟⎠
(25.21)
μ1 α ⎡ ⎛ ⎤ q− p ⎞ h3 ( t ) = μ0 X ⎢1 − ⎜ − pt ⎥ − qt ⎟ ⎣ ⎝ qe − pe ⎠ ⎦
(25.22)
h2 ( t ) =
μ2 ⎛ q− p ⎪⎧ t ⎡⎛ ⎞ h4 ( t ) = μ0 X ⎜ 1 − exp ⎨∫0μ1 ⎢⎜ − p(t − u) ⎟ − pe − q(t − u) ⎠ ⎝ ⎪⎩ ⎣⎝ qe
α
⎤ ⎪⎫⎞ − 1⎥ du ⎬⎟ ⎦ ⎪⎭⎠
(25.23)
25.1.3.2. Age-Dependent Parameters. The hazard and survival functions of the MSCE model with age-dependent parameters cannot be evaluated in closed form. However, they can be calculated using numerical techniques. See Little (1995), Little and Wright (2003) and Crump et al. (2005) for more details.
25.1.4. Modeling Dose–Response in the TSCE and MSCE Models The effects of carcinogenic agents on cancer risk can be evaluated by estimating the dose–response of TSCE or MSCE model parameters. In general, it is not convenient to assume a priori a specific mechanism of action of an agent. Therefore, in any analysis we should allow in principle all identifiable parameters in a model to be dose-dependent and should then use statistical analysis to identify the relevant effects. In the following we describe a standard approach to modeling the dose– response in biologically based models. 25.1.4.1. Single Agent. Let d(s) denote the exposure of an individual to a specific agent at age s. We can then assume that each of the identifiable parameters of the model has a dose–response given by
(
θ car ( d ( s )) = θ 1 + θ c d ( s ) e θ
)
(25.24)
where θ is the background parameter, and θc and θe are the dose–response coefficients. This functional form of the dose–response (power law) is quite general and covers a wide range of possible behaviors, and it should suffice in most cases. For example, previous analyses using the TSCE model of the relationship between smoking (cigarettes per day) and lung cancer incidence or mortality rates suggested that power laws are good models for smoking dose–response (Hazelton et al. 2001, 2005; Meza et al. 2008a). However, other functional forms can be also implemented if supported by additional information.
25.1. INTRODUCTION
643
25.1.4.2. Multiple Agents. When we have information about the exposure levels of several agents, we can assume that the model parameters are functions of all the exposures simultaneously. In particular, let’s assume that we have information about the exposure of an individual to m agents. Let dj(s) denote the exposure to the jth agent at age s, j = 1, … , m. The following dose–response function has been used in the past to evaluate the effects of several agents combined (Hazelton et al. 2001, 2006):
(
)
θ car ( d1 ( s ) , d2 ( s ) , … , dm ( s )) = θ 1 + θ c1d1 ( s ) e1 + θ c 2 d2 ( s ) e 2 + + θ cm dm ( s ) em (25.25) θ
θ
θ
where θ is the background parameter, and θcj and θej are the dose–response coefficients corresponding to the jth agent, j = 1, … , m. This functional form of the dose–response assumes that all the agents act independently from each other at the cellular level. However, this doesn’t necessarily imply that the risks due to exposure to multiple agents predicted by the TSCE or MSCE models would behave in an additive way. 25.1.4.3. Effects on Cell Division and Cell Death Rates. The cell division and cell death rates in the TSCE and MSCE model cannot be estimated independently when using incidence data alone; however, the approximate net cell proliferation rate, − (p + q) = (α − β − μk−1) is an identifiable parameter (Heidenreich et al. 1997). The effects of nongenotoxic agents on cancer risk can be evaluated by estimating the effect of such agents on the cell proliferation (or promotion) rate in the TSCE or MSCE model.
25.1.5.
Analysis of Epidemiological Data
25.1.5.1. Cancer Registry Data. Secular trends in cancer data, like cancer registry data, can be analyzed using traditional age–period–cohort (APC) models. However, these models are limited by a well-known nonidentifiability problem by which any of the estimated trends can be transformed linearly on the log scale (Clayton and Schifflers 1987a,b; Holford 1991). Replacing the age effects in an APC model with the hazard function of a biologically based model solves, at least in principle, the nonidentifiability issue (Luebeck and Moolgavkar 2002). In addition, if exposure data are available, the use of a biologically based model may allow the estimation of the effects of agents exposure on cancer risk. Here we describe how to fit a biologically based model to tabular data while adjusting for calendar year (i.e., period) and birth cohort effects. 25.1.5.1.1. Likelihood Function (Poisson Regression). Assume we have cancer incidence (or mortality) data in tabular form for j = 1, … , J calendar years, covering i = 1, … , I age groups. For each age group, the number of cancer cases diagnosed during calendar year j can be assumed to follow a Poisson distribution with mean: Λ i , j = bi , j c j PYi , j h ( ai )
(25.26)
644
CHAPTER 25
QUANTITATIVE CANCER RISK ASSESSMENT OF NONGENOTOXIC CARCINOGENS
where ai is the mean age of the ith age group, bi,j and cj are coefficients that adjust for birth cohort and calendar year effects, respectively, PYi,j is the person years at risk, and h(ai) represents the hazard function of a biologically based model evaluated at age ai. The overall likelihood L for the observed incidence in all age-calendar year groups is given by L=∏ i, j
Λ i ,ij, j e − Λi , j oi , j! o
(25.27)
where oi,j is the number of cases in the ith age group during calendar year j. The likelihood can be used to estimate not only the secular terms (birth cohort and period effects), but also the parameters of the biologically based model. In addition, if the data can be broken further by carcinogen(s) exposure (dose and duration) by age and period, one can also estimate in principle dose–response parameters. 25.1.5.2. Individual Data. In this section we briefly describe how to compute the likelihood function in case we have cancer incidence (or mortality) data with detailed individual exposure information. 25.1.5.2.1. Likelihood Function. For prospective cohort data, the likelihood function is the product of individual likelihoods over all the subjects in the cohort. Assuming that participants are cancer free at their age of entry the study (aei), and that we censor individuals in case of death from other causes or in case they survive cancer until the end of follow-up, the individual likelihoods are given by ⎧ S ′ ( ali; θ ( di )) h ( ali; θ ( di )) S ( ali; θ ( di )) = ⎪− S ( aei; θ ( di )) ⎪ S ( aei; θ ( di )) Li ( ali , aei ; θ ( di )) = ⎨ ⎪ S ( ali; θ ( di )) ⎪ S ( ae ; θ ( d )) i i ⎩
for cancer cases otherwise (25.28)
where ali is the individual’s censoring or failure (cancer diagnosis) age, h ( t; θ ( di )) and S ( t; θ ( di )) denote the hazard and survival function at age t of an individual with exposure history di, respectively, and θ ( di ) denotes the vector of identifiable model parameters given the exposure history di [see Eqs. (25.24) and (25.25)]. Note: The prime denotes derivative with respect to t. The overall likelihood is then L = ∏Li ( ali, aei; θ ( di ))
(25.29)
i
where the product is taken over all the subjects in the cohort(s).
25.1.6. Analysis of Premalignant Lesions Using the TSCE Model In several cancers, premalignant lesions appear long before the clinical onset of the disease. Examples of these are the aberrant crypts foci (ACF) and the adenomatous polyps that appear in the colon and which are considered to be precursors of colorec-
25.1. INTRODUCTION
645
tal carcinoma. Premalignant lesions provide important information about carcinogenesis. They clearly show the multistage nature of the process, and their study has provided a lot of insight about the mechanisms of malignant transformation. Since the experiments of Berenblum and Shubik (1947), the analysis of premalignant lesions in animal experiments constitutes one of the principal frameworks to investigate the carcinogenesis process. In particular, initiation–promotion animal experiments where tumors are induced by chemical agents have been particularly useful for elucidating the specific effects of carcinogens on the occurrence and growth dynamics of premalignant lesions (Moolgavkar et al. 1999). More recently, the analysis of premalignant lesions in humans has provided significant information about the genetic transformations required for malignant transformation in different tissues (Barrett et al. 1999; Fearon and Vogelstein 1990). Moreover, epidemiological analysis of the characteristics of premalignant lesions found during screening has also been crucial in the design of prevention, and intervention strategies against cancer (Brenner et al. 2006; Doria-Rose et al. 2004; Jeon et al. 2008; Macari et al. 2004). One of the big advantages of using multistage models is that they provide a unified framework for analyzing epidemiological and experimental data. Cancer incidence (or mortality) data alone may not be sufficient to evaluate in great detail the mechanisms of carcinogenic agents. For example, the cell division and cell death rates cannot be estimated independently when using cancer incidence data alone. However, information about the number and size distribution of premalignant clones can elucidate the specific role of some agents for cancer development (Moolgavkar and Luebeck 1990). For example, in rat hepatocarcinogenesis experiments, good quantitative information is available on the number and size of enzymealtered foci, which are considered premalignant clones in liver cancer. And this information may allow the estimation of both the cell division and cell death rates in the TSCE model. Several expressions for the number and size of premalignant lesions under different scenarios have been derived in the past (Dewanji et al. 1989, 2005; Jeon et al. 2008; Moolgavkar and Luebeck 1990). We briefly describe some of the mathematical formulas for the number and size distributions of premalignant lesions and the total number of premalignant cells under the TSCE model. 25.1.6.1.
Number and Size Distribution
25.1.6.1.1. Constant Parameters. Let N(t) represent the number of nonextinct premalignant lesions at time t. Then N(t) is a Poisson random variable with expectation Λ(t), where (Dewanji et al. 1989) Λ (t ) =
μ0 X α
β ⎡ ⎛ ⎞⎤ ⎢ ln ⎜⎝ β − α p(t ) ⎟⎠ ⎥ ⎣ ⎦
(25.30)
and ⎧ β − β e −(α −β )t ⎪⎪ α − β e −(α −β )t p (t ) = ⎨ ⎪ αt ⎪⎩ 1 + α t
if α ≠ β (25.31) if α = β
646
CHAPTER 25
QUANTITATIVE CANCER RISK ASSESSMENT OF NONGENOTOXIC CARCINOGENS
For a nonextinct premalignant lesion, say W(t), at time t, the probability of it having m ≥ 1 cells is (Dewanji et al. 1989) ⎛α ⎞ ⎜⎝ β p ( t )⎟⎠ Pr [W ( t ) = m W ( t ) > 0 ] = β ⎛ ⎞ m ln ⎜ ⎝ β − α p ( t ) ⎟⎠ m
E [W ( t ) W ( t ) > 0 ] =
e(α −β )t − 1 α ⎡ ⎤ ⎢ α − β ⎣ ln ( β ( β − α p ( t ))) ⎥⎦
(25.32)
(25.33)
25.1.6.1.2. Minimum Detection Threshold. Since in reality, premalignant lesions can be detected only when they are larger than a certain size, Dewanji et al. also derived expressions for the number and size of detectable premalignant lesions assuming a detection threshold size n0 (Dewanji et al. 1989). Assuming that α and β are constant, the number of detectable premalignant lesions at time t, say N*(t), is a Poisson random variable with expectation Λ*(t): Λ* ( t ) = ∫0μ0 (u ) X (u ) (1 − p* ( t − u )) du t
(25.34)
where p*(t − u) is the probability that the size of the lesion is less than or equal to n0 at time t, and is given by ⎛α ⎞ p* ( t − u ) = 1 − (1 − p ( t − u )) ⎜ p ( t − u )⎟ ⎝β ⎠
n0
(25.35)
where p(t − u) is the same as in Eq. (25.31). The distribution of the size of a detectable premalignant lesion at time t is given by Pr[W ( t ) = m W ( t ) > n0 ] =
1 t ⎛α ⎞ λ* (u ) ⎜ p ( t − u )⎟ ∫ 0 ⎝ ⎠ Λ* ( t ) β
m − n0 −1
⎛ α ⎞ ⎜⎝ 1 − β p ( t − u )⎟⎠ du
(25.36)
where λ*(u) = μ0(u)X(u)(1 − p*(t − u)). The expected size of a detectable premalignant lesion is given by E [W ( t ) W ( t ) > n0 ] = n0 +
0 1 t ⎛α ⎞ (α −β )(t −u) μ u X u p t − u du ( ) ( ) ( ) ∫ 0 ⎜⎝ β ⎟⎠ e Λ* ( t ) 0
n
(25.37)
25.1.6.1.3. General Time-Dependent Parameters. Let W(u, t) denote the size of the premalignant lesion at time t, given that the first premalignant cell in the lesion is generated at time u ≤ t. The model assumes that a premalignant lesion grows according to a birth–death process with rates α(u, t) and β(u, t). Let t
(α (u ,s )− β (u ,s )) ds g (u, t ) = e ∫u t G (u, t ) = ∫uα (u, s ) g (u, s ) ds −
(25.38) (25.39)
25.1. INTRODUCTION
647
The number of nonextinct premalignant lesions follows Poisson distribution with expectation (Luebeck and Moolgavkar 1991): Λ ( t ) = ∫0 t
μ0 (u ) X (u ) du G (u, t ) + g (u, t )
(25.40)
And the probability of having m ≥ 1 cells in a nonextinct premalignant lesion at time t is Pr [W ( t ) = m W ( t ) > 0 ] =
m ⎡ g (u, t ) ⎛ 1 t G (u, t ) ⎞ ⎤ λ u , t ( ) ⎢ ∫ ⎜ ⎟ ⎥ du Λ (t ) 0 ⎣ G (u, t ) ⎝ G (u, t ) + g (u, t ) ⎠ ⎦
(25.41)
where
λ (u, t ) =
μ0 (u ) X (u ) G (u, t ) + g (u, t )
(25.42)
25.1.6.1.4. Piecewise Constant Parameters. If we assume that the model parameters are piecewise constant, the number of nonextinct premalignant clones is Poisson distributed with mean (Dewanji et al. 1989; Kopp-Schneider 1992; Luebeck et al. 2005):
μ0, j X ⎡ α j e{(α j −β j )(t j −t j −1 )} − β j ⎤ ln ⎢ ⎥ α j − β j j =1 α j ⎦ ⎣ n
Λ (t ) = ∑
(25.43)
where n is the number of age–periods with different parameter values before age t; [tj−1, tj], j = 1, … , n, denote the endpoints of the jth age–period; t0 = 0, and μ0,j, αj, βj denote the parameter values during the jth age–period. The probability of having m ≥ 1 cells in a nonextinct premalignant lesion at time t satisfies (Dewanji et al. 1989; Kopp-Schneider 1992; Luebeck et al. 2005) Pr [W ( t ) = m W ( t ) > 0 ] =
1 n ∑p j (m), Λ ( t ) j =1
(25.44)
where − α −β t −t 1 ⎪⎧⎛ μ0, j X μ0, j −1 X ⎞ α j − α j e{ ( j j )( j j −1 )} ⎪⎫ p j ( m ) = ⎨⎜ − ⎬ , m ⎪⎩⎝ α j α j −1 ⎟⎠ α j − β j e{−(α j −β j )(t j −t j −1 )} ⎪⎭ m
j = 1, … , n (25.45)
and μ0,0X/α0 ≡ 0. The parameters α j and β j are defined recursively by
α n = α n α j = α j −
(α j − β j ) (α − α e{−(α (α j +1 − β j +1 ) j +1 j +1
β n = β n n β j = α j − (α j − β j ) ∏ e{−(αl −βl )(tl −tl −1 )}
(25.46) j +1 − β j +1
)(t j +1 − t j )}
)
(25.47) (25.48) (25.49)
l = j +1
j = n − 1, … , 1
(25.50)
648
CHAPTER 25
QUANTITATIVE CANCER RISK ASSESSMENT OF NONGENOTOXIC CARCINOGENS
25.1.6.2. Distribution of Total Number of Premalignant Cells. Let Y(t) denote the total number of premalignant cells (in all lesions present) at time t. Let Pm(t) denote the probability that Y(t) = m. Then (Dewanji et al. 2005) P0 ( t ) = exp ( − Λ ( t )) m −1 m−i Pm ( t ) = ∑ Pi ( t ) pm −i ( t ) m i =0
(25.51) (25.52)
where pk ( t ) = ∫0μ0 (u ) X (u ) t
g (u, t ) G (u, t ) 1 ⎛ ⎞ du ⎜ ⎝ G (u, t ) G (u, t ) + g (u, t ) G (u, t ) + g (u, t ) ⎟⎠ k
(25.53)
and Λ(t) denotes the expected number of nonextinct premalignant lesions at time t. The mean and variance of Y(t) are t t E [Y ( t )] = ∫0μ0 (u ) X (u ) exp ⎡⎣ ∫u (α (u, s ) − β (u, s )) ds ⎤⎦ du
(25.54)
and Var [Y ( t )] = ∫0μ0 (u ) X (u ) t
2G (u, t ) + g (u, t ) du g 2 (u, t )
(25.55)
25.1.6.2.1. Constant Parameters. When the parameters X, μ0, α, and β are constant, we can evaluate explicitly the distribution of Y(t): Pr [Y ( t ) = n ] =
Γ ( μ0 X α + n ) (1 − αζ )μ0 X α (αζ )n Γ ( n + 1) Γ ( μ0 X α )
(25.56)
where
ζ=
e(α −β )t − 1 α e(α −β )t − β
(25.57)
25.1.6.2.2. Distribution of Total Number of Premalignant Cells Conditional on No Prior Malignancies. To model screening for premalignant lesions in asymptomatic individuals with no prior history of cancer, Jeon et al. (2008) derived mathematical expressions for the size distribution of premalignant lesions, conditional on no prior malignancy in the tissue of interest. Let Z(t) be the indicator variable for clinical detection of cancer. For n ≥ 0, assuming constant parameters X, μ0, α, β, the size distribution of the total number of premalignant cells conditioned on no malignancy satisfies (Jeon et al. 2008) P*[Y ( t ) = n ] ≡ Pr [Y ( t ) = n Z ( t ) = 0, Y ( 0 ) = 0 ] =
(25.58)
Γ ( μ 0 X α + n) (1 − αζ *)μ0 X α (αζ *)n (25.59) Γ ( n + 1) Γ ( μ0 X α )
where
ζ* =
e − pt − e − qt (q + α ) e− pt − ( p + α ) e− qt
(25.60)
25.2. SOME EXAMPLES AND APPLICATIONS
649
and 1⎡ 2 − (α − β − μ1 ) − (α − β − μ1 ) + 4αμ1 ⎤ ⎦ 2⎣ 1 2 q = ⎡ − (α − β − μ1 ) + (α − β − μ1 ) + 4αμ1 ⎤ ⎦ 2⎣ p=
(25.61) (25.62)
25.1.6.3. Other Developments. In addition to the formulas presented here, expressions for the joint analysis of premalignant and malignant lesions under the TSCE model have been derived previously (De Gunst and Luebeck 1994; Dewanji et al. 1991). Furthermore, Dewanji et al. (1999) derived expressions for the analysis of longitudinal data on the number and size of premalignant clones. In particular, the joint distribution of the number of premalignant clones observed at different points in time in the same subject was derived. In addition, De Gunst and Luebeck (1994, 1998) developed methods to estimate the number and size distribution of premalignant lesions assuming that transectional data are available (stereological problem). We refer the reader to the literature for further details.
25.2.
SOME EXAMPLES AND APPLICATIONS
25.2.1. Smoking, Radon, and Arsenic Exposures and Lung Cancer 25.2.1.1. Smoking and Lung Cancer. The TSCE model has been used to describe the lung cancer incidence and mortality in several cohort and case–control studies (Hazelton et al. 2001, 2005; Heidenreich et al. 2002; Luebeck et al. 1999; Meza et al. 2008a; Schöllnberger et al. 2006). In all of them, smoking-related promotion has been found to be the primary etiological mechanism of lung carcinogenesis. Interestingly, analyses of older datasets have shown also an effect of smoking on lung cancer initiation and no effect on malignant conversion (Hazelton et al. 2001, 2005). However, exactly the opposite has been found in more recent datasets (Heidenreich et al. 2002; Meza et al. 2008a; Schöllnberger et al. 2006). In particular, in a case–control study in Germany Heidenreich et al. (2002) found that smoking has significant effects on promotion and malignant conversion and no effects on initiation. Schöllnberger et al. (2006) found similar patterns in a large cohort study carried out in 10 European countries, and Meza et al. (2008a) found similar patterns in two large cohort studies in the United States, namely, the Nurses’ Health Study and the Health Professional’s Follow-Up Study. On the other hand, Hazelton et al. (2005) also found a limited effect of tobacco on the lung cancer initiation in the CPS-I study; however, no effect on malignant conversion was seen in that cohort. The differences found between older and more recent studies may in part be explained by changes in cigarette composition, with higher levels of nitrosamines in the newer cigarettes acting as promoters, while the lower tar levels may be associated with the lower apparent initiation rate. Additionally, the smoking
650
CHAPTER 25
QUANTITATIVE CANCER RISK ASSESSMENT OF NONGENOTOXIC CARCINOGENS
information available in the older cohorts may not have been detailed enough to detect an effect on malignant conversion. Background rates do not seem to differ significantly between different genders (Meza et al. 2008a; Schöllnberger et al. 2006). However, there is suggestion of minor differences in tobacco-related promotion and malignant conversion (Meza et al. 2008a). Schöllnberger et al. (2006) reported that a common model described lung cancer incidence in males and females in the European Prospective Investigation into Cancer and Nutrition (EPIC) and concluded that gender differences in lung cancer risk are due entirely to differences in smoking habits. However, Meza et al. (2008a) found a somewhat higher effect of smoking on promotion among women and a somewhat lower effect on malignant conversion. In summary, smoking-related promotion appears to be the primary mechanism of lung carcinogenesis. 25.2.1.2. Radiation and Lung Cancer. High-linear-energy-transfer (highLET) or ionizing radiation is known to directly induce genomic defects in critical target genes leading to initiation of premalignant lesions. However, it is not clear that this is the most relevant mechanism of radiation induced carcinogenesis. In particular, analyses of the effects of radon (daughters) exposure, a type of high-LET radiation, on lung cancer risk in animal experiments and in epidemiological studies using the TSCE model have suggested the possibility of a strong effect on promotion of high-LET radiation. In particular, analyses of lung cancers in rats exposed to radon and radon progeny at various dose levels and durations showed significant effects of high-LET radiation on the initiation and promotion of premalignant lesions (Heidenreich et al. 1999; Luebeck et al. 1996; Moolgavkar et al. 1990a). Analysis of the lung cancer incidence in the Colorado Plateau Uranium Miners (CPUM) cohort showed a significant effect of radon exposure on the cell proliferation rate of premalignant lesions (Luebeck et al. 1999). Analyses of the lung cancer incidence in the Chinese Tin Miners (CTM) cohort showed significant effects of radon exposure on both promotion and malignant conversion rates (Hazelton et al. 2001). Interestingly, no significant effect of radon exposure on lung cancer initiation was found in either the CPUM or the CTM cohort. Analysis of low-dose X-ray, gamma, and other low-linear-energy-transfer (low-LET) radiation in the Canadian National Dose Registry (CNDR) cohort using the TSCE and MSCE models indicated a significant nonlinear effect of this type of radiation on the promotion rate as well (Hazelton et al. 2006). The so-called inverse dose-rate or protraction effect, where a protracted dose produces in the long run a larger effect than an acute exposure to the same dose, has been found in both the analysis of high-LET (CPUM and CTM) and low-LET (CNDR) radiation effects on lung cancer risk using the TSCE model. In terms of the model, the inverse dose-rate effect can be explained by the increase in promotion that appears to saturate with increasing dose rate. Interestingly, as shown in Figure 25.3, under some scenarios the TSCE model predicts a transition from a direct doserate effect, seen for a few years after exposure, to an inverse dose-rate effect thereafter, a phenomenon that is difficult to detect and address using conventional epidemiological methods (Hazelton et al. 2006; Little et al. 2008).
25.2. SOME EXAMPLES AND APPLICATIONS
651
0.0020 0.0015 0.0010
100 mSv/yr for 1 yr
0.0005
Excess Probability
0.0025
0.0030
Transition from direct to inverse dose rate or protraction effect using the full Canadian male TSCE model
50 mSv/yr
0.0000
for 2 yrs
50
51
52
53
54
55
56
age
Figure 25.3. Transition from direct to inverse dose rate or protraction effect using the full Canadian male TSCE model. [Reproduced from Hazelton et al. (2006).]
In summary, all these studies strongly suggest that in addition to its wellknown genotoxic effects, radiation may also be a cancer promoter. 25.2.1.3. Interaction Between Different Agents. Interaction of smoking with other lung carcinogens has been studied extensively using the TSCE model. In particular, analysis of the lung cancer incidence in the CPUM found significant interaction between smoking and radiation (between additive and multiplicative effects). Strong interactions between smoking, radiation, and arsenic were found in the CTM cohort analysis. Figure 25.4 shows the exposure pattern for a typical subject in the CTM cohort. In this analysis, the interaction between agents was modeled using dose-response functions like formula (25.25). Figure 25.5 shows the estimated attributable risks to tobacco, radon, and arsenic in the CTM cohort. Tobacco and arsenic dominate the attributable risk of lung cancer. Interestingly, all three agents appear to increase the cell division, death, and malignant conversion rates of premalignant cells; however, significant differences in net cell proliferation rates in response to the different exposures were found.
652
CHAPTER 25
QUANTITATIVE CANCER RISK ASSESSMENT OF NONGENOTOXIC CARCINOGENS
Exposures For ID #19530
As = 0.415 mg./m3
-Arbitrary Scale -
As = 0.333 mg./m3 Rn = 2.56 WLM/month Rn = 1.97 WLM/month Rn = 2.24 WLM/month
Cig = 19 cig/day Pipe = 15 liang/month
Age 0 11 15 Born First work, 1917 Begin smoking
26
29
34
51 Stop Smoking
68.2 Death - Other cause -
Figure 25.4. Exposure pattern for a typical subject in the CTM cohort. [Reproduced from Hazelton et al. (2001).]
Arsenic = 15.8% 133 cases Radon + Arsenic = 11.0% 93 cases
Radon-5.5% 46 cases
Radon + Arsenic + Tobacco = 8.7% 74 cases
Radon + Tobacco = 9.2% 77 cases
Background = 9.2% 73 cases
Arsenic + Tobacco = 19.7% 166 cases
Tobacco = 21.4% 180 cases
Total Deaths = 842
Figure 25.5. Estimated attributable risks to tobacco, radon, and arsenic in the CTM cohort. [Modified from Hazelton et al. (2001).]
25.2. SOME EXAMPLES AND APPLICATIONS
25.2.2.
653
Folate and Colorectal Cancer
Multistage carcinogenesis models are useful in investigating agents with potentially antagonistic modes of action (carcinogenic and anticarcinogenic), like folate. Low folate status has been associated with higher mutation rates. Low folate levels in pregnant women are strongly associated with the risk of developmental neural tube defects, spina bifida, and anencephaly (Czeizel and Dudás 1992; Mason et al. 2007). For this reason, folate supplementation through fortification of enriched uncooked cereal grains with folic acid began in the United States and Canada in 1996, leading to a significant decrease in the incidence of neural tube defects in newborns. Low levels of folate have also been associated with increased risk of colon, cervix, esophagus, and breast cancer (McCullough and Giovannucci 2004). For this reason, folate has been considered as a potential chemopreventive agent (Brekelmans 2003; Kelloff et al. 1995; Majumdar et al. 2004). However, experimental and epidemiological evidence also suggest that folate increases cell proliferation and may act as a promotor of premalignant lesions (Cole et al. 2007; Farber 1949). In particular, folic acid supplementation has been associated with a recent increase in the colon cancer rates in the United States and Canada (Mason et al. 2007). What are the overall effects of agents that may have antagonistic mechanisms, like folate, and how can we measure such effects? Multistage models allow us to investigate such questions. For example, Luebeck et al. (2008) recently evaluated the possible trade-offs between (a) the protective effects of folate due to decreased mutation rates in the colon and (b) the possible harmful effects due to increased cell proliferation, using a four-stage model for colon carcinogenesis. The model predicts that colon cancer rates should be expected to increase due to folic acid supplementation, unless the supplementation is started early in life. The decrease in mutation rates due to folate supplementation would reduce cancer risk mainly by slowing down the initiation rate of adenomas, the premalignant lesions associated with a large majority of colorectal cancers. However, if adenomas are already present at the time when supplementation starts, the risk due to the promotion of preexisting adenomas seems to outweight the benefits due to the reduction of mutation rates. Figure 25.6 shows the predicted excess number of colon cancer cases due to folate supplementation as a function of age for different supplementation starting ages (Luebeck et al. 2008). Clearly the only scenario where supplementation reduces colon cancer risk is the one where supplementation starts very early in life. Further clinical and experimental studies are definitely needed to corroborate or refute these findings; however, this type of analysis provides insights that would not be possible without the use of a mechanistic model.
25.2.3.
Enzyme-Altered Foci in the Rat Liver
The analysis of experiments that measure the number and size of premalignant lesions in animals exposed to carcinogenic agents is possibly the “purest” application of multistage models to the investigation of promotion. In particular, the mathematical expressions discussed in Section 25.1.6 have been used extensively to analyze experimental data on enzyme-altered foci (EAF) in the rat liver. EAFs are clonal
654
CHAPTER 25
QUANTITATIVE CANCER RISK ASSESSMENT OF NONGENOTOXIC CARCINOGENS
Figure 25.6. Predicted excess number of colon cancer cases due to folate supplementation as a function of age for different supplementation ages. [Reproduced from Luebeck et al. (2008).]
outgrowths in the rat liver that presumably consist of premalignant cells expressing a particular aberrant enzyme phenotype. In a typical experiment, rats are exposed briefly to an initiator followed by long-term exposure to a promotor. Groups of rats exposed to varying doses of either initiator or promotor are sacrificed at various times, and their livers are sectioned and examined for enzyme-altered foci. The data consist of (a) information on the number of (two-dimensional) foci seen per unit area in the liver cross section and (b) the sizes (radii in microns) of each of the two-dimensional sections of foci. Given that the data are two-dimensional, but the foci are three-dimensional, stereological methods are required to fit the model to the data. In particular, methods to transform the three-dimensional quantities, number and size (the later given in cell numbers) distributions, of foci derived from the TSCE model into two-dimensional expressions (for the number of foci per unit area and the foci size radii in microns) have been developed (De Gunst and Luebeck 1994, 1998). Details of the likelihood construction for this type of experiments are described by Moolgavkar et al. (1990b), Luebeck et al. (1991), Moolgavkar et al. (1996), and Luebeck et al. (2000). In the first paper, a single-agent regimen was analyzed. Rats were chronically exposed to N-nitrosomorpholine (NNM) in their drinking water. The objective was to investigate the initiating and promoting effects of NNM exposure. In Moolgavkar et al. (1996) and Luebeck et al. (2000), initiation–promotion (two-agent) protocols were analyzed. The objective was to investigate the promoting activities of various polychlorinated biphenyls (PCBs) and dioxin. One of the interesting conclusions of the analyses was that a large number (perhaps 90%) of initiated cells never develop into foci, but become extinct. In the last paper, the effects of tetrachlorodibenzo-p-dioxin (TCDD) on initiation and promotion of EAFs were
REFERENCES
655
investigated. Interestingly, the estimated promotional effect of TCDD was accounted by the selective outgrowth of glutathione transferase-P positive clones in response to inhibitory effects of TCDD on apoptosis.
25.3.
CONCLUDING REMARKS
In this chapter, we discussed the analysis of the effects of carcinogenic agents on cancer risk using multistage carcinogenesis models. Multistage carcinogenesis models are clearly a simplification of the underlying biology, and their predictions should be always used with caution. However, multistage models, particularly those explicitly incorporating clonal expansion of intermediate cells, provide a powerful and unifying framework for analyses of premalignant and malignant lesions in experimental and epidemiologic data. The use of such models for data analyses can make predictions and generate hypotheses that traditional statistical approaches cannot. Ideally, the investigation of carcinogenic agents should be an iterative (cyclic) process, where mechanistic models based on current biological knowledge are fitted to data, leading to new hypotheses about specific mechanisms. These hypotheses suggest the design of new experiments, which are in turn tested in the laboratory and provide new information that can be used to refine the mechanistic models.
REFERENCES Armitage, P., and Doll, R. (1954). The age distribution of cancer and a multi-stage theory of carcinogenesis. Br J Cancer 8, 1–12. Armitage, P., and Doll, R. (1957). A two-stage theory of carcinogenesis in relation to the age distribution of human cancer. Br J Cancer 11, 161–169. Barrett, M. T., Sanchez, C. A., Prevo, L. J., Wong, D. J., Galipeau, P. C., Paulson, T. G., Rabinovitch, P. S., and Reid, B. J. (1999). Evolution of neoplastic cell lineages in Barrett oesophagus. Nat Genet 22, 106–109. Berenblum, I., and Shubik, P. (1947). A new, quantitative, approach to the study of the stages of chemical carcinogenesis in the mouse’s skin. Br J Cancer 1, 383–391. Bodmer, W. F., Bailey, C. J., Bodmer, J., Bussey, H. J., Ellis, A., Gorman, P., Lucibello, F. C., Murday, V. A., Rider, S. H., and Scambler, P. (1987). Localization of the gene for familial adenomatous polyposis on chromosome 5. Nature 328, 614–616. Brekelmans, C. T. (2003). Risk factors and risk reduction of breast and ovarian cancer. Curr Opin Obstet Gynecol 15, 63–68. Brenner, H., Chang-Claude, J., Seiler, C. M., Sturmer, T., and Hoffmeister, M. (2006). Does a negative screening colonoscopy ever need to be repeated? Gut 55, 1145–1150. Clayton, D., and Schifflers, E. (1987a). Models for temporal variation in cancer rates. I: Age–period and age–cohort models. Stat Med 6, 449–467. Clayton, D., and Schifflers, E. (1987b). Models for temporal variation in cancer rates. II: Age–period– cohort models. Stat Med 6, 469–481. Cole, B. F., Baron, J. A., Sandler, R. S., Haile, R. W., Ahnen, D. J., Bresalier, R. S., McKeown-Eyssen, G., Summers, R. W., Rothstein, R. I., Burke, C. A., Snover, D. C., Church, T. R., Allen, J. I., Robertson, D. J., Beck, G. J., Bond, J. H., Byers, T., Mandel, J. S., Mott, L. A., Pearson, L. H., Barry, E. L., Rees, J. R., Marcon, N., Saibil, F., Ueland, P. M., and Greenberg, E. R. (2007). Folic acid for the prevention of colorectal adenomas: A randomized clinical trial. JAMA 297, 2351–2359.
656
CHAPTER 25
QUANTITATIVE CANCER RISK ASSESSMENT OF NONGENOTOXIC CARCINOGENS
Crump, K. S., Subramaniam, R. P., and Van Landingham, C. B. (2005). A numerical solution to the nonhomogeneous two-stage MVK model of cancer. Risk Anal 25, 921–926. Czeizel, A. E., and Dudás, I. (1992). Prevention of the first occurrence of neural-tube defects by periconceptional vitamin supplementation. N Engl J Med 327, 1832–1835. De Gunst, M. C., and Luebeck, E. G. (1994). Quantitative analysis of two-dimensional observations of premalignant clones in the presence or absence of malignant tumors. Math Biosci 119, 5–34. De Gunst, M. C., and Luebeck, E. G. (1998). A method for parametric estimation of the number and size distribution of cell clusters from observations in a section plane. Biometrics 54, 100–112. Dewanji, A., Goddard, M. J., Krewski, D., and Moolgavkar, S. H. (1999). Two stage model for carcinogenesis: Number and size distributions of premalignant clones in longitudinal studies. Math Biosci 155, 1–12. Dewanji, A., Luebeck, E. G., and Moolgavkar, S. H. (2005). A generalized Luria–Delbruck model. Math Biosci 197, 140–152. Dewanji, A., Moolgavkar, S. H., and Luebeck, E. G. (1991). Two-mutation model for carcinogenesis: Joint analysis of premalignant and malignant lesions. Math Biosci 104, 97–109. Dewanji, A., Venzon, D. J., and Moolgavkar, S. H. (1989). A stochastic two-stage model for cancer risk assessment. II. The number and size of premalignant clones. Risk Anal 9, 179–187. Doria-Rose, V. P., Levin, T. R., Selby, J. V., Newcomb, P. A., Richert-Boe, K. E., and Weiss, N. S. (2004). The incidence of colorectal cancer following a negative screening sigmoidoscopy: Implications for screening interval. Gastroenterology 127, 714–722. Farber, S. (1949). Some observations on the effect of folic acid antagonists on acute leukemia and other forms of incurable cancer. Blood 4, 160–167. Fearon, E. R., and Vogelstein, B. (1990). A genetic model for colorectal tumorigenesis. Cell 61, 759–767. Hazelton, W. D., Clements, M. S., and Moolgavkar, S. H. (2005). Multistage carcinogenesis and lung cancer mortality in three cohorts. Cancer Epidemiol Biomarkers Prev 14, 1171–1181. Hazelton, W. D., Luebeck, E. G., Heidenreich, W. F., and Moolgavkar, S. H. (2001). Analysis of a historical cohort of Chinese tin miners with arsenic, radon, cigarette smoke, and pipe smoke exposures using the biologically based two-stage clonal expansion model. Radiat Res 156, 78–94. Hazelton, W. D., Moolgavkar, S. H., Curtis, S. B., Zielinski, J. M., Ashmore, J. P., and Krewski, D. (2006). Biologically based analysis of lung cancer incidence in a large Canadian occupational cohort with low-dose ionizing radiation exposure, and comparison with Japanese atomic bomb survivors. J Toxicol Environ Health Part A 69, 1013–1038. Heidenreich, W. F., Jacob, P., Paretzke, H. G., Cross, F. T., and Dagle, G. E. (1999). Two-step model for the risk of fatal and incidental lung tumors in rats exposed to radon. Radiat Res 151, 209–217. Heidenreich, W. F., Luebeck, E. G., and Moolgavkar, S. H. (1997). Some properties of the hazard function of the two-mutation clonal expansion model. Risk Anal 17, 391–399. Heidenreich, W. F., Wellmann, J., Jacob, P., and Wichmann, H. E. (2002). Mechanistic modelling in large case–control studies of lung cancer risk from smoking. Stat Med 21, 3055–3070. Holford, T. R. (1991). Understanding the effects of age, period, and cohort on incidence and mortality rates. Annu Rev Public Health 12, 425–457. Jeon, J., Luebeck, E. G., and Moolgavkar, S. H. (2006). Age effects and temporal trends in adenocarcinoma of the esophagus and gastric cardia (United States). Cancer Causes Control 17, 971–981. Jeon, J., Meza, R., Moolgavkar, S. H., and Luebeck, E. G. (2008). Evaluation of screening strategies for pre-malignant lesions using a biomathematical approach. Math Biosci 213, 56–70. Jones, S., Chen, W. D., Parmigiani, G., Diehl, F., Beerenwinkel, N., Antal, T., Traulsen, A., Nowak, M. A., Siegel, C., Velculescu, V. E., Kinzler, K. W., Vogelstein, B., Willis, J., and Markowitz, S. D. (2008). Comparative lesion sequencing provides insights into tumor evolution. Proc Natl Acad Sci USA 105, 4283–4288. Kelloff, G. J., Boone, C. W., Crowell, J. A., Nayfield, S. G., Hawk, E., Steele, V. E., Lubet, R. A., and Sigman, C. C. (1995). Strategies for phase II cancer chemoprevention trials: Cervix, endometrium, and ovary. J Cell Biochem Suppl 23, 1–9. Kopp-Schneider, A. (1992). Birth–death processes with piecewise constant rates. Stat Probabil Lett 13, 121–127.
REFERENCES
657
Little, M. P. (1995). Are two mutations sufficient to cause cancer? Some generalizations of the twomutation model of carcinogenesis of Moolgavkar, Venzon, and Knudson, and of the multistage model of Armitage and Doll. Biometrics 51, 1278–1291. Little, M. P., Heidenreich, W. F., Moolgavkar, S. H., Schöllnberger, H., and Thomas, D. C. (2008). Systems biological and mechanistic modelling of radiation-induced cancer. Radiat Environ Biophys 47, 39–47. Little, M. P., and Li, G. (2007). Stochastic modelling of colon cancer: Is there a role for genomic instability? Carcinogenesis 28, 479–487. Little, M. P., and Wright, E. G. (2003). A stochastic carcinogenesis model incorporating genomic instability fitted to colon cancer data. Math Biosci 183, 111–134. Luebeck, E. G., Buchmann, A., Schneider, D., Moolgavkar, S. H., and Schwarz, M. (2005). Modulation of liver tumorigenesis in Connexin32-deficient mouse. Mutat Res 570, 33–47. Luebeck, E. G., Buchmann, A., Stinchcombe, S., Moolgavkar, S. H., and Schwarz, M. (2000). Effects of 2,3,7,8-tetrachlorodibenzo-p-dioxin on initiation and promotion of GST-P-positive foci in rat liver: A quantitative analysis of experimental data using a stochastic model. Toxicol Appl Pharmacol 167, 63–73. Luebeck, E. G., Curtis, S. B., Cross, F. T., Moolgavkar, S. H., and Curtis, S. B. (1996). Two-stage model of radon-induced malignant lung tumors in rats: effects of cell killing. Radiat Res 145, 163–173. Luebeck, E. G., and Hazelton, W. D. (2002). Multistage carcinogenesis and radiation. J Radiol Prot 22, A43–A49. Luebeck, E. G., Heidenreich, W. F., Hazelton, W. D., Paretzke, H. G., and Moolgavkar, S. H. (1999). Biologically based analysis of the data for the Colorado uranium miners cohort: Age, dose and doserate effects. Radiat Res 152, 339–351. Luebeck, E. G., and Moolgavkar, S. H. (1991). Stochastic analysis of intermediate lesions in carcinogenesis experiments. Risk Anal 11, 149–157. Luebeck, E. G., and Moolgavkar, S. H. (2002). Multistage carcinogenesis and the incidence of colorectal cancer. Proc Natl Acad Sci USA 99, 15095–15100. Luebeck, E. G., Moolgavkar, S. H., Buchmann, A., and Schwarz, M. (1991). Effects of polychlorinated biphenyls in rat liver: Quantitative analysis of enzyme-altered foci. Toxicol Appl Pharmacol 111, 469–484. Luebeck, E. G., Moolgavkar, S. H., Liu, A. Y., Boynton, A., and Ulrich, C. M. (2008). Does folic acid supplementation prevent or promote colorectal cancer? Results from model-based predictions. Cancer Epidemiol Biomarkers Prev 17, 1360–1367. Macari, M., Bini, E. J., Jacobs, S. L., Naik, S., Lui, Y. W., Milano, A., Rajapaksa, R., Megibow, A. J., and Babb, J. (2004). Colorectal polyps and cancers in asymptomatic average-risk patients: Evaluation with CT colonography. Radiology 230, 629–636. Majumdar, A. P., Kodali, U., and Jaszewski, R. (2004). Chemopreventive role of folic acid in colorectal cancer. Front Biosci 9, 2725–2732. Mason, J. B., Dickstein, A., Jacques, P. F., Haggarty, P., Selhub, J., Dallal, G., and Rosenberg, I. H. (2007). A temporal association between folic acid fortification and an increase in colorectal cancer rates may be illuminating important biological principles: A hypothesis. Cancer Epidemiol Biomarkers Prev 16, 1325–1329. McCullough, M. L., and Giovannucci, E. L. (2004). Diet and cancer prevention. Oncogene 23, 6349–6364. Meza, R. (2006). Some Extensions and Applications of Multistage Carcinogenesis Models. University of Washington. Meza, R., Hazelton, W. D., Colditz, G. A., and Moolgavkar, S. H. (2008a). Analysis of lung cancer incidence in the nurses’ health and the health professionals’ follow-up studies using a multistage carcinogenesis model. Cancer Causes Control 19, 317–328. Meza, R., Jeon, J., Moolgavkar, S. H., and Luebeck, E. G. (2008b). Age-specific incidence of cancer: Phases, transitions, and biolog ical implications. Proc Natl Acad Sci USA 105, 16284–16289. Meza, R., Luebeck, E. G., and Moolgavkar, S. H. (2005). Gestational mutations and carcinogenesis. Math Biosci 197, 188–210. Moolgavkar, S. H., Cross, F. T., Luebeck, G., and Dagle, G. E. (1990a). A two-mutation model for radon-induced lung tumors in rats. Radiat Res 121, 28–37.
658
CHAPTER 25
QUANTITATIVE CANCER RISK ASSESSMENT OF NONGENOTOXIC CARCINOGENS
Moolgavkar, S. H., Dewanji, A., and Venzon, D. J. (1988). A stochastic two-stage model for cancer risk assessment. I. The hazard function and the probability of tumor. Risk Anal 8, 383–392. Moolgavkar, S. H., and Knudson, A. G. (1981). Mutation and cancer: A model for human carcinogenesis. J Natl Cancer Inst 66, 1037–1052. Moolgavkar, S. H., Krewski, D., and Schwarz, M. (1999). Mechanisms of carcinogenesis and biologically based models for estimation and prediction of risk. IARC Sci Publ 131, 179–237. Moolgavkar, S. H., and Luebeck, E. G. (1990). Two-event model for carcinogenesis: Biological, mathematical, and statistical considerations. Risk Anal 10, 323–341. Moolgavkar, S. H., Luebeck, E. G., Buchmann, A., and Bock, K. W. (1996). Quantitative analysis of enzyme-altered liver foci in rats initiated with diethylnitrosamine and promoted with 2,3,7,8tetrachlorodibenzo-p-dioxin or 1,2,3,4,6,7,8-heptachlorodibenzo-p-dioxin. Toxicol Appl Pharmacol 138, 31–42. Moolgavkar, S. H., Luebeck, E. G., de Gunst, M., Port, R. E., and Schwarz, M. (1990b). Quantitative analysis of enzyme-altered foci in rat hepatocarcinogenesis experiments—I. Single agent regimen. Carcinogenesis 11, 1271–1278. Moolgavkar, S. H., and Venzon, D. J. (1979). Two-event models for carcinogenesis: incidence curves for childhood and adult tumors. Math Biosci 47, 55–77. Muller, H. J. (1951). Radiation damage to the genetic material. Sci Progress 7, 93–165 and 481–493. Nordling, C. O. (1952). Theories and statistics of cancer. Nord Med 47, 817–820. Nordling, C. O. (1953). A new theory on cancer-inducing mechanism. Br J Cancer 7, 68–72. Schöllnberger, H., Manuguerra, M., Bijwaard, H., Boshuizen, H., Altenburg, H. P., Rispens, S. M., Brugmans, M. J. P., and Vineis, P. (2006). Analysis of epidemiological cohort data on smoking effects and lung cancer with a multi-stage cancer model. Carcinogenesis 27, 1432–1444.
CH A P TE R
26
NONLINEAR LOW-DOSE EXTRAPOLATIONS Ari S. Lewis Barbara D. Beck
26.1.
INTRODUCTION
The use of high-dose toxicity data to inform risk management decisions has been one of the major challenges of modern-day human health risk assessment. While characterization of chemical toxicity in animals at high doses is an important part of hazard identification, using such data to predict health effects in humans at typical low environmental exposures is highly uncertain. The general view that a high-dose exposure simply causes the same effect as a low-dose exposure, but with a higher incidence, has been demonstrated as overly simplistic and incorrect for many chemicals (Slikker et al. 2004). Noncancer and cancer risk assessment procedures diverged in the early 1980s, when risk assessment paradigms were first taking form (Bogdanffy et al. 2001). A guiding principle of noncancer risk assessment is the understanding that chemicals have a threshold of toxicity; that is, below a certain level, a chemical is unlikely to cause adverse effects. This view is premised on the idea that living organisms contain an exquisite amount of redundancy in biological processes (e.g., multiples of the same organelle within a cell, multiple cells of similar type in tissues, etc.), and the flexible intra- and intercellular communication pathways that function to restore body homeostasis (Rhomberg 2004). As such, a single interaction between a compound molecule and its molecular target is unlikely to produce any measurable change in body function (Rhomberg 2004). It is only when a certain dose threshold is exceeded and an organism is unable to adapt to external stresses that measurable effects begin to manifest. While the difficulties in identifying a precise threshold have been a challenge for noncancer risk assessment, this uncertainty has been addressed through the use of health-protective approaches (i.e., the application of uncertainty factors) that allow for a margin of error when estimating threshold doses. Traditional cancer risk assessment has taken a different approach. As a default assumption, carcinogens have been considered to act without a threshold of toxicity
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
659
660
CHAPTER 26 NONLINEAR LOW-DOSE EXTRAPOLATIONS
and to have a linear dose–response at low doses. In practice, this means that at doses below the observable dose–response data there is a proportional relationship between exposure and risk, with the implication that any increase in dose can cause a biological change that will increase cancer risk. Based on the premise of low-dose linearity, regulatory agencies have established risk targets aimed to reflect a de minimis cancer risk—that is, a cancer risk too small with which to be concerned. Most U.S. and international agencies have converged on target cancer risk levels ranging from 1 × 10−6 to 1 × 10−4. The actual risk target or range, however, varies by agency (Beck et al. 2007). Determining the chemical exposures associated with these target risk levels is profoundly uncertain. The standard carcinogenic animal bioassays that form the basis of most carcinogenicity assessments have resolution only down to the 1 × 10−2 risk level (Kodell 2001). Even when human data are available, determining chemical exposures associated with incremental cancer risk over background is still a challenge. In the United States, men have about a one-in-two lifetime risk of developing cancer, and women have about a one-in-three lifetime risk (ACS 2009). Epidemiological studies often do not have sufficient power to quantify small incremental increases (1 × 10−6 to 1 × 10−4) over such large background risks. Thus, for the development of toxicity criteria, which often form the basis of regulatory determinations, mathematical extrapolations are almost always required to estimate potential risks at lower doses. The low-dose linearity assumption has largely been considered a healthprotective policy decision reflecting an upper bound on cancer risk estimates. One of the primary hypotheses supporting low-dose linearity has been the concept that a single mutagenic agent is capable of interacting with DNA and transforming a cell. In addition, however, some scientists have proposed that low-dose linearity is supported based on the assumption of additivity to background. The “additivity to background” theory is a hypothetical construct that carcinogens have a linear dose– response relationship if the carcinogenic process operates by the same mechanisms as an existing background process (Crump et al. 1976). The genesis of the linearity assumption and concerns with such a model are discussed in more depth later in this chapter. Based on a biological understanding of carcinogenesis, we propose that lowdose linearity is unlikely for many carcinogenic compounds, even for some of those that act as direct DNA mutagens. As with noncancer insults, biological systems have adaptive strategies that limit the ability of genotoxic lesions to transform cells and develop into tumors. Moreover, there is greater recognition that the biological processes that underlie carcinogenic events can be the same as those that lead to noncancer effects (Bogdanffy et al. 2001). The implication of this is that some carcinogens may also have a threshold (i.e., a dose level below which tumor formation would not be expected to occur). The United States Environmental Protection Agency’s (EPA’s) most recent Guidelines for Carcinogenic Risk Assessment have recognized that for certain carcinogenic chemicals, a threshold may be possible, and a departure from the linearity assumption is warranted (US EPA 2005a). As described in more detail later in this chapter, departure from low-dose linearity requires an understanding of the carcinogenic mode of action (MOA) and that the MOA is
26.2. MECHANISTIC ASPECTS OF NONLINEAR CARCINOGENESIS
661
consistent with a nonlinear dose–response. The EPA defines an MOA as a “sequence of key events and processes, starting with interaction of an agent with a cell, proceeding through operational and anatomical changes, and resulting in cancer formation” (US EPA 2005a). The following sections examine some of these issues more in depth. We focus on general mechanistic considerations that suggest a nonlinear dose–response for the carcinogenic process. Additionally, we discuss the application of nonlinear extrapolation methods, including those harmonized with current noncancer approaches as well those that rely on more flexible, biologically-based methods. We also address the regulatory implications of nonlinear extrapolation, and we give examples where environmental regulatory agencies have used nonlinear extrapolation to develop toxicity criteria.
26.2. MECHANISTIC ASPECTS OF NONLINEAR CARCINOGENESIS An understanding of the carcinogenesis process up through the 1980s resulted in the theory called the clonal model of carcinogenesis. Under this model, scientists proposed that a single molecule of a carcinogen was capable of interacting with a single molecule of DNA to produce a single mutated cell (Ames and Yanofsky 1971; Higginson 1997). This, consequently, could yield a tumorigenic cell. The theory further held that certain mutations could cause the cell to divide uncontrollably and produce a large number of genetically identical daughter cells (clonal cells), which could become invasive and develop into metastatic cancer. Based on this model, the general understanding was that any carcinogenic exposure, no matter how small, would be associated with some incremental risk of developing cancer. While the current understanding of carcinogenesis retains some elements of the clonal growth model (i.e., DNA mutations can cause clonal expansion and ultimately cancer), it is now more widely held that carcinogenesis is a multistep process requiring the accumulation of multiple heritable DNA mutations (see Chapters 5 and 16). Also more widely understood are the cellular protective strategies that act to maintain DNA integrity. Broadly speaking, these strategies include cellular functions that detoxify reactive compounds, repair DNA mutations before the mutations become fixed, and also induce cell death when DNA damage is too extensive to repair. These background and inducible cellular processes that work to limit, repair, or eradicate DNA damage (and thereby ensure mutations are not propagated) provide a biologically plausible explanation for nonlinearity in response to carcinogen exposures.
26.2.1.
Pre-DNA Damage Mechanisms
Many cellular responses protect DNA against initial genotoxic insults. Most notably, a battery of endogenous antioxidants, including glutathione, catalase, superoxide dismutase, and metallathione, can become induced in the presence of free radical formation (Cerutti et al. 1994; Boelsterli 2003). Protection by these enzymes plays
662
CHAPTER 26 NONLINEAR LOW-DOSE EXTRAPOLATIONS
an essential role in cell protection against compounds that can cause oxidative DNA damage. It is known that Phase I enzymes, including the ubiquitous cytochrome P450 enzymes, serve to metabolically activate xenobiotics into reactive intermediates (electrophiles). In contrast, Phase II enzymes, including glutathione S-transferases (GST) enzymes, convert reactive electrophiles to inactive conjugates and inhibit formation of reactive intermediates that can ultimately damage DNA. Evidence for the importance of these enzymes in modulating cancer comes from studies demonstrating that genetic polymorphisms that alter the balance of these endogenous enzymes are associated with differential cancer risks (Holland et al. 2003). Additionally, transporter proteins such as P-glycoprotein (Pgp), which is expressed in many different cell types, including hepatic cells, renal proximal tubular cells, and capillary endothelial cells, may further act to limit cellular and DNA damage. Pgp acts as a pump, exporting chemicals out of cells, providing protection against potential genotoxic agents (Leonard et al. 2003). Because of the effectiveness of Pgp proteins in transporting harmful chemicals out of cells, cells can even become resistant to chemotherapeutic agents, which usually act as potent cytotoxic and mutagenic agents (Leonard et al. 2003).
26.2.2.
Post-DNA Damage Mechanisms
Cellular processes that can protect DNA integrity once a DNA lesion has occurred are well-characterized. Broadly, these responses can be categorized as DNA repair, cell-cycle delay, apoptosis (i.e., programmed cell death), and cellular differentiation (Pedraza-Farina 2006; Schulte-Hermann et al. 2000). The key DNA repair mechanisms such as mismatch repair, base excision repair, nucleotide excision repair, homologous recombination, and nonhomologous end-joining can remedy DNA damage before modifications are permanently fixed and propagated through cell division (Fleck and Nielsen 2004). Cell-cycle checkpoints are molecular surveillance mechanisms that can monitor DNA integrity and orchestrate cell-cycle delays to allow for DNA repair if needed (Shackelford et al. 1999; Schmitt et al. 2007). Apoptosis is a process that allows cells with DNA damage that cannot be repaired to undergo a controlled cell death, thereby eliminating the possibility of a heritable mutation (Marsman and Barrett 1994; Schmitt et al. 2007). Immune responses can also play a role in eradicating tumor cells (Gasser and Raulet 2006). While these processes are active in controlling endogenous DNA insults, they also become induced in response to chemical insults. The processes that limit heritable DNA damage do not occur in isolation, but are controlled by intricate signaling pathways that can coordinate a multifaceted response. As an example, in response to an outside stress, the mitogen-activated protein (MAP) kinase pathway differentially controls gene expression related to DNA repair, cell-cycle delay, and apoptosis. As an example, activation of the MAP kinase protein JNK has been associated with tumor suppressor activities, DNA repair, apoptosis, and cell proliferation depending on stress, dose, and cell type (Johnson and Nakamura 2007). Andersen et al. (2003) suggests that by virtue of the complex “circuitry” in living organisms, biological responses are nonlinear, and that by extension the “transition between different states of cellularity,” which can be
26.2. MECHANISTIC ASPECTS OF NONLINEAR CARCINOGENESIS
663
controlled by signaling pathways in response to exogenous chemicals, is also nonlinear (Andersen et al. 2003). Based on the processes discussed above, cells have the ability to counteract a certain amount of chemical and environmental stress to maintain an overall level of organismal homeostasis and health. Thus, it is unlikely that any carcinogenic exposure, no matter how small, would always lead to an increased cancer risk (see Chapters 5 and 8).
26.2.3.
Hormesis
Hormesis is a theory that proposes that biological responses to chemicals at high versus low doses differ (see Chapter 7). High doses are associated with an inhibitory response due to overt toxicity, while low-dose responses are stimulatory and may be potentially beneficial to the cell, tissue, or organism. Hormetic dose–responses are commonly depicted as low-dose stimulatory and high-dose inhibitory responses that take on a characteristic J- or inverted U-shaped dose–response. This curve results from an increase in response (on the ordinate axis) to chemical agents, with endpoints such as increased cell viability or cell proliferation on the cellular level and increased health (e.g., tumor incidence reduction) on the organismal level (Figure 26.1). These types of dose–responses have been observed with respect to a multitude of chemical agents, cell types, and experimental endpoints (Calabrese and Blain 2005). For example, in a study of tumor cell lines, hormetic-type dose– responses were observed in 136 different cell lines, representing 30 tissue types, following exposure to over 120 different agents (Calabrese 2005). Calabrese and Baldwin (1998) specifically examined hormesis during the carcinogenic process and found evidence of hormesis at all stages of carcinogenesis (initiation, promotion, tumor development, and progression). While most traditional rodent cancer bioassays are not designed to test hormetic properties of tumor development, studies in recent years have used doses that allow for such investigations. For example, Kinoshita et al. (2003) found that while high-dose phenobarbital (500 ppm) acted as a liver promoter in N-diethynitrosomine (DEN)-initiated rats, low-dose penobarbital (2 ppm) caused a decrease in oxidative DNA lesions, GSTP-positive liver foci, and tumor incidence and multiplicity compared to rats treated only with DEN. The decrease in DNA damage and ultimate tumorigencity appeared to be associated with an induction (i.e., hormetic stimulation) in DNA repair enzymes. Kodell (2001) has proposed mathematical models that support the hormetic dose–response at the DNA level, suggesting that genotoxic compounds do not necessarily add to any background mutation rate (i.e., exhibit additivity to background). Instead, low doses of genotoxic compounds may trigger DNA repair mechanisms, such that endogenous levels of DNA damage decrease, thereby decreasing the probability of a heritable mutation. The extent to which hormetic information may be useful in cancer risk assessment deserves consideration. At present, it is unclear how hormetic dose–responses might influence the MOA information necessary for a departure from the linearity assumption. In practice, the hormetic relationships may provide a useful backdrop for understanding that carcinogenic thresholds can exist, but it still may not be
664
CHAPTER 26 NONLINEAR LOW-DOSE EXTRAPOLATIONS
Figure 26.1. Hormesis may be defined as a dose–response relationship that is characterized by a biphasic (J- or inverted U-shaped) response. The quantitative features of the typical hormetic responses are similar, with the magnitude of the maximum stimulatory response typically being 30–60% greater than controls, the width of the stimulatory or hormetic zone averaging approximately 10-fold, and the interval from the zero equivalent point to the maximum stimulation averaging four- to fivefold (Calabrese and Blain 2005). [Figure adapted from Nascarella et al. (2009).]
feasible to use hormetic data quantitatively to set permissible exposure limits, at least at this time.
26.3. DNA-REACTIVE CARCINOGENS AND NONLINEARITY DNA-reactive compounds include compounds that test positive in standard in vitro mutagenic assays, such as the S. Typhimurium reverse mutation assay, the mouse lymphoma thymidine kinase assay, and the micronucleus test in human peripheral lymphocytes (WHO 2007). Based on the adaptive cellular responses described earlier, the assumption of linearity, even for DNA-reactive compounds, has been challenged (e.g., Kodell 2001). Ionizing radiation (IR) is perhaps the most wellunderstood mutagenic agent, but a substantial amount of literature argues in favor of a threshold (and perhaps hormetic characteristics) for this direct-acting mutagen (e.g., Pollycove and Feinendegen 1999; ICRP 2006; Chen et al. 2004). Arguments are based on both mechanistic and epidemiological considerations (ICRP 2006;
26.3. DNA-REACTIVE CARCINOGENS AND NONLINEARITY
665
Cuttler and Pollycove 2009). Pollycove (1998) has noted that while a linear doseresponse between radiation and human cancer has been observed in atomic bomb survivors exposed to high levels of radiation, this linear relationship does not bear out in low-dose studies. Pollycove (1998) has reviewed several studies where lowlevel exposure to IR has been associated with lower-than-background rates of cancer (a hormetic relationship). In one example, he noted that one researcher observed a tendency for lung cancer mortality to decrease as radon levels in homes increased up to a certain level (about 6 picocuries or pCi/liter) (Cohen et al. as cited in Pollycove 1998). On the molecular level, Pollycove (1998) pointed out that compared to background rates of DNA damage, DNA damage caused by low-level radiation (0.1 centigrays or cGy/year) is negligible; and at levels of radiation below 20 cGy, cellular adaptive processes increase DNA damage control. A report by the International Commission on Radiological Protection (ICRP 2006) acknowledged a potential threshold for radiation-induced carcinogenesis, noting that “When considered as a whole, the emerging results with regard to radiation-related adaptive response, genomic instability, and bystander effects suggest that the risk of low-level exposure to IR is uncertain and a simple extrapolation from high dose effects may not be wholly justified in all instances.” Despite some evidence for a threshold, however, the report ultimately concluded that “The LNT [linear no-threshold] model remains the most prudent risk model for the practical purposes of radiological protection” (ICRP 2006). Vinyl chloride (VC), a known human carcinogen specifically with respect to angiosarcoma of the liver, is clearly mutagenic with metabolic activation, but nonetheless may operate with a carcinogenic threshold in humans. The most reactive metabolite of VC, chloroethylene oxide (CEO), creates DNA adducts (specifically N2-,3-ethenoguanime) that generate base pair substitutions, and it is considered the ultimate carcinogenic metabolite of VC (ATSDR 2006; Dogliotti 2006). CEO is formed mainly by the cytochrome P-450 (CYP) isoenzyme CYP2E1 (Dogliotti 2006). The adducts formed by VC increase are also formed endogenously in unexposed animals and humans. Because of the presence of background lesions, it is important to consider whether VC-induced adducts cause a significant increase over background levels and whether this increase will cause ultimate tumor formation. Based on experiments that show exposure to 10 ppm VC increase N2-,3-ethenoguanime adducts only 2.2-fold above control, Swenberg et al. (2000) have suggested that it is unlikely that modest increases in adduct formation over controls at lower doses would result in tumor formation (Swenberg et al. 2000). Formaldehyde is another example of a mutagenic compound with evidence of nonlinearity. Based on limited evidence of respiratory cancer in humans and evidence of nasal tumors in rats and mice, EPA classifies formaldehyde as a probable human carcinogen. In rats, the most well-studied animal model for formaldehyde carcinogenesis, there is a threshold for tumor formation at 6 ppm. At levels below 6 ppm, formaldehyde induces DNA crosslinks in a dose-dependent manner with apparent low-dose linearity (Conolly et al. 2000; Slikker et al. 2004). At doses above 6 ppm, concurrent with tumor formation, formaldehyde causes marked cytotoxicity in the nasal passages of rats.
666
CHAPTER 26 NONLINEAR LOW-DOSE EXTRAPOLATIONS
Based on these mechanistic studies, scientists have developed a biologically based dose–response (BBDR) modeling that demonstrates that formaldehydeinduced nasal tumors in rats have a nonlinear relationship with dose that is dependent on cytotoxicity and is regenerative (and not a DNA crosslink formation). Modeling of the data in rats has actually showed the dose–response for tumor formation to be “J-shaped,” despite the assumed low-dose linear component of DNA–protein crosslinks (Conolly 2002; Conolly et al. 2004). While specifics of this evaluation are explained in more depth later, it should be noted that this biologically based approach identifying cytotoxicity followed by regeneration as the key event in tumor formation has led to the understanding that current assessments of formaldehyde risks in humans, which are based on linear extrapolation, could be overestimated by more than 1000-fold (Conolly et al. 2004).
26.4. NONMUTAGENIC CARCINOGENS AND NONLINEARITY Many carcinogenic compounds do not cause direct DNA damage, and are termed “nonmutagenic carcinogens.” A key feature of nonmutagenic carcinogens is their ability to increase a cell’s proliferative capacity. These compounds have generally been termed “promoters.” Promoters may act in concert with mutagens allowing chemically induced DNA damage to propagate more efficiently. Alternatively, by simply increasing the level of cell proliferation, a promoter can increase the probability of fixing an endogenous DNA error into a heritable mutation. In either case, cancer promotion response would be expected to exhibit nonlinearity. The US National Research Council (NRC 2006) noted that “there is general consensus in the scientific community that nongenotoxic carcinogens that act as tumor promoters exhibit nonlinear dose–response relationships, and that thresholds (doses below which the expected response would be zero) are likely to be present.” Increasing cell proliferation can occur through a number of different mechanisms. The well-characterized mechanisms include direct receptor binding and alteration of signal transduction pathways. Receptor-mediated binding and induction of signaling cascades are both thought to be nonlinear processes (Andersen et al. 2003). The nonlinearity of receptor-mediated responses is due to the nonlinearity of binding kinetics. As noted by NRC, “EPA determined in previous evaluations of receptor-mediated carcinogens that a nonlinear, low-dose model, that may accommodate a threshold, is appropriate” (NRC 2006). 2,3,7,8-Tetrachloro-dibenzo-dioxin (TCDD) and other coplanar polychlorinated dioxins (PCDDs, described here as “dioxins”) are ligands for the aryl hydrocarbon receptor (AhR) ligand. TCDD specifically is an example of a well-studied receptor-mediated carcinogen, although other coplanar dioxins are believed to operate through the same MOA (Andersen et al. 2003; NRC 2006). Other AhR ligands include some polychlorinated biphenyls (PCBs) and polybrominated biphenyls. The evidence for a nonlinear response for dioxin at low doses includes the nonlinear nature of receptor binding kinetics in general, which often leads to disruption of cell signaling, decreased apoptosis and/or increased mitogenesis, and
26.4. NONMUTAGENIC CARCINOGENS AND NONLINEARITY
667
increased oxidative stress (which in turn can activate other signaling pathways). More specifically, studies of dioxin activity have shown that binding of dioxin to the AhR causes the induction of CYP enzymes, along with the activation of many downstream genes, through complex signaling pathways that are still largely unknown (Gasiewicz and Park 2003). Andersen et al. (2003) described the initiation of cellular changes by AhR agonist binding as a “switch,” an all-or-none response that depends on a threshold dose of agonist. In this model, individual cells are either in the induced state or in the basal state, rather than exhibiting a continuous dose– response. Individual cells, for example, rodent liver cells in different acinar regions, may have different thresholds of sensitivity to dioxin. Thus, although ligand binding to the AhR is the initial event leading to a spectrum of biological responses, this event may not reflect the dose–response relationship observed for tumorigenesis on the organism level. Although the law of mass action predicts that a single molecule of ligand can interact with a receptor, thereby inducing a biological response, it is also understood that at very low doses, receptor occupancy is trivial, and no discernable response can be observed (Andersen et al. 2003; Motulsky and Christopoulos 2004). An increasingly large number of chemicals are believed to cause tumors through an MOA that entails cytotoxicity followed by regenerative hyperplasia. This MOA is generally accepted to exhibit a threshold and provides a basis for explaining high-dose carcinogenicity (Cohen et al. 2004; Schulte-Hermann et al. 2000; Clewell et al. 1995). Exposure to certain chemicals at high doses leads to overt cellular toxicity and tissue necrosis. The remaining cells respond with increased proliferation, sometimes leading to hyperplasia. Because cell replication is not 100% error-free, increased cell proliferation increases the probability of mutations and subsequent tumorigenesis (Cohen and Ellwein 1990, 1991; Ames and Gold 1990). It also increases the probability that previously initiated cells will undergo clonal expansion and form lesions. Examples of lesions believed to be caused by a cytotoxic MOA include chloroform-induced liver tumors, dimethylarsenic acid (DMA)-induced bladder tumors, and formaldehyde-induced nasal tumors (all in rodents) (see Section 26.6 for more information). Interestingly, although the carcinogenic MOA for these compounds involves cytotoxicity followed by regenerative hyperplasia at high doses, some of the compounds also cause direct DNA damage at lower doses (e.g., formaldehyde); the cytotoxicity and increased reparative cell proliferation, however, appear to be a critical step in the overall carcinogenic MOA (Conolly 2002). There are other carcinogenic processes that do not involve direct DNA reactivity, but either facilitate DNA damage or cause increased cell proliferation (and thereby increase potential for propagation of DNA damage) indirectly. These mechanisms include inhibition of DNA repair, alteration of DNA methylation patterns, oxidative damage, and modulation of signal transduction pathways, among others (Pratt and Barron 2003). Involvement of these mechanisms as part of the MOA is likely be associated with tumor development that has a nonlinear dose response; however, because it is frequently difficult to establish a precise MOA for these processes, linear extrapolation is often invoked as a default assumption.
668
CHAPTER 26 NONLINEAR LOW-DOSE EXTRAPOLATIONS
26.5. 26.5.1.
CANCER RISK ASSESSMENT Basis for the Linearity Assumption
Agencies concerned with cancer risk assessment [e.g., US EPA, European Union (EU), International Programme for Chemical Safety (IPCS)] have often distinguished between direct DNA reactive carcinogens (assumed to have low-dose linearity) and carcinogens that may produce tumors through alternative mechanisms, such as increased cell proliferation (where nonlinear extrapolation is warranted) (US EPA 2005b; Pratt and Barron 2003; Bolt and Degen 2004; Seeley et al. 2001). Low-dose modeling assumptions for indirect genotoxins have been less clear. Indirect genotoxins have non-DNA biological targets, but produce DNA damage as a secondary result. For example, inorganic arsenic, a confirmed human carcinogen, is thought to cause DNA damage indirectly, likely through the generation of oxidative stress. Additionally, inorganic arsenic can inhibit DNA repair enzymes, which results in further DNA damage indirectly (Schoen et al. 2004). The linearity assumption for direct acting mutagens is a policy decision with some biological roots (see Chapters 2 and 3). Before the advent of the recent Cancer Guidelines, EPA policy advocated use of the linearized multistage (LMS) model to quantitatively evaluate cancer potency (US EPA 1986). The upper confidence limits of curves generated using this model yielded linear relationships that were relatively stable and could be considered to conservatively represent the upper bounds of potential cancer risk (US EPA 1986). The 1986 Cancer Guidelines stated the following: “It should be emphasized that the linearized multistage procedure leads to a plausible upper limit to the risk that is consistent with some proposed mechanisms of carcinogenesis.” Because LMS did not allow for a transparent understanding of where the modeling of observable data ended and extrapolation to lower doses began, the more recent EPA Guidelines recommended abandoning the LMS model. Instead, the revised Guidelines promoted the LNT model for mutagenic carcinogens or compounds when the MOA is not established. This approach involves modeling observable data to a specified effect level, called the point of departure (POD), and extrapolating linearly from the POD to 0. This approach provides a clear division between observable and extrapolated data. The LMS model and linear extrapolation from a POD, however, do not produce substantially different results. In a recent examination of 130 datasets for 60 carcinogens listed in the EPA’s Integrated Risk Information System (IRIS), the ratio of the slope generated by the LMS model to that generated by linear extrapolation from a POD was between 0.9 and 1.1 for 82% of the chemicals (Subramaniam et al. 2006). The “additivity to background” hypothesis, which holds that a carcinogenic exposure will increase risk linearly if the carcinogen is operating by mechanisms that also occur under background conditions, has been considered as supportive of low-dose linearity. Under this hypothesis, mutagenic compounds, by default, exhibit linearity at low doses because these compounds add to endogenous DNA damage present in all biological systems. In theory, a nonmutagenic compound could also be “additive to background” if the exposure causes cancer with an MOA that overlaps with background processes that lead to the same cancer type.
26.5. CANCER RISK ASSESSMENT
669
The “additivity to background” concept is found in the 2005 Cancer Guidelines, which stated that “Agents that are generally considered to be linear … include … agents for which human exposures or body burdens are high and near doses associated with key precursor events in the carcinogenic process, so that background exposures to this and other agents operating through a common mode of action are in the increasing, approximately linear, portion of the dose–response curve” (US EPA 2005a). Additionally, more recently, some scientists have also suggested that even when compounds exhibit nonlinearity on the cellular level, they are likely to manifest as a linear dose–response relationship on the population level. This hypothesis, as presented by White et al. (2009), holds that modifying factors (such as nutritional status) and biological variability cause the dose–response relationship to be linear across the population (White et al. 2009). “Additivity to background,” as well as the epidemiological considerations in support of low-dose linearity, relies on statistical arguments over accepted biological principles. As discussed in depth throughout this chapter, every individual has a complex array of biological processes that operate through intricate cellular pathways to maintain homeostasis. Heterogeneity in response to chemical insults across a population will tend to broaden the dose–response relationship, but will not linearize it (Rhomberg 2009). Over the past decades there has been substantial scientific progress in understanding the disease process on genetic and cellular levels. Assuming all external and background disease processes are additive is an oversimplification and a retreat from a biological understanding of disease and its application to risk assessment. It should also be recognized that even if a compound has an MOA that is coincidental with a background disease process, the compounds may not, at any plausible dose, contribute significantly to the background process, and therefore may not cause a measurable increase in cancer incidence above background.
26.5.2. EPA Cancer Risk Assessment and Low-Dose Extrapolation The present EPA approach is to use linear extrapolation from high doses to low doses under two distinct scenarios: (1) when a mutagenic compound is known to directly interact with DNA; and (2) when the MOA of a carcinogen is not welldefined. The Cancer Guidelines (US EPA 2005b) stated: When the weight of evidence evaluation of all available data are insufficient to establish the mode of action for a tumor site and when scientifically plausible based on the available data, linear extrapolation is used as a default approach, because linear extrapolation generally is considered to be a health-protective approach. Nonlinear approaches generally should not be used in cases where the mode of action has not been ascertained. Where alternative approaches with significant biological support are available for the same tumor response and no scientific consensus favors a single approach, an assessment may present results based on more than one approach.
Thus, the EPA has recognized the nonlinearity of indirect genotoxins, but does not depart from the assumption of linearity without a well-defined understanding of
670
CHAPTER 26 NONLINEAR LOW-DOSE EXTRAPOLATIONS
the key events in the carcinogenic process. In practice, the MOA of most chemical carcinogens is not well-established, and the default assumption of linearity is often invoked in risk assessments. The EPA does not list MOAs with probable low-dose nonlinearity. While several MOAs have some acceptance as being nonlinear or having a threshold (e.g., cytotoxicity followed by regeneration, and receptor-mediated processes), the establishment of nonlinear dose–responses for specific carcinogenic MOAs is still controversial (Andersen et al. 2003). More discussion on implementation of nonlinear extrapolation by regulatory agencies for nonlinear carcinogens is presented later in this chapter.
26.5.3.
Low-Dose Extrapolation Outside the United States
Although this chapter is focused on cancer risk assessment in the United States, outside of the EPA, there is increasing acceptance that carcinogenic chemicals operating through indirect DNA-damaging mechanism MOAs may have a threshold. Bolt (2003) reported that the German Commission for the Investigation of Health Hazards of Chemical Compounds in the Work Area (MAK Commission) has proposed a category of carcinogens that have genotoxic potential, but the potency is so low that below a certain level, no contribution to human cancer risk is expected. Similarly, several agencies within the EU have recognized a distinction between mutagens and indirectly genotoxic compounds and that risk-based regulations should accommodate this understanding (Pratt and Barron 2003). The Netherlands National Institute for Public Health and the Environment (RIVM), an environmental regulatory agency, has long made a distinction between threshold and nonthreshold carcinogens. The Agency (Baars et al. 2001) stated: In evaluating the toxicity of chemical substances, distinction must be made between two fundamental different approaches. Genotoxic carcinogens are assumed to exert their activity also at the smallest dose, i.e., by definition a threshold for genotoxic activity does not exist. Toxic effects other than genotoxic carcinogenicity, however, are assumed to occur via receptor interaction, which implies that a certain threshold needs to be exceeded before a toxic effect will occur.
Based on this approach, the RIVM has determined that because inorganic arsenic compounds are “inactive or weak mutagens,” “the carcinogenic action of inorganic arsenic is based on a nongenotoxic mechanism” and should be evaluated using a threshold approach (Baars et al. 2001). By contrast, the EPA has used a lowdose linear approach for inorganic arsenic risk assessment, stating that although it is clear inorganic arsenic does not cause direct genotoxicity, there is insufficient information on the MOA to depart from the linear default assumption (US EPA 2005c).
26.6.
NONLINEARITY PRINCIPLES INTO PRACTICE
If sufficient evidence is available to support a nonlinear MOA, information must be translated to a form useful for quantitative risk assessment. There are currently two
26.6. NONLINEARITY PRINCIPLES INTO PRACTICE
671
general approaches to estimate cancer potency for nonlinear carcinogens. The first approach involves modeling data in the observable range using generic dose– response models, identifying a POD, and using uncertainty factors to establish a threshold of concern. As noted earlier, according to the EPA Cancer Guidelines, the POD is an estimated dose (usually expressed in human-equivalent terms) near the lower end of the observed range, without significant extrapolation to lower doses. Identification of a POD and application of uncertainty factors for carcinogenic evaluation is fundamentally and computationally similar to estimating noncancer risks using benchmark doses (BMDs). In fact, some EPA analyses have termed some nonlinear cancer evaluations as “reference dose” (RfD) approaches (e.g., US EPA 2005d). In a similar manner, a margin of exposure (MOE) analysis can be performed, where the potential for unacceptable risk can be judged by the ratio between the POD and human exposure. A more refined approach to quantifying low-dose cancer risk relies on BBDR modeling. In this method, quantitative pharmacokinetic and mechanistic data are used to project tumor incidence without extrapolation. More details on these approaches are presented below.
26.6.1.
Using an RfD or MOE Approach
In the first approach, quantifying low-dose risk for nonlinear carcinogens mirrors current noncancer risk assessment methods and begins with the identification of a POD. Finding the point near the lower end of the observed range, without extrapolation to lower doses, is a challenge common to both linear and nonlinear extrapolation. In general, this approach relies on statistical dose–response models to define a dose at which there is a specific increase in response (1%, 5%, 10%, etc.)— known as the benchmark dose (BMD01, BMD5, BMD10, etc.) or the effective dose (ED01, ED05, ED10, etc.). Although these models are used to define the BMD, it should be noted that, depending on the input data, these models can yield linear, sublinear, and supralinear dose–response curves at low doses. Once the BMD is identified, the one-sided lower 95th confidence interval on the BMD called the BMDL (benchmark dose lower bound) is used as the POD. Similar to noncancer risk assessment, the POD is then divided by uncertainty factors to account for potential interspecies differences, intraindividual variability, and so on. The current risk assessment paradigm has accepted that, by accounting for uncertainty through use of the BMDL and other uncertainty factors, the resulting dose is either below a toxic threshold or so low as to constitute a “virtually safe dose” (Bogdanffy et al. 2001). Selecting a model to define the POD involves several factors, most importantly the nature of the available data, the desired risk metric, and the size and statistical power of the study. Depending on whether the data is quantal (based on incidence data) or continuous (based on a continuous biological parameter), as well the nature or severity of the adverse outcome, different modeling decisions may be appropriate. For quantal data, excess risk is usually examined, while for continuous data, several other metrics may be more useful (e.g., metrics that measure relative and absolute differences in mean responses, changes in mean relative to the standard deviation of controls, changes above specified value, etc.). Using information on cancer
672
CHAPTER 26 NONLINEAR LOW-DOSE EXTRAPOLATIONS
precursor events in conjunction with a BMD obviates the need for quantifying low tumor incidence risk increases, which can’t be observed directly in any practical experiment. Instead, it allows for determination of a dose at which a more measurable critical event (to which an increased risk of cancer is secondary) becomes inconsequential. The decision to use a response level of 1%, 5%, or 10% may produce substantially different results, but there is no specific guidance on response level selection; the optimal benchmark is case-specific. The 10% response level is typically selected from animal tumor bioassays because it reflects a low response rate while still providing a reliable estimate of the observed data. Clearly, some experiments involve large amounts of data and may have greater statistical power to estimate a lower response with reliability (e.g., 1% or 5%). For quantal epidemiological data, EPA recommends using the 1% response rate because the data typically have greater sensitivity in the low-dose regions (US EPA 2008). The DMA cancer risk assessment exemplifies the modeling decisions required for the evaluation of carcinogens using a nonlinear approach and the impact of these decisions on the POD. DMA is a rat bladder carcinogen that operates with an MOA consistent with cytotoxicity, followed by regenerative hyperplasia, and therefore operates with a threshold (Cohen et al. 2006). At the time of the DMA cancer evaluation, data on bladder cell toxicity, compensatory cell proliferation, bladder hyperplasia, and tumor formation were all available. Bladder cell cytotoxicity, bladder hyperplasia, and tumor incidence data were expressed as quantal variables (i.e., animals were categorized as exhibiting the toxic endpoint or not). Compensatory cell proliferation, which was measured as a BrdU labeling index, was the only continuous variable considered. Selection of the key endpoint as well as the response level greatly affects the magnitude of the POD. As shown in Table 26.1, the magnitude of the POD can vary by several orders of magnitude, depending on the selected key endpoint and response level. For example, the BMDL1 based on cell cytotoxicity is 8500 times lower than BMDL10 based on tumor incidence.
TABLE 26.1. Benchmark Dose Calculation for Various Endpoints from Rat Bladder Cancer Studies at 10% and 1% Response Levels
10%
Biological Event Tumor Hyperplasia BrdU labeling (proliferation) Cytotoxicity
1%
Duration
BMD (mg/kg/day)
BMDL (mg/kg/day)
BMD (mg/kg/day)
BMDL (mg/kg/day)
104 weeks 10 weeks 104 weeks 10 weeks
7.74 2.00 1.97 0.92
5.96 1.54 1.61 0.43
6.80 0.62 0.93 0.75
2.22 0.48 0.66 0.10
3 weeks 10 weeks
0.68 0.02
0.18 0.006
0.31 0.002
0.02 0.0007
26.6. NONLINEARITY PRINCIPLES INTO PRACTICE
673
After evaluating the merits of the available data, the EPA Office of Pesticide Programs (OPP) concluded that increased cell proliferation was the key cancer precursor and should serve as the basis for the POD. It stated that “Among the several key events, all of which are necessary for tumor formation, cell proliferation is proposed for deriving a point of departure because it is needed for increasing the likelihood of chromosome mutation and for the perpetuation of genetic errors, as well as for hyperplasia.” Additionally, OPP selected the 10% response level stating that the BMDL10 was sufficiently protective because it was (a) an order of magnitude below the experimental doses that resulted in increased cell proliferation and (b) two orders of magnitude lower than the experimental doses that caused hyperplasia.
26.6.2. Other Nonlinear Cancer Evaluations: Captan and Chloroform The EPA has also used nonlinear extrapolation for the cancer evaluation of captan (a pesticide) and chloroform. The analysis for captan was less quantitative (in terms of deriving a POD) than for DMA, but after an MOA evaluation, OPP concluded “captan to be a potential carcinogen at prolonged high doses that cause cytotoxicity and regenerative cell hyperplasia. These high doses of captan are many orders of magnitude above those likely to be consumed in the diet, or encountered by individuals in occupational or residential settings. Therefore, captan is not likely to be a human carcinogen nor pose cancer risks of concern when used in accordance with approved product labels.” Chloroform is the only carcinogen evaluated using a nonlinear approach in IRIS database as of today. In 2001, the EPA determined the “carcinogenic responses observed in animals [from chloroform] are associated with regenerative hyperplasia that occurs in response to cytolethality.” Specifically, the EPA concluded that “because the carcinogenicity of chloroform is secondary to cytotoxicity which has a threshold, and because chloroform and its metabolites are not strongly genotoxic and are unlikely to cause cancer by a genotoxic mechanism (Lohman et al. 1992; ILSI 1997; Golden et al. 1997; WHO 1998),* a nonlinear approach is more appropriate for low-dose extrapolation” (US EPA 2001). To quantify the low-dose risk for chloroform (via the oral route), EPA employed an RfD/MOE approach based on kidney tumor incidence data in male rats. The EPA determined that the modeled BMDL10 had an MOE of 2000 compared to the noncancer oral RfD for chloroform. The EPA reasoned that even after accounting for uncertainty in the BMDL10 estimate (e.g., intraindividual variation, interspecies pharmacokinetic and pharmacodynamic differences, and use of cancer incidence instead of cancer precursor information), the large MOE indicated that “the RfD is adequately protective of public health for cancer effects, based on the nonlinear dose–response for chloroform and the mode of action for both cancer and noncancer effects having a common link through cytotoxicity.” *As cited by US EPA (2001).
674
CHAPTER 26 NONLINEAR LOW-DOSE EXTRAPOLATIONS
26.6.3.
BBDR Modeling
Rather than using an RfD approach, there has been intensive research interest characterizing dose–responses on the cellular level in target organs and using that information to estimate human tumor incidence through BBDR models. To the extent possible, these models use pharmacokinetic and pharmacodynamic information from low-dose studies to estimate tumor incidence. Implementation of pharmacokinetic information, however, has outpaced pharmacodynamic models, which still seem to involve more general assumptions and simplifications. For example, pharmacodynamic cancer models [such as the Moolgavkar–Venzon–Knudson (MVK) or twostage clonal growth model] predict the percentage of cancer cells as a function of time and dose by empirically estimating dose-dependent birth, death, and mutation rates in cells. While biologically based, by simply accounting for cell birth, death, and mutation rates, these models are still clearly an oversimplification of the complex biological process of carcinogenesis. The most extensive work in this type of BBDR modeling has been in characterizing formaldehyde cancer risks (Conolly 2002). The formaldehyde BBDR model consists of three parts: a dosimetry model, a description of the cell-specific dose and corresponding formation of DNA–protein crosslinks, and finally the MVK or two-stage clonal growth model for cancer cell formation (Chin et al. 2007). While many of the pharmocodynamic parameters for the MVK model for formaldehyde have been determined experimentally, some parameters are still estimated from the data, leading to unavoidable uncertainty. Due to this uncertainty, it is important to evaluate the sensitivity of the model to uncertainties and variability in the selected parameters. For example, Crump et al. (2008) specifically investigated the BBDR for formaldehyde (Conolly et al. 2004) and found that slight changes in the cell division and death rates assumptions could produce estimates of human risk ranging anywhere from negative up to 10,000 times higher than those calculated by Conolly et al. (2004). Thus, while advances have been made in BBDR modeling to characterize the shape of the dose–response curve in the low-dose region, some scientists and regulators believe that further model characterization is necessary before integrating these models into the current risk assessment framework. Interestingly, while EPA’s National Center for Environmental Assessment (NCEA) has not used the formaldehyde BBDR model for toxicity criteria derivation for IRIS, the Toxic Air Programs of EPA have used the model in toxicity assessments under the National Scale Air Toxics Assessment (US EPA 2005e). More sophisticated approaches involving computational models of whole biological systems are also being explored (Conolly and Thomas 2007). These models will attempt to use pharmacokinetic, mechanistic, and MOA information to simulate the complex biological circuitry of signal transduction pathways and will aim to reconcile information from dose-dependent changes in gene induction, resulting protein induction, cellular responses, and organ toxicity. This “bottom-up” approach (i.e., predicting the toxic response from mechanistic data) is the focus of the “Toxicity Testing in the 21st Century” initiative. Specifically, this NRC initiative calls for a paradigm shift away from animal testing, which often has questionable
26.6. NONLINEARITY PRINCIPLES INTO PRACTICE
675
human relevance, especially at the high doses tested, and instead promotes the use of a battery of high-output in vitro tests that can be used to define the doses that lead to perturbations in signaling pathways and ultimately human disease (NRC 2007).
26.6.4.
Harmonization of Cancer and Noncancer Risks
Over the past decade there has been a movement to harmonize cancer and noncancer risk assessment (Gaylor 1997; Bogdanffy et al. 2001) based on the premise that cancer and noncancer events share similar pharmacokinetic dependencies and overlapping MOAs and thus have similar dose–response relationships. The benchmark dose approach lends itself to the evaluation of both linear and nonlinear dose– response. In fact, one of the stated purposes of EPA’s formalization of the benchmark dose process was to provide a standardized approach to chemical dose–response assessment, regardless of whether the chemical is a carcinogen. Interestingly, while much of the focus over the past several years has been applying traditionally noncancer approaches, such as the BMD analysis, to nonlinear carcinogens, a recent NRC report has advocated an alternative approach to unifying dose–response assessments for cancer and noncancer endpoints. This report reflects some of the positions summarized by White et al. (2009) discussed earlier in the context of the low-dose linearity of carcinogens. Specifically, this report suggests that noncancer effects do not necessarily have a threshold and that a linear model for both cancer and noncancer effects should be considered (NRC 2009). This supposition is based on the same considerations described earlier for mutagenic carcinogens, namely “additivity to background” on the mechanistic levels and interindividual variability in sensitivity to toxic effects on the population level. Specifically, NRC recommended that when selecting a dose–response model for both noncancer and cancer compounds, risk assessors should consider linearity by taking into account any background exposure to the chemical in question, as well as to any other chemicals, exogenous or endogenous, that may affect the same biological or toxicological processes. Additionally, the report stated that linearity in the dose that could result from potentially sensitive individuals (e.g., based on age, gender, or health status) on the population level should be carefully examined. Use of this proposed approach represents a significant paradigm shift in noncancer risk assessment and is inconsistent with basic biological principles. Earlier in this chapter we discussed the fact that it is unlikely for many carcinogens to operate with a linear dose–response. Although it is possible for a single molecule to interact with DNA to increase cancer risk, carcinogenesis is not a multistep process, and protective controls are in place to limit tumor formation. Under this same line of reasoning, it is even less likely for a single-molecule interaction to increase the risk of noncancer effects. As noted earlier, biological systems have several levels of redundancy, such that insults to one cell or even many cells is not sufficient to lead to a loss of function. In response to the assertion that noncancer risk assessment should default to a linear approach, Rhomberg (2009) has commented that “there are still fundamental differences between carcinogenicity
676
CHAPTER 26 NONLINEAR LOW-DOSE EXTRAPOLATIONS
on the one hand—in which the probability of constellations of rare events (that get rarer with lower doses) drives the dose–response function—and most noncancer responses on the other—in which the dose–response function hinges on the degree of perturbation of physiological and homeostatic processes (which becomes less pronounced and less efficacious with lower doses)” (Rhomberg 2009).
26.7.
SUMMARY AND CONCLUSION
While an early understanding of carcinogenesis hinged on the idea of a single molecule being able to create a single mutation that could cause cancer, we now understand that carcinogenesis is much more complex, involving the accumulation of multiple mutations and not always resulting from direct chemical interaction with DNA. For example, a carcinogen may act by increasing cell proliferation and thereby increasing the probability of fixing an endogenous DNA mutation. Moreover, scientists have characterized a vast array of overlapping protective mechanisms that work to maintain DNA integrity and limit tumor formation. Protective mechanisms such as antioxidant induction, DNA repair, and apoptosis often function on coordinated fronts through complex signaling pathways. The early, more simplistic views of cancer were consistent with a linear dose– response relationship, but as our understanding of the complexity of the carcinogenic process and protective mechanisms increases, it is clear that a linear dose–response relationship may not be appropriate for some, if not most, carcinogens. For example, chemicals that cause cancer though overt cellular cytotoxicity and resulting regenerative hyperplasia clearly have a threshold; one chemical molecule is not capable of producing a cytotoxic condition of sufficient magnitude to mount a regenerative response. Many regulatory agencies, including EPA, have come to accept the changing views on carcinogenesis and have applied this understanding to risk assessment. In several instances, EPA agencies have recommended an MOE or RfD approach when there was sufficient evidence of a nonlinear carcinogenic MOA. Nonlinear risk assessments have been performed for chemicals such as DMA and chloroform. While the EPA has successfully used nonlinear extrapolation to describe lowdose cancer risk, there are some roadblocks in moving from a linear to an MOE approach for carcinogens in terms of regulatory implementation. More research needs to be focused on understanding MOA and the key steps required in tumor formation. BBDR models provide a means to incorporate MOA information (along with pharmacokinetics information), to better translate how both nonlinear and linear activities at low doses translate into observable health effects. These models, however, require further refinement and validation before they are used more routinely. The “Toxicity Testing in the 21st Century” initiative is focused on guiding research to make these models more functional and applicable to risk assessment. Using the MOA information in BBDR models will also offer an opportunity to test the “additivity to background” argument that has formed the cornerstone of the default linear assumption for decades. Another challenge is that under an MOE approach, the benefit of an exposure reduction cannot be easily quantified. Thus, using an MOE approach can stymie
REFERENCES
677
efforts to conduct cost–benefits analyses, which, to guide public health policy, are concerned with understanding how decreases in exposure relate to the number of individuals who will be disease-free. Through an MOE-based risk assessment, we come to understand how great the distance is between expected human exposure and potential adverse health effects, but lose resolution for calculating the probability of disease in a given individual. Despite the perceived difficulties in conducting cost–benefit analyses for nonlinear carcinogens, several economists have offered methodologies for handling such nonlinear data such that alternatives to the current cost–benefit framework should not be viewed as untenable (e.g., Griffiths et al. 2002; Axelrad et al. 2005). Despite the challenges that nonlinear extrapolation for carcinogens offers in terms of regulatory implementation, it is important that the best available science guide risk assessment practices. Default assumptions designed to be healthprotective, such as low-dose linear assumption, are useful when no other alternatives are available, but public health initiatives will be best served when grounded in biologically based approaches.
ACKNOWLEDGMENTS Special thanks to Jennifer Leary for her editorial support.
REFERENCES American Cancer Society (ACS) (2009). Lifetime Probability of Developing or Dying from Cancer. http://www.cancer.org/docroot/CRI/content/CRI_2_6x_Lifetime_Probability_of_Developing_or_ Dying_From_Cancer.asp (accession date, May 11, 2009). Agency for Toxic Substances and Disease Registry (ATSDR) (2006). Toxicological Profile for Vinyl Chloride. National Technical Information Service (NTIS) NTIS PB2007-100671, July, 329 pages. Ames, B. N., and Yanofsky, C. [University of California (Berkeley, CA); Stanford University (Stanford, CA); Academia] (1971). The detection of chemical mutagens with enteric bacteria. In Chemical Mutagens: Principles and Methods for Their Detection, Vol. I, Hollaender, A., ed., Plenum Press, New York, pp. 267–282. Ames, B. N., and Gold, L. S. (1990). Too many rodent carcinogens: Mitogenesis increases mutagenesis. Science 249(4972), 970–971. Andersen, M. E., Yang, R. S. H., French, C. T., Chubb, L. S., and Dennison, J. E. (2003). Molecular circuits, biological switches, and nonlinear dose–response relationships. Environ Health Perspect 110(Suppl 6), 971–978. Axelrad, D. A., Baetcke, K., Dockins, C., Griffiths, C. W., Hill, R. N., Murphy, P. A., Owens, N., Simon, N. B., and Teuschler, L. K. (2005). Risk assessment for benefits analysis: Framework for analysis of a thyroid-disrupting chemical. J Toxicol Environ Health A 68(11–12), 837–855. Baars, A. J., Theelen, R. M. C., Janssen, P. J. C. M., Hesse, J. M., van Apeldoorn, M. E., Meijerink, M. C. M., Verdam, L., and Zeilmaker, M. J. (2001). Re-evaluation of human-toxicological maximum permissible risk levels. RIVM Report 711701025, March, Netherlands National Institute of Public Health and the Environment (RIVM), pages 8–15, 25–29. Beck, B. D., Calabrese, E. J., Slayton, T. M., and Rudel, R. (2007). The use of toxicology in the regulatory process. In Principles and Methods of Toxicology, 5th edition, Hayes, A. W., ed., Taylor & Francis, Philadelphia, pp. 45–102. Boelsterli, U. A. (2003). Mechanistic Toxicology, Taylor & Francis, London.
678
CHAPTER 26 NONLINEAR LOW-DOSE EXTRAPOLATIONS
Bogdanffy, M. S., Daston, G., Faustman, E. M., Kimmel, C. A., Kimmel, G. L., Seed, J., and Vu, V. (2001). Harmonization of cancer and noncancer risk assessment: Proceedings of a consensus-building workshop. Toxicol Sci 61, 18–31. Bolt, H. M. (2003). Genotoxicity—Threshold or not? Introduction of cases of industrial chemicals. Toxicol Lett 140/141, 43–51. Bolt, H. M., and Degen, G. H. (2004). Human carcinogenic risk evaluation, Part II: Contributions of the EUROTOX specialty section for carcinogenesis. Toxicol Sci 81(1), 3–6. Calabrese, E. J. (2005). Cancer biology and hormesis: Human tumor cell lines commonly display hormetic (biphasic) dose responses. Crit Rev Toxicol 35(6), 463–582. Calabrese, E. J., and Baldwin, L. A. (1998). Can the concept of hormesis be generalized to carcinogenesis? Regul Toxicol Pharmacol 28, 230–241. Calabrese, E. J., and Blain, R. (2005). The occurrence of hormetic dose responses in the toxicological literature, the hormesis database: An overview. Toxicol Appl Pharmacol 202(3), 289–301. Cerutti, P., Ghosh, R., Oya, Y., and Amstad, P. (1994). The role of the cellular antioxidant defense in oxidant carcinogenesis. Environ Health Perspect 102(Suppl 10), 123–129. Chen, W. W., Luan, Y. C., Shieh, M. C., Chen, S. T., Kung, H. T., Soong, K. L., Yeh, Y. C., Chou, T. S., Mong, S. H., Wu, J. T., Sun, C. P., Deng, W. P., Wu, M. F., and Shen, M. L. (2004). Is chronic radiation an effective prophylaxis against cancer? J Am Physicians Surgeons 9(1), 6–10. Chin, W. A., Chen, C., Hogan, K., Lipscomb, J. C., Scott, C. S., and Subramaniam, R. (2007). High-tolow dose extrapolation: Issues and approaches. Hum Ecol Risk Assess 13, 46–51. Clewell, H. J., Gentry, P. R., Gearhart, J. M., Allen, B. C., and Andersen, M. E. (1995). Considering pharmacokinetic and mechanistic information in cancer risk assessments for environmental contaminants: Examples with vinyl chloride and trichloroethylene. Chemosphere 31(1), 2561–2578. Cohen, S. M., and Ellwein, L. B. (1990). Cell proliferation in carcinogenesis. Science 249(4972), 1007–1011. Cohen, S. M., and Ellwein, L. B. (1991). Genetic errors, cell proliferation, and carcinogenesis. Cancer Res 51, 6493–6505. Cohen, S. M., Arnold, L. L., Eldan, M., Schoen, A. S., and Beck, B. D. (2006). Methylated arsenicals: The implications of metabolism and carcinogenicity studies in rodents to human risk assessment. Crit Rev Toxicol 36, 99–133. Cohen, S. M., Klaunig, J., Meek, M. E., Hill, R. N., Pastoor, T., Lehman-McKeeman, L., Bucher, J., Longfellow, D. G., Seed, J., Dellarco, V., Fenner-Crisp, P., and Patton, D. (2004). Evaluating the human relevance of chemically-induced animal tumors. Toxicol Sci 78, 181–186. Conolly, R. B. (2002). The use of biologically based modeling in risk assessment. Toxicology 181–182, 275–279. Conolly, R. B., and Thomas, R. S. (2007). Biologically motivated approaches to extrapolation from high to low doses and the advent of systems biology: The road to toxicological safety assessment. Human and Ecological Risk Assessment 13(1), 52–56. Conolly, R. B., Kimbell, J. S., Janszen, D., Schlosser, P. M., Kalisak, D., Preston, J., and Miller, F. J. (2004). Human respiratory tract cancer risks of inhaled formaldehyde: Dose–response predictions derived from biologically-motivated computational modeling of a combined rodent and human dataset. Toxicol Sci 82(1), 279–296. Conolly, R. B., Lilly, P. D., and Kimbell, J. S. (2000). Simulation modeling of the tissue disposition of formaldehyde to predict nasal DNA–protein cross-links in Fischer 344 rats, rhesus monkeys, and humans. Environ Health Perspect 108(Suppl 5), 919–924. Crump, K. S., Chen, C., Fox, J. F., Van Landingham, C., and Subramaniam, R. (2008). Sensitivity analysis of biologically motivated model for formaldehyde-induced respiratory cancer in humans. Ann Occup Hyg 52(6), 481–495. Crump, K. S., Hoel, D. G., Langley, C. H., and Peto, R. (1976). Fundamental carcinogenic processes and their implications for low dose risk assessment. Cancer Res 36(7), 2973–2979. Cuttler, J. M., and Pollycove, M. (2009). Nuclear energy and health and the benefits of low-dose radiation hormesis. Dose–Response 7, 52–89. Dogliotti, E. (2006). Molecular mechanisms of carcinogenesis by vinyl chloride. Ann Ist Super Sanita 42(2), 163–169. Fleck, O., and Nielsen, O. (2004). DNA repair. J Cell Sci 117, 515–517.
REFERENCES
679
Gasiewicz, T. A., and Park, S. (2003). Ah receptor: Involvement in toxic responses. In Dioxins and Health, 2nd edition, Schecter A., and Gasiewicz, T. A., eds., John Wiley & Sons, Hoboken, NJ, pp. 493–532. Gasser, S., and Raulet, D. H. (2006). The DNA damage response arouses the immune system. Cancer Res 66(8), 3959–3962. Gaylor, D. W. (1997). Some current procedures and issues in cancer risk assessment. Hum Ecol Risk Assess 3(4), 513–519. Griffiths, C. W., Dockins, C., Owens, N., Simon, N. B., and Axelrad, D. A. (2002). What to do at low doses: A bounding approach for economic analysis. Risk Anal 22(4), 679–688. Higginson, J. (1997). Historical comments on regulation, public health, and cancer prevention. Regul Toxicol Pharmacol 26, 2–6. Holland, J. F., Frei, E., III, Bast, R. C., Jr., Kufe, D. W., Pollock, R. E., and Weichselbaum, R. R., eds. (2003). Section 4: Cancer epidemiology, prevention, and screening. In Cancer Medicine, 6th ed, Decker Inc., Hamilton, Ontario, Canada. International Commission on Radiological Protection (ICRP) (2006). Low-Dose Extrapolation of Radiation-Related Cancer Risk, Valentin, J., ed., ICRP Publication 99, Elsevier, Amsterdam, 141 pages. International Life Sciences Institute (ILSI) (1997). An evaluation of EPA’s proposed guidelines for carcinogen risk assessment using chloroform and dichloroacetate as case studies: report of an expert panel. ILSI Health and Environmental Sciences Institute, Washington, DC. Johnson, G. L., and Nakamura, K. (2007). The c-Jun kinase/stress-activated pathway: Regulation, function and role in human disease. Biochim Biophys Acta 1773(8), 1341–1348. Kinoshita, A., Wanibuchi, H., Morimura, K., Wei, M., Shen, J., Imaoka, S., Funae, Y., and Fukushima, S. (2003). Phenobarbital at low dose exerts hormesis in rat hepatocarcinogenesis by reducing oxidative DNA damage, altering cell proliferation, apoptosis and gene expression. Carcinogenesis 24(8), 1389–1399. Kodell, R. L. (2001). U-shaped dose–response relationships for mutation and cancer. Hum Ecol Risk Assess 7(4), 909–919. Leonard, G. D., Fojo, T., and Bates, S. E. (2003). The role of ABC transporters in clinical practice. The Oncologist 8, 411–424. Lohman, P. H. M., Mendelsohn, M. L., Moore, D. H. II, et al. (1992). A method for comparing and combining short-term genotoxicity text data: the basic system. Mutat Res 266, 7–25. Marsman, D. S., and Barrett, J. C. (1994). Apoptosis and chemical carcinogenesis. Risk Anal 14(3), 321–326. Motulsky, H., and Christopoulos, A. (2004). Fitting Models to Biological Data Using Linear and Nonlinear Regression: A Practical Guide to Curve Fitting. Oxford University Press, New York. Nascarella, M. A., Stanek, E. J., Hoffmann, G. R., and Calabrese, E. J. (2009). Quantification of hormesis in anticancer-agent dose–responses. Dose–Response 7(2), 160–171. National Research Council (NRC). (2006). Health Risks from Dioxin and Related Compounds: Evaluation of the EPA Reassessment, Committee on EPA’s Exposure and Human Health Reassessment of TCDD and Related Compounds, National Academies Press, Washington, D.C. National Research Council (NRC). (2009). Science and Decisions: Advancing Risk Assessment, Committee on Improving Risk Analysis Approaches Used by the US EPA, National Academies Press, Washington, D.C. National Research Council of the National Academies (NRC). (2007). Toxicity Testing in the TwentyFirst Century: A Vision and a Strategy, National Academies Press, Washington, D.C. Pedraza-Farina, L. G. (2006). Mechanisms of oncogenic cooperation in cancer initiation and metastasis. Yale J Biol Med 79, 95–103. Pollycove, M. (1998). Nonlinearity of radiation health effects. Environ Health Perspect 106(Suppl 1), 363–368. Pollycove, M., and Feinendegen, L. E. (1999). Molecular biology, epidemiology, and the demise of the linear no-threshold (LNT) hypothesis. CR Acad Sci III 322, 197–204. Pratt, I. S., and Barron, T. (2003). Regulatory recognition of indirect genotoxicity mechanisms in the European Union. Toxicol Lett 140–141, 53–62. Rhomberg, L. R. (2004). Mechanistic considerations in the harmonization of dose-response methodology: The role of redundancy at different levels of biological organization. In Risk Analysis and Society: An
680
CHAPTER 26 NONLINEAR LOW-DOSE EXTRAPOLATIONS
Interdisciplinary Characterization of the Field, McDaniels, T., and Small, M. J., eds. Cambridge University Press, Cambridge, UK, pp. 46–73. Rhomberg, L. R. (2009). Linear low-dose extrapolation for noncancer responses is not generally appropriate (Letter). Environ Health Perspect 117(4), A141–A142. Schmitt, E., Paquet, C., Beauchemin, M., and Bertrand, R. (2007). DNA-damage response network at the crossroads of cell-cycle checkpoints, cellular senescence and apoptosis. J Zhejiang Univ Sci B 8(6), 377–397. Schoen, A., Beck, B., Sharma, R., and Dube, E. (2004). Arsenic toxicity at low doses: epidemiological and mode of action considerations. Toxicol Appl Pharmacol 198, 253–267. Schulte-Hermann, R., Grasl-Kraupp, B., and Bursch, W. (2000). Dose–response and threshold effects in cytotoxicity and apoptosis. Mutat Res 464(1), 13–18. Seeley, M. R., Tonner-Navarro, L. E., Beck, B. D., Deskin, R., Feron, V. J., Johanson, G., and Bolt, H. M. (2001). Procedures for health risk assessment in Europe. Regul Toxicol Pharmacol 34, 153–169. Shackelford, R. E., Kaufmann, W. K., and Paules, R. S. (1999). Cell cycle control, checkpoint mechanisms, and genotoxic stress. Environ Health Perspect 107(1), 5–24. Slikker, W., Andersen, M. E., Bogdanffy, M. S., Bus, J. S., Cohen, S. D., Conolly, R. B., David, R. M., Doerrer, N. G., Dorman, D. C., Gaylor, D. W., Hattis, D., Rogers, J. M., Setzer, R. W., Swenberg, J. A., and Wallace, K. (2004). Dose-dependent transitions in mechanisms of toxicity: Case studies. Toxicol Appl Pharmacol 201(3), 226–294. Subramaniam, R. P., White, P., and Cogliano, V. J. (2006). Comparison of cancer slope factors using different statistical approaches. Risk Anal 26(3), 825–830. Swenberg, J. A., Ham, A., Koc, H., Morinello, E., Ranasinghe, A., Tretyakova, N., Upton, P. B., and Wu, K. (2000). DNA adducts: Effects of low exposure to ethylene oxide, vinyl chloride and butadiene. Mutat Res 464(1), 77–86. US EPA (1986). Guidelines for carcinogen risk assessment. Fed Reg 51(185), 33992(12). US EPA (2001). Toxicological Review of Chloroform (CASRN 67–66–3.) Washington, D.C., EPA/635/R-01/001, 112 pages, October. US EPA (2005a). Guidelines for Carcinogen Risk Assessment (Final). Risk Assessment Forum, Washington, DC, EPA/630/P-03/001B, http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=116283, 166 pages, March (accession date, March 30, 2005). US EPA (2005b). Guidelines for Carcinogen Risk Assessment. Risk Assessment Forum, Washington, D.C., EPA/630/P-03/001F. March. US EPA (2005c). Toxicological Review of Ingested Inorganic Arsenic (Draft), 61 pages, July. US EPA (2005d). Science Issue Paper: Mode of Carcinogenic Action for Cacodylic Acid (Dimethylarsinic Acid, DMA(V)) and Recommendations for Dose Response Extrapolation. Office of Pesticide Programs, Health Effects Div, http://www.epa.gov/oppsrrd1/reregistration/cacodylic_acid/, 201p., July 26 (accession date, August 18, 2005). US EPA (2005e). Health Effects Information Used In Cancer and Noncancer Risk Characterization For the 1999 National-Scale Assessment, http://www.epa.gov/ttn/atw/nata1999/99pdfs/healtheffectsinfo. pdf (accession date, May 1, 2009). US EPA (2008). Benchmark Dose (BMD) Methodology, Benchmark Dose Software (BMDS), http:// www.epa.gov/ncea/bmds/bmds_training/methodology/intro.htm (accession date, February 15, 2009). White, R. H., Cote, I., Zeise, L., Fox, M., Dominici, F., Burke, T. A., White, P. D., Hattis, D. B., and Samet, J. M. (2009). State-of-the-Science Workshop Report: Issues and approaches in low-dose– response extrapolation for environmental health risk assessment. Environ Health Perspect 117(2), 283–287. World Health Organization (WHO) (2007). Harmonization Project DRAFT Document for Public and Peer Review: Mutagenicity Testing for Chemical Risk Assessment. September. World Health Organization (WHO) (1998). Guidelines for Drinking-Water Quality. Second edition. Addendum to vol. 2. Health criteria and other supporting information. Chloroform. WHO, Geneva, pp. 255–275.
CH A P TE R
27
CANCER RISK ASSESSMENT: MORE UNCERTAIN THAN WE THOUGHT Edmund A. C. Crouch
27.1.
INTRODUCTION
This chapter continues the long history of empirical evaluation of quantitative extrapolation between species of measures of carcinogenicity. Such extrapolations underlie much of quantitative risk assessment for carcinogens, and current default paradigms for such extrapolations rely more on theoretical hypotheses than empirical tests. The attempt here is to use the most extensive available database summarizing results of rodent bioassays, coupled with a more thorough analysis, to evaluate the uncertainties in extrapolating a quantitative measure of carcinogenicity obtained at high doses (that is, at doses that cause readily measurable increases in tumor rates) between animal species. There is no discussion here of the many additional problems that may be present in making such extrapolations, such as the problem of relying on results observed in high-dose settings to predict risks to people exposed to low doses. For a discussion on these topics, see Chapters 24–26.
27.2.
SUMMARY OF PREVIOUS ANALYSES
All the analyses discussed here require many selections that are glossed over—for example, the tumor type(s) and site(s) to evaluate, or the combinations of such types and sites, the treatment of early mortality, extrapolation to standard lifetimes, and so forth. Efforts at performing quantitative extrapolations between species generally choose for evaluation the tumor type and site that gives the largest response, as defined in some convenient way, and do not attempt to claim concordance between tumor types or sites in the species examined. They also generally examine only those chemicals that have been declared carcinogenic in the two species examined or that satisfy some statistical test that serves the same purpose. The following summary provides an overview of the relevant literature on the subject.
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
681
682
CHAPTER 27 CANCER RISK ASSESSMENT: MORE UNCERTAIN THAN WE THOUGHT
Several analysts have evaluated concordance of carcinogenicity declarations and of tumor sites or types. Purchase (1980), for example, examined 250 chemicals in two species (usually rat and mouse) for concordance of declarations of carcinogenicity (82%) and also for concordance of the site of the tumor (64%) among those chemicals carcinogenic in both species under the conditions examined. Gold et al. (1989) reported 76% overall concordance between carcinogenicity in rats and mice among 392 compounds tested in both species, and they discussed various factors affecting concordance prediction accuracy. Subsequently, Gold et al. (1991) also evaluated whether the target site in rats predicted those results in mice and vice versa, showing that between 48% and 52% of chemicals were carcinogenic at the same site in both species. Freedman et al. (1996) present simulations based on the United States National Cancer Institute/National Toxicology Program (NCI/NTP) bioassays reported in the Carcinogenic Potency Database [CPDB; see Gold et al. (2008), for the current version] and using a one-hit dose–response model to illustrate that a declared concordance of around 75% could be due to true concordances anywhere in the broad range of 20% to 100%. Crouch and Wilson (1979) quantitatively compared the results of NCI bioassays of approximately 70 chemicals in rats (Osborne-Mendel and Fisher 344) and mice (B6C3F1), using a carcinogenic “potency” defined as the parameter β in a one-hit dose–response model:
βd ⎞ p = 1 − (1 − α ) exp ⎛⎜ − ⎟ ⎝ 1−α ⎠ where p is the lifetime probability for a tumor, d is the lifetime average dose rate (in mg/kg-day), and α is the background probability for tumor. They also attempted similar comparison between humans and animals using epidemiological evidence. With the selections they made, Crouch and Wilson (1979) presented graphical interspecies comparisons and indicated that the data were consistent with the potency parameter being equal in mice and rats, but with an uncertainty factor ranging from about 1/3 to 3. The similar interspecies factor for extrapolation to humans was indicated to be less than about 5. I extended this analysis (Crouch 1983) by providing statistics for the distribution of estimated potency ratios between rats and mice using NCI/NTP bioassay data. Anderson et al. (1983) published the methods used by the U.S. Environmental Protection Agency (EPA) for quantitative extrapolations from animals to humans. They based the extrapolations on the linear term of a multistage model (see Section 27.3), assuming that an allometric scaling law applied to that linear term such that equal doses per unit body surface area would produce equivalent lifetime probabilities of tumor in animals and humans. However, no empirical data were presented to support this view. Gaylor and Chen (1986) used the TD50 measure of carcinogenic potency (the chronic dose rate expressed in mg/kg body weight/day that halves the fraction of tumor-free animals at the end of a standard lifetime) defined and reported in the CPDB [see Gold et al. (2008); this measure is obtained using a one-hit dose– response model] to compare rats, mice, and hamsters. They chose the lowest available TD50 to characterize the effect of a chemical in each species, computed
27.2. SUMMARY OF PREVIOUS ANALYSES
683
interspecies ratios by chemical, and estimated the geometric mean values of these ratios and their geometric standard deviations, with specialization to routes of exposure (diet, gavage, water, i.p. injection, inhalation) and for specific tumor sites (all sites, total tumor-bearing animals, liver). Chen and Gaylor (1987) compared rats and mice using the NCI/NTP carcinogenesis bioassays exclusively. They defined a virtually safe dose (VSD) based on an upper confidence limit of a multistage dose– response model, and they evaluated the ratio of VSDs in mice and rats for 39 chemicals. They provided estimates of geometric mean and the logarithm of the geometric standard deviation (ln(GSD)), and they presented a stem-and-leaf plot giving some indication of the shape of the distribution of results. Allen et al. (1987, 1988), and Crump et al. (1989) provided an analysis of the epidemiological data on human carcinogens and results from rodent bioassays on the same carcinogens. They reported results in terms of a TD25 measure of carcinogenic potency (analogous to the TD50, but using 25% decrement in tumor-free fraction rather than 50%), generally using the multistage model for evaluation of the animal data, while for epidemiological observations various methods were used, matched to the available data. The authors carried out a wide-ranging quantitative evaluation of the various choices, some of which are mentioned at the beginning of this section, which can affect how interspecies comparisons are made. Some results of one of their baseline analyses are included below (Figures 27.7 and 27.8). The EPA summarized previous empirical attempts at interspecies comparisons, including those listed above, and discussed at length various observations and hypotheses about pharmacokinetics and pharmacodynamics that lead to an expectation of allometric scaling (EPA 1992). The EPA concluded that the empirical evidence at that time was consistent with allometric scaling such that lifetime cancer risks would be approximately equal for mammalian species when daily amounts administered are in proportion to body weight raised to any power between 2/3 and 1. A consensus was reached by the EPA, the U.S. Food and Drug Administration (FDA), and the U.S. Consumer Product Safety Commission (CPSC) to consider lifetime cancer risks to be equal using an allometric scaling power of 3/4, and this is the current default assumption of those agencies (this leads to the default scaling power of 1/4 for the measures of carcinogenicity discussed here). Gaylor et al. (1993) continued examination of the TD50 measures provided by the Gold et al. database (see above). Examining 30 near-replicate bioassays, they observed an approximate lognormal distribution of minimum TD50s that they attributed to experimental variation. For 229 sets of data for chemicals administered to two or more strains in the same species and sex and by the same route, the minimum TD50s were again lognormally distributed, a variation the authors attributed to a combination of experimental and genetic variation. The rat-to-mouse interspecies ratios of minimum TD50s for 190 chemicals administered in the diet were found to be lognormally distributed with a geometric standard deviation of a factor of 5.7 and a median consistent with 2/3 power allometric scaling. It was also observed that the data reported by Allen et al. (1988) corresponded to (a) ratios of TD25s being lognormally distributed with a geometric standard deviation of 10.5, (b) a median that was a factor of 3.7 away from 2/3 power allometric scaling, and (c) a further factor of 6 or 12 away from unity power allometric scaling.
684
CHAPTER 27 CANCER RISK ASSESSMENT: MORE UNCERTAIN THAN WE THOUGHT
I also (Crouch 1996) evaluated the data of Gold et al. (see above) while illustrating the use of uncertainty distributions for interspecies extrapolations, but used the multistage model (rather than the one-hit model used to derive TD50s) and a different adjustment to a standard lifetime. I demonstrated the lognormality of rat-to-mouse, mouse-to-hamster, and hamster-to-rat interspecies ratios of a carcinogenic potency parameter, and I also demonstrated that the available information was inconsistent with allometric scaling between these rodents with any power law. The work presented here extends the analyses described above by using more extensive data and taking account more completely of the uncertainties involved.
27.3. SELECTION OF CARCINOGENICITY MEASURE—THE CD10 Evaluation of the uncertainty in cancer risk assessment requires specification of what measure of carcinogenicity is to be evaluated. Current U.S. guidelines for risk assessment (EPA 2005) call for selection of a point of departure (POD) on the dose– response curve, defined as an estimated dose that is near the lower end of the observed range without significant extrapolation to lower doses. Typically for rodent bioassays the POD will be selected at a response level of 1–10%, below which a linear extrapolation would be performed in the absence of evidence contradicting such an approach. As discussed above, previous analyses have used various measures of carcinogenicity, with generally similar results obtained for all reasonable definitions. To correspond approximately with current US procedures, the analysis reported here will use the CD10 (10% cancer dose) measure, defined (as for the TD50 and TD25) as the dose giving a 10% increase in (p − p0)/(1 − p0), where p is the probability for tumor at a given dose and p0 is the probability at zero dose. The dose–response curve used here is the multistage model defined by p = 1 − exp ( − ( q0 + q1d + q2 d 2 + + qn−1d n−1 ) ) where d is the dose and q0, q1, … , qn−1 are parameters estimated from the data (see Section 27.8.1 for further details). This model was chosen for its flexibility in fitting a wide range of dose–response shapes; its selection is not expected to have a material effect on the results obtained, since other empirical dose–response curves that fit the data adequately should give similar estimates at the CD10; at this response there should be little or no extrapolation outside the experimental observations for practically all cancer bioassays.
27.4.
THE VARIATION OF CD10 WITHIN A SPECIES
In the work presented here, the CPDB compiled by Gold et al. (2008) has been used. From this, 2670 estimates for CD10 (and its uncertainty) in 660 chemicals in rats, mice, and hamsters were obtained as described in Section 27.8.1. Each estimate corresponds to an experiment categorized by chemical, species, strain, sex, and certain recorded experimental conditions, and these experiments have been assumed to be independent. Initial investigation indicated that attempting to separately discern
27.4. THE VARIATION OF CD10 WITHIN A SPECIES
685
effects of particular strains (there are 144 distinct strains of rat, mouse, and hamster included in the set of 2670 estimates), sex, or the recorded experimental conditions would not be productive. Instead, the variation of CD10 measurements within a species for each chemical was examined. There are a total of 1032 chemical–species combinations, of which 260 have just one experimental measurement of CD10, 506 have 2, 100 have 3, 66 have 4, 30 have 5, 23 have 6, 9 have 7, 10 have 8, 4 have 9, 4 have 10, 4 have 11, 3 have 12, 4 have 13, 2 have 14, 1 has 15, 1 has 16, 1 has 19, 2 have 21, 1 has 31, and finally 1 has 44 experimental measurements (2acetylaminofluorene in the mouse). Examining the distributions of the CD10 measurements within each chemical–species combination suggests that the intraspecies variation in CD10 is lognormal; Figure 27.1 shows the distributions of measurements for the four chemical–species combinations with most measurements. In Figure 27.1 and subsequent such figures, the ordinate for the ith measurement of n taken in ascending order is an approximation to a normal order statistic given by (Cunnane 1978) ⎛ i −3 8 ⎞ Φ −1 ⎜ ⎟ ⎝ n +1 4 ⎠ where Φ−1 is the inverse of the standard normal function. On such plots, random samples from a lognormal distribution will plot approximately in straight lines. Similar figures are obtained for the other chemical–species combinations; and the near uniformity (Figure 27.2) of the distribution of Shapiro–Wilk p values (Royston 1993, 1995) obtained by testing for lognormality all 166 chemical– 2.5 2 1.5
Normal order statistic
1 0.5 0 -0.5 -1 -1.5 -2 -2.5 -2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
3
log(CD10/(mg/kg-d)) 2-acetylamonofluorene/mouse
vinyl chloride/rat
DDT/mouse
N-methyl-N'-nitro-N-nitrosoguanidine/rat
Figure 27.1. The distributions of CD10 measurements (±1 SD) for the four chemical/ species combinations with most available measurements. See insert for color representation of this figure.
686
CHAPTER 27 CANCER RISK ASSESSMENT: MORE UNCERTAIN THAN WE THOUGHT 1
0.9 0.8
Cumulative fraction
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Shapiro-Wilk p value
Figure 27.2. Cumulative distribution of Shapiro–Wilk p values for the 166 chemical/ species combinations with four or more measurements.
species combinations with four or more measurements shows that a lognormal assumption for the distribution of CD10 values adequately explains the observations (this use of the Shapiro–Wilk statistic is heuristic insofar as the uncertainties of each measurement differ).
27.5. EXTRAPOLATION OF THE MEDIAN CD10 BETWEEN SPECIES Since there appears to be a lognormal distribution of CD10 for a given chemical within a given species, it is fruitless attempting to select one or another particular experiment as being in some way most representative for that species. Instead, some representation of the distribution is needed; the most obvious choice is the median and geometric standard deviation. That is the approach adopted here; an estimate of μ, the logarithm of the median CD10 and its standard deviation (see Section 27.8.2) was obtained for each of the 1032 unique chemical–species combinations. These estimates for each chemical were then compared between any two of the three species (rats, mice, and hamsters). Figure 27.3 shows the comparison for the 331 chemicals that have been assayed in both mice and rats. The distribution across chemicals of the difference between the logarithms of median CD10 estimates in rat and mouse is shown in Figure 27.4; the Shapiro–Wilk p value for normality of this difference is 0.78, indicating that the distribution of differences cannot be distinguished from normal. The straight line shows a maximum likelihood fit to this normal distribution. Figure 27.5 and Figure 27.6 show the same sort of comparisons for mouse versus hamster
27.5. EXTRAPOLATION OF THE MEDIAN CD10 BETWEEN SPECIES
687
5 4 3
Log(median CD10) - Mouse
2 1 0 -1 -2 -3 -4 -5 -5
-4
-3
-2
-1
0
1
2
3
4
5
Log(median CD10) - Rat
Figure 27.3. Comparison of median (±1 SD) CD10 estimates for 331 chemicals between rat and mouse. See insert for color representation of this figure.
3
2
Normal order statistic
1
0 -3
-2
-1
0
1
2
-1
-2
-3 Log(median CD10 rat) - Log(median CD10 mouse)
Figure 27.4. The difference (±1 SD) between logarithms of median estimates of CD10 in rat and mouse.
3
688
CHAPTER 27 CANCER RISK ASSESSMENT: MORE UNCERTAIN THAN WE THOUGHT
5 4
Log(median CD10) - Hamster
3 2 1 0 -1 -2 -3 -4 -5 -5
-4
-3
-2
-1
0
1
2
3
4
5
Log(median CD10) - Mouse 3
2
Normal order statistic
1
0 -3
-2
-1
0
1
2
-1
-2
-3 Log(median CD10 mouse) - Log(median CD10 hamster)
Figure 27.5. hamster.
Comparison of median CD10 estimates for 23 chemicals between mouse and
3
27.5. EXTRAPOLATION OF THE MEDIAN CD10 BETWEEN SPECIES
689
5 4 3
Log(median CD10) - Rat
2 1 0 -1 -2 -3 -4 -5 -5
-4
-3
-2
-1
0
1
2
3
4
5
Log(medianCD10) - Hamster
3
2
Normal order statistic
1
0 -3
-2
-1
0
1
2
-1
-2
-3 Log(median CD10 hamster) - Log(median CD10 rat)
Figure 27.6. and rat.
Comparison of median CD10 estimates for 34 chemicals between hamster
3
690
CHAPTER 27 CANCER RISK ASSESSMENT: MORE UNCERTAIN THAN WE THOUGHT
5 4 3
Human log(TD25)
2 1 0 -5
-4
-3
-2
-1
0
1
2
3
4
5
-1 -2 -3 -4 -5 Animal log(TD25)
3
2
Normal order statistic
1
0 -3
-2
-1
0
1
2
-1
-2
-3 log(TD25-human)-log(TD25-animal)
Figure 27.7. Comparison of TD25 estimates in humans and animals for 20 chemicals (Allen et al. 1987, 1988). Top panel: Error bars assumed to represent 80% confidence interval. Bottom panel: Error bars ±1 SD.
3
27.5. EXTRAPOLATION OF THE MEDIAN CD10 BETWEEN SPECIES
691
(p = 0.53 for normality of the differences) and for hamster versus rat (p = 0.61 for normality of the differences). For comparison, Figure 27.7 shows a similar comparison for individual TD25 estimates (not medians of distributions) in humans and animals derived by Allen et al. (1987, 1988, analysis 3(b); for definiteness, the semiquantitative uncertainty ranges presented by Allen et al. have been assumed to represent 80% confidence intervals). Maximum likelihood estimates for the parameters of the normal distributions of differences between log(median CD10) estimates in different species shown in Figures 27.4–27.6 are shown in Table 27.1. The same comparisons are shown graphically in Figure 27.8, plotted against the logarithm of body weight ratio for the TABLE 27.1.
Statistics for the Distribution Across Chemicals of Differences of log(CD10)
Rat Minus Mouse
Mean difference in log(CD10) Standard deviation of difference in log(CD10)
Mouse Minus Hamster
Hamster Minus Rat
Value
SD
Value
SD
Value
SD
−0.10 0.79
0.05 0.03
−0.46 0.84
0.18 0.13
0.42 0.76
0.14 0.10
1.50 H - hamster R - rat M - mouse
Log(median CD10 ratio)
1.00
HumanAnimal
H-R
0.50
0.00 R-M
-0.50 M-H
-1.00
-1.50 -1
0
1
2
3
4
Log(body mass ratio) Observed
1/4 power
1/3 power
Equal
Figure 27.8. Comparison of log(median CD10) differences (±1 SD) between rat, mouse, and hamster, and the same for human-to-animal individual comparisons, compared with default allometric scalings.
692
CHAPTER 27 CANCER RISK ASSESSMENT: MORE UNCERTAIN THAN WE THOUGHT
species involved. The current default approach for CD10 extrapolation between species (strictly, from animals to humans) is an allometric extrapolation based on the 1/4 power of body weight (EPA 1992), replacing the previous default assumption of 1/3 power body weight scaling, both of which default extrapolations are shown as lines in Figure 27.8. Some of the results of Allen et al. (1987, 1988, analysis 3(b)) are also shown in Figure 27.8. Clearly, as shown, the available evidence on mouse, rat, and hamster is not consistent with any allometric scaling approach; and the average of the individual estimates of Allen et al. (1987, 1988) is not consistent with the current default approach either.
27.6. EXTRAPOLATION OF THE INTRASPECIES VARIATION IN CD10 Figures 27.1 and 27.2 illustrate that the distribution of CD10 for a single chemical within a species is consistent with lognormal, and the previous section has examined the extrapolation of the median of that lognormal distribution between species. The analyses used to derive the median and ln(GSD) of the CD10 distribution may be extended to test hypotheses about the intraspecies variation (see Section 27.8.3). First, examination of data like those in Figure 27.1 indicates that the within-species ln(GSD) is not the same for all chemicals, and a likelihood ratio test confirms this observation (p < 10−100 on 771 degrees of freedom). Moreover, the ln(GSD) is not the same for all chemicals within each species separately (p < 10−100 on 669 degrees of freedom), nor is the ln(GSD) the same for all three species for each chemical individually (p = 4 × 10−14 on 182 degrees of freedom). These observations show that extrapolation of the ln(GSD) between species is not trivial; any such extrapolation can only be made on a probabilistic basis, so the distribution (across chemicals) of the ln(GSD) estimates were examined by species. Figure 27.9 shows the distributions across chemicals of ln(GSD) estimates obtained for rat (389 estimates), mouse (356 estimates) and hamster (27 estimates). The point estimates shown in Figure 27.9 are maximum likelihood estimates, but the profile likelihood for ln(GSD) is highly non-Gaussian in form, so the error bars shown correspond to changes of 0.5 in profile loglikelihood. The point estimates are also likely to be substantially biased, because most are based on only two or three measurements (Harville 1977). The distributions of ln(GSD) appear to be reasonably consistent with lognormal,* and fitting lognormal distributions gives parameter estimates shown in Table 27.2. Testing these parameters shows that it is plausible that those for the hamster are equal to those for the mouse, but equalities between any of the others is implausible (p < 0.05, likelihood ratio tests).
*No formal test is available. The curves have a visual appearance more consistent with a mixed distribution, with some probability for zero mixed with a lognormal. However, visual appearance is deceptive; fitting such mixed distributions indicates that a single lognormal has higher likelihood.
27.7. CONCLUSIONS
693
1
0.9
0.8
Cumulative probability
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0 0.01
0.1
1
10
ln(GSD) = SD of ln(CD10) Hamster
Mouse
Rat
Hamster Fit
Mouse Fit
Rat Fit
Figure 27.9. Distributions of ln(GSD) for hamster, mouse, and rat. Error bars correspond to changes of 0.5 in loglikelihood. See insert for color representation of this figure.
TABLE 27.2. Parameters Obtained for Fits of Lognormal Distributions to the ln(GSD) Estimates
Median of the distribution across chemicals of ln(GSD) estimates ln(GSD) of the distribution across chemicals of ln(GSD) estimates
27.7.
Rat
Mouse
Hamster
0.49
0.36
0.23
0.79
0.88
0.68
CONCLUSIONS
The results of the analysis performed here are generally in agreement with previous reports (see Section 27.2). However, they extend those reports by using a measure of high-dose carcinogenicity, the CD10, that is similar to that currently used for risk
694
CHAPTER 27 CANCER RISK ASSESSMENT: MORE UNCERTAIN THAN WE THOUGHT
assessment purposes, by taking account of the uncertainty in the CD10 carcinogenicity measure, by explicitly showing that the intraspecies and interspecies variabilities in CD10 are lognormal, and by showing that the intraspecies variability differs even in distribution between species and chemicals. The median CD10 for a chemical within a species may be extrapolated between species, but with an associated lognormal uncertainty distribution (Table 27.1). The median of the uncertainty distribution depends on the species pair, but does not appear to be predictable using any allometric scaling rule. The width of the uncertainty distribution, as given by the geometric standard deviation, is a factor of about 6.2 and does not differ significantly with species pair. The intraspecies variability of CD10 for any chemical depends on both the chemical and the species and does not appear to be predictable in one species given observations in another, even in distribution. In particular, the potential variability of CD10 between human population groups that might differ in ways corresponding somehow to the differences between laboratory strains of rats, mice, and hamsters does not appear predictable from rodent experimental data. These results challenge some current default approaches to interspecies extrapolation used for risk assessment. In particular, the selection of index animal experiment(s) on which to base the estimation of carcinogenic potency is inconsistent with the observation that there are lognormal distributions of CD10 for each chemical in each animal species; in fact, no one experiment can be singled out as representative in such circumstances. Furthermore, the use of allometric scaling for extrapolating quantitative carcinogenicity measures from animals to humans is not supported by the observations in animals, nor the limited information on humans. Additional work may clarify whether this rather bleak assessment is entirely correct. While selection of a single index experiment is inconsistent with the observations made here, it is not difficult to propose the use of a median estimate derived from multiple bioassays. However, the optimum procedure to follow when there are bioassays available in two or more rodent species is not clear. Allometric scaling is ruled out by the observations, but estimation of a better (perhaps entirely empirical) replacement would substantially benefit from an updated examination of epidemiological data and incorporation of corresponding more recent, standardized, rodent bioassays. Extrapolation of intraspecies variability directly between species pairs appears impossible; but that does not rule out the possibility of a relationship between species triples, nor the possibility of correlations with other endpoints. Even if no such extrapolation procedure can be devised, it is nevertheless possible to devise a probabilistic extrapolation that takes account of the lack of correlation.
ACKNOWLEDGMENTS I should like to thank Cambridge Environmental Inc. for funding this work and Sara Hendrix for producing an annotated and categorized bibliography.
27.8. APPENDIX
27.8. 27.8.1.
695
APPENDIX Obtaining the CD10 Estimates
Each database entry was fitted with the multistage model (Anderson et al. 1983; Crump 1984) starting with six parameters. A recursive procedure was followed to obtain an adequate fit, remove unnecessary parameters, and estimate the likelihood that the observations were random (i.e., had no dose–response). In each recursion, the probability for a non-dose-related curve was estimated using the change in likelihood from the current fit versus a one-parameter fit, assuming a number of degrees of freedom equal to max(1, min(5, D) − 1), where D is the number of dose groups with nonzero response. The smallest such probability obtained during the recursive procedure was taken as an estimate of the probability that the observed results were not dose-related, and the database entry was deleted if this probability equaled or exceeded 0.05 (of the 2670 results used here, 468 have minimum probability exceeding 0.01, 496 between 0.001 and 0.01, 308 between 0.0001 and 0.001, and the remaining 1398 below 0.0001). If the multistage dose–response curve did not adequately fit the observations, as judged by p ≤ 0.01 using the standard chi-squared approach (Anderson et al. 1983), the highest dose was recursively removed until an adequate fit was obtained or just two doses remained. Then parameters were recursively removed from the multistage model while leaving the fit adequate (p > 0.01) or until only two parameters remained (if the six-parameter fit was always inadequate, two parameters were used). At this point, the MLE and the upper 95th percentile (UCL95) on the linear term of the multistage model were estimated (Anderson et al. 1983), and the MLE CD10 estimate was obtained. In addition, an estimate of the standard deviation (SD) of the CD10 estimate was obtained as the square root of the inverse of the second derivative of the profile likelihood for CD10. Subsequently, the SD estimates were interpreted as being symmetrical on a logarithmic scale, since typically confidence limits are substantially less skewed on that scale. All CD10 and SD estimates were then corrected to a 104-week “standard lifetime” using a cubic power law applied to the length of time on test. Each entry in the database corresponds to a tumor type. Those coded as “Berkeley mix,” “mandatory mix,” and “total tumor bearing” were omitted (the first two were constructed by the CPDB team; the last was omitted as being relatively uninformative). Results for species other than rats, mice, and hamsters were omitted; there were too few results on monkeys to be analyzed. With this approach, there were 14,859 records that were not excluded (by the exclusions of the last paragraphs). However, 133 of these records were omitted from further examination here because the MLE for CD10 was zero (100% response in all the dosed animals, after any censoring of high doses described above). Each of the records corresponds to one tumor type and tumor site in an “experiment” characterized by chemical name, source of the result (paper, NTP experiment, and so forth), species, strain, sex, route, dosing time, and observation time. Each such “experiment” was assumed to be an independent measurement of carcinogenicity. The single record (tumor response site and type) selected for each experiment was
696
CHAPTER 27 CANCER RISK ASSESSMENT: MORE UNCERTAIN THAN WE THOUGHT
that with the largest UCL95 on the linear term of the multistage model selected by the recursive procedure described (this selection is almost equivalent to selecting the record with lowest lower confidence limit on CD10). This gave 3070 selected experimental CD10 measures for 1059 distinct chemicals. Since the concern here is with variation among and extrapolation of results for any given chemical, all chemicals for which only one CD10 result was available were omitted, leaving 2671 estimates for 660 chemicals; and finally, one estimate for which the CD10 estimation procedure described above failed was omitted, giving 2670 estimates for 660 chemicals.
27.8.2.
Median and Geometric Standard Deviation
The median and geometric standard deviation for the distribution of CD10 values in each chemical–species combination were obtained by using a mixed-effects model of the form ln ( di ) = μ + Si
where Si = σ 2 + si2
where di is the observed CD10 for experiment i, with estimated standard deviation si, μ is the logarithm of median CD10 in the chemical/species combination, and σ is the logarithm of the geometric standard deviation for the chemical–species combination. The loglikelihood for this model is just 2 k ⎛ 1 ⎛ μ − ln ( di ) ⎞ ⎞ −∑ ⎜ ln Si + ⎜ ⎟ ⎟ 2⎝ Si i =1 ⎝ ⎠ ⎠
for the k measurements of CD10 in the chemical–species combination. An MLE estimate of μ was obtained using this likelihood, and its standard deviation was estimated by using the square root of the inverse of the second derivative of the profile likelihood for μ. It is well known that this estimate is biased low, so the value obtained was multiplied by k ( k − 1) as a correction, although the effect on the analysis here is very small. For the interspecies comparisons in those cases where k = 1, the standard deviation of μ was estimated as the quadrature sum of s1 and the mean estimate for σ derived below (see Section 27.8.3). The maximum likelihood estimate for σ, the ln(GSD), was also obtained, and is used in some of the figures in the text. However, the profile likelihood for the σ is not well-approximated by a Gaussian form, so in further analysis the full profile likelihood was used.
27.8.3.
Testing Hypotheses about ln(GSD)
The loglikelihoods used in Section 27.8.2 were summed across all chemical–species combinations and used to test hypotheses about σ using likelihood ratio tests. Care was taken to select only those chemical–species combinations that contributed information to the test; for example, those chemical–species combinations with only one measurement have to be excluded (since they contain no information on the σ). There are 772 chemical–species combinations with more than one measurement, 3
REFERENCES
697
species involved, 660 chemicals, and 349 experiments in 167 chemicals for which 2 or more measurements are available in 2 or more species. The distributions of σ shown in Figure 27.9 were fitted using the full profile likelihoods derived from the likelihoods given in Section 27.8.2. The likelihood for the lognormal approximations to the distributions of σ were thus convolution integrals of such profile likelihoods with a lognormal and were evaluated numerically. Mean estimates for σ for the three rodent species were obtained from the parameters of the fitted lognormal distributions (Table 27.2).
REFERENCES Allen, B. C., Shipp, A. M., Crump, K. S., Kilion, B., Hogg, M. L., Tudor, J., and Keller, B. (1987). Investigation of cancer risk assessment methods (4 parts: Summary and Vols. 1–3) EPA/600/6-87/007a,b,c,d. Allen, B. C., Crump, K. S., and Shipp, A. M. (1988). Correlation between carcinogenic potency of chemicals in animals and humans. Risk Anal 8(4), 531–544. Anderson, E. L., and the Carcinogen Assessment Group of the US Environmental Protection Agency. (1983). Quantitative approaches in use to assess cancer risk. Risk Anal 3(4), 277–295. Chen, J. J., and Gaylor, D. W. (1987). Carcinogenic risk assessment: Comparison of estimated safe doses for rats and mice. Environ Health Perspect 72, 305–309. Crouch, E., and Wilson, R. (1979). Interspecies comparison of carcinogenic potency. J Toxicol Environ Health 5(6), 1095–1118. Crouch, E. A. C. (1983). Uncertainties in interspecies extrapolations of carcinogenicity. Environ Health Perspect 50, 321–327. Crouch, E. A. C. (1996). Uncertainty distributions for cancer potency factors: Laboratory animal carcinogenicity bioassays and interspecies extrapolation. Human Ecol Risk Assess 2(1), 103–129. Crump, K. S. (1984). An improved procedure for low-dose carcinogenic risk assessment from animal data. J Environ Pathol Toxicol Oncol 5(4–5), 339–348. Crump, K., Allen, B., and Shipp, A. (1989). Choice of dose measure for extrapolating carcinogenic risk from animals to humans: an empirical investigation of 23 chemicals. Health Physics 57(Suppl 1), 387–393. Cunnane, C. (1978). Unbiased plotting positions—A review. J Hydrol 37(3–4), 205–222. EPA (US Environmental Protection Agency) (1992). Draft report: A cross-species scaling factor for carcinogen risk assessment based on equivalence of mg/kg3/4/day; Notice. Fed Reg 57, 24152–24173. EPA (US Environmental Protection Agency) (2005). Guidelines for carcinogen risk assessment. EPA/630/P-03/001F. Freedman, D. A., Gold, L. S., and Lin, T. H. (1996). Concordance between rats and mice in bioassays for carcinogenesis. Regul Toxicol Pharmacol 23(3), 225–232. Gaylor, D. W., and Chen, J. J. (1986). Relative potency of chemical carcinogens in rodents. Risk Anal 6(3), 283–290. Gaylor, D. W., Chen, J. J., and Sheehan, D. M. (1993). Uncertainty in cancer risk estimates. Risk Anal 13(2), 149–154. Gold, L. S., Ames, B. N., Bernstein, L., Blumenthal, M., Chow, K., Da Costa, M., de Veciana, M., Eisenberg, S., Garfinkel, G. B., Haggin, T., Havender, W. R., Hooper, N. K., Levinson, R., Lopipero, P., Magaw, R., Manley, N. B., MacLeod, P. M., Peto, R., Pike, M. C., Rohrbach, L., Sawyer, C. B., Slone, T. H., Smith, M., Stern, B. R., and Wong, M. (2008). The Carcinogenic Potency Database (CPDB), http://potency.berkeley.edu/ (accession date, 1/29/2009; last data update August 2007). Gold, L. S., Bernstein, L., Magaw, R., and Slone, T. H. (1989). Interspecies extrapolation in carcinogenesis: prediction between rats and mice. Environ Health Perspect 81, 211–219. Gold, L. S., Slone, T. H., Manley, N. B., and Bernstein, L. (1991). Target organs in chronic bioassays of 533 chemical carcinogens. Environ Health Perspect 93, 233–246.
698
CHAPTER 27 CANCER RISK ASSESSMENT: MORE UNCERTAIN THAN WE THOUGHT
Harville, D. A. (1977). Maximum likelihood approaches to variance component estimation and to related problems. J Am Stat Assoc 72(358), 320–338. Purchase, I. F. H. (1980) Inter-species comparisons of carcinogenicity. Br J Cancer 41(3), 454–468. Royston, P. (1993). A toolkit for testing for non-normality in complete and censored samples. The Statistician 42(1), 37–43. Royston, P. (1995). Remark AS R94. A remark on algorithm AS 181: The W-test for normality. J R Stat Soc Ser C Appl Stat 44(4), 547–551.
CH A P TE R
28
COMBINING NEOPLASMS FOR EVALUATION OF RODENT CARCINOGENESIS STUDIES Amy E. Brix Jerry F. Hardisty Ernest E. McConnell
28.1.
INTRODUCTION
It is well-accepted that the pathophysiology of some of the more common neoplastic processes observed in rodent carcinogenicity studies are a progression from hyperplastic precursor lesions through benign lesions to malignant lesions. Furthermore, several significant neoplastic processes in the rodents are systemic neoplasms, such as the lymphomas and leukemias. Should these neoplasms be recorded as a single entity, or as a separate diagnosis, depending on the organ where they reside? Despite our knowledge of the progression or systemic nature of selected neoplasms, however, many times the most important endpoint of the chronic experimental study in rats or mice is the incidence of a malignant neoplastic response in the treated groups relative to that of the control group. When is it acceptable to combine the incidences of the hyperplastic lesions with those of the neoplastic lesions for evaluation of the study results? When is it acceptable to combine benign neoplasms with malignant ones? When is it acceptable to combine various sites throughout the body in the case of systemic neoplasms? In 1986, Ernest E. McConnell and colleagues published an article that provided guidelines for combining benign and malignant neoplasms and hyperplasias with neoplasias (McConnell et al. 1986). The work of McConnell and colleagues is still valid today, and this book chapter, with permission from Dr. McConnell, will draw heavily on the recommendations from the 1986 publication. Regulatory decisions concerning the carcinogenic potential of chemicals are based primarily on the results of long-term studies in experimental animals, particularly rats and mice. During the past decade many chemicals have been studied
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
699
700
CHAPTER 28
EVALUATION OF RODENT CARCINOGENESIS STUDIES
in rodents, and the study results have shown that carcinogenicity is not simply an all-or-nothing phenomenon (IARC 1987). The present chapter deals with one facet of the pathologic evaluation process—that is, the combination of neoplasms. Since the introduction of the process, the toxicologist and the pathologist are often confronted with the following question: “To evaluate the statistical and biological significance of observed neoplasms, can the incidence of neoplasms in the same organ or tissue or in different organs or tissues in which the morphology of the tumors is comparable be combined?” The relevance of this question is obvious if one realizes that combining or not combining neoplasms may affect carcinogenic evaluation of a given chemical. This can best be illustrated by the following hypothetical examples. In carcinogenesis studies, treated animals may exhibit increased incidences of benign and malignant neoplasms, each originating from the same cell type, but without the increases being statistically significant when analyzed separately. However, when the incidence data are combined, the increases may be statistically significant. For example, four benign and four malignant neoplasms of the same cell type are found in a group of 50 treated animals versus none in 50 control animals. When the incidence data are analyzed separately, the increase in benign and malignant neoplasms does not reach significance with the use of Fisher ’s exact probability test (one-sided; P = 0.059), but, when combined (0/50 versus 8/50), the increase is highly significant (one-sided; P = 0.003). However, an opposite effect may occur as well. For example, if the incidence of benign neoplasms is zero in the control group and six in the treated group, the comparison would be statistically significant (P < 0.05). But if in the same control group there were two malignant neoplasms of the same type and one in the treated group, the combination would be 2 in 50 versus 7 in 50 and, therefore, not statistically significant (P > 0.05). A particular chemical that induces a statistically significant shift in tumor expression from benign to malignant without the total incidence increasing may be regarded as a carcinogen. However, when the benign and malignant neoplasms are combined in such a case, the study would be classified as negative, implying that the chemical is noncarcinogenic. Thus combining neoplasms in this case would result in a false-negative effect. These hypothetical examples show that there are good arguments for and against the combining of neoplasms in the evaluation of rodent carcinogenesis studies; thus the reasons for and results of such combinations must be evaluated on a study-by-study basis. A good policy would be that benign and malignant neoplasms of the same cell of origin should be analyzed both separately and in combination. The same applies to neoplasms that have the same histogenesis but show different morphologic or cellular features. However, since there are a number of well-reasoned arguments favoring and opposing combination of tumor data (NTP 1984), guidelines for combining neoplasms are helpful when evaluating the incidence of neoplasms in individual or related studies. The purpose of this chapter is to provide the rationale and criteria for combining certain neoplasms and to provide guidelines for combining neoplasms and sites.
28.2. RATIONALE FOR COMBINING NEOPLASMS
28.2.
701
RATIONALE FOR COMBINING NEOPLASMS
In order to consider whether or not to combine neoplasms, consistent nomenclature must be developed and used, based on commonly applied diagnostic criteria. Much effort has gone into the standardization of diagnosing proliferative lesions in rodents by the various Societies of Toxicologic Pathology and by the International Agency for Research on Cancer (IARC). Four major considerations provided the basis for the decision to combine certain neoplasms for the evaluation of evidence of the carcinogenicity of a given chemical: (a) The terms “benign” and “malignant” are important for estimation of clinical prognosis, but they may represent an artificial division of neoplastic development and progression. This consideration is based on the following two facts: (1) The morphologic criteria for differentiating borderline benign and malignant neoplasms are often subjective and arbitrary. In fact, for certain lesions it is difficult to differentiate hyperplasia from neoplasia. Examples are pituitary gland lesions in rats (ICRT 1994a; Lee et al. 1982; Majka et al. 1990; Trouillas et al. 1982), thyroid follicular cell lesions in mice (Capen et al. 2001; Frith and Heath 1984), thyroid C-cell lesions in rats (Botts et al. 1991; DeLellis et al. 1979; ICRT 1994a), adrenocortical lesions in mice and rats (Brown et al. 1995; Capen et al. 2001; ICRT 1994a), adrenal medullary lesions in rats (Brown et al. 1995; ICRT 1994a), pancreatic acinar cell lesions in rats (Boorman and Eustis 1984; Hansen et al. 1995), pancreatic islet cell neoplasms in mice (Capen et al. 2001), and renal tubular cell lesions in mice and rats (Hard et al. 1995, 2001; ICRT 1992c). (2) For some neoplasms, there is substantial evidence for the sequential progression from the hyperplastic to the benign stage and from the benign to the malignant stage. Progression has been suggested for the following lesions: epidermal skin lesions in mice (Bruner et al. 2001), alveologenic lesions in mice (Dungworth et al. 2001), esophageal and forestomach lesions in rats (Frantz et al. 1991; ICRT 1997c), bladder urothelial lesions in rats (Cohen 1983; Frith et al. 1995; ICRT 1992c; Kunze 1979), testicular interstitial cell lesions in rats (ICRT 1997a; McConnell et al. 1992; Turek and Desjardins 1979), and prostatic lesions in rats (Bosland et al. 1998; ICRT 1997a; Reznik et al. 1981; Ward et al. 1980). A direct transition from the hyperplastic or dysplastic to the malignant stage has been suggested for nasal cavity, glandular stomach, and thyroid C-cell lesions in rats (DeLellis et al. 1979; Kunze et al. 1979; Takano et al. 1982), although others do recognize adenoma as an intermediate stage in the development of thyroid C-cell carcinomas (Botts et al. 1991; ICRT 1994a). Transition from a benign to a malignant neoplasm has been suggested for harderian gland neoplasms in mice (Krinke et al. 2001a; Sheldon et al. 1983), mammary gland neoplasms in rats (Mann et al. 1996), preputial and clitoral gland neoplasms in rats (Evans et al. 1997; Reznik and Ward 1981; Yoshida 1983), Clara cell neoplasms of the lung in mice (Kauffman et al. 1979), and small intestine neoplasm in rats (ICRT 1997c; Whiteley et al. 1996). In addition, it has been suggested
702
CHAPTER 28
EVALUATION OF RODENT CARCINOGENESIS STUDIES
that hemangiomas are the first stage of hemangiosarcomas, although this viewpoint is not universally accepted (Ruben et al. 1997). Therefore, the classification of the various stage of neoplastic formation into distinct categories in isolation from each other in cases where sequential progression is known may obfuscate a true effect. (b) Neoplasms arising from one cell type may show features of cellular differentiation, transformation, or various morphologic patterns. Subclassifying these neoplasms according to their predominant cell type or pattern and evaluating them independently may make a possible carcinogenic effect more or less apparent. Examples of neoplasms in which cells may differentiate in various directions are epithelial basal cell, Zymbal’s gland, preputial gland, and clitoral gland neoplasms. For example, epithelial basal cells are capable of differentiating into any of the various types of epidermal or adnexal cells (Bruner et al. 2001; Evans et al. 1997). In addition, there are many examples of neoplasms of the same cell of origin showing different morphologic growth patterns; for example, mammary gland and endocrine organ neoplasms may show cystic, papillary, or solid features and may, in fact, exhibit more than one pattern in the same neoplasm. (c) In most carcinogenesis studies in which rodents are used, time and resources do not allow for step-sectioning (multiple sections) of a given “benign” lesion to determine whether malignant areas are present. Also, the use of multiple time-sequential scheduled sacrifices to determine the progression from nonneoplastic lesions through the neoplastic stage is rarely included in carcinogenicity studies. (d) The categorization of a neoplasm as benign or malignant may bear little relevance to its adverse biologic potential. Some benign neoplasms are as lifethreatening as their malignant counterparts. Examples are squamous cell papillomas in the oral cavity or esophagus that may prevent food intake and ultimately lead to death due to starvation; benign neoplasms of the central nervous system or vertebrae that cause compression giving rise to paresis and paralysis, which in turn may lead to death due to starvation and/or uremia; thymomas causing compression of the lung and displacement of the heart leading to respiratory and cardiac failures; internal hemangiomas that rupture leading to lethal blood loss; mammary gland fibroadenomas that may reach such a large size that animals must be killed for humane reasons; and cholangiofibrosis/ cholangiocarcinomas in the liver of rats exposed to furans and dioxin-like compounds.
28.3. USEFULNESS OF DIFFERENTIATING BENIGN FROM MALIGNANT NEOPLASMS AND OF SUBCLASSIFYING NEOPLASMS On the basis of the above reasons for combining neoplasms, one might get the impression that any attempt to differentiate benign from malignant lesions and to
28.3. USEFULNESS OF DIFFERENTIATING
703
subclassify tumors would be useless; instead, one could simply diagnose the lesions as neoplasms or record only the most severe one. This would be a mistake because: (a) Differentiating benign and malignant neoplasms provides at least circumstantial evidence for the aggressiveness of a particular type of neoplasm, thereby providing better data for interpretation of carcinogenicity and for use in hazard identification and risk assessment. Malignancy implies a more extensive disease process and by definition irreversibility. (b) A chemical that induces a shift in tumor expression from benign to clearly malignant without the total incidence increasing might be misclassified as noncarcinogenic if the benign and malignant tumors were not recorded separately but simply recorded as neoplasms, or only malignant neoplasms were recorded when both may be present in the organ. For example, if both a hepatocellular adenoma and hepatocellular carcinoma are present in the same animal, both should be recorded. (c) Studies on multistage carcinogenesis have demonstrated that sequential mutational events may be involved in the progression from benign to malignant neoplasms (Foley et al. 1991; Hennings et al. 1983; Potter 1984). Thus differentiating benign and malignant neoplasms may provide information relevant to the mechanistic action of a chemical. (d) Several types of malignant neoplasms appear to arise de novo. Examples are leukemia, rhabdomyosarcoma, certain epithelial neoplasms of the gastrointestinal tract, and primitive neoplasms such as nephroblastomas, neuroblastomas, and teratomas. Therefore, the relevance of such an observation would be lost by merely calling them neoplasms of their respective organs. (e) Some benign neoplasms have the capacity to regress to nonneoplastic lesions, particularly when exposure to the inciting chemical ceases. This is a commonly observed phenomenon for papillomas occurring in skin carcinogenesis studies (Burns et al. 1976; Iversen 1982). The mechanisms underlying regression are unknown, but they are probably complex; a high cellular turnover rate, vascular insufficiency, physical removal due to scratching, fighting, or cannibalism, and immune mechanisms may all be involved (Iversen 1982). It is also known that premalignant mammary gland lesions in mice may regress (Sass et al. 1982). The occurrence of remission due to encapsulation and tumor necrosis has also been suggested for harderian gland adenomas (Sheldon et al. 1983). (f) Subclassification of neoplasms has the advantage that strain, time, and treatment-related differences in tumor types will become apparent, and it facilitates a comparison of results among laboratories. (g) Neoplasia represents a complex disease involving many modes and mechanisms of action. To diagnose a lesion merely as neoplastic would ignore this complexity and would not communicate to other investigators the type of information needed for decision-making (Haseman 1983).
704
CHAPTER 28
28.4.
EVALUATION OF RODENT CARCINOGENESIS STUDIES
CRITERIA FOR COMBINING NEOPLASMS
Criteria for combining neoplasms are based on the following: (a) Substantial evidence exists for progression of benign to malignant neoplasms of the same histomorphogenic type. Progression is considered more important if demonstrated within the study in question than if comparisons must be made with past experience (although this knowledge is valuable). (b) The occurrence of hyperplasia may be used as supporting evidence alone, but more so when the criteria for differentiating hyperplasia from benign neoplasia are not clear (i.e., borderline lesions) or when they are arbitrary and do not reflect the biologic potential of a given lesion. (c) Most neoplasms of the same histomorphogenic type are combined even if they occur in different anatomic sites. (d) Neoplasms of different morphologic classification may be combined when their histomorphogenesis is comparable. These criteria are used as a first step in combining neoplasms. Other combinations are possible and are used at times, but these have ordinarily been given less scientific weight in the interpretation of carcinogenic potential.
28.4.1.
Combinations According to Organ And Tissue
Table 28.1 contains a list of organs and tissues where combining neoplasms is or is not appropriate to obtain a better understanding of the evidence of carcinogenicity. This list comprises those organs and tissues in which neoplasia is most often observed in F344 rats and B6C3F1 mice and may or may not be appropriate for use in other strains or species. Organs and tissues not currently on the list would be considered case by case. As the depth of knowledge increases in regard to the biologic behavior of neoplasms in a given organ or tissue, certain combinations may become inappropriate or more appropriate in the future. In addition, there may be occasions where lesions not normally combined would be examined in combination if thought to be appropriate—for example, neoplasms of different morphologic types in the same organ to establish a target organ effect. This would be most appropriate; however, when a mode(s) of action or biological mechanism(s) of action are identified.
28.4.2.
Combinations by Site
Table 28.2 contains a list of organs and tissues where combining of neoplasms at different sites is or is not appropriate in the F344 rat and B6C3F1 mouse. Realizing that exceptions are possible for a given tissue site, one should view this table as a general guideline. Some of the more notable exceptions are the following:
28.4. CRITERIA FOR COMBINING NEOPLASMS
TABLE 28.1.
Organ System
Guidelines for Combining Neoplasms in the Rat and the Mouse
Neoplasms
Integumentary System Skin Basal cell neoplasms—all types Squamous cell papillomas and squamous cell carcinomas Squamous cell neoplasms and keratoacanthomas Squamous cell neoplasms and adnexal neoplasms Subcutis Benign mesenchymal neoplasms—various types Malignant mesenchymal neoplasms—various types Benign and malignant mesenchymal neoplasms of the same type Mammary gland Adenomas and fibroadenomas Carcinomas—various types Adenoma, adenocarcinoma, and carcinoma Mammary gland fibroadenomas and fibromas–fibrosarcomas of the subcutis Preputial, Adenomas—various types clitoral, or Zymbal’s Carcinomas—various types gland Adenomas and carcinomas
Respiratory System Nasal cavity Squamous cell papillomas and squamous cell carcinomas
Lung
705
Squamous cell neoplasms and glandular cell neoplasms Esthesioneural epithelial neoplasms and other neoplasms Bronchioalveolar adenomas and bronchioalveolar carcinomas Squamous cell neoplasms and bronchioalveolar neoplasms
Combine Yesa Yesa Yes Sometimesb No No Sometimesb
Reference
Bruner et al. (2001), Evans et al. (1997), ICRT (1993b) Bruner et al. (2001), Evans et al. (1997), ICRT (1993b) Bruner et al. (2001), Evans et al. (1997), ICRT (1993b) Bruner et al. (2001), Evans et al. (1997), ICRT (1993b) Bruner et al. (2001), Evans et al. (1997), ICRT (1993b) Bruner et al. (2001), Evans et al. (1997), ICRT (1993b) Bruner et al. (2001), Evans et al. (1997), ICRT (1993b)
Sometimesc Yes Sometimesb
Mann et al. (1996) Mann et al. (1996) Mann et al. (1996)
Sometimesd
Bruner et al. (2001), Evans et al. (1997), ICRT (1993b), Mann et al. (1996) Bruner et al. (2001), Evans et al. (1997), ICRT (1993b) Bruner et al. (2001), Evans et al. (1997), ICRT (1993b) Bruner et al. (2001), Evans et al. (1997), ICRT (1993b), Reznik and Ward (1981), Yoshida (1983)
Yes Yes Yes
Sometimesb
No No
Yesa
No
Dungworth et al. (2001), ICRT (1992a), Schwartz et al. (1994), Takano et al. (1982) Dungworth et al. (2001), ICRT (1992a), Schwartz et al. (1994) Dungworth et al. (2001), ICRT (1992a), Schwartz et al. (1994) Dungworth et al. (2001), ICRT (1992a), Kauffman et al. (1979), Schwartz et al. (1994) Dungworth et al. (2001), ICRT (1992a), Schwartz et al. (1994)
(Continued)
706
CHAPTER 28
EVALUATION OF RODENT CARCINOGENESIS STUDIES
TABLE 28.1. (Continued)
Organ System
Neoplasms
Cardiovascular System Vascular Hemangiomas and endothelium hemangiosarcomas Hematopoietic System Rat Large granular lymphoma (mononuclear cell leukemia) and leukemia—other types Malignant lymphomas-all types Large granular lymphoma (Mononuclear cell leukemia) and malignant lymphomas— all types Malignant lymphomas—all types—and histiocytic sarcomas Mouse Malignant lymphomas—all types Malignant lymphomas—all types—and histiocytic sarcomas Malignant lymphomas—all types—and lymphocytic leukemia Leukemias—all types Digestive System Oral cavity, esophagus, and forestomach Glandular stomach Small and large intestines
Combine
Reference
Yes
Elwell et al. (2004), Ruben et al. (1997)
Yes
Frith et al. (1996), ICRT (1993a)
Yes Sometimes
Frith et al. (1996), ICRT (1993a), Squire et al. (1981), Stromberg et al. (1983)
No
Frith et al. (1996, 2001), ICRT (1993a), Squire et al. (1981)
Yes
Frith et al. (2001)
No
Frith et al. (2001), Pattengale and Frith (1983)
Yes
Frith et al. (2001), Pattengale and Frith (1983)
No
Frith et al. (2001)
Squamous cell papillomas and squamous cell carcinomas
Yesa
Glandular adenomas and adenocarcinomas
Yes
Adenomatous polyps, adenomas, and adenocarcinomas Benign mesenchymal neoplasms—various types Malignant mesenchymal neoplasms—various types Benign and malignant mesenchymal neoplasms of the same time
Yes
Betton et al. (2001), Frantz et al. (1991), ICRT (1997c), Whiteley et al. (1996) Betton et al. (2001), Frantz et al. (1991), ICRT (1997c), Kunze et al. (1979) Betton et al. (2001), ICRT (1997c), Whiteley et al. (1996)
No No Sometimesb
Betton et al. (2001), ICRT (1997c), Whiteley et al. (1996) Betton et al. (2001), ICRT (1997c), Whiteley et al. (1996) Betton et al. (2001), ICRT (1997c), Whiteley et al. (1996)
(Continued)
28.4. CRITERIA FOR COMBINING NEOPLASMS
707
TABLE 28.1. (Continued)
Organ System Liver
Exocrine pancreas
Urinary System Kidney
Urinary bladder
Endocrine System Pituitary gland
Thyroid gland
Pancreatic islets
Neoplasms Hepatocellular adenomas, hepatocellular carcinomas, and hepatoblastomas Bile duct neoplasms and hepatocellular neoplasms Hepatocellular neoplasms and vascular endothelial neoplasms Acinar cell adenomas and acinar cell carcinomas
Tubular cell adenomas and tubular cell carcinomas Transitional cell adenomas and transitional cell carcinomas Tubular cell neoplasms and transitional cell neoplasms Mesenchymal neoplasms and epithelial neoplasms Transitional cell papillomas and transitional cell carcinomas
Combine
Reference
Yes
Deschl et al. (2001), Goodman et al. (1990), ICRT (1997c)
No
Deschl et al. (2001), Goodman et al. (1990), ICRT (1997c) Deschl et al. (2001), Goodman et al. (1990), ICRT (1997c)
No
Yesa
Boorman and Eustis (1984), Deschl et al. (2001), Goodman et al. (1990), ICRT (1997c)
Yesa
Hard et al. (1995, 2001), ICRT (1992c) Cohen (1983), Hard et al. (1995, 2001), ICRT (1992c), Kunze (1979) Hard et al. (1995, 2001), ICRT (1992c) Hard et al. (1995, 2001), ICRT (1992c) Cohen (1983), Frith et al. (1995), Hard et al. (2001), ICRT (1992c), Kunze (1979)
Yesa
No No Yesa
Adenomas and carcinomas
Yesa
Follicular cell adenomas and follicular cell carcinomas
Yesa
C-cell adenomas and C-cell carcinomas
Yesa
Follicular cell neoplasms and C-cell neoplasms Adenomas and carcinomas
No Yesa
Capen et al. (2001), ICRT (1994a), Lee et al. (1982), Majka et al. (1990), Trouillas et al. (1982) Botts et al. (1991), Capen et al. (2001), Frith and Heath (1984), ICRT (1994a) Botts et al. (1991), Capen et al. (2001), DeLellis et al. (1979), ICRT (1994a) Botts et al. (1991), Capen et al. (2001), ICRT (1994a) Capen et al. (2001), ICRT (1994a), Riley et al. (1990)
(Continued)
708
CHAPTER 28
EVALUATION OF RODENT CARCINOGENESIS STUDIES
TABLE 28.1. (Continued)
Organ System Adrenal gland
Genital System Ovary and testicle
Uterus and cervix
Prostate gland
Neoplasms
Combine
Reference
a
Boorman and Eustis (1984), Brown et al. (1995), Capen et al. (2001), ICRT (1994a) Brown et al. (1995), Capen et al. (2001), ICRT (1994a)
Yesa
Boorman and Eustis (1984), Brown et al. (1995), Capen et al. (2001), ICRT (1994a) Brown et al. (1995), Capen et al. (2001), ICRT (1994a) Brown et al. (1995), Capen et al. (2001), ICRT (1994a)
Cortical adenomas and cortical carcinomas
Yes
Cortical neoplasms and subscapular neoplasms (mouse) Pheochromocytomas and malignant pheochromocytomas Cortical neoplasms and medullary neoplasms Subscapular neoplasms and medullary neoplasms (mouse)
No
No No
Germ cell neoplasms-all types
Yes
Stromal neoplasms-all types
Yes
Germ cell neoplasms and stromal neoplasms
No
Glandular adenomas and glandular carcinomas Stromal polyps and stromal sarcomas Stromal neoplasms and glandular neoplasms Adenomas and carcinomas
Yes
Yesa
Gliomas—various types
Yes
Gliomas and medulloblastomas
No
Yes No
Davis et al. (2001), Dixon et al. (1999), ICRT (1997a,b), McConnell et al. (1992), Rehm et al. (2001) Davis et al. (2001), Dixon et al. (1999), ICRT (1997a,b), McConnell et al. (1992), Rehm et al. (2001) Davis et al. (2001), Dixon et al. (1999), ICRT (1997a,b), McConnell et al. (1992), Rehm et al. (2001) Davis et al. (2001), Dixon et al. (1999), ICRT (1997b) Davis et al. (2001), Dixon et al. (1999), ICRT (1997b) Davis et al. (2001), Dixon et al. (1999), ICRT (1997b) Bosland et al. (1998), ICRT (1997a), Rehm et al. (2001), Reznik et al. (1981), Ward et al. (1980)
Nervous System ICRT (1994b), Krinke et al. (2001b), Solleveld et al. (1991) ICRT (1994b), Krinke et al. (2001b), Solleveld et al. (1991)
(Continued)
28.4. CRITERIA FOR COMBINING NEOPLASMS
709
TABLE 28.1. (Continued)
Organ System
Neoplasms
Combine
Granular cell neoplasms and gliomas Nerve cell neoplasms and gliomas Meningiomas—all types and other central nervous system neoplasmse
No No No
Reference ICRT (1994b), Krinke et al. (2001b), Solleveld et al. (1991) ICRT (1994b), Krinke et al. (2001b), Solleveld et al. (1991) ICRT (1994b), Krinke et al. (2001b), Solleveld et al. (1991)
Skeletal System Bone neoplasm and cartilage neoplasms
Sometimesb
Ernst et al. (2001), ICRT (1992b), Long et al. (1993)
a
Neoplasms where the incidence of hyperplasia is taken into consideration in the evaluation of a carcinogenic response or mixture of related types of cells. b
If a continuum is observed in a given study.
c
In general, these tumors should not be combined, because they are thought to arise from different parts of the mammary gland. However, when an adenoma or carcinoma arises from a fibroadenoma, it should be combined with other adenomas and carcinomas of the mammary gland.
d
When the fibroma–fibrosarcoma arises in the mammary gland region.
e
Malignant reticulosis is a diagnosis that is no longer in general use; however, if it is recorded, it should not be considered a nervous system neoplasm and should not be combined with any other type of nervous system neoplasm, including gliomas.
TABLE 28.2. Guidelines for Combining Neoplasms of the Same Histomorphogenic Type from Different Anatomic Sites in the Rat and the Mouse
Tissue Skin Adnexa Fibrous connective tissue Smooth muscle Skeletal muscle Bone Cartilage Endothelium Respiratory tract Alimentary tract Lymphoreticular Urothelium Nervous system Glial Nerve cell Nerve sheath a
See text for explanation.
Combine Yes Yes Yes Yesa Yes Yes Yes Yes Sometimesa Sometimesa Yes Yes Yes Yes Yes
710
CHAPTER 28
EVALUATION OF RODENT CARCINOGENESIS STUDIES
Respiratory Tract. While neoplasms of the same cell of origin in the larynx, trachea, and major bronchi would usually be combined, neoplasms of the peripheral lung (bronchiolar–alveolar neoplasms) would be analyzed separately. Similarly, neoplasms of the nasal cavity would not normally be combined with those found in the rest of the respiratory tract. Alimentary Tract. Squamous cell neoplasms of the tongue, esophagus, and forestomach (nonglandular) are often combined for evaluation. Likewise, epithelium-derived neoplasms from various portions of the small intestine are combined, as are neoplasms from different areas of large intestine (cecum and colon). Morphologically similar neoplasms of the small and large intestines are combined at times to evaluate the intestinal tract as a whole. Neoplasms of the glandular stomach are usually evaluated independently. Smooth Muscle. Smooth muscle neoplasms are combined for all sites of the body, except the gastrointestinal and reproductive tracts, where they are evaluated independently. One combination of the organs sites that should be considered inappropriate for evaluating the carcinogenic potential of a given chemical is the grouping of all animals showing neoplasia within a treatment group, without regard for the specific type of neoplasm (i.e., counting the number of animals showing any morphologic type of neoplasm, benign or malignant). The reason for this being inappropriate is that most groups of rats and mice (control as well as treated) in 2-year studies will show a very high incidence of neoplasms (approaching 80–100% in rats and 50–70% in mice based on the NTP historical database), so that only rarely would a carcinogenic chemical be identified with the use of such a comparison. Interestingly, the incidence of tumors in aged humans also approaches 100% when a set of tissuesorgans comparable to that examined in rodent studies is evaluated and the history of surgical intervention is taken into account (Shubik 1984).
28.4.3. Combining Neoplasms of a Common Cell Type in Different Tissues Another issue that the toxicologic pathologist encounters is what to do with neoplasms that have a common histogenesis but arise in small numbers in various tissues. Such systemic neoplasms sometimes occur in statistically insignificant numbers in individual organs; but when the occurrences of the neoplasm is tallied independent of the organ, statistical significance is sometimes achieved. An example of this might be malignant lymphoma. Malignant lymphoma might be recorded in treated animals in various organs, with no single organ having a statistically significant increase in the incidence of lymphoma; but when the incidence of lymphoma is examined in all organs, the incidence is significantly increased when compared to controls. Such systemic neoplasms that might be considered in this fashion would include all blood cell neoplasms, including histiocytic sarcoma, as well as neoplasms in which the cell type is present in many different organs, such as hemangiosarcomas and malignant mesotheliomas.
REFERENCES
28.5.
711
SUMMARY
This chapter gives the rationale, criteria, and guidelines for the combining of neoplasms occurring at the same time and different sites in long-term toxicology and carcinogenesis studies. In determining the “weight of evidence” that a chemical possesses carcinogenic potential, one must evaluate several criteria. These criteria include the dose–response relationship, variability of the endpoint, effects on tumor latency, and the tumor type. The greatest weight of evidence consists of a doserelated induction of multiple malignant neoplasms with a shortened latent period. Less weight of evidence would be given to a chemical that caused an increase in the high-dose group only, of a benign neoplasm whose incidence is normally quite variable with no change in latency. By evaluating malignant and benign tumors both separately and combined, one is able to identify more clearly the weight of evidence for a given chemical. In principle, one can combine any group of neoplasms or sites that are considered scientifically appropriate. However, the degree of biologic importance attached to such groupings must be considered in the context of the rationale for such a procedure. While it seems clear that a weight-of-evidence approach to qualitative risk assessment requires consideration of benign and malignant neoplasms, there is considerable controversy over the proper use of such data in quantitative risk assessment. For the reasons outlined in this chapter—that is, benign neoplasms may regress or progress or be incidental or fatal—data on benign and malignant tumors must be evaluated on a case-by-case basis to determine the most appropriate use in quantitative risk assessment.
REFERENCES Betton, G. R., Whiteley, L. O., Anver, M. R., Brown, R., Deschl, U., Elwell, M., Germann, P. G., Hartig, F., Küttler, K., Mori, H., Nolte, T., Püschner, H., and Tuch, K. (2001). Gastrointestinal tract. In International Classification of Rodent Tumors: The Mouse, Mohr, U., ed., Springer-Verlag, Berlin, pp. 23–58. Boorman, G. A., and Eustis, S. L. (1984). Proliferative lesions of the exocrine pancreas in male F344/N rats. Environ Health Perspect 56, 213–217. Bosland, M. C., Tuomari, D. L., Elwell, M. R., Shirai, T., Ward, J. M., and McConnell, R. F. (1998). Proliferative lesions of the prostate and other accessory sex glands in male rats, URG-4. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–20. Botts, S., Jokinen, M. P., Isaacs, K. R., Meuten, D. J., and Tanaka, N. (1991). Proliferative lesions of the thyroid and parathyroid glands, E-3. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–12. Brown, W. R., Gough, A., Hamlin, M. H., 2nd, Hottendorf, G. H., and Patterson, D. R. (1995). Proliferative lesions of the adrenal glands in rats, E-4. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–12. Bruner, R., Küttler, K., Bader, R., Kaufmann, W., Boothe, A., Enomoto, M., Holland, J. M., and Parish, W. E. (2001). Integumentary system. In International Classification of Rodent Tumors: The Mouse, Mohr, U., ed., Springer-Verlag, Berlin, pp. 2–22. Burns, F. J., Vanderlaan, M., Sivak, A., and Albert, R. E. (1976). Regression kinetics of mouse skin papillomas. Cancer Res 36, 1422–1427.
712
CHAPTER 28
EVALUATION OF RODENT CARCINOGENESIS STUDIES
Capen, C. C., Karbe, E., Deschl, U., George, C., Germann, P. G., Gopinath, C., Hardisty, J. F., Kanno, J., Kaufmann, W., Krinke, G., Küttler, K., Kulwich, B., Landes, C., Lenz, B., Longeart, L., Paulson, I., Sander, E., and Tuch, K. (2001). Endocrine system. In International Classification of Rodent Tumors: The Mouse, Mohr, U., ed., Springer-Verlag, Berlin, pp. 269–322. Cohen, S. M. (1983). Pathology of experimental bladder cancer in rodents. In The Pathology of Bladder Cancer, Vol. II, Bryant, G. T., and Cohen, S. M., eds., CRC Press, Boca Raton, FL, pp. 1–40. Davis, B., Harleman, J. H., Heinrichs, M., Maekawaq, A., McConnell, R. F., Reznik, G., and Tucker, M. (2001). Female genital system. In International Classification of Rodent Tumors: The Mouse, Mohr, U., ed., Springer-Verlag, Berlin, pp. 211–268. DeLellis, R. A., Nunnemacher, G., Bitman, W. R., Gagel, R. F., Tashjian, A. H., Jr., Blount, M., and Wolfe, H. J. (1979). C-cell hyperplasia and medullary thyroid carcinoma in the rat. An immunohistochemical and ultrastructural analysis. Lab Invest 40, 140–154. Deschl, U., Cattley, R. C., Harada, T., Küttler, K., Hailey, J. R., Hartig, F., Leblanc, B., Marsman, D. S., and Shirai, T. (2001). Liver, gallbladder, and exocrine pancreas. In International Classification of Rodent Tumors: The Mouse, Mohr, U., ed., Springer-Verlag, Berlin, pp. 59–86. Dixon, D., Leininger, J. R., Valerio, M. G., Johnson, A. N., Stabinski, L. G., and Frith, C. H. (1999). Proliferative lesions of the ovary, uterus, vagina, cervix and oviduct in rats, URG-5. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–32. Dungworth, D. L., Rittinghausen, S., Schwartz, L., Harkema, J. R., Hayashi, Y., Kittel, B., Lewis, D., Miller, R. A., Mohr, U., Morgan, K. T., Rehm, S. S., and Slayter, M. V. (2001). Respiratory system and mesothelium. In International Classification of Rodent Tumors: The Mouse, Mohr, U., ed., Springer-Verlag, Berlin, pp. 87–136. Elwell, M. R., Mahler, J. F., and Ruecker, F. A. (2004). Proliferative and non-proliferative lesions in the heart and vaculature in mice, MCV-1. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–14. Ernst, H., Long, P. H., Wadsworth, P. F., Leininger, J. R., Reiland, S., and Konishi, Y. (2001). Skeletal system and teeth. In International Classification of Rodent Tumors: The Mouse, Mohr, U., ed., Springer-Verlag, Berlin, pp. 389–415. Evans, M. G., Cartwright, M. E., Sahota, P. S., and Clifford, C. B. (1997). Proliferative lesions of the skin and adnexa of rats, IS-1. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–14. Foley, J. F., Anderson, M. W., Stoner, G. D., Gaul, B. W., Hardisty, J. F., and Maronpot, R. R. (1991). Proliferative lesions of the mouse lung: Progression studies in strain A mice. Exp Lung Res 17, 157–168. Frantz, J. D., Betton, G., Cartwright, M. E., Crissman, J. W., Macklin, A. W., and Maronpot, R. R. (1991). Proliferative lesions of the non-glandular and glandular stomach in rats, GI-3. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–20. Frith, C. H., Eighmy, J. J., Fukushima, S., Cohen, S. M., Squire, R. A., and Chandra, M. (1995). Proliferative lesions of the lower urinary tract in rats, URG-2. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–13. Frith, C. H., and Heath, J. E. (1984). Morphological classification and incidence of thyroid tumors in untreated aged mice. J Gerontol 39, 7–10. Frith, C. H., Ward, J. M., Brown, R. H., Tyler, R. D., Chandra, M., and Stromberg, P. C. (1996). Proliferative lesions of the hematopoietic and lymphatic systems in rats, HL-1. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–20. Frith, C. H., Ward, J. M., Harleman, J. H., Stromberg, P. C., Halm, S., Inoue, T., and Wright, J. A. (2001). Hematopoietic system. In International Classification of Rodent Tumors: The Mouse, Mohr, U., ed., Springer-Verlag, Berlin, pp. 417–451. Goodman, D. G., Maronpot, R. R., Newberne, P. M., Popp, J. A., and Squire, R. A. (1990). Proliferative and selected other lesions of the liver in rats, GI-5. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–24. Hansen, J. F., Ross, P. E., Makovec, G. T., Eustis, S. L., and Sigler, R. E. (1995). Proliferative and other selected lesions of the exocrine pancreas in rats, GI-6. Guides for Toxicologic Pathology, STP/ARP/ AFIP, Washington, D.C., pp. 1–7.
REFERENCES
713
Hard, G. C., Alden, C. L., Stula, E. F., and Trump, B. F. (1995). Proliferative lesions of the kidney in rats, URG-1. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–19. Hard, G. C., Durchfeld-Meyer, B., Short, B., Bube, A., Krieg, K., Creasey, D., Everitt, J., Frith, C. H., Glaister, J., Seely, J. C., Tsuda, H., and Turusov, V. S. (2001). Urinary system. In International Classification of Rodent Tumors: The Mouse, Mohr, U., ed., Springer-Verlag, Berlin, pp. 139–162. Haseman, J. K. (1983). A reexamination of false-positive rates for carcinogenesis studies. Fundam Appl Toxicol 3, 334–339. Hennings, H., Shores, R., Wenk, M. L., Spangler, E. F., Tarone, R., and Yuspa, S. H. (1983). Malignant conversion of mouse skin tumours is increased by tumour initiators and unaffected by tumour promoters. Nature 304, 67–69. IARC (1987). Overall Evaluations of Carcinogenicity: An Updating of IARC Monographs Volumes 1 to 42. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, Supplement 7, http:// monographs.iarc.fr/ENG/Monographs/suppl7/Suppl7.pdf. ICRT (1992a). No. 122. Part I: The Rat. 1. Respiratory system. In International Classification of Rodent Tumors, Mohr, U., ed., Springer-Verlag, Berlin. ICRT (1992b). No. 122. Part I: The Rat. 2. Soft tissue and musculoskeletal system. In International Classification of Rodent Tumors, Mohr, U., ed., Springer-Verlag, Berlin. ICRT (1992c). No. 122. Part I: The Rat. 3. Urinary system. In International Classification of Rodent Tumors, Mohr, U., ed., Springer-Verlag, Berlin. ICRT (1993a). No. 122. Part I: The Rat. 4. Haematopoietic system. In International Classification of Rodent Tumors, Mohr, U., ed., Springer-Verlag, Berlin. ICRT (1993b). No. 122. Part I: The Rat. 5. Integumentary system. In International Classification of Rodent Tumors, Mohr, U., ed., Springer-Verlag, Berlin. ICRT (1994a). No. 122. Part I: The Rat. 6. Endocrine system. In International Classification of Rodent Tumors, Mohr, U., ed., Springer-Verlag, Berlin. ICRT (1994b). No. 122. Part I: The Rat. 7. Central nervous system; heart; eye; mesothelium. In International Classification of Rodent Tumors, Mohr, U., ed., Springer-Verlag, Berlin. ICRT (1997a). No. 122. Part I: The Rat. 8. Male genital system. In International Classification of Rodent Tumors, Mohr, U., ed., Springer-Verlag, Berlin. ICRT (1997b). No. 122. Part I: The Rat. 9. Female genital system. In International Classification of Rodent Tumors, Mohr, U., ed., Springer-Verlag, Berlin. ICRT (1997c). No. 122. Part I: The Rat. 10. Digestive system. In International Classification of Rodent Tumors, Mohr, U., ed., Springer-Verlag, Berlin. Iversen, O. H. (1982). Hairless mouse skin in two-stage chemical carcinogenesis. Virchows Arch B Cell Pathol Incl Mol Pathol 38, 263–272. Kauffman, S. L., Alexander, L., and Sass, L. (1979). Histologic and ultrastructural features of the clara cell adenoma of the mouse lung. Lab Invest 40, 708–716. Krinke, G., Fix, A., Jacobs, M., Render, J., and Weisse, I. (2001a). Eye and harderian gland. In International Classification of Rodent Tumors, Mohr, U., ed., Springer-Verlag, Berlin, pp. 347–359. Krinke, G., Fix, A., Kaufmann, W., Ackerman, L. J., Garman, R. H., George, C., Jortner, B. S., Leininger, J. R., Mitsumori, K., Morgan, K. T., Paulson, I., and Robinson, M. (2001b). Central nervous system. In International Classification of Rodent Tumors, Mohr, U., ed., Springer-Verlag, Berlin, pp. 323–345. Kunze, E. (1979). Development of urinary bladder cancer in the rat. In Carcinogenesis, Grundmann, E., ed., Springer-Verlag, Berlin, pp. 145–232. Kunze, E., Schauer, A., Eder, M., and Seefeldt, C. (1979). Early sequential lesions during development of experimental gastric cancer with special reference to dysplasias. J Cancer Res Clin Oncol 95, 247–264. Lee, A. K., DeLellis, R. A., Blount, M., Nunnemacher, G., and Wolfe, H. J. (1982). Pituitary proliferative lesions in aging male Long–Evans rats. A model of mixed multiple endocrine neoplasia syndrome. Lab Invest 47, 595–602. Long, P. H., Leininger, J. R., Nold, J. B., and Lieuallen, W. G. (1993). Proliferative lesions of bone, cartilage, tooth, and synovium in rats, MST-2. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–18.
714
CHAPTER 28
EVALUATION OF RODENT CARCINOGENESIS STUDIES
Majka, J. A., Solleveld, H. A., Barthel, C. H., and Van Zwieten, M. J. (1990). Proliferative lesions of the pituitary in rats, E-2. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–8. Mann, P. C., Boorman, G. A., Lollini, L. O., McMartin, D. N., and Goodman, D. G. (1996). Proliferative lesions of the mammary gland in rats, IS-2. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–11. McConnell, E. E., Solleveld, H. A., Swenberg, J. A., and Boorman, G. A. (1986). Guidelines for combining neoplasms for evaluation of rodent carcinogenesis studies. J Natl Cancer Inst 76, 283–289. McConnell, R. F., Weston, H. H., Ulland, B. M., Bosland, M. C., and Ward, J. M. (1992). Proliferative lesions of the testes in rats with selected examples from mice, URG-3. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–32. NTP (1984). Report of the NTP Ad Hoc Panel on chemical carcinogenisis testing and evaluation. Board of Scientific Counselors, National Toxicology Program, pp. 1–280. Pattengale, P. K., and Frith, C. H. (1983). Immunomorphologic classification of spontaneous lymphoid cell neoplasms occurring in female BALB/c mice. J Natl Cancer Inst 70, 169–179. Potter, V. R. (1984). Use of two sequential applications of initiators in the production of hepatomas in the rat: An examination of the Solt–Farber protocol. Cancer Res 44, 2733–2736. Rehm, S., Harleman, J. H., Cary, M., Creasey, D., Ettlin, R. A., Eustis, S. L., Foley, G. L., LeNet, J. L., Maekawaq, A., Mitsumori, K., McConnell, R. F., and Reznik, G. (2001). Male genital system. In International Classification of Rodent Tumors, Mohr, U., ed., Springer-Verlag, Berlin, pp. 163–210. Reznik, G., Hamlin, M. H., 2nd, Ward, J. M., and Stinson, S. F. (1981). Prostatic hyperplasia and neoplasia in aging F344 rats. Prostate 2, 261–268. Reznik, G., and Ward, J. M. (1981). Morphology of hyperplastic and neoplastic lesions in the clitoral and preputial gland of the F344 rat. Vet Pathol 18, 228–238. Riley, M. G. I., Boorman, G. A., McDonald, M. M., Longnecker, D., Solleveld, H. A., and Giles, H. D. (1990). Proliferative and metaplastic lesions of the endocrine pancreas in rats, E-1. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–7. Ruben, Z., Arceo, R. J., Bishop, S. P., Elwell, M. R., Kerns, W. D., Mesfin, G. M., Sandusky, G. E., and Van Vleet, F. (1997). Proliferative lesions of the heart and vasculature in rats, CV-1. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–9. Sass, B., Vlahakis, G., and Heston, W. E. (1982). Precursor lesions and pathogenesis of spontaneous mammary tumors in mice. Toxicol Pathol 10, 12–21. Schwartz, L. W., Hahn, F. F., Keenan, K. P., Keenan, C. M., Brown, H. R., and Mann, P. C. (1994). Proliferative lesions of the rat respiratory tract, R-1. Guides for Toxicologic Pathology, STP/ARP/ AFIP, Washington, D.C., pp. 1–24. Sheldon, W. G., Curtis, M., Kodell, R. L., and Weed, L. (1983). Primary harderian gland neoplasms in mice. J Natl Cancer Inst 71, 61–68. Shubik, P. (1984). Progression and promotion. J Natl Cancer Inst 73, 1005–1011. Solleveld, H. A., Gorgacz, E. J., and Koestner, A. (1991). Central nervous system neoplasms in the rat, NS-1. Guides for Toxicologic Pathology, STP/ARP/AFIP, Washington, D.C., pp. 1–17. Squire, R. A., Brinkhous, K. M., Peiper, S. C., Firminger, H. I., Mann, R. B., and Strandberg, J. D. (1981). Histiocytic sarcoma with a granuloma-like component occurring in a large colony of Sprague-Dawley rats. Am J Pathol 105, 21–30. Stromberg, P. C., Rojko, J. L., Vogtsberger, L. M., Cheney, C., and Berman, R. (1983). Immunologic, biochemical, and ultrastructural characterization of the leukemia cell in F344 rats. J Natl Cancer Inst 71, 173–181. Takano, T., Shirai, T., Ogiso, T., Tsuda, H., Baba, S., and Ito, N. (1982). Sequential changes in tumor development induced by 1,4-dinitrosopiperazine in the nasal cavity of F344 rats. Cancer Res 42, 4236–4240. Trouillas, J., Girod, C., Claustrat, B., Cure, M., and Dubois, M. P. (1982). Spontaneous pituitary tumors in the Wistar/Furth/Ico rat strain. An animal model of human prolactin adenoma. Am J Pathol 109, 57–70. Turek, F. W., and Desjardins, C. (1979). Development of Leydig cell tumors and onset of changes in the reproductive and endocrine systems of aging F344 rats. J Natl Cancer Inst 63, 969–975.
REFERENCES
715
Ward, J. M., Reznik, G., Stinson, S. F., Lattuada, C. P., Longfellow, D. G., and Cameron, T. P. (1980). Histogenesis and morphology of naturally occurring prostatic carcinoma in the ACI/segHapBR rat. Lab Invest 43, 517–522. Whiteley, L. O., Anver, M. R., Botts, S., and Jokinen, M. P. (1996). Proliferative lesions of the intestine, salivary glands, oral cavity, and esophagus in rats, GI-1/2/4. Guides for Toxicologic Pathology, STP/ ARP/AFIP, Washington, D.C., pp. 1–18. Yoshida, H. (1983). Preputial tumors induced by intragastric intubations of 7,12-dimethylbenz(a)anthracene in gonadectomized female and male rats. J Cancer Res Clin Oncol 105, 299–302.
CH A P TE R
29
CANCER RISK BASED ON AN INDIVIDUAL TUMOR TYPE OR SUMMING OF TUMORS Andrew G. Salmon Lindsey A. Roth
29.1.
INTRODUCTION
The standard quantitative analyses used in developing cancer potency estimates from animal data have been developed without any specific reference to the site or type of tumor, but it has generally been assumed that these analyses would be applied to a single site and type of tumor. It is presumed that all instances of the tumor share a common mechanism of origin and therefore relate to a single dose-response relationship. Standard practice has been to use dose–response data for the most sensitive tumor site as the basis of the estimate (Anderson et al. 1983; CDHS 1985, US EPA 1986). However, protection of public health generally is concerned with assessing and reducing the impact of cancer in humans at any site, which is related to the chemical exposure being considered. For example, the overall assessment of cancer risk from cigarette smoking (US DHHS 1982) or ionizing radiation (NRC 1990) is not based on risk at one site, such as lung cancer. Instead, total cancer risk is estimated from all the sites at which agent-induced tumors are observed (lung, bladder, leukemia, etc.), combined. For many carcinogens, the tumors observed in animal studies occur principally at a particular site (Huff et al. 1991), and are of a specific histological type and presumed cellular origin. In such cases, selection of the most sensitive site in the animal studies is recognized as providing a risk estimate, which is appropriate to protect human health. Nevertheless, the interpretation of these animal studies is complicated by the evidence that although concordance of site or tumor type between animal models and human health effects may occur, it cannot be assumed (Wilbourn et al. 1984; NRC 1994). In addition, there are many examples where more than one tumor type is observed following exposure to a single chemical (Haseman et al. 1987). Thus, for chemicals that induce tumors at multiple sites, the single-site
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
716
29.2. SUMMING OF TUMORS OF RELATED TYPES
717
approach may underestimate the true carcinogenic potential. In cases where the potency at one site predominates (for instance, by an order of magnitude or more), this potential underestimation has generally been ignored with the assumption that it can be covered by more health-protective assumptions in other aspects of the assessment, including the use of 95% confidence limits as potency estimates. However, where a chemical shows sufficient potency to contribute substantially to the overall tumor incidence at more than one site, a procedure is necessary to calculate a combined potency estimate. It would appear reasonable to develop a combined potency on the basis that the independent actions at various sites would be additive. Conceptually, this is not unlike the assumption that, all other things being equal, the contributions to the overall carcinogenic potency of a mixture from various component chemicals would be additive, at least for carcinogens whose primary action is as an initiator (Zielinski et al. 2001). By this procedure, one would simply add together the potency values for the different sites. An alternative approach that has been used routinely for analysis of multiple tumor types meeting specified criteria was laid out in the description of the basic procedures for carcinogen risk assessment (Anderson et al. 1983). Combined incidence data are developed by counting all animals affected by one or more of the specified tumors, as a proportion of the total group sizes. As described below, this approach has become standard only when considering the combined impact of tumors having a common cellular origin, such as benign and malignant tumors of a related cell type and at a common site. However, no such restriction was actually specified by Anderson et al. (1983) or the original U.S. Environmental Protection Agency risk assessment guidelines (US EPA, 1986). A preferred approach for tumors at different sites, or with different cellular origin and presumably therefore independent mechanisms of causation, has not so far been universally accepted. The divergent recommendations on how to deal with estimates of combined potency for multisite carcinogens were noted, without detailed explanation or recommendation, by the EPA’s revised cancer risk assessment guidelines (US EPA 2005).
29.2.
SUMMING OF TUMORS OF RELATED TYPES
Where tumors of more than one histological type but of a presumed common cellular origin are observed at a single site, the combined incidence—that is, proportion of animals affected with at least one tumor of any of the relevant types—is used for dose–response assessment. This requires access to the individual animal data. These data are routinely provided in National Toxicology Program reports, although they are not provided in the public reports of early National Cancer Institute studies. Many publications in the scientific literature also do not provide sufficient data to complete this analysis unless suitable combined incidences were calculated by the authors. This approach is appropriate to the case of related tumor types, where there is expected to be a marked correlation between the incidences of the different tumor types. This is often readily observed in actual data sets, which may show, for instance, a greater proportion of carcinomas but fewer adenomas at high doses of
718
CHAPTER 29 CANCER RISK BASED ON AN INDIVIDUAL TUMOR TYPE OR SUMMING OF TUMORS
the carcinogen. The same rules for combining tumor types are generally applied that are described in determining statistical significance for carcinogen identification (McConnell et al. 1986; IARC 2006). In particular, tumor types considered to represent different stages of progression following mutation of a common normal cell type are usually combined. US EPA (2005) has also suggested that it may be appropriate to combine the incidences of tumors resulting from a common mode of action. Most of the debate in this area has concerned the decision as to whether to include benign as well as malignant lesions. On the basis of results from the National Toxicology Program’s bioassays, Huff et al. (1989) argued that the appearance of benign neoplasms was an important indicator of carcinogenic potential and that these should be considered in evaluating the overall weight of evidence for hazard identification of carcinogens. Similar recommendations appear for dose–response assessment, in both the earlier (US EPA 1986; CDHS 1985) and current (US EPA 2005) cancer risk assessment guidelines.
29.3. SUMMING OF UNRELATED TUMOR TYPES Several alternative approaches have been proposed to determine an overall potency for carcinogens that induce tumors at multiple sites and/or with different cell types in a particular species and sex.
29.3.1.
Affected-Animal Count
This is the same combined incidence basis as is used for tumors of presumed common cellular origin at a single site. Anderson et al. (1983) and the original EPA risk assessment guidelines (US EPA 1986) endorse this procedure for multiple tumor sites or types; however, it has not been widely used to estimate the combined risk from tumors at multiple sites or of evidently different histological origin. This preference may be related to the perceived problems in combining tumor types that are clearly independent from the biological as well as statistical point of view (NRC 1994), even if resulting from a common chemical exposure. Bogen (1990) argued that this procedure may create errors by randomly excluding the information in multiple-tumor-bearing cases where the tumor incidences are independent. Also there may be numerical problems based on different background rates and shapes of the dose–response curve, which may actually reduce the sensitivity of the combined estimate. These difficulties are related to the disadvantages of combined incidence measures in the hazard identification phase of the process, as described by Haseman et al. (1986).
29.3.2.
Addition of Independent Potency Values
NRC (1994) argued that where the tumor incidence rates are low, the simple procedure of adding individual potency estimates is preferable when considering disparate tumor types, whose incidence is statistically independent (i.e., there is no synergism, progression from one type to another, or other such quantitative interactions). This also has the advantage of requiring only standard software [GLOBAL in its various
29.3. SUMMING OF UNRELATED TUMOR TYPES
719
versions (Crump and Watson 1979), ToxRisk (Crump et al. 1991) or MSTAGE (Crouch 1985)] being the most commonly used) to provide the likelihood-optimized polynomial fit to the dose–response data required both for the linearized multistage model, and the normal implementation of the newer benchmark dose procedure (US EPA 2005). However, simply adding the potency values calculated for each major site may produce errors, for several reasons. First, because the potency estimates are cited as 95% upper confidence limits (q1*) rather than estimates of central tendency, the sum of the two estimates may overestimate the true 95% confidence limit of the combined distribution. Second, at high incidence rates particularly, there may be mis-estimation of the impact from cases where tumors appear at two or more sites in a single individual; such multiple occurrences can only contribute to a single fatality or other measure of overall cancer incidence. This may well be significant in the analysis of animal bioassay data, but is less likely to be a problem with studies of cancer incidence in human populations [or at least, one hopes this is the case; tumors from cigarette smoking might be an exception based on the data reported by US DHHS (1982)]. The effort to predict the confidence limit on a combined distribution is further complicated by the fact that the likelihood distributions of the estimates for two independent sites are neither symmetrical nor, necessarily, similar in shape. Some investigators have studied the effect of addition or combination of potencies in the parallel case of exposure to mixtures containing more than one carcinogen. Thus Kodell et al. (1995) reported that Monte Carlo simulations for the combined effect of multiple carcinogens showed, unexpectedly, that under some circumstances the simple procedure of adding single-chemical potency estimates may produce a result which is insufficiently health-protective.
29.3.3.
Distribution-Based Methods
In order to avoid the disadvantages, seen or inferred, of the simple addition of q1* values, various analysts have either calculated or assumed a distribution (for each tumor type) representing the likelihood for the plausible range of estimates of the linear term (q1) of the multistage model (q1), and then they used the Monte Carlo procedure to add the distributions rather than merely adding specific points on the distributions such as the maximum likelihood estimate (MLE) or 95% confidence limit. A combined potency estimate (q1* for all sites) is then obtained as the 95% confidence limit on the summed distribution. This resembles the approach for multiple carcinogens by Kodell et al. (1995) noted above. Some analysts, such as US EPA (2002) for the analysis of 1,3-butadiene carcinogenicity examined in more detail below, assumed a distribution shape (normal in that case, with a standard deviation calculated from the estimated MLE and confidence limits of the q1 estimates). However, as will be shown below, such assumptions are approximate at best even in favorable cases. Zeise et al. (1991) demonstrated a procedure to generate the required distribution by tracing the likelihood profile of the linear term (q1) for a linearized multistage fit to dichotomous tumor data. This used a modified version (Zeise and Salmon 1991) of the MSTAGE program developed by Crouch (1985). More recently, Crouch (2006) developed a similar program that provides the likelihood distribution for estimates of q1 when
720
CHAPTER 29 CANCER RISK BASED ON AN INDIVIDUAL TUMOR TYPE OR SUMMING OF TUMORS
the multistage-in-dose, Weibull-in-time model is fitted to time-to-tumor data from animal bioassays. The distributions of q1 for each of the treatment-related sites are then statistically summed using a Monte Carlo approach and assuming independence. The sum is created by adding the linear term for each tumor site, according to its distribution, through random sampling (usually with at least 100,000 trials). The upper 95% confidence limit on the summed distribution is taken as the multisite animal cancer potency estimate (McDonald et al., 2003; McDonald and Komulainen 2005). Using the combined potency distribution takes into account the multisite tumorigenicity and provides a basis for estimating the cumulative risk of all treatment-related tumors, while avoiding some of the objections to the alternatives previously described. The California Environmental Protection Agency’s Office of Environmental Health Hazard Assessment (OEHHA) has applied this approach in several recent dose–response analyses, including that for naphthalene (OEHHA 2004; Hoover et al. 2005). The revised EPA cancer risk assessment guidelines (U.S. EPA 2005) recommend that rather than estimating the potency as q1* in a fit of the linearized multistage model, the benchmark dose approach should be used. In this method, the 95% lower confidence limit on the dose producing a tumor incidence of 10% (LED10) is estimated as a point of departure (POD) by fitting the tumor incidence data to a suitable model. The choice of model is arbitrary from the conceptual point of view: The only criterion of applicability is whether it fits the data in the observed range, and the model is not used to inform the low-dose extrapolation procedure. Low-dose extrapolation is performed separately by linear extrapolation from the POD to zero dose (i.e. the potency slope factor = 0.1/LED10); or if there is specific evidence to support a threshold mechanism, an uncertainty factor approach may be used. In practice, although the multistage model is no longer assumed to be necessarily a biologically accurate model of the entire dose–response relationship, it is generally the best accessible model from the mathematical point of view, and this constrained polynomial model is generally used as the default model for fitting cancer data. Linear extrapolation from a POD obtained with this model produces very similar results for the potency slope factor to the linearization of the multistage model (i.e., slope factor = q1*) for most datasets [see, for example, OEHHA (2007)]. A Monte Carlo addition of the distribution of risk predicted for each site at a chosen POD could be used to provide a combined estimate of the potency slope for all sites, by analogy with the methods used for adding q1* for multiple sites. However, examples of this approach have not so far appeared in the literature, probably because of the lack of readily available software solutions to provide the necessary distributions. A partial approach to the problem of adding potency values without undertaking the Monte Carlo addition of the entire distributions of estimates has used the addition of fixed points on the distributions other than the 95% confidence limit. Unlike confidence limits, the MLE values of q1 can be simply added to generate an MLE for the combined distribution [see US EPA (2002), for example]. However, it is generally recognized that the MLE is an unsatisfactory parameter for describing estimated potency slopes, as will be demonstrated below. This value is unstable for polynomial fits of variable order such as those used in the multistage model. Except
29.4. EXAMPLE: 1,3-BUTADIENE
721
in the case of simple datasets where the linear term is the only one contributing significantly to the dose–response curve, it is not even a plausible estimate of the central tendency of the distribution, as it is sometimes supposed to be (in fact it may well be zero for strongly curved dose–response curves). Moreover, it is then necessary to develop some measure of the distribution, or at least the variance, of the combined q1 value in order to estimate a combined q1*, which is the value generally used in risk assessment and is required for comparison with published potency values for other carcinogens.
29.4.
EXAMPLE: 1,3-BUTADIENE
In order to illustrate the various possible approaches to estimating a combined potency for a multisite carcinogen, an example is offered here. The carcinogen selected, 1,3-butadiene, is an important industrial chemical and a widespread air pollutant. Risk assessments for this chemical have been published (OEHHA 1992; US EPA 2002). The more recent estimate by US EPA (2002) depends primarily on estimates of risk from the various human epidemiology studies of industrial exposures to this chemical, but it is their analysis of the animal bioassay data and also that by OEHHA (1992) which are of interest for the present purpose.
29.4.1.
Source Data
1,3-Butadiene has been the subject of two inhalation bioassays by the National Toxicology Program (NTP 1984, 1993). The mouse is notably susceptible to 1,3-butadiene carcinogenesis at multiple sites, as well as other types of toxicity. The results of the first NTP study, although valuable in identifying the status of 1,3-butadiene as a carcinogen (IARC 1999), are of limited value in quantitative risk assessment because exposure concentrations were used where the dose–tumor incidence curves plateau, possibly because of saturation of metabolism as well as intercurrent and tumor-related mortality. In the second study, mortality was also severe at the higher exposure levels, but lower levels were also used which allowed observation of the shape of the dose–response curves for the various tumors. The tumor incidences reported by NTP (1993) are shown in Table 29.1. OEHHA (1992) did not have the final technical report of this study available when preparing their risk assessment, but instead used an earlier published report of the study (Melnick et al. 1990). Analysis of this dataset has been particularly challenging due to the multiple tumor sites, some with very high incidence and early onset, creating a problem of competing risks both from other tumors and from overall mortality. No mice in the highest-dose group (625 ppm) survived to the end of the study (week 103), and in the second-highest-dose group (200 ppm) only 8% of the males and no females survived that long. Mortality data were extensively reported an analyzed by NTP (1993), as summarized in Table 29.2. As was seen in the earlier NTP (1984) bioassay, the metabolism of 1,3-butadiene tends to show saturation at higher exposures. OEHHA (1992) used
722
CHAPTER 29 CANCER RISK BASED ON AN INDIVIDUAL TUMOR TYPE OR SUMMING OF TUMORS
TABLE 29.1.
Tumor Incidences in Mice Exposed to 1,3-Butadiene (NTP 1993)
Exposure Concentration (ppm) 0
6.25
20
62.5
200
625
50 0 4 0 21
50 0 2 0 23
50 1 4 4 19
49 5 6 5 31
50 20 2 7 35
73 4 51 4 3
6 21 1 0
7 23 0 0
9 30 0 0
20 25 1 0
31 33 8 5
6 5 4 0
50 0 12 2 15 10 14 0 1 2
50 0 11 7 19 7 15 3 1 4
50 1 7 4 24 15 19 2 9 12
50 21 9 7 24 20 16 4 7 15
80 23 32 4 22 9 2 20 6 12
Male Mice Total animals Heart Hematopoietic system Lung Harderian gland Liver Forestomach Preputial gland
Hemangiosarcoma All malignant lymphomas Histiocytic sarcoma Alveolar and bronchiolar neoplasms Adenoma or carcinoma Adenoma or carcinoma Papilloma or carcinoma Carcinoma
Female Mice Total animals Heart Hematopoietic system Lung Harderian Liver Forestomach Ovary Mammary
Hemangiosarcoma All malignant lymphomas Histiocytic sarcoma Adenoma or carcinoma Adenoma or carcinoma Adenoma or carcinoma Papilloma or carcinoma Granulosa cell neoplasm Carcinoma or adenoacanthoma
49 0 6 3 4 8 15 0 1 0
pharmacokinetic modeling (based on Hattis and Wasson, 1987) to address this complexity in their analysis of the NTP (1993) data, although US EPA (2002) decided that that endeavor was too uncertain and complex.
29.4.2.
Affected-Animal Count
US EPA (2002) presented a linearized-multistage analysis of multiple tumor incidence in female mice in the NTP (1993) study, as an example of the application of this methodology. They made no specific allowance for nonlinear pharmacokinetics. In fact (as may be seen from the data presented in Table 29.3), the incidence of all tumors was high even in the control groups, and approached 100% in several of the treated groups, so dose–response analysis of the data for all tumors is of limited value. They therefore considered only a subset of the observed tumors in the female mice, counting animals with malignant lymphomas, heart hemangiosarcomas, lung tumors (alveolar/bronchiolar adenomas or carcinomas), mammary gland tumors (carcinomas, adenocanthomas, or malignant mixed tumors), or benign or malignant
29.4. EXAMPLE: 1,3-BUTADIENE
723
TABLE 29.2. Survival of Male and Female B6C3F1 Mice Exposed to 1,3-Butadiene by Inhalation for 103 Weeks (NTP 1993)
Exposure Concentration (ppm) 0
6.25
20
62.5
200
625
50 24 50 49
50 22 49 46
50 4 50 8
72 0 69 0
50 24 50 50
50 11 50 23
50 0 49 0
72 0 69 0
Male Mice Animals initially in studya Animals surviving to end of study Animals surviving at first tumorb Percent survival at end of studyc
50 35 50 70
50 39 50 78
Female Mice Animals initially in studya Animals surviving to end of study Animals surviving at first tumorb Percent survival at end of studyc
50 37 50 74
50 33 49 66
a
Excluding those in the interim evaluations at 9 and 15 months.
b
Number of animals surviving to the time of appearance of the first tumor. This was a malignant lymphoma in both sexes, appearing at 161 days (23 weeks) in a male and 203 days (29 weeks) in a female. c
Kaplan–Meier determinations. Survival rates were adjusted for interim evaluations, accidental deaths, and missing animals.
TABLE 29.3.
Total Tumor Incidences in Mice Exposed to 1,3-Butadiene (NTP 1993)
Exposure concentration (ppm) Metabolized dose (mg/kg-day)a
0 0.00
6.25 0.62
20 1.97
62.5 6.12
200 17.80
625 54.30
50 45
49 48
50 49
69 62
Male Mice Animals at risk Tumor-bearing animals
50 44
50 40
Female Mice Animals at risk Tumor-bearing animals All tumors Selected tumorsb
50
49
50
50
49
69
35 6
47 19
43 26
48 31
49 46
69 —
a
From OEHHA (1992).
b
Animals with malignant lymphomas, heart hemangiosarcomas, lung tumors (alveolar/bronchiolar adenomas or carcinomas), mammary gland tumors (carcinomas, adenocanthomas, or malignant mixed tumors), or benign or malignant ovary granulosa cell tumors, as reported by US EPA (2002). They did not evaluate the 625 ppm group.
724
CHAPTER 29 CANCER RISK BASED ON AN INDIVIDUAL TUMOR TYPE OR SUMMING OF TUMORS
ovary granulosa cell tumors. The number of animals at risk was taken as the number surviving to the time of appearance of the first tumor (a malignant lymphoma at 203 days), which provides a limited adjustment for intercurrent mortality. The high-dose group was excluded in view of the extremely high mortality in this group and poor overall fit. For this selected subset of the overall tumors in female mice, the GLOBAL86 program (Crump and Watson 1979) produced a relatively good fit of a linear model (q0 and q1 only) to the remaining dose groups, with a q1* of 0.1 ppm−1. This does not require a study duration correction since the study continued to the full nominal two-year lifetime, at least for those groups where there were survivors at this time point. US EPA (2002) also obtained the same value by the benchmark method [i.e., estimating the 95 % lower confidence limit on the dose producing a tumor incidence of 10% (LEC10) and extrapolating linearly from that POD]. This is as expected; for a pure linear model the linearized multistage and benchmark methods produce exactly the same result. Slight differences may be noticed for response curves modeled using higher-order polynomials, although in that case GLOBAL86 should not be used to predict the ED10 or LED10. Following the original U.S. EPA guidance for low-dose risk estimation but in contrast to more recent software such as ToxRisk (Crump et al. 1991), GLOBAL versions use only q1 to predict risk-specific doses; higherorder terms are ignored (Crump and Watson 1979). This is entirely reasonable for estimates in the low-risk range (e.g., 1 in 1000 or below) for which this software feature was originally designed, but is contrary to the intentions of the benchmark method where a dose at a relatively high risk such as 1 in 10 is to be estimated. Since other analyses using a pharmacokinetic exposure metric [metabolized dose, as described by OEHHA 1992)] will be reported below, the analysis of these data using metabolized dose as the exposure metric was performed for comparison. Using the MSTAGE program (Crouch 1985), a linear model fitted the data with q1 (MLE) = 1.5 and q1* = 0.19 (mg/kg-day)−1.
29.4.3.
Distribution-Based Methods
In order to illustrate the distribution-based methods for various specific sites that were described above, it is necessary for this dataset to develop incidence data that are to some degree corrected for the relatively severe competing risk from other early-onset tumors and other intercurrent mortality. The preferred way of doing this is by using a time-to-tumor analysis, which will be presented below. However, in order to illustrate the application of the distributional approach to total incidence (dichotomous) data, an attempt was made to correct for the competing risks. This was done by using the analysis presented by NTP in which they calculated the endof-study competing risks using the poly-3 model (Portier and Bailer 1989). This was preferred to the Kaplan–Meier estimates that were also presented as being similar in general form to the Weibull model used in the full time-to-tumor analysis, although in practice the poly-3 and Kaplan–Meyer corrections do not differ substantially in this case. The corrected incidence rates were applied to the numbers of animals at risk in each group as determined above. The resulting corrected incidences are shown in Table 29.4.
TABLE 29.4. Adjusted Tumor Incidences at End of Study (Using Poly-3) in Mice Exposed to 1,3-Butadiene (NTP, 1993)
Male Mice Nominal dose (ppm in air) Metabolized dose (mg/kg-day)a Number of animals at riskb Heart Hemangiosarcoma Hematopoietic system
All malignant lymphomas
Lung
Alveolar and bronchiolar neoplasms
Harderian gland
Adenoma or carcinoma
Liver
Adenoma or carcinoma
Forestomach
Papilloma and carcinoma
Preputial gland
Carcinoma
Histiocytic sarcoma
0.00 0.00 50 0.0% 0 9.0% 5 0.0% 0 47.5% 24 13.5% 7 44.6% 22 2.3% 1 0.0% 0
6.25 0.62 50 0.0% 0 4.4% 2 0.0% 0 49.0% 25 15.2% 8 48.2% 24 0.0% 0 0.0% 0
20.00 1.97 50 2.6% 1 10.0% 5 2.6% 1 44.9% 22 22.4% 11 65.2% 33 0.0% 0 0.0% 0
62.50 6.12 49 13.5% 7 15.3% 7 13.5% 7 74.2% 36 50.8% 25 61.6% 30 2.7% 1 0.0% 0
200.00 17.80 50 64.1% 32 7.4% 4 64.1% 32 87.8% 44 80.6% 40 85.9% 43 28.7% 14 18.9% 9
625.00 54.30 69 52.9% 37 97.3% 67 52.9% 37 45.1% 31 64.2% 44 61.2% 42 53.5% 37 0.0% 0
20.00 1.97 50 0.0% 0 27.5% 14 17.2% 9 46.5% 23 17.4% 9 36.4% 18 7.8% 4 2.6% 1 10.2% 5
62.50 6.12 50 3.1% 2 20.2% 10 11.8% 6 61.1% 31 41.2% 21 51.4% 26 6.1% 3 26.3% 13 32.6% 16
200.00 17.80 49 71.9% 35 40.1% 20 34.0% 17 81.5% 40 70.9% 35 64.9% 32 22.5% 11 41.1% 20 56.4% 28
625.00 54.30 69 83.4% 58 85.5% 59 35.0% 24 82.4% 57 58.0% 40 21.7% 15 82.6% 57 46.5% 32 66.8% 46
Female Mice Nominal dose (ppm in air) Metabolized dose (mg/kg-day)a Number of animals at riskc Heart Hemangiosarcoma Hematopoietic system
All malignant lymphomas Histiocytic sarcoma
Lung
Alveolar and bronchiolar neoplasms
Harderian gland
Adenoma or carcinoma
Liver
Adenoma or carcinoma
Forestomach
Papilloma and carcinoma
Ovary
Granulosa cell neoplasm
Mammary gland
Carcinoma or adenoacanthoma
0.00 0.00 50 0.0% 0 13.1% 7 6.5% 3 8.8% 4 17.5% 9 33.3% 17 0.0% 0 2.3% 1 0.0% 0
6.25 0.62 49 0.0% 0 27.2% 13 4.4% 2 33.0% 16 22.7% 11 30.3% 15 0.0% 0 0.0% 0 4.5% 2
a
From Hattis and Wasson (1987).
b
First incidence MLL 161 days = 23 weeks.
c
First incidence MLL 203 days = 29 weeks.
725
726
CHAPTER 29 CANCER RISK BASED ON AN INDIVIDUAL TUMOR TYPE OR SUMMING OF TUMORS
The linearized multistage model fits to these incidence data, with metabolized dose as the exposure metric, were then performed using the modified MSTAGE program (Zeise and Salmon 1991), with calculation of the likelihood profile values ranging from 0.1% to 99.9%. For male mice, the data at the highest dose level (625 ppm, 54.3 mg/kg-day) were excluded, since the confidence in this incidence rate is low in view of the high mortality at that dose. The same procedure was applied to the data for female mice, with the exception of two datasets where the fit was in fact improved by inclusion of the highest dose level. The maximum likelihood estimates and 95% upper confidence limits of the linear terms (q1 and q1*) for the fit to this parameter are presented in Table 29.5. The sums of these estimates of q1 for all sites are also shown as estimates of the overall potency. In addition to determining values for q1 and q1*, the program reports the estimates of q1 for cumulative likelihood values between 0.1% and 99.9%, allowing examination of the actual shape of the distribution for each fit. In simple cases where the majority of the variation in the data is explained by a linear model, the resulting TABLE 29.5. Carcinogenic Potency Estimates Using Dichotomous Tumor Incidence Data for Mice Exposed by Inhalation to 1,3-Butadiene
Site
Lesion
q1
q1*
3.340 × 10−3 3.931 × 10−4 3.340 × 10−3 8.907 × 10−2 8.372 × 10−2 6.506 × 10−2 0 0 2.45 × 10−1
2.718 × 10−2 6.776 × 10−3 2.718 × 10−2 1.257 × 10−1 1.081 × 10−1 1.052 × 10−1 6.595 × 10−3 3.244 × 10−3 4.10 × 10−1
3.210 × 10−2 5.791 × 10−4 1.575 × 10−2 1.052 × 10−1 4.143 × 10−2 4.084 × 10−2 1.511 × 10−2 3.078 × 10−2 2.813 × 10−2 3.10 × 10−1
4.273 × 10−2 2.525 × 10−2 2.935 × 10−2 1.381 × 10−1 7.578 × 10−2 6.148 × 10−2 2.176 × 10−2 4.082 × 10−2 3.462 × 10−2 4.70 × 10−1
Male Mice Heart Hematopoietic system Lung Harderian gland Liver Forestomach Preputial gland Sum for all sites
Hemangiosarcoma All malignant lymphomas Histiocytic sarcoma Alveolar & bronchiolar neoplasms Adenoma or carcinoma Adenoma or carcinoma Papilloma or carcinoma Carcinoma
Female Mice Heart Hematopoietic system Lung Harderian gland Liver Forestomach Ovary Mammary gland Sum for all sites a
a
Hemangiosarcoma All malignant lymphomas Histiocytic sarcoma Alveolar and bronchiolar neoplasms Adenoma or carcinoma Adenoma or carcinoma Papilloma or carcinoma Granulosa cell neoplasm Carcinoma or adenoacanthomaa
For all datasets except the heart hemangiosarcomas and mammary gland tumors in female mice, the incidences at the highest dose were excluded from the fit. For those two sites only, a poor fit was obtained with the first five groups only, which was improved by including the highest dose group.
29.4. EXAMPLE: 1,3-BUTADIENE
727
distribution shape is a single peak, although this may have a long tail at the high end of the estimate range depending on how much unexplained variation is present. Where the separate tumor incidences to be added together all fit this description, the procedure used by US EPA (2002) of assuming a normal shape of the likelihood density function in order to predict a 95% confidence limit on the combined distribution will produce reasonable, if not exact, results. At least, it could be argued that any uncertainties resulting from this assumption are slight in comparison with the other uncertainties in the estimate of human cancer risk. However, where higherorder polynomial terms contribute substantially either to the optimal model or to alternative plausible models, the likelihood function is far from a normal shape, showing long shoulders or tails and even multiple peaks. In such situations, q1 may even be zero (i.e., the optimal model does not include a linear term). Reliance on the assumption of normality for the combined distribution presents an unknown and potentially considerable uncertainty in such cases. This uncertainty may be somewhat reduced for low-dose estimates of risk since then the higher-order terms will make a much lower contribution to the overall model than the linear terms. However, attempts to estimate risk-specific doses at higher risks (e.g., 10%) as used in the LED10 method of potency estimation do not have this helpful property of largely excluding the higher-order terms (in other words, the measured dose–response curve at risks as high as 10% may show significant curvature). As an illustration of the complexity of the likelihood functions for fit to real datasets, a selection of graphs showing the relative likelihood as a function of the estimated value of q1 are shown below. (Some fine detail in the curves shown is noise induced by rounding errors in the numerical routine used to derive the curves, and it should be ignored.) Figure 29.1 shows the curve for the fit to the data for data 1.80 1.60
Percent contribution
1.40 1.20 1.00 0.80 0.60 0.40 0.20 0.00 0.00
0.02
0.04
0.06
0.08
0.10
Estimate of q1
Figure 29.1. Likelihood density function for fit to the data on liver adenoma or carcinoma in female mice (dichotomous models).
728
CHAPTER 29 CANCER RISK BASED ON AN INDIVIDUAL TUMOR TYPE OR SUMMING OF TUMORS
on liver adenoma or carcinoma in female mice. This is a typical example of the case where a single model dominated by the linear term (q0 = 3.881 × 10−1, q1 = 4.084 × 10−2, higher terms set to zero) fits the data reasonably well. The shape of the curve approximates a normal distribution, although there is some departure from this ideal on the high side of the MLE. A few of the other datasets for 1,3-butadiene also follow that pattern, and some (lung and mammary tumors in the female mice) show fairly symmetrical normal shapes. In contrast, the data for Harderian gland carcinoma in female mice (Figure 29.2) show a case where although the optimal fit (q0 = 1.919 × 10−1, q1 = 4.143 × 10−2, q2 = 1.084 × 10−3, higher terms set to zero) is dominated by the linear term, an alternative fit with a greater contribution from higher-order terms (and thus lower values of q1) is also plausible, as indicated by a second broader peak on the low side of the MLE. Even more extreme deviations from the normal distribution shape appear in cases where the dose–response function for a particular tumor is strongly curved and the MLE of q1 for the optimal fit is in fact zero. An example of this situation is seen in Figure 29.3 for the preputial gland carcinomas in male mice. Here the data were fit with q4 = 1.947 × 10−6 and all other parameters set to zero. In this extreme case the q1* is finite, although small (3.244 × 10−3). In a case such as these datasets for 1,3-butadiene where some of the tumor sites show fairly linear dose responses, while others are highly curved, those with linear responses will dominate the low-dose risk based on a combined potency estimate. However, the diversity of likelihood distributions is quite considerable even where
1.80 1.60
Percent contribution
1.40 1.20 1.00 0.80 0.60 0.40 0.20 0.00 0.00
0.02
0.04
0.06
0.08
0.10
0.12
Estimate of q1
Figure 29.2. Likelihood density function for fit to the data on Harderian gland carcinoma in female mice (dichotomous models).
29.4. EXAMPLE: 1,3-BUTADIENE
729
1.80 1.60
Percent contribution
1.40 1.20 1.00 0.80 0.60 0.40 0.20 0.00 0.0000
0.0010
0.0020
0.0030
0.0040
0.0050
0.0060
Estimate of q1
Figure 29.3. Likelihood density function for fit to the data on preputial gland carcinoma in male mice (dichotomous models).
the optimal fit includes a substantial linear term, and it is not obvious which category will apply from a simple inspection of the data for any particular site. Therefore there is a significant potential for error when using a method which assumes, rather than measures, the shape of the likelihood distributions. This argues in favor of the application of the Monte Carlo procedure for addition of the whole distributions, rather than relying on addition of fixed points such as the MLE or 95% confidence limit. US EPA (2002) assumed normal shapes of the likelihood distributions and used the ratio of the MLE and 95% upper confidence limit to estimate the standard deviations (σ) of the site-specific distributions; the upper 95% confidence limit of the combined distribution was then estimated by adding the MLEs and also estimating the standard deviation of the combined distribution from the sum of the variances (σ2) for the individual sites, since 95% UCL = MLE + 1.645σ Using this method for the data calculated here suggested overall q1* values of 0.314 for male mice and 0.371 for female mice (based on all tumor sites, with the high dose excluded except where noted). Use of the Crystal Ball add-in to Microsoft’s Excel spreadsheet provides a convenient way of adding the complete distributions by the Monte Carlo procedure. This was done for both male and female mice, using as input the distributions for all the sites with significant tumor incidences, calculated as described above, and 100,000 trials. For male mice, the combined distribution for q1 had an MLE of 0.307
730
CHAPTER 29 CANCER RISK BASED ON AN INDIVIDUAL TUMOR TYPE OR SUMMING OF TUMORS
and an upper 95% confidence limit of 0.364. For female mice the values were MLE = 0.380 and upper 95% confidence limit = 0.422. Both combined distributions were slightly skewed toward the high end, but the departure from a normal shape was not large. The preceding calculations have relied on the dichotomous (whole-life incidence) version of the linearized multistage model to estimate the carcinogenic potency at individual sites and overall. These data for 1,3-butadiene are perhaps an unusual case of carcinogenicity at numerous independent sites, which also presents the problem of competing risks where some tumors are fatal. The solution adopted here to allow for these competing risks (which had a severe impact on the unadjusted tumor incidence data) of using the poly-3 survival analysis to predict end-of life tumor incidences was chosen to allow the dichotomous model approaches to be used for illustration and comparison, without complicating effects of mortality. However, in a real risk assessment project it would be preferable to use a time-dependent version of the multistage tumor incidence model, which determines the time dependence of tumor incidence independently at each site by analysis of the data, rather than assuming the time-related exponent of 3 for all datasets. Since both the poly-3 analysis and the time-dependent model require the same individual time-to-tumor data, the only further limitation that would prevent application of a time-dependent analysis to the current problem is that, until recently, software to list out the individual likelihood distributions was not available. This deficiency has now been remedied by Crouch (2006). Results obtained using this program for the NTP (1993) data on 1,3-butadiene are shown in Table 29.6. The time-dependent analysis allows consideration of whether observed tumors are considered fatal or incidental: this impacts the treatment of early mortality. For this analysis, lymphomas were considered fatal tumors, but tumors at all other sites were considered incidental; this is consistent with the description of tumor incidence and mortality by NTP (1993). The time-dependent model fit all dose groups adequately, although in some cases the fit without the high-dose group was better. Results for analyses with and without the high-dose group are shown in Table 29.6. Dropping the high-dose group had a variable effect: For some sites the q1* was increased, whereas for others it decreased. From a procedural point of view the fits with the high dose included might be considered preferable provided the fit was adequate, but the values with the high dose dropped are a better direct comparison to the q1* values obtained by the previous methods. Shapes of the likelihood density functions for estimates of q1 with the time-dependent model were in general terms similar to those observed with the dichotomous models, although actual parameter estimates and quality of fit measures naturally differed. This is illustrated in Figure 29.4, showing the time-to tumor model fit for the Harderian gland carcinomas in female mice. As with the dichotomous data, this distribution shows multiple peaks representing different orders of polynomial, although in this case a higher-order polynomial resulted in a better fit (and a somewhat lower estimate of q1) than the pure linear model. Comparison of these two figures also illustrates how relatively small changes in input data or model constraints can produce large and sudden changes in the maximum likelihood estimate of q1 due to “switching” of the distribution mode from one peak to another.
29.4. EXAMPLE: 1,3-BUTADIENE
731
TABLE 29.6. Multisite Time-Dependent Analysis of Tumor Incidence in the NTP(1993) Study of Mice Exposed to 1,3-Butadiene
Male Mice Administered Dose (ppm) Metabolized Dose (mg/ kg-day) Total Animals
0
0.00
6.25
0.62
20
1.97
62.5
6.12
200
625
17.80
54.30
q1*
q1* (mg/kg-day)−1 (All Dose Groups)
(mg/kg-day)−1 (High Dose Dropped)
50
50
50
49
50
73
Heart
0
0
1
5
20
4
2.173 × 10−2
3.360 × 10−2
Lymphomas
4
2
4
6
2
51
1.089 × 10−2
1.393 × 10−2
Histiocytic sarcoma
0
0
4
5
7
4
1.602 × 10−2
2.056 × 10−2
21
23
19
31
35
3
2.344 × 10−1
1.876 × 10−1
6
7
9
20
31
6
1.547 × 10−1
1.272 × 10−1
21
23
30
25
33
5
1.242 × 10−1
9.157 × 10−2
−3
9.326 × 10−3
Lung Harderian gland Liver Forestomach
1
0
0
1
8
4
9.300 × 10
Preputial
0
0
0
0
5
0
6.206 × 10−3
4.966 × 10−3
4.593 × 10
3.799 × 10−1
−1
Multisite
Female Mice Administered Dose (ppm) Metabolized Dose (mg/ kg-day) Total Animals Heart
0
0.00
6.25
0.62
20
1.97
62.5
6.12
200
625
17.80
54.30
q1*
q1*
(mg/kg-day)−1 (All Dose Groups)
(mg/kg-day)−1 (High Dose Dropped)
49
50
50
50
50
80
0
0
0
1
21
23
1.013 × 10−2
7.809 × 10−3
6.538 × 10
−2
8.466 × 10−2
Lymphomas
6
12
11
7
9
32
Histiocytic sarcoma
3
2
7
4
7
4
2.048 × 10−3
1.378 × 10−2
Lung
4
15
19
24
24
22
2.750 × 10−1
2.500 × 10−1
Harderian gland
8
10
7
15
20
9
1.164 × 10−1
8.899 × 10−2
15
14
15
19
16
2
1.409 × 10−1
1.060 × 10−1
20
3.757 × 10
−2
4.354 × 10−2
7.109 × 10
−2
6.213 × 10−2
8.262 × 10−2
6.006 × 10−2
6.126 × 10−1
5.171 × 10−1
Liver Forestomach
0
0
3
2
4
Ovary
1
1
1
9
7
6
Mammary gland
0
2
4
12
15
12
Multisite
732
CHAPTER 29 CANCER RISK BASED ON AN INDIVIDUAL TUMOR TYPE OR SUMMING OF TUMORS
1.8% 1.6%
Percent Contribution
1.4% 1.2% 1.0% 0.8% 0.6% 0.4% 0.2% 0.0% 0.0E+00
5.0E-02
1.0E-01
1.5E-01
q animal
Figure 29.4. Likelihood density function for fit to the data on Harderian gland carcinoma in female mice (time-to-tumor models).
29.5.
CONCLUSIONS
In the final analysis it seems evident that the time-dependent version of the linearized multistage model is the best way of fitting a complex dataset such as that for 1,3-butadiene where time-to-tumor data are available. This approach maximizes the extent to which the actual data are used to determine both the time and dose dependence of tumor incidence. This avoids, to the extent possible, the complicating effects on potency estimation at each site of intercurrent nontumor mortality and competing risks from other fatal tumors. The total incidence—that is, number of affected animals—has long been accepted as a proper basis for estimating the overall risk from tumors at a single site and of types related by progression from a common cellular origin. However, at least in the case of 1,3-butadiene, this approach appears to fail on two counts: First, it is not feasible to include all the tumor types of interest due to the high doserelated incidences and in some cases background rates. In particular, the elevated background rate for “all tumors” reduces the sensitivity of the analysis in the same way as it dilutes the sensitivity for hazard identification as noted by Haseman et al. (1986). In order to represent the total risk at low doses from several independent types of tumors, it is evidently necessary to separately determine the potency at each site and then use an appropriate procedure to add the individual potencies. A comparison of the results obtained by the approaches reviewed here is shown in Table 29.7. All the values shown are animal potency values (i.e., not corrected for human body weight or pharmacokinetics), using the rodent metabolized dose as the exposure metric, and were calculated as q1* values (95% upper confidence limits) or q1 (MLE)
733
29.5. CONCLUSIONS
TABLE 29.7. Overall Potency Estimates for Cancer Risk of Exposure to 1,3-Butadiene, Based on Tumors Observed in the NTP(1993) Bioassay in Mice
Total qanimal Estimate: (mg/kg-day)−1 Males
Females
Not determined
0.19
Basis All animals bearing selected tumors (US EPA 2002)
Dichotomous Model (Whole-Life Incidences, Poly-3 Corrected) Sum of q1 (MLE) Sum of q1* (95% UCL) Sum of q1 + 1.645σ q1* combined distribution
0.25 0.41 0.31 0.36
0.31 0.47 0.37 0.42
Time-Dependent Model (Multistage/Weibull, Time-to-Tumor Data) Sum of q1 + 1.645σ q1* combined distribution
0.36 0.38
0.50 0.52
from tumor incidence data excluding the high-dose group (except where constrained by poor fits, as noted earlier). In comparing these results it is interesting to note that the q1* values from the combined distributions are not greatly different between the full time-dependent model and the poly-3 corrected dichotomous model. This suggests that although clearly the full time-dependent model is preferable from a procedural point of view, the use of the poly-3 mortality correction is a reasonable substitute if needed, such as in cases where suitable software for the full multisite, time-dependent approach is unavailable. For the purposes of this discussion it is assumed that the Monte Carlo addition procedure to produce a combined distribution generates a “true” value for the combined q1 and q1*, against which other methods may be compared. As expected, the estimate obtained by adding upper confidence limits (q1*) exceeds the true value, while the sum of the q1 MLEs is significantly lower. The value obtained by US EPA (2002) using the tumor-bearing animals approach seriously underestimates the combined site potency due to its inability to consider all the sites of interest, as well as the limitations of the method. Their method of estimating the combined potency by adding the MLE’s and variances for individual sites (thus assuming a normal likelihood distribution) produces results closer to the true combined distribution values than simply adding MLEs or upper 95% confidence limits. However, this method consistently underestimates the true combined q1* to some degree for both the dichotomous data and the time-to-tumor data. Importantly, however, the differences involved are relatively small and are clearly less than the many other uncertainties involved in the risk assessment.
734
CHAPTER 29 CANCER RISK BASED ON AN INDIVIDUAL TUMOR TYPE OR SUMMING OF TUMORS
REFERENCES Anderson, E. L. and the Carcinogen Assessment Group of the US Environmental Protection Agency. (1983). Quantitative approaches in use to assess cancer risk. Risk Anal 3, 277–295. Bogen, K. T. (1990). Uncertainty in Environmental Health Risk Assessment, Garland Publishing, New York. California Department of Health Services (CDHS) (1985). Guidelines for chemical carcinogen risk assessments and their scientific rationale. CDHS, Health and Welfare Agency, Sacramento, CA. Crouch, E. C. (1985). MSTAGE program. E. A. C. Crouch, Cambridge Environmental Inc., 58 Buena Vista Road, Arlington, MA 02141. Crouch, E. C. (2006). Personal communication: Email from Dr. Edmund Crouch (Cambridge Environmental, Inc.) to Dr. Lauren Zeise (OEHHA). March 26, 2006. Crump, K. S., and Watson, W. W. (1979). GLOBAL79: A FORTRAN program to extrapolate dichotomous animal carcinogenicity data to low doses. National Institute of Environmental Health Sciences, Contract No. 1-ES-2123. Crump, K. S., Howe, R. B., Van Landingham, C., and Fuller, W. G. (1991). TOXRISK Version 3. TOXicology RISK Assessment Program. KS Crump Division, Clement International Division, 1201 Gaines Street, Ruston, LA 71270. Haseman, J. K., Huff, J. E., Zeiger, E., and McConnell, E. E. (1987). Comparative results of 327 chemical carcinogenicity studies. Environ Health Perspect 74, 229–235. Haseman, J. K., Tharrington, E. C., Huff, J. E., and McConnell, E. E. (1986). Comparison of site-specific and overall tumor incidence analyses for 81 recent National Toxicology Program carcinogenicity studies. Regul Toxicol Pharmacol 6, 155–170. Hattis, D. and Wasson, J. (1987). A pharmacoknetic/mechanism-based analysis of the carcinogenic risk of butadiene. US National Technical Information Service No. NTIS/PB 88-202817, M.I.T. Center for Technology, Policy and Industrial Development, CTPID 87-3. Hoover, S., Brown, J. P., Salmon, A. G., Sandy, M. S., Zeise, L., and Marty, M. A. (2005). Cancer risk estimation for exposure to naphthalene. Society of Toxicology Annual Meeting, New Orleans, LA, USA, March 2005. Abstract No.1521. The Toxicologist 84(S-1), 310. Huff, J. E., Eustis, S. L., and Haseman, J. (1989). Occurrence and relevance of chemically induced benign neoplasms in long-term carcinogenicity studies. Cancer Metast Rev 8, 1–21. Huff, J., Cirvello, J., Haseman, J., and Bucher, J. (1991). Chemicals associated with site-specific neoplasia in 1394 long-term carcinogenesis experiments in laboratory rodents. Environ Health Perspect 93, 247–270. International Agency for Research on Cancer (IARC). (1999). IARC Monographs on the evaluation of carcinogenic risk to humans, Vol. 71, part 1: Re-evaluation of Some Organic Chemicals, Hydrazine and Hydrogen Peroxide, pp. 109–122. International Agency for Research on Cancer (IARC). (2006). Monographs on the Evaluation of Carcinogenic Risks to Humans—Preamble. http://monographs.iarc.fr/ENG/Preamble/CurrentPreamble. pdf, Lyon, France. Kodell, R. L., Ahn, H., Chen, J. J., Springer, J. A., Barton, C. N., and Hertzberg, R. C. (1995). Upper bound risk estimates for mixtures of carcinogens. Toxicology 105, 199–208. McConnell, E. E., Solleveld, H. A., Swenberg, J. A., and Boorman, G. A. (1986). Guidelines for combining neoplasms for evaluation of rodent carcinogenesis studies. JNCI 76, 283–289. McDonald, T., Hoover, S., Faust, J., Rabovsky, J., MacGregor, M. K., Sherman, C., Sandy, M., and Zeise, L. (2003). Development of cancer potency estimates for California’s Proposition 65. Society of Toxicology Annual Meeting. March 2003, Salt Lake City, UT. Abstract No. 687. Toxicol Sci 72(S-1), 142. McDonald, T. and Komulainen, H. (2005). Carcinogenicity of the chlorination disinfection by-product MX. J Environ Sci Health Part C 23, 163–214. Melnick, R. L., Huff, J. E., Chou, B. J., and Miller, R. A. (1990). Carcinogenicity of 1,3-butadiene in C57BL/6 × C3H F1 mice at low exposure concentrations. Cancer Res 50, 6592–6599. National Research Council (NRC) (1990). Health Effects of Exposure to Low Levels of Ionizing Radiation. BEIR V. Committee on the Biological Effects of Ionizing Radiation, National Academies Press, Washington, D.C.
REFERENCES
735
National Research Council (NRC) (1994). Science and Judgment in Risk Assessment. Committee on Risk Assessment of Hazardous Air Pollutants, Board on Environmental Studies and Toxicology, Commission on Life Sciences, National Academies Press, Washington, D.C., 651 pages. National Toxicology Program (NTP) (1984). Toxicology and carcinogenesis studies of 1,3-butadiene (CAS No. 106-99-0) in B6C3F1 mice (Inhalation Studies). Publication No. NIH 84-2544, NTP Technical Report No. 288. National Toxicology Program, National Institutes of Health, Washington, D.C. National Toxicology Program (NTP) (1993). Toxicology and carcinogenesis studies of 1,3-butadiene (CAS No. 106-99-0) in B6C3F1 mice (inhalation studies). Publication No. NIH 84-2544, NTP Technical Report No. 434. National Toxicology Program, National Institutes of Health, Washington, D.C. Office of Environmental Health Hazard Assessment (OEHHA) (1992). Proposed identification of 1,3-butadiene as a toxic air contaminant. Part B. Health assessment. Hazard Identification and Risk Assessment Branch, Berkeley, CA. Office of Environmental Health Hazard Assessment (OEHHA) (2004). Air toxic hot spots: Adoption of a unit risk value for naphthalene. http://www.oehha.ca.gov/air/hot_spots/naphth.html Office of Environmental Health Hazard Assessment (OEHHA) (2007). Long-term health effects of exposure to ethylbenzene. http://www.oehha.ca.gov/air/hot_spots/pdf/Ethylbenzene_FINAL110607. pdf Portier, C. J. and Bailer, A. J. (1989). Testing for increased carcinogenicity using a survival-adjusted quantal response test. Fundam Appl Toxicol 12, 731–737. U.S. Department of Health and Human Services (US DHHS) (1982). The health consequences of smoking: Cancer. A report of the Surgeon General. US DHHS Pub. No. (PHS) 82-50179, Washington D.C. U.S. Environmental Protection Agency (US EPA) (1986). Guidelines for carcinogen risk assessment. Fed Reg 51, 33992–34003. U.S. Environmental Protection Agency (US EPA) (2002). Health assessment of 1,3-butadiene. EPA/600/ P-98/001F. National Center for Environmental Assessment—Washington Office, Office of Research and Development, US EPA, Washington, D.C. U.S. Environmental Protection Agency (US EPA) (2005). Guidelines for carcinogen risk assessment. EPA/630/P-03/001F. Risk Assessment Forum, Washington, D.C. Wilbourn, J. D., Haroun, L., Vainio, H., and Montesano, R. (1984). Identification of chemicals carcinogenic to man. Toxicol Pathol 12, 397–398. Zeise, L. and Salmon, A. G. (1991). Modified version of MSTAGE. Dr. A. G. Salmon, OEHHA, 1515 Clay St, 16th floor, Oakland, CA 94612. Zeise, L., Salmon, A. G., McDonald, T., and Painter, P. (1991). Cancer potency estimation. In Risks of Carcinogenesis from Urethane Exposure, Salmon, A. G., and Zeise, L., eds., CRC Press, Boca Raton, FL, pp. 97–112. Zielinski, J. M., Kodell, R. L., and Krewski, D. (2001). Interaction between two carcinogens in the twostage clonal expansion model of carcinogenesis. J Epidemiol Biostat 6, 219–228.
CH A P TE R
30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION Shannon Gaffney Jennifer Sahmel Kathryn D. Devlin Dennis J. Paustenbach
30.1.
INTRODUCTION
Historical exposures to chemicals in occupational and nonoccupational settings have, at times, been linked to increased health risks and disease. A risk assessment is typically required to understand the relationship between exposure and disease. A classic risk assessment model contains four key components: hazard identification, dose–response assessment, exposure assessment, and risk characterization (Paustenbach 2000). In order to understand the dose–response relationship and, ultimately, the risk for exposure to a given chemical, a dose reconstruction is often necessary. To effectively reconstruct doses, particularly when evaluating and characterizing historical exposures (critical in order to quantify risk to many carcinogens), it is often necessary to reconstruct past exposures. The concept of exposure reconstruction has been recognized for decades, but methods of such reconstruction have improved greatly over the years (Esmen 1979). The first step of an exposure assessment includes quantifying, to the extent possible, the amount of exposure via all possible routes of entry into the human body. Routes of exposure may include inhalation, dermal absorption, or ingestion of a substance of concern. Other key steps in an exposure assessment include validating or confirming exposure values when possible (such as by biological monitoring) and characterizing the uncertainties in the exposure assessment process (Paustenbach 2000; Williams et al. 2003). There is no single approach that will be effective in assessing all exposure conditions or situations; sometimes historical industrial hygiene measurements are available, while at other times, in the absence of data, exposure estimation may be necessary. Occasionally, simulation studies can be conducted to reconstruct exposures. Most
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
736
30.2. EXPOSURE RECONSTRUCTION METHODOLOGY
737
commonly, a number of methods or data sources are combined to generate the best possible reconstructed exposure estimates. Such combination approaches may take into account, for example, the following different sources of information: a small or incomplete historical industrial hygiene dataset, exposure modeling, qualitative information about experienced exposures, and a statistical uncertainty analysis. Developing quantitative exposure or dose estimates can be approached in three different ways, depending upon the requirements of the assessment. First, if personal exposures are of highest interest, exposures should be estimated or measured at the point of contact with the body, such as the lungs or skin. Second, if exposures from a particular scenario or location are the major interest, the intake or dose can be estimated or measured and then later combined with exposure duration information. Third, if body burden or the presence of a particular biomarker is the focus, a total body dose can be estimated or measured (US EPA 1992a). Any of these approaches may be appropriate for exposure reconstruction, depending on the type of information available. Ideally, direct measurement data should be used whenever possible to evaluate exposures. However, when such data are lacking or insufficient, there are a number of established methods or method combinations that have proven valuable for reconstructing past exposures, including specific quantitative, semiquantitative, and qualitative techniques. While many of these have been used for some time, others are relatively new. The use of computer-assisted advanced modeling techniques has also greatly assisted the exposure reconstruction field. Advanced statistical methods can now be applied to both measured and modeled estimates to significantly improve the understanding of, and confidence in, exposure reconstructions. This chapter consists of two primary sections. The first section presents a framework or methodology for reconstructing exposures of interest, which can then be used to produce exposure estimates appropriate for conducting cancer risk assessments. The second section discusses the application of reconstructed exposures to the determination of dose estimates and cancer potency. Although the methods discussed here can be used in both occupational and nonoccupational exposure scenarios, the focus of this chapter is primarily on occupational exposure reconstruction methods.
30.2. EXPOSURE RECONSTRUCTION METHODOLOGY This section will outline a framework for approaching exposure reconstruction for risk assessment purposes, with a focus on deriving quantitative or semiquantitative exposure estimates that are needed in order to complete a quantitative risk assessment. There are five general steps in the exposure reconstruction process, including: (1) addressing the goals of the exposure or dose reconstruction; (2) organizing all available exposure data or other information and ranking it according to its value and relative contribution to the exposure reconstruction process (typically by quantitative, semiquantitative, and qualitative types of data); (3) identifying key data gaps in the available information; (4) selecting the appropriate method or combination of
738
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
methods needed to reconstruct the exposure values or estimates of interest; and (5) conducting an uncertainty analysis or statistical characterization of the reconstructed exposure values in order to understand the quality of the values that have been generated.
30.2.1. Addressing the Goals of the Exposure Reconstruction Exposure reconstructions can be developed using several different types of techniques, and can be conducted using widely varying levels of scientific rigor or certainty, depending upon how the results will be used. Estimates may be generated for use in a number of different types of investigations, including epidemiology studies, evaluation of cancer or other illness clusters, concerns or questions regarding worker exposure, or other types of risk assessment (Viet et al. 2008). Exposure estimates developed even from robust existing exposure measurements often require additional task-specific information in order to understand the ultimate internal dose of interest and the effects on the human body, including outcomes such as cancer (Agency for Toxic Substances and Disease Registry 2005). An exposure reconstruction can be structured to provide simple qualitative results using an exposure classification scheme such as exposed versus not exposed, or it may employ more specific exposure classifications such as quantitative exposure estimates by work location, job title, or individual task. The health outcome of interest will also affect the goals of the reconstruction. If the outcome of concern is the result of acute exposure, then short-term or peak exposures will be the focus of the reconstruction. For chronic outcomes such as cancer, longer-term exposures in the form of time-weighted average (TWA) exposures and lifetime cumulative exposures are likely to be of greatest interest (Viet et al. 2008). The most robust exposure reconstruction method possible should generally be selected to meet the needs of the assessment. At a minimum, a quantitative or semiquantitative method should be used for cancer risk assessment purposes. Quantitative methods for reconstructing exposures may improve risk estimates by reducing nondifferential exposure misclassification, and can also provide the opportunity to investigate the toxicological mode of action. Investigating the mode of action can be particularly helpful for substances or scenarios in which the mechanism of carcinogenicity is not well understood. However, even when basing an exposure reconstruction on quantitative exposure data or estimates, exposure misclassification can potentially be a significant problem, and may affect study results more than any other confounding factor (Blair et al. 2007; Blair and Stewart 1992; Grandjean et al. 2004). Therefore, careful consideration must be made when selecting any exposure reconstruction method or combination of methods to ensure that it will be appropriate both for the scenario and the outcome of interest.
30.2.2. Organizing and Ranking Available Exposure Information When preparing to conduct an exposure reconstruction, data organization is critical, particularly for reconstructions that potentially involve hundreds or perhaps
30.2. EXPOSURE RECONSTRUCTION METHODOLOGY
739
thousands of workers performing many different job tasks. All available information potentially associated with the exposure scenarios of interest should first be collected, including actual exposure measurements for the applicable scenarios (where available), exposure measurements for similar scenarios during similar time periods, current exposure data if the exposure scenario is still in existence, information on workplace or environmental conditions and characteristics, typical practices of individual workers, characteristics and descriptions of processes, key process changes over time, a description of all chemicals or substances of interest used in the environment or workplace, exposure controls in place during the time period of interest, changes in exposure controls over time, assessments by safety and health professionals or the workers themselves regarding exposures or exposure potential, medical monitoring data, safety reports, and any other relevant information associated with exposure potential. A key step in the exposure reconstruction process is to determine the level of detail to be used to reconstruct the exposures. The reconstruction may be as simple as grouping data according to exposed populations versus nonexposed populations (i.e., qualitative exposure reconstruction estimates), or it may include quantitative estimates for individual worker tasks within the same job title or worker group. The level of detail used to reconstruct exposures is sometimes referred to as the exposure classification strategy. Although many different classification schemes have been used in exposure reconstructions for the workplace, in general, classifying exposures according to the most specific exposure grouping possible is ideal. For example, while exposures can be reconstructed by department, division, or work location, it may be more informative to differentiate exposures by individual job title with an emphasis on 8-hr time-weighted average (TWA) exposures, or by specific tasks conducted during a work shift, even when a job title is the same among a group of individuals. However, while the use of job title as a primary means of exposure classification is common, it should be noted that variability in exposures among workers within the same job title can be significant (Esman 1979; Gaffney et al. 2009; Ignacio and Bullock 2006; Panko et al. 2009; Stewart et al. 1996). Large exposure variability within job titles can be caused by such issues as a wide variety of specific tasks completed by individual employees or differences in the way employees carry out the same types of tasks (Friesen et al. 2006; Proctor et al. 2004; Rappaport 1991). Furthermore, classifications based on historical job title alone may result in a lower exposure contrast between groupings than classifications based on location, such as plant and department (Vermeulen and Kromhout 2005). Job title and other exposure determinants (such as worker location) may need to be used together to appropriately categorize exposures. Approaches using exposure zones defined by a combination of job title and work location have also been employed (Corn and Esmen 1979; Panko et al. 2009; Stewart et al. 1998). The use of exposure zones rather than the more traditional task-based exposure classification has been shown to be a statistically valid method of classification (Smith et al. 1997). Exposure zones, rather than job titles, have also been used to estimate noise exposures (Burgess et al. 2004). There can be a trade-off between using broader exposure classes—which reduces misclassification but results in poorer resolution between exposure distributions—and using narrower exposure classes, in which the opposite effect occurs (Esmen et al. 2007c).
740
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
The available data and information to be used in the reconstruction should be ranked from most to least robust. This step is one of the most important in the process, because once a full understanding of the available information is achieved, a strategy can be developed to generate the most appropriate reconstructed values. For a majority of scenarios, quantitative data will be most important and relevant to the assessment, followed by semiquantitative data or estimates, and finally, qualitative data, information, or estimates.
30.2.3. Identifying Key Data Gaps in the Available Exposure Information Once all available and relevant information has been assembled, it can be critically analyzed for missing data elements or significant data gaps. The exposure assessor must determine where existing exposure measurements can be used directly and where other techniques must be applied to extrapolate or estimate the exposures of interest. Key types of data gaps that may need to be addressed include insufficient existing exposure data, inadequate information to characterize exposure scenarios for individual job tasks or titles, limited inputs for exposure models, or a lack of certainty regarding appropriate exposure classification. Data gaps can often be addressed by using a combination of exposure reconstruction techniques simultaneously; this approach is common, and there are many examples of it in the peer-reviewed literature. A number of these studies are discussed and compared below. The use of a combination of methods may also be the most effective approach for exposure reconstruction for a particular scenario, given a unique dataset with certain robust elements and other relatively weak elements. Uncertainty analyses, such as a Monte Carlo statistical assessment, can also be used to address data gaps generated by uncertainty in existing data or information, as well as to increase the likelihood that true exposures are captured (Cohen Hubal et al. 2000). Once key data gaps have been identified, an appropriate method or combination of methods can be selected to ensure that they are adequately addressed.
30.2.4. Selecting the Appropriate Methodology to Reconstruct Exposure Values A number of specific techniques can be used to address the identified data gaps in an exposure or dose reconstruction. Quantitative techniques can include estimating exposure based on similar exposure data (either historical or current), data collected during simulation studies, or biological monitoring data. Semiquantitative techniques can include the use of exposure data matrices and the use of extrapolation, estimation, or modeling based on similar data or available exposure information. Qualitative techniques can include the use of some types of exposure determinants, the use of expert or professional judgment, and the collection of self-reported exposures. Similar groupings of exposure reconstruction categories have been described by the U.S. Agency for Toxic Substances and Disease Registry (ATSDR) (Maslia et al. 2001). Combination techniques can include various permutations of the above techniques, including the consideration of mixed types of data together, or the use
30.2. EXPOSURE RECONSTRUCTION METHODOLOGY
741
of newer techniques for combining different types of data such as Bayesian decision analysis, for example. The appropriate selection of a method for reconstruction and estimation is extremely important, whether for an exposure assessment, risk assessment, or epidemiological study. Some of the factors affecting the selection of an exposure reconstruction method include: the characteristics of the outcome of interest (such as carcinogenic latency and dose of concern); the effectiveness of analytical methods to identify and measure the substance of interest (such as ability to distinguish from similar substances or interferences caused by mixtures of substances); and the consideration of exposures by multiple routes, including inhalation and dermal exposures. 30.2.4.1.
Quantitative Reconstruction Methods
30.2.4.1.1. Use of Historical Exposure Data to Reconstruct Exposures. Employing existing contemporaneous measurement data to characterize historical exposure scenarios is always the most appropriate and straightforward method for conducting an exposure or dose reconstruction. Quantitative reconstructions can provide exposure estimates with a unit of measurement (i.e., ppm, mg/m3, ppmyears, etc.) rather than a qualitative descriptor of the magnitude of exposure (such as low, medium, or high). Studies have shown that the arithmetic mean of available data for a specific exposure category is typically the most appropriate metric for characterizing cumulative worker exposures (Checkoway and Rice 1992; Crump 1998; Seixas et al. 1988; Williams and Paustenbach 2005), rather than base calculations for similar groups using the geometric mean. While a direct data analysis method is potentially the most accurate way to reconstruct exposures, a number of caveats may exist when using measured data. Depending on the number of years that have passed between when monitoring data were collected and the timeframe of the exposure of interest, even historical measurements collected for a similar scenario may not be entirely representative of true exposures unless the specific differences between the scenarios are fully understood. Some important factors to consider when using similar data to reconstruct exposures include process changes, exposure control changes, or differences in personal practices regarding general hygiene or the use of personal protective equipment. Changes in sampling techniques and analytical methods over time may also affect exposure values. There are four important considerations when using historical industrial hygiene measurements to reconstruct similar exposures. First, such historical measurements were often collected as part of a targeted sampling campaign for the purposes of exposure control and compliance with existing exposure standards. As a result, those jobs or tasks for which a known exposure concern existed (and which may have had the highest exposures) might have been disproportionately sampled (Gaffney et al. 2009; Panko et al. 2009; Stewart et al. 1996). Second, industrial hygiene measurements typically were or are taken on the outside of personal protective equipment; therefore, those samples will not be representative of actual exposures to an individual (Hall et al. 2007). Third, certain types of exposure measurements, including industrial hygiene measurements, were not historically
742
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
collected using random sampling techniques or a systematic approach (Esmen et al. 2007c; Stewart et al. 1996). Fourth, the number of samples or measurements upon which an exposure reconstruction is based can affect the level of confidence in the results because exposure variability may not be well-characterized (Ignacio and Bullock 2006). Some historical exposure datasets have shown significant within- and betweenworker variability (Rappaport 1991). Six to ten exposure measurements from a specific operation or task (over a relatively narrow time window) have been recommended to approach a level of confidence at the 95th percentile, and possibly more, depending on the sample variability (Ignacio and Bullock 2006). In one example of an exposure reconstruction using historic data to estimate beryllium exposures and associated beryllium sensitization thresholds, the authors reconstructed the TWA exposures for workers in a number of different ways to determine potential differences in the estimates generated. These different methods included calculation of the TWA by highest year exposed based on available industrial hygiene data, by highest job title exposed, and by pooled data for year and job title, as well as by lifetime averages, among other groupings. The results demonstrated that exposure estimations based on shorter averaging times better represent upper-bound exposures and, therefore, better capture variability in the dataset (Madl et al. 2007). 30.2.4.1.2. Use of Current Exposure Data to Reconstruct Exposures. The use of present-day quantitative exposure measurements can be acceptable for estimating historical exposures in certain situations. Retrospective estimation of exposures using existing data should only be applied to scenarios where the facility, operations, and specific job positions have remained stable over the entire historical period of interest, and these situations are rare (Smith et al. 1984). Since there were typically dramatic improvements in the quality of safety and health in workplaces after the passage of the U.S. Occupational Safety and Health Act of 1970 (OSH Act; Public Law 91-596), reconstructions of exposures prior to this date may be more difficult. When attempting to use current data to estimate historical exposures, work histories and other exposure determinants can be used to classify persons by job tasks, and information on exposure duration for task or job categories could also be potentially significant (Smith et al. 1984). For those situations where little has changed over time, cumulative past exposures might then be estimated using the current concentration data multiplied by the exposure duration (Smith et al. 1984). 30.2.4.1.3. Use of Simulated Exposure Data to Reconstruct Exposures. Often, when there is a lack of historic data, or current exposure data are not deemed representative of historic scenarios, simulation studies are conducted to recreate a historical exposure scenario in order to reconstruct the exposure or dose. When conducting an exposure simulation, it is important to gather as much information as possible regarding the past scenario. This information can be obtained through reports, notes, records, and interviews with retired workers (Fallentin and Kamstrup 1993). If necessary, equipment and materials representative of the time of interest should be found or created. Although it is unlikely that a simulation will exactly duplicate the original exposure scenario, it can often provide the most representative
30.2. EXPOSURE RECONSTRUCTION METHODOLOGY
743
estimates of historical exposures. If nothing else, simulation studies can provide an indication of the quality of historical exposure estimates or validate small or incomplete sets of data, and they are likely to provide substantive evidence regarding the order of magnitude of the original exposure level (Esmen and Corn 1998; Fallentin and Kamstrup 1993; Madl et al. 2009). A number of studies have used simulations to reconstruct historical occupational exposures (see Table 30.1). Examples of occupational exposure simulation studies include side-by-side measurements using both historical sampling instruments and contemporary instrumentation to compare the results for consideration in historical reconstruction of dust exposures (Dahmann et al. 2008); measuring exposures to methanol during the process of semiconductor wafer cleaning (Gaffney et al. 2008); and characterizing the absorbed dose of hexavalent chromium via all routes of exposure following contact with tap water (Paustenbach et al. 2003). Perhaps the biggest benefit of simulation studies is that they attempt to recreate an exact situation, and use modern, generally highly sensitive measuring devices that provide accurate results, rather than relying on estimation or personal opinion to determine past exposure. Although simulation studies can be very costly to conduct, especially if all variables are well-controlled, some have reported that representative results can be obtained in some situations for a reasonable cost (Esmen and Corn 1998). 30.2.4.1.4. Use of Biological Monitoring Data to Reconstruct Exposures. The use of biological monitoring data can be an effective way to reconstruct historical exposures to chemicals (Aylward et al. 1996; Hayes et al. 2008). Biomonitoring data can also be particularly useful when the exposures of interest may have occurred dermally (Paustenbach and Madl 2008). Current biological monitoring data may also be used in certain circumstances to help reconstruct past exposures, particularly for substances that are known to bioaccumulate. Biological monitoring has most commonly been conducted using blood or urine samples. When chemical-specific biological monitoring data for the time period of interest are available, they can help to complete the exposure and dose picture and provide a comparison to air sampling or skin surface exposure measurements alone (Hayes et al. 2008; Leung and Paustenbach 1988). However, the list of chemicals for which reference doses for biological data are available is limited. The American Conference of Governmental Hygienists has developed quantitative Biological Exposure Indices (BEIs) for 43 chemicals (American Conference of Industrial Hygienists 2008), while the American Industrial Hygiene Association has developed one Biological Environmental Exposure Limit (BEEL), with a number of additional BEELs under development (American Industrial Hygiene Association 2008). In addition, a Biomonitoring Equivalents (BE) Pilot Project is underway, jointly sponsored by several governmental and private institutions from which five example BEs have already been published (Aylward et al. 2009a,b; Aylward and Hayes 2009; Hayes et al. 2008, 2009a,b; Hayes and Aylward 2009). Furthermore, the U.S. Center for Disease Control and Prevention’s (CDC)’s National Center for Environmental Health (NCEH) has reported biomonitoring data for 116 environmental chemicals as part of the National Health and Nutrition Examination Survey (NHANES) (Centers for Disease Control and Prevention 2005).
TABLE 30.1.
Selected Examples of Simulation Studies in the Peer reviewed Literature
744
Study Author (Country)
Year
Gaffney (USA)
2008
Cleaning semiconductor wafers
Methanol
Sheehan (USA)
2008
Benzene
Dahmann (Germany) Mowat (USA)
2008 2007
Mangold (USA)
2006
Paustenbach (USA)
2006
Cleaning of elevator bearing housings using mineral spirit solvent Uranium mining industry Occupational exposure to asbestos from asphaltbased roofing products Installation and removal of asbestos gaskets and packing Removal of automobile exhaust systems
Paustenbach (USA)
2003
Hexavalent chromium [Cr(VI)]
Madl (USA) Madl (USA)
2002 2002
Esmen (USA) Gitelman (USA) Schultz (USA)
1998 1996 1995
Fallentin (Denmark) McKinnery (USA)
1993 1992
Keyes (USA)
1991
Characterization of absorbed dose of Cr(VI) following contact with tap water via all routes of exposure Cleaning of locomotive generator and traction motor Concentrations of benzene due to diesel exhaust from a locomotive in a roundhouse Splitting and boxing bags Auto mechanic performing brake repair Exposure of hospital emergency personnel during decontamination of chemically exposed patients Slag wool production factory Removal and installation of valve gaskets and packing Ceiling cable installation in presence of asbestoscontaining fireproofing
Scenario
Chemical(s)
Dust, crystalline silica, arsenic Asbestos Asbestos Chrysotile asbestos
Benzene, mineral spirits Benzene Chrysotile asbestos 1,1,1, Trichloroethane Acetone, p-xylene, iron oxide, zinc oxide, total dust PAH Asbestos Asbestos
Type of Estimate Generated Short-term (15-min) and long-term (2 to 4-hr) personal and area samples Breathing zone, area, and background airborne concentrations (mg/m3) Dust measurements (mg/m3) Airborne fiber concentrations (fibers/m3) Personal and area airborne fiber concentrations (fibers/m3) Personal, bystander, and area airborne fiber concentrations (fibers/m3) Red blood cell uptake and sequestration of chromium along with total urinary excretion (mg/L) Range of 8-hr TWA concentrations (ppm) 1-hr TWA (ppm) 15-min TWA (fibers/m3) Air concentrations (ppm) Breathing zone concentrations (ppm and mg/m3) TWA (μg/m3) Airborne fiber concentrations (fibers/m3; structures/m3) Area and personal samples (structures/m3)
References: Dahmann et al. (2008), Esmen and Corn (1998), Gaffney et al. (2008), Gitelman and Dement (1996), Keyes et al. (1991), Madl and Paustenbach (2002a,b), Mangold et al. (2006), McKinnery and Moore (1992), Mowat et al. (2007), Paustenbach et al. (2003), Paustenbach et al. (2006), Schultz et al. (1995), Sheehan et al. (2008), Fallentin and Kamstrup (1993).
30.2. EXPOSURE RECONSTRUCTION METHODOLOGY
745
In the future, biological monitoring will likely play a larger role in exposure and dose reconstruction studies as additional data are collected and additional reference doses are developed (Agency for Toxic Substances and Disease Registry 2005; American Conference of Industrial Hygienists 2008; American Industrial Hygiene Association 2008; US CDCP 2008a,b). Examples of studies that have used biomonitoring to evaluate exposure include exposure reconstruction using blood sampling for dioxins, furans, and polychlorinated biphenyls (PCBs) (Aylward et al. 1996); using tooth enamel spectroscopy to estimate exposure to radioactivity following the Chernobyl accident (Ivannikov et al. 2004); using blood and hair samples to estimate methylmercury exposures versus self-reported intakes through food sources (Gosselin et al. 2006); using chromosome painting, tooth enamel spectroscopy, and the glycophorin A mutation assay to reconstruct exposures (Kleinerman et al. 2006); comparing pesticide urine concentrations with an exposure algorithm for estimating lifetime average occupational exposure intensity (Acquavella et al. 2006); and comparing measured benzene metabolites against measured airborne exposures (Carrieri et al. 2006). Table 30.2 provides selected examples of studies that have used biological monitoring to reconstruct past exposures. 30.2.4.2. Semiquantitative Reconstruction Methods. In scenarios with limited exposure data or no exposure data, exposures can be reconstructed in a semiquantitative manner through techniques such as extrapolation or estimation. Extrapolation or interpolation approaches are a scientific way of estimating unknown values from known values and/or observations. Extrapolation techniques include using exposure data matrices, which may derive semiquantitative estimates using a variety of data sources; considering information obtained through interviews with long-term workers and other personnel (Deadman et al. 1997); using statistical techniques such as analysis of variance (ANOVA) and regression models (Seixas et al. 1993; Romundstad et al. 1999; Hallock et al. 1994; Hornung et al. 1994; Hein et al. 2008); applying multiplication or modifying factors with existing data (Lewis et al. 1997; van Tongeren et al. 1998; Henn et al. 2007; Laden et al. 2006); or combining any of the above (Sivulka and Seilkop 2009). Alternatively, estimation methods are typically used when insufficient information is available for extrapolation, interpolation, or exposure data matrices. Estimation methods include deterministic and stochastic (probabilistic) mathematical models (Kauppinen et al. 1994; Keil 2000; World Health Organization 2005; Yu et al. 1990), exposure algorithms (Armstrong et al. 1996; Glass et al. 2000), and the source-receptor model (Smith et al. 1991; Kauppinen et al. 1994; Smith et al. 1993). 30.2.4.2.1. Use of Exposure Data Matrices to Reconstruct Exposures. Exposure data or job exposure matrices have been used since the 1940s, and continue to be a valuable method for exposure reconstruction (Coughlin and Chiazze 1990). A job exposure matrix (JEM) aims to incorporate all sources of available data in order to link information about job categories and likely exposures (Coughlin and Chiazze 1990). The types of information used in a JEM, if available, include: exposure data, current data combined with other information extrapolated to previous job tasks or working conditions, exposure time durations, job titles, work locations or
746 TABLE 30.2.
Selected Examples of Biological Monitoring Studies in the Peer Reviewed Literature
Study Author (Country)
Year
Scenario
Acquavella (USA)
2006
Pesticide application by farmers
Gosselin (Canada)
2006
Kleinerman (USA)
2006
Carrieri (Italy)
2006
Methymercury intakes were reconstructed for a fish-eating indigenous population using total mercury concentrations in hair segments and a toxicokinetic model; corresponding mercury blood concentrations were then simulated Comparison of three biodosimeters used to estimate past radiation exposure (chromosome painting, electron paramagnetic resonance with teeth, and glycophorin A somatic mutation assay) Comparison of biological exposure indexes to assess exposure to gasoline filling-station attendants
Aylward (USA)
2005
Egeghy (USA)
2005
Evaluation of concentration-dependent elimination kinetics for 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) by varying the hepatic elimination rate parameter for introduction of TCDD to the liver Mixed effects regression models developed to better understand the temporal variability in total human exposure for metals, PAHs, and pesticides
Chemical(s)
Type of Estimate Generated
Urinary pesticide concentrations (glyphosate, 2,4-D, and chlorpyifos) Organic and inorganic mercury used as biomarker for methylmercury exposure
Range of concentrations (ppb) Reconstructed MeHg intakes (μg/kg/day)
Ionizing radiation
Gy and mGy
Trans, trans-muconic acid and S-phenylmercapturic acid excretion compared to evaluate exposure to benzene Serum lipid TCDD
Creatinine (μg/g)
Metals (lead), PAHs (phenanthrene), and pesticides (chlorpyrifos)
Serum lipid concentrations (ppt)
ng/m3, ng/m2, ng/wipe, ng/dl blood, μg/liter urine
Study Author (Country)
Year
Scenario
Koo (South Korea)
2005
Ivannikov (Russia)
2004
Chang (Taiwan)
2004
Kawasaki (Japan)
2004
Mage (USA)
2004
Olsen (USA)
2003
Weisel (USA)
1996
Examination of urinary levels of phthalates and their metabolites to assess exposure among Korean women and children Estimation of individual irradiation doses determined using electron paramagnetic resonance spectroscopy of the tooth enamel for inhabitants of the most contaminated inhabited settlement not evacuated following Chernobyl accident Examination of the effects of dermal and respiratory exposure to N,N-dimethylformamide (DMF) on the total body burden Assess changes in concentrations of cadmium and its markers in the blood, urine, and serum of workers at a cadmium pigment factory Analysis of the urinary excretion of 12 chemicals, which were either pesticides or their metabolites, to estimate pesticide dose Investigation of exposure levels of employees to PFOS and PFOA according to their jobs and work areas by measuring serum levels Time-series measurements of benzene in exhaled breath (biomarker) were used to evaluate a PBPK model
Chemical(s)
Type of Estimate Generated
Phthalates (DEHP, DBP, BEP, BBP) and the metabolite for DEHP (MEHP) Ionizing radiation
μg/kg body weight/day
Urinary biomarkers of DMF
ppm, μg/cm2, and mg/liter
Cadmium and urine and serum markers of effects on renal tubular function Various pesticides and their metabolites
μg/g, mg/liter, and mg/g
mGy
μg/kg/day
Serum concentrations of PFOS and PFOA
Geometric means (ppm)
Benzene
μg/m3
References: Acquavella et al. (2006), Aylward et al. (2005), Carrieri et al. (2006), Chang et al. (2004), Egeghy et al. (2005), Gosselin et al. (2006), Ivannikov et al. (2004), Kawasaki et al. (2004), Kleinerman et al. (2006), Koo and Lee (2005), Mage et al. (2004), Olsen et al. (2003), Weisel et al. (1996).
747
748
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
zones, working time periods, historical work and facility procedures (including changes in these procedures over time), and using personal protective equipment. Once a JEM is created, the resulting exposure estimates can be linked to an individual’s work history to estimate cumulative exposure (Harber et al. 2003; Macaluso et al. 1996). The validity of the exposure estimates resulting from a JEM is dependent on the level of completeness and accuracy of the information sources used (Harber et al. 2003; Macaluso et al. 1996). Databases of similar exposure groups along with associated exposures for each group have been shown to be effective tools for creating exposure estimates for use in epidemiology studies (Drummond et al. 2006). A number of studies have applied JEMs using different approaches to estimate past exposures for particular situations (see Table 30.3). Dement et al. (2009) developed a JEM for historical asbestos fiber exposures in three asbestos textile plants to estimate fiber-size-specific exposures to airborne asbestos for use in epidemiological investigations (Dement et al. 2009). Additional examples of scenarios in which JEMs have been used include assessments of silica exposures in mine and pottery workers, organophosphate exposures in farm workers, and organic solvent exposures at paint manufacturing sites, among others (Dosemeci et al. 1993; Gerin et al. 1993; Glass et al. 1994; Harber et al. 2003; Herrick et al. 2005; London and Myers 1998; Macaluso et al. 1996; Owen et al. 1992; Romundstad et al. 1998; Smith et al. 2001; Stewart et al. 1986). Studies have also evaluated JEMs and their strengths and weaknesses (Bouyer and Hemon 1993; Coughlin and Chiazze 1990; Friesen et al. 2003; Goldberg et al. 1993; McNamee 1996; Seixas and Checkoway 1995; Vermeulen and Kromhout 2005). The strengths of JEMs include enhanced statistical power to detect associations, low cost (economical to develop), avoidance of certain biases, including differential information bias, and the ability for validation through comparison with existing industrial hygiene data when available. The weaknesses of JEMs include possibly misclassifying exposure, depending on the accuracy of historical data and records, forming inaccurate, statistically significant associations due to the large number of comparisons made, and generating summary, rather than specific, exposure measurements. 30.2.4.2.2. Use of Extrapolation Methods to Reconstruct Exposures. The scientific approach that is employed in most methods of extrapolation or interpolation provides a structured format for estimating past exposures while allowing personal input and logic; however, there are some limitations to this approach. In this type of approach, knowledge of exposure trends or patterns over time can be used to complete gaps in existing datasets. Extrapolations can be conducted using statistical models, modifying factors (Sivulka and Seilkop 2009), nearest-neighbor interpolation, inverse-distance-squared weighting, kriging (Wartenberg et al. 1991), or task-TWA models (Smith et al. 1991, 1993). However, the accuracy of the estimated exposures is limited by the reliability and robustness of available exposure data (whether past or current), regardless of the scientific rigor of the extrapolation method used (Hornung et al. 1994; Stewart et al. 1996; van Tongeren et al. 1998). It can also be difficult and imprecise to estimate trends in historical exposure from current data, for the reasons discussed above, including changes in
TABLE 30.3.
Selected Examples of Studies Using Job Exposure Matrices in the Peer Reviewed Literature
749
Study Author (Country)
Year
Scenario
Dement (USA)
2009
Couch (USA)
2005
Harber (USA)
2003
Smith (USA)
2001
Sanderson (USA)
2001
London (South Africa)
1998
Romundstad (Norway)
1998
Development of fiber-size-specific estimates of exposure to airborne asbestos in asbestos textile plants for use in epidemiological investigations Assessment of historical exposures of workers at the Portsmouth Naval Shipyard using weighting factors based on knowledge of work activity, actual data, engineering controls, and PPE for each year, job title, and shop Use of industrial hygiene measurements, a formal process survey identifying dates and process changes, and a historical relative exposure rating scale to estimate exposures to carbon black production Reconstruction of historical exposures at various fiberglass, glass filament, and rock/slag wool production plants using exposure data and information about operations and job titles Estimation of worker exposure to beryllium in a lung cancer case–control study using historical data from a beryllium manufacturing plant and detailed job histories Estimation of long term exposure of farm workers to agrichemicals based on secondary data from industry and expert opinions Estimation of historical exposure levels at a coke plant using relevant industrial hygiene data and job exposure profiles
Chemical(s)
Type of Estimate Generated
Asbestos fibers
Fibers/cm3
Benzene and carbon tetrachloride
Exposure scores normalized to current occupational limits
Carbon black
mg years/m3
Fiberglass, mineral wool, formaldehyde, trace metals, silica dust, and multiple other chemicals Beryllium
fibers/cm3, ppm, mg/m3
Daily weighted average (μg/cm3)
Organophosphate
kg organophosphate
PAHs, carbonaceous particulates, quartz, benzene, asbestos, arsenic, carbon monoxide
8-hr TWA (μg/m3, mg/m3)
(Continued)
750
TABLE 30.3. (Continued)
Study Author (Country)
Year
Scenario
Kauppinen (Finland)
1998
Description of a Finnish JEM (FINJEM) constructed for exposure assessment in large register-based studies; estimates exposures based on definitions, inferences, exposure data, and references
Macaluso (USA)
1996
Glass (England)
1994
Dosemeci (USA)
1993
Gerin (Canada)
1993
Use of work areas, historical changes, and mathematical models to obtain cumulative exposure estimates at six North American rubber manufacturing plants Examination of solvent exposures at two paint making sites; investigation of neuropsychological effects of exposure Quantitative estimation of exposure to silica dust in mine and pottery workers in China using historical exposure data and current exposure profiles Application of a welding process exposure matrix to welding history of 11,092 welders
Owen (USA)
1992
Stewart (USA)
1986
Investigation of a potentially increased occurrence of colorectal cancer among employees in a polypropylene unit using available industrial hygiene data, records, work histories, and employee recollections Comparison of historical exposures, current exposures, and job titles to develop a final exposure estimate
Chemical(s)
Type of Estimate Generated
Variety of physical, chemical, and microbial agents, ergonomical or physiological stress factors, and psychosocial stress factors 1,3-Butadiene, styrene, benzene
Mean level of exposure (various units and scores)
Organic solvents
8-hr TWA (ppm)
Silica
mg/m3
Welding fumes, total chromium, chromium (VI), nickel
Cumulative estimates (mg yr/m3) Average concentrations (mg/m3) High, medium, low, none
Butylated hydroxyl toluene, xylene, calcium stearate, methanol, sodium hydroxide
Formaldehyde
ppm-yr
Cumulative exposure (ppm)
References: Couch et al. (2005), Dement et al. (2009), Dosemeci et al. (1993), Gerin et al. (1993), Glass et al. (1994), Harber et al. (2003), Kauppinen et al. (1998), London and Myers (1998), Macaluso et al. (1996), Owen et al. (1992), Romundstad et al. (1998), Sanderson et al. (2001), Smith et al. (2001), Stewart et al. (1986).
30.2. EXPOSURE RECONSTRUCTION METHODOLOGY
751
the workplace (such as process modifications and the use of different raw materials) and changes in exposure controls over time (Yu et al. 1990). Additionally, many uncertainties are possible, the largest of which are in the lowest exposure estimates (Smith et al. 1993; Williams and Paustenbach 2003, 2005). 30.2.4.2.3. Use of Estimation Methods to Reconstruct Exposures Deterministic Models. Deterministic models have often been used to estimate exposures when characteristics of a particular operation or process are known, but exposure measurements have not been collected or are not available. By definition, a deterministic model is one that provides point estimates of exposure based on the best available model inputs. By comparison, stochastic or probabilistic models use a likely range for one or more of the model inputs in order to address any uncertainty in the model inputs. Critics of deterministic models have found them simplistic, and have noted that appropriate model outcomes are dependent upon carefully selected inputs. Deterministic models may be used to characterize exposure or dose via inhalation, dermal, or ingestion routes. In the absence of more specific data, the US ATSDR recommends using exposure factors or model parameters that have been derived from population studies, such as those found in the U.S. Environmental Protection Agency (EPA)’s Exposure Factors Handbook (Agency for Toxic Substances and Disease Registry 2005; US EPA 1997). The Exposure Factors Handbook presents summary data, including ranges, for common exposure assessment parameters from population studies that can be used in exposure models and other semiquantitative approaches. Some of the data presented in this handbook that may be particularly helpful for exposure reconstruction studies include breathing rates, body weights, skin surface areas, and time typically spent doing common activities. The most basic deterministic model to characterize inhalation exposure is the well-mixed box model or the single-zone mixing factor model. While relatively easy to apply, this model can underestimate exposures close to a contaminant emission source and will typically use qualitative mixing factors that can defy the law of conservation of mass (Keil 2000). However, they can be valuable for screening purposes or for understanding exposures to individuals working away from a contaminant emission source when well-mixed room conditions are present. A number of studies have used single-zone models to reconstruct exposures in occupational exposure scenarios (see Table 30.4). As an example, in order to predict emission rates for formaldehyde from cadavers in gross anatomy laboratories, Keil et al. (2001) used concentration measurements and ventilation rates in a mass balance model (Keil et al. 2001). Their results showed that the total emission rate along with the number of dissecting tables can be used to develop an emission factor and appropriate control strategies for laboratories of various sizes (Keil et al. 2001). The results of estimates made using a steady-state mixing factor model have been validated with direct measurements and have been found to be good surrogates for measurements when the model inputs are carefully selected (Cherrie 1999; Fehrenbacher and Hummel 1996; Gaffney et al. 2008; Matthiessen 1986; Williams et al. 2008). In another study, Gaffney et al. (2008) compared personal and area
752 TABLE 30.4.
Selected Single-Zone Model Applications Reported in the Peer Reviewed Literature
Study Author (Country)
Year
Scenario
Gaffney (USA)
2008
EPA (USA)
2005
Groah (USA)
2006
Suh (USA)
2004
Keil (USA) Riley (USA)
2001 2001
EPA (USA)
1986
Cleaning semiconductor wafers; simulation study results compared to steady-state, single compartment model Assess the fate and transport of mercury vapors associated with cultural uses of elemental mercury Projection of indoor-air formaldehyde concentrations from up to six sources Determination of concentrations of and exposures to pollutants in Los Angeles homes Evaporation of formaldehyde from cadavers Estimation of mercury exposure from use in cultural and religious practices General one-zone exposure model used to model formaldehyde exposures in residential settings
References: Gaffney et al. (2008), Groah (2006), Keil et al. (2001), Riley et al. (2001), Suh (2004), US EPA (1986, 2005).
Chemical(s) Methanol
Mercury
Type of Estimate Generated Short-term (15-min) and long-term (2 to 4-hr) personal and area samples (ppm) Airborne concentrations (μg/m3)
Formaldehyde
Airborne concentrations (ppm)
Black carbon, nitrate Formaldehyde Mercury
Exposure estimate ranges (μg/m3) TWA (mg/m3) TWA (μg/m3)
Formaldehyde
Airborne concentrations (ppm)
30.2. EXPOSURE RECONSTRUCTION METHODOLOGY
753
exposure values determined through a simulation study of methanol exposure in the semiconductor industry to a steady-state, single-compartment model with an appropriate mixing factor, and concluded that the model-predicted airborne concentrations of methanol were reasonably consistent with measured values (Gaffney et al. 2008). A more robust deterministic model for estimating occupational exposures is the transient two-zone model described by Nicas in Keil (2000). This model provides more accurate exposure estimates for situations where exposures may be highest in the area of a contaminant emission source (Keil 2000). The two-zone modeling approach has been used in many published studies to estimate exposure, including assessments of microbial exposures, solvent exposures during many different scenarios, occupational exposure to dry cleaners, and exposures during chemical spills (Armstrong and Haas 2007; Keil and Nicas 2003; Nicas et al. 2006; Spencer and Plisko 2007; von Grote et al. 2006). In one notable example, Nicas et al. (2006) applied a near-field/far-field model with an exponentially decreasing emission rate to estimate benzene exposure at a parts washer using a petroleum distillates solvent containing benzene (Nicas et al. 2006). The results showed that the modeled benzene concentrations were within a multiplicative range of one-half to twofold the measured concentrations determined through a simulation study (Nicas et al. 2006). Table 30.5 below provides a summary of key published studies that have used the transient two-zone model to estimate occupational exposures. Stochastic (Probabilistic) Models. One of the most significant advances in exposure estimation in the past 15 to 20 years has been the application of probabilistic statistical methods to many types of data analyses (Duan and Mage 1997; Finley and Paustenbach 1994; Morgan and Henrion 1990; US EPA 1995, 1997, 2000a). Stochastic or probabilistic techniques can help quantify variability and uncertainty in model inputs and outputs, can be used to better characterize the possible range of exposures for a particular scenario when measured data are minimal, and can be employed to better understand the uncertainty inherent in estimates developed from many different types of sources, whether quantitative or qualitative. Monte Carlo analysis is a specific probabilistic assessment method that can be used to characterize health risks and their likelihood of occurrence based on a wide range of parameters (Shade and Jayjock 1997). The U.S. EPA’s Stochastic Human Exposure and Dose Simulation (SHEDS) model allows for the quantification of exposures based on a probabilistic assessment of multiple exposure pathways and multiple routes of exposure (Mokhtari et al. 2006; US EPA 2003b). Additional applications of probabilistic techniques will be discussed in the section below on conducting an uncertainty analysis of reconstructed exposure values. 30.2.4.2.4. Semiquantitative Techniques for Reconstructing Dermal or Ingestion Exposures. Dermal exposures can be estimated using available models such as the SkinPerm model developed by ten Berge, which estimates the permeation rate and mass uptake through the skin for a number of chemical substances (ten Berge 2008; Wilschut et al. 1995). The CONSEXPO model also estimates dermal
TABLE 30.5.
Selected Transient Two-Zone Model Applications Reported in the Peer Reviewed Literature
754
Study Author (Country)
Year
Scenario
Chemical(s)
Gaffney (USA)
2008
Cleaning semiconductor wafers; simulation study results compared to transient two-zone model
Methanol
Armstrong (USA)
2007
Legionella
Eickmann (Germany)
2007
Quantitative microbial risk assessment (QMRA) model for Legionnaires’ disease at whirlpool spa Generic two-zone model description
Spencer (USA) Keil (USA)
2007 2006
Nicas (USA) Vernez (Switzerland)
2006 2006
Solvent exposure during metal part disassembly Chemical exposure at a university teaching laboratory Solvent parts washer usage Application of waterproofing sprays
Von Grote (Switzerland) Vernez (Switzerland)
2006 2004
Occupational dry cleaning exposure Waterproofing of a tiled surface
Keil (USA) Nicas (USA)
2003 2003
Nicas (USA) Von Grote (Switzerland)
2003 2003
Organic solvent spill Fumigation of commodities—manual processing (e.g., foods) Splash loading of gasoline Occupational metal degreasing
Nicas [in Keil (2000), USA] Keil (USA)
2000 1998
Potential for multiple applications including spray emissions Solvent (cyclohexane) Ethyl ether, n-hexane, methylene chloride Benzene Respirable aerosol particles containing solvent Perchloroethylene (PERC) Solvent vapors, aerosols n-Pentane Methyl bromide
Type of Estimate Generated Short-term (15-min) and long-term (2 to 4-hr) personal and area samples (ppm) Exposure estimate range (CFU/m3) Particle concentration in air (mg/m3) Point estimate (mg/m3) mg/m3 1-hr TWA and 4-hr TWA (ppm) Exposure estimate range (mg/m3) g/m3 Exposures to mist (mg/m3) and solvent vapors (ppm) 5-min TWA (mg/m3) TWA (ppm) 15-min TWA (ppm) g/m3
Application of adhesive
Benzene Trichloroethylene (TCE), perchloroethylene (PCE) Toluene
Parts washing operation
Toluene
Exposure estimate range (mg/m3)
mg/m3
References: Armstrong and Haas (2007), Eickmann et al. (2007), Gaffney et al. (2008), Keil and Murphy (2006), Keil (1998), Keil and Nicas (2003), Nicas (2000, 2003a,b), Nicas et al. (2006), Spencer and Plisko (2007), Vernez et al. (2004, 2006), von Grote et al. (2003, 2006).
30.2. EXPOSURE RECONSTRUCTION METHODOLOGY
755
uptake (van Veen 2001). A method for developing semiquantitative exposure values for dermal exposure assessment based on deposition and transfer to the skin has also been presented (van Wendel de Joode et al. 2005). Other references and critical reviews provide detailed information and guidance on developing reconstructed dermal exposure values or estimates (Paustenbach and Madl 2008; Paustenbach 2000; Sahmel and Boeniger 2006). Ingestion exposures can be estimated using the Dietary Exposure Evaluation Model (DEEM), a computer-based model developed by Novigen Sciences, Inc. to estimate dietary intake of chemical residues (California Environmental Protection Agency 2007). The outputs from this model include estimates of dietary exposure over multiple averaging times so that both acute and chronic exposures can be considered. Exposures can also be considered on a population or individual level. The U.S. EPA’s Office of Pesticide Programs has used DEEM in its exposure and risk assessments. The Dietary Exposure Potential Model (DEPM) is another ingestion model that facilitates the correlation of food consumption data and chemical residue data (Cherrie et al. 2006; Tomerlin et al. 1997; US EPA 2003a). 30.2.4.3. Qualitative Estimation Methods. Qualitative techniques have also been used to conduct or refine exposure reconstructions. The use of exposure determinants to classify exposures is a technique that has been performed qualitatively in some cases based on available data. Other qualitative techniques can include the use of expert or professional judgment and self-reported exposures. While selfreported exposures have been found to be convenient and helpful when no other exposure information is available, the method has been shown in at least one validation study to be inadequate for use as the sole data source in an exposure reconstruction. When self-reported exposures have been compared to professional judgments by experts, they have been found to have high specificity but low sensitivity (Fritschi et al. 1996). 30.2.4.3.1. Use of Exposure Determinants for Reconstructing Exposures. An exposure determinant is a factor influencing the magnitude of exposure and can be either qualitative or semiquantitative in nature. Examples include characteristics of work practice, type of equipment used, ventilation characteristics, and personal hygiene practices. Many studies have used exposure determinants to help reconstruct exposures. For example, in their exposure assessment of benzene and toluene in two shoe factories in China, Vermeulen et al. (2004) used degree of contact, chemical weight fraction, and air movement or ventilation patterns to help determine benzene and toluene exposures (Vermeulen et al. 2004). Laden et al. (2006) used processspecific weighting factors such as fuel type and emission rates to reduce exposure misclassification in their study of historical diesel exhaust exposure to railroad workers (Laden et al. 2006). Other approaches that have combined exposure determinants include using task and time-dependent weighting factors (Henn et al. 2007); the effect of level of risk acceptance and self-reported activity factors on exposure classification (Nieuwenhuijsen et al. 2005); the effect of particle size characteristics on exposure classification (Johnson and Esmen 2004); using technology
756
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
factors, PPE, field reentry, pesticide storage techniques, and personal hygiene factors to reconstruct pesticide dermal exposure intensities (Monge et al. 2005); using census data and cancer registry data to help classify exposures (Johansen et al. 2005); and the impact of the specificity of the exposure metric on exposure–response relationships (Friesen et al. 2007). 30.2.4.3.2. Use of Expert/Professional Judgment for Reconstructing Exposures. Expert judgment is another common method of estimating unknown historical exposures. In this method, a group of experts, typically consisting of industrial hygienists, chemical engineers, occupational physicians, chemists, or a variety of other professionals, use their training, experience, knowledge, and supplemental information to assess historical exposures to chemicals. This supplemental information can be gathered in the form of questionnaires, interviews, and/or historical records. The more detailed the information gathered regarding the exposures, the better the outcome (Mannetje et al. 2003). Many studies have used panels of experts to perform dose reconstructions (see Table 30.6). For example, Seel et al. (2007) used expert judgment to characterize exposures to asbestos, welding fumes, chromium, and nickel at the Portsmouth Naval Shipyard. Specifically, three experienced industrial hygienists from the shipyard independently assessed exposures for over 3500 shop–job–time-period combinations; these estimates were linked with employment histories to calculate cumulative exposures (Seel et al. 2007). While a few studies, such as Seel et al. (2007), have included only industrial hygienists on their panel (Kudla 1997; Nelson et al. 1995; Seel et al. 2007; Stewart et al. 2000), several others have incorporated a variety of professionals to determine exposure outcomes (Benke et al. 1997; Clavel et al. 1993; Dosemeci et al. 1997; Fritschi et al. 2003; Mannetje et al. 2003; Siemiatycki et al. 1997). In their study involving historical benzene exposures of workers in China, Dosemeci et al. (1997) elicited the opinions of an exposure assessment team consisting of industrial hygienists, safety officers, supervisors, and longterm employees (Dosemeci et al. 1997). This team used available exposure data to develop a summary estimate of benzene exposure for each factory–work-unit–jobtitle–calendar-year combination (Dosemeci et al. 1997). Bayesian decision analysis can also use professional judgments to reconstruct exposures by statistically combining these judgments with measured or modeled exposure values (Ramachandran 2001; Ramachandran and Vincent 1999). A detailed discussion of Bayesian decision analysis is provided below. Although the resulting information is subject to professional judgment, the use of experts in the exposure reconstruction process can provide useful estimations of past exposures if a few guidelines are followed. First, the maximum amount of supplemental information should be gathered and used during the expert’s decisionmaking process (Nelson et al. 1995). Second, experts should be screened for their suitability in particular situations, with the goal of using experts with the most appropriate range of knowledge or greatest amount of relevant experience (Benke et al. 1997). Third, teams of experts should be used instead of individual experts when possible (Siemiatycki et al. 1997). Calibration studies of industrial hygiene professional judgments have also been conducted. Industrial hygienists have
TABLE 30.6.
Selected Examples of Studies Using Expert Judgment in the Peer Reviewed Literature
Study Author (Country)
Year
Scenario
Seel (USA)
2007
Fritschi (Australia)
2003
Mannetje (France)
2003
Ramachandran (USA)
2001
Stewart (USA)
2000
Ramachandran (USA)
1999
Cherrie (United Kingdom)
1999
Assessment of historical exposures at the Portsmouth Naval Shipyard for a study of lung cancer risk from external ionizing radiation Comparison of the ratings of experienced raters with previously recorded industrial hygiene measurements for occupations in Australia Estimation of the levels of exposure misclassification by expert assessment in a study of lung cancer in central and eastern Europe and Liverpool Application of Bayesian framework for retrospective exposure assessment of workers in a nickel smelter Determination of the level of information required by industrial hygienists to develop reliable exposure estimates Explanation of new framework to obtain exposure estimates through the Bayesian approach Validation of a new method for structured subjective assessment of past concentration
Chemical(s)
Supplemental Information Available
Type of Estimate Generated
Asbestos, chromium, nickel, welding fumes
Employment histories and exposure data
fibers/cm3; mg/m3; μg/m3
Several chemicals
Exposure data and interview information
70 putative lung carcinogens
Occupational questionnaire
Qualitative frequency (% work week) and exposure level (high, medium, low) ratings Low, medium, high
Nickel
Exposure data
Inhalable dust and nickel (mg/m3)
Formaldehyde
Job title, department title, industry, dates, plant process report Exposure data, deterministic models
Exposure categories (ppm ranges)
Nickel
Man-made mineral fibers, asbestos, styrene, toluene, and mixed respirable dust
Descriptive information about work activities and the work environment
Inhalable dust and nickel (mg/m3) Fibers/ml, ppm, mg/m3
757
(Continued)
758
TABLE 30.6. (Continued)
Study Author (Country)
Year
Scenario
Benke (Great Britain)
1997
Dosemeci (USA)
1997
Siemiatycki (Canada)
1997
Assessment of the validity, reliability, and feasibility of using an occupational hygiene panel to estimate exposures in a brain tumor case-control study Validation study for a retrospective assessment procedure to evaluate historical benzene exposures of workers in China Application of expert judgment in a study of cancer and occupation in Montreal between 1979 and 1986
Kudla (USA)
1997
Nelson (USA)
1995
Clavel (Great Britain)
1993
Analysis of historical exposures to two workers with bone marrow disorders in the printing trade Assessment of historical exposure for a study of renal cell disease and total hydrocarbon exposure
Description of rules that experts can use to identify and quantify exposures
Chemical(s)
Supplemental Information Available
Type of Estimate Generated
21 different chemicals
Exposure data available for comparison
Sensitivity and specificity relative to actual estimates
Benzene
Exposure data available for comparison
Concentration ranges (ppm)
294 workplace chemicals
Interviews and job histories
Benzene, toluene
Detailed worker interviews and workplace descriptions Interview data used to create standard industrial classification (SIC) and standard occupational classification (SOC) codes Information for questionnaires
Qualitative frequency (% work week) and exposure level (high, medium, low) ratings Qualitative estimates generated
Hydrocarbons
Organic solvents
ppm
Low, medium, high
References: Benke et al. (1997), Cherrie and Schneider (1999), Clavel et al. (1993), Dosemeci et al. (1997), Fritschi et al. (2003), Kudla (1997), Mannetje et al. (2003), Nelson et al. (1995), Ramachandran (2001), Ramachandran and Vincent (1999), Seel et al. (2007), Siemiatycki et al. (1997), Stewart et al. (2000).
30.2. EXPOSURE RECONSTRUCTION METHODOLOGY
759
demonstrated an ability to estimate the mean for a specific exposure scenario with a substantial degree of accuracy, indicating that these values can be potentially used in place of modeled estimates or other types of exposure estimates based on minimal data (Hawkins and Evans 1989; Walker et al. 2001). Industrial hygienists have also been shown to be effective at estimating 90th percentile exposure values, in some cases even more so than for the mean (Walker et al. 2003), although they were less effective at estimating median exposure values (Hawkins and Evans 1989). Industrial hygienists were also found to be underconfident overall in their ability to qualitatively judge exposures, although they were overconfident in their judgments of the mean of an exposure group (Walker et al. 2003). 30.2.4.4.
Combined Methodologies for Exposure Reconstruction
30.2.4.4.1. Exposure Reconstruction Using a Combination of Best Available Data, Estimates, and Information. A common technique in exposure reconstruction is to use measured or simulated data wherever possible in an assessment and then to fill in the remaining gaps for particular routes of exposure or exposure pathways using semiquantitative estimates or qualitative judgments as appropriate. Stewart and Stenzel (2000) point to the substantial overlap between exposure assessment methods, and indicate that more than one method may be necessary (Stewart and Stenzel 2000). Certain types of measured data may be available for an entire period of historical interest, whereas other critical information may be available only indirectly by semiquantitative calculation or qualitative estimation. For example, the U.S. ATSDR describes a historical exposure reconstruction for a groundwater study in which groundwater well location and production were well-characterized with direct data, but other information, such as total consumption, was estimated using semiquantitative techniques, and the spatial distribution of consumption was estimated using qualitative information or judgments (Maslia et al. 2001). In such cases, a transparent description of the source of each type of data is important. Government guidance documents for dose reconstruction of ionizing radiation exposures often recommend a similar approach; the U.S. National Institute for Occupational Safety and Health (NIOSH) provides a hierarchy of data sources for dose reconstruction of ionizing radiation exposures. Where gaps in data exist, specific methods for interpolation or extrapolation of data are given, including: (1) averages of existing personal data to cover gaps, (2) co-worker data, (3) data from nearby sites, (4) radiation survey data, or (5) source term information (NIOSH 2007). NIOSH also recommends that when personal exposure information is incomplete, site profiles developed by the U.S. EPA be used. These documents describe a specific work site, including the physical appearance and layout of the work site, the work processes used, the types of materials used, potential sources of radiation, and other relevant details. More detailed documents, known as technical basis documents, make up a site profile, and typically contain information on work processes used and descriptions of work performed. These are described on NIOSH’s website, and methods for using the information in dose reconstruction are included in technical documentation for NIOSH’s Interactive RadioEpidemiological Program (NIOSH-
760
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
IREP) (National Institute for Occupational Safety and Health 2002; US CDCP 2008b). A similar hierarchy can be applied to the assessment of chemical exposures. Individual studies have used such combination techniques as the consideration of both comprehensive process analysis and modeling estimates (Esmen et al. 2007a); the consideration of measured data and modeling estimates together (Esmen et al. 2007b); the combined use of JEMs and physiologically based pharmacokinetic (PBPK) modeling (Aylward et al. 2005); the use of historical exposure models to develop empirical Bayes estimates (Friesen et al. 2006); the use of modeling to refine and increase the utility of general dust samples (Friesen et al. 2005); the use of a combination of job and task information, historical exposure factors, and mathematical models to develop exposure estimates (Macaluso et al. 2004); and the combined use of biomarker measurements with pharmacokinetic modeling (Pleil et al. 2007). Table 30.7 provides a summary of representative studies that used a combination of techniques to estimate historical exposures.
30.2.4.4.2. Bayesian Decision Analysis for Exposure Reconstruction. A potentially valuable technique that can be used for retrospective exposure assessment and reconstruction is Bayesian decision analysis (BDA). BDA is based on inductive reasoning, and it statistically combines prior information about an exposure scenario with quantitative or semiquantitative data. Historically, retrospective exposure assessments have relied on qualitative exposure determinants such as duration of exposure and other descriptors or exposure factors that do not allow for quantitative consideration of available information. In retrospective exposure assessments using BDA, prior information, such as expert judgments about exposure or historical determinants of exposure, can be quantitatively expressed and statistically combined with existing data or generated estimates. Together, these two pieces of information create a refined exposure value (Ramachandran 2001; Ramachandran and Vincent 1999). As one example, this method can be applied in retrospective exposure assessment by having industrial hygiene or other exposure experts create a “prior” probability distribution of exposures for particular tasks or scenarios. The experts might provide a numeric “prior” best estimate of the likely exposure concentration for a specific job or exposure classification based on their knowledge of the processes, worker practices, and exposure controls. Experts should typically also provide a numeric level of confidence in their estimates for each of their judgments. These estimates can then be analyzed in combination with limited measured data, modeled estimates, Monte Carlo estimates, or other exposure information to obtain a final “posterior” distribution function that statistically considers all available information (Ramachandran 2001; Ramachandran and Vincent 1999). BDA techniques have also been used to consider within-worker and between-worker variability in retrospective exposure assessments (Friesen et al. 2006), to analyze datasets for exposure misclassification (Chu et al. 2006; Gustafson and Greenland 2006), and to confirm exposure classifications based on measured data (Hewett et al. 2006).
TABLE 30.7.
Study Author (Country)
Selected Examples of Studies Using Combination Techniques in the Peer Reviewed Literature
Year
Scenario
Chemical(s)
Techniques Used
761
Dahmann (Germany) Esmen (USA)
2008
Uranium mining industry
Dust, crystalline silica, arsenic
Simulation study and JEM
2007a
Mortality patterns in industrial workers
Chloroprene
Esmen (USA)
2007b
Chloroprene, vinyl chloride
Pleil (USA)
2007
Friesen (Canada)
2006
Yokley (USA)
2006
Aylward (USA)
2005
Four-facility occupational epidemiology study of chloroprene monomer and polymer production workers Demonstration of the value of spot measurements of breath or blood for reconstructing an individual’s recent average exposure to MTBE Assessment of three models for predicting historical exposures of sawmill workers Description of a PBPK model that quantifies tissue doses of benzene and its key metabolites US chemical manufacturing workers (NIOSH cohort)
Process analysis and modeling based exposure reconstruction were used to augment, extrapolate, or interpolate available exposure data Comprehensive process analysis and modeling estimates
Friesen (Canada)
2005
Prediction of historical dust and wood dust exposure in sawmills
Type of Estimate Generated Dust measurements (mg/m3) ppm
Semiquantitative exposure classes
Methyl tertiary butyl ether (MTBE)
Biological monitoring and pharmacokinetic model
μg/L and ng/L
Dust
Mixed models and empirical Bayes estimates of exposure
Dust concentrations (mg/m3)
Benzene and its key metabolites, benzene oxide, phenol, and hydroquinone TCDD
PBPK modeling and Bayesian analysis
Benzene concentration (μM) ng/kg
Wood dust
Actual exposure data and modeling estimates to estimate serum lipid AUC / JEM used to compute exposure scores Available exposure data and modeling estimates
Dust levels (mg/m3)
(Continued)
762
TABLE 30.7. (Continued)
Study Author (Country)
Year
Scenario
Chemical(s)
Coble (USA)
2005
Herbicides (MCPA and 2,4-D)
Abadie (France) Macaluso (USA) Chen (Taiwan)
2004
Evaluation of an algorithm designed to calculate pesticide exposure using biological monitoring results Evaluation of particle pollution in French high-speed train smoker cars Worker exposure at synthetic rubber plants Statistical assessment of the relationship between biomarker concentrations and environmental exposures Reconstruction of exposures to trichloroethylene (TCE) through use of biomonitoring data and Bayesian interface Examination of a reliable estimate of the metabolic capacity for styrene in humans Investigation of potential human exposure to groundwater contaminants through development of a water-distribution model specific to township of interest Application of Bayesian framework for retrospective exposure assessment of workers in a nickel smelter
2004 2004
Sohn (USA)
2004
Jonsson (Sweden)
2002
Maslia (USA)
2001
Ramachandran (USA)
2001
Cigarette smoke particles 1,3-Butadiene, styrene, dimethyldithiocarbamate Benzene, methyl chloroform
Techniques Used Exposure algorithm based on a questionnaire and biological monitoring Actual exposure data and one-zone model Exposure matrix and modeling estimates Modeling estimates and biological monitoring
Type of Estimate Generated Urinary concentrations (μg/liter) Particle concentrations TWA (ppm) ppm
Trichloroethylene
Biomonitoring data and Bayesian analysis
g/L
Styrene
Parmacokinetic model and Bayesian analysis
Metabolic capacity (μmol/L)kg−1
Groundwater contaminants
Actual data, models of historical water-distribution system operations, and qualitative judgments
Concentration in water
Nickel
Professional judgment, Monte Carlo analysis, exposure data, and Bayesian analysis
Inhalable dust and nickel (mg/m3)
Study Author (Country)
Year
Scenario
Chemical(s)
Fenske (USA)
2000
2000
Two OP pesticides used most frequently in area (azinphosmethyl and phosmet) Isocyanates
Biological monitoring and deterministic steady-state model
Meredith (United Kingdom) Ramachandran (USA)
Nickel
1997
Professional judgment, deterministic models, exposure data, and Bayesian analysis Measured data and modeling estimates
Inhalable dust and nickel (mg/m3)
Duan (USA)
Ripple (USA)
1996
Plutonium, uranium
Airborne effluent monitoring data and modeling techniques
nBq m−3 and kBq km−2
Hallock (USA)
1994
Machining fluids
1992
Historical data, work history information, interviews, and modeling techniques Previous exposure data and new plant information from interviews and historical records
mg/m3
Paustenbach (USA)
Eisen (USA)
1984 1983
Previous exposure data and modeling techniques Historical data combined with information on processes, dust control methods, and job assignments were used to create linear statistical models
μg/m3
Dement (USA)
Estimation of organophosphorus (OP) pesticide exposure to children in an agricultural community Examination of the quantitative relation between exposure to isocyanates and occupational asthma Explanation of new framework to obtain exposure estimates through the Bayesian approach Combined direct (personal air samples) and indirect (activity pattern model) approaches used in human air pollution exposure assessment Reconstruction of contaminant doses to the public from operations at the Rocky Flats nuclear weapons facility Estimation of historical exposures to machining fluids in the automotive industry Evaluation of prior exposure estimates for the pliofilm cohort and creation of new exposure estimates based on new available information for this cohort Estimation of long-term dust exposures in the Vermont granite sheds Estimation of exposures to asbestos in an asbestos textile plant processing chrysotile
1999
Carbon monoxide
Benzene
Granite dust Chrysotile asbestos
Techniques Used
Actual data, job title information, and dates of employment
Type of Estimate Generated Urinary metabolite concentrations (μg/kg/day) 8-hr TWA and peak (ppb)
ppm
ppm
fibers/cm3
763
References: Abadie et al. (2004), Aylward et al. (2005), Chen et al. (2004), Coble et al. (2005), Dahmann et al. (2008), Dement et al. (1983), Duan and Mage (1997), Eisen et al. (1984), Esmen et al. (2007a,b), Fenske et al. (2000), Friesen et al. (2005, 2006), Hallock et al. (1994), Jonsson and Johanson (2002), Macaluso et al. (2004), Maslia et al. (2001), Meredith et al. (2000), Paustenbach et al. (1992), Pleil et al. (2007), Ramachandran (2001), Ramachandran and Vincent (1999), Ripple et al. (1996), Sohn et al. (2004), Yokley et al. (2006).
764
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
30.2.5. Conducting an Uncertainty Analysis of the Reconstructed Exposure Values As introduced above, probabilistic statistical techniques allow the user to account for the uncertainty in select parameters by considering the plausible range and probability of exposure levels. In addition to improving the quality of modeling outputs or data extrapolations, probabilistic techniques can be used to combine available data to generate a more robust health risk assessment. Monte Carlo, one method of probabilistic analysis, has existed as an engineering analytical tool for many years, but the development of specialized computer software (e.g., Crystal Ball [Oracle, Redwood Shores, CA], @RISK [Palisades Corp., Newfield, NY]) has allowed this method to be more easily applied to new areas of analysis, such as risk assessment. Probabilistic techniques can be used to characterize uncertainty or variability in the results of an exposure or dose reconstruction, whether the assessment is conducted using measured exposure data, model outputs, or a combination of data and estimates, or to validate the results obtained by any single method (Burmaster and Anderson 1994; Burmaster and von Stackelberg 1991). The Monte Carlo statistical simulation is a method in which the individual input parameters or available data are varied simultaneously and randomly during many successive model runs. The values are chosen from the parameter distributions, with the frequency of a particular value being equal to the relative frequency of the parameter in the distribution. This technique generates distributions that describe the uncertainty associated with the exposure, dose, or risk estimate. The predicted estimate for the 50th percentile and the 95th percentile of the exposed population along with the true mean are calculated. When an uncertainty analysis is conducted, the assessor is not forced to rely solely on a single exposure parameter or the repeated use of conservative assumptions to identify the plausible exposure estimates. Instead, the full range of possible values and their likelihood of occurrence is incorporated into the analysis to produce the range and probability of expected exposure levels, decreasing the potential for exposure misclassification (Anderson and Yuhas 1996). Studies using Monte Carlo analysis techniques as part of an exposure reconstruction have used a variety of methods, including recreating datasets from estimated geometric means and geometric standard deviations and reevaluating previous exposure estimates using available data (Lavoue et al. 2007; Williams and Paustenbach 2003). Studies that have used uncertainty analysis as part of an exposure reconstruction are listed in Table 30.8. Like traditional exposure analysis, one challenge to properly performing an uncertainty analysis is having appropriate distributions for use in the analysis. Numerous studies on individual variables, such as exposure frequencies, durations, and personal habits, have been published in the risk assessment literature (Beck and Cohen 1997; Burmaster 1998a,b; Copeland et al. 1993, 1994; Finley and Paustenbach 1994; Gargas et al. 1999; Sedman et al. 1998; Smith et al. 1992; Stanek and Calabrese 1995; US EPA 1997), and the impacts of the distributions employed on the outcome (sensitivity analyses) have also been discussed (Bukowski et al. 1995; Cooper et al. 1996; Haas 1997; Hamed and Bedient 1997; Hattis and Burmaster
TABLE 30.8.
Selected Examples of Exposure Reconstruction Studies Using Monte Carlo Analysis in the Peer Reviewed Literature
Study Author (Country)
Year
Lavoue (Canada)
2007
Numerical synthesis of exposure data in reconstituted wood panels industry
Formaldehyde
Zartarian (USA)
2006
Arsenic
Williams (USA)
2003
Estimation of exposure to children while playing on chromated copper arsenate (CCA)-treated playsets and decks using the SHEDS model Reconstruction of exposure for Pliofilm cohort
Nestorov (United Kingdom) Stanek (USA) Smith (USA)
1999 1995 1994
Copeland (USA)
1994
Copeland (USA)
1993
Scenario
Sensitivity analysis of pharmacokinetic and pharmacodynamic systems Re-estimation of soil ingestion rates for children Exposure to volatile solvents by drinking water ingestion and showering near a superfund site Use of probabilistic methods to understand the conservatism in California’s approach to assessing health risks posed by air contaminants Assessment of exposures at a wood treatment site by comparing Monte Carlo analysis and EPA’s Reasonable Maximum Exposed Individual (RMEI) approach
Chemical(s)
Benzene
Cyclosporine
Type of Estimate Generated Geometric mean and standard deviation (mg/m3) Absorbed dose of arsenic (mg/kg/day) Range of percentiles for estimated 8-hr TWA (ppm) ml/min
Soil Volatile solvents (1,1-dichloroethane, 1,1-dichloroethene, tetrachloroethene, 1,1,1-trichloroethane, trichloroethene) PCDDs, PCDFs
mg/day mg/kg/day
mg/kg/day
Soil samples of PCDDs and PCDFs
mg/kg/day
References: Copeland et al. (1993, 1994), Lavoue et al. (2007), Nestorov (1999), Smith (1994), Stanek and Calabrese (1995), Williams and Paustenbach (2003), Zartarian et al. (2006).
765
766
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
1994; Hoffman and Hammonds 1994). In addition, standard data distributions have been proposed for a variety of exposure variables, such as age-specific distributions for soil ingestion rates, inhalation rates, body weights, skin surface area, tap water and fish consumption, residential occupancy and occupational tenure, and soilon-skin adherence (Finley et al. 1994). It should also be pointed out that these techniques can be combined with other advanced risk assessment methods (i.e., PBPK modeling) to further reduce uncertainty in exposure estimates (Cronin et al. 1995; Simon 1997; Nestorov 1999, 2003). Recently, two-dimensional Monte Carlo analyses as well as other types of probabilistic approaches have been developed that take into account both variability and uncertainty. These have been utilized as methods for quantifying uncertainty in sensitivity analyses (Cullen and Frey 1999; Vose 1996; Frey and Patil 2002; Greenland 2001; Helton and Davis 2002; Saltelli 2002). Information appropriate to probabilistic analyses can often be found in published papers in fields quite distant from the environmental sciences. Probabilistic analyses have, in recent years, been recognized in regulatory guidance (US EPA 1992a, 1997, 2000b), and the U.S. EPA’s Risk Assessment Forum has published a document of principles for conducting Monte Carlo analyses (Cox 1996). The U.S. EPA and a number of states (and other countries) have published guidance documents on conducting Monte Carlo assessments (US EPA 1999b, 1997, 2000b).
30.3. APPLICATION OF ESTIMATED HISTORICAL EXPOSURE VALUES TO CANCER RISK ESTIMATES Each of the above exposure reconstruction methods provides exposure estimates for an individual to a substance. In most occupational settings, results will be in the form of an air concentration, or an amount of a certain chemical in a given volume of air. However, exposure estimates may also be in the form of the amount of chemical contacting the skin or potentially ingested, or the amount of chemical in a given body fluid or tissue. Depending on the exposure reconstruction method used, results may be quantitative or qualitative. For example, when estimating historical exposures based on current data, either simplistically or using extrapolation techniques, actual exposure estimates will result. In the case of using expert judgment or a JEM, however, the results may be qualitative (e.g., high, low) unless the exposure rankings are accompanied by ranges of concentrations or confirmed with actual measurements, in which case the results will be semiquantitative. If cancer risk estimation is a final goal of the exposure reconstruction, at a minimum, semiquantitative estimates of exposure are necessary because a dose estimate or range must be calculated. The U.S. ATSDR makes distinctions regarding different types of dose. It is to use the correct type of dose metric for the scenario of interest. The following terms have been defined by the U.S. ATSDR for use in considering chemical (nonradiological) exposures or doses (Agency for Toxic Substances and Disease Registry 2005):
30.3. APPLICATION OF ESTIMATED HISTORICAL EXPOSURE VALUES TO CANCER RISK ESTIMATES
767
Exposure or Administered Dose. The mathematical estimation of the amount of a substance encountered in the environment per unit of body weight and time. Absorbed or Internal Dose. The amount of the exposure dose that actually enters the body (i.e., penetrates barriers such as the skin, gastrointestinal tract, lung tissue). The route of exposure, type, and form of a substance, among other factors, influence how much of a substance is absorbed into the bloodstream. Levels of internal dose may be measured in some body compartments through biological sampling (e.g., medical testing for biological markers of exposure in blood or urine). Target Tissue Dose. The amount of the absorbed dose reaching the cells or target sites where an adverse effect occurs. Biologically Effective Dose. The amount of the target tissue dose needed to produce a biologic response.
30.3.1.
Estimating Dose
For risk assessment purposes, dose estimates should be expressed in a manner that can be compared with available dose–response data from animal or human studies. For example, if data on human exposure are measured in micrograms of lead per deciliter of blood (μg/dL), it would be best to use the blood concentrations in an animal study to predict the risk to humans. Frequently, dose–response relationships are based on potential dose (called administered dose in animal studies), although dose–response relationships are sometimes based on internal dose; these differences need to be taken into account. The measure of dose selected should be based on the mode of action of the adverse effect (Andersen et al. 1987, 1995; Andersen and Conolly 1998; Aylward et al. 1996; Hallock et al. 1994; US EPA 1992b, 1996a). For example, to assess a nasal irritant, the airborne concentration of the chemical is a relevant dose, and an even better dose metric would be milligrams of chemical contacting a square centimeter of nasal mucosa. In this case, dose may be estimated as a dose rate over the time period of interest. The dose per unit time is the dose rate, which has units of mass/time. The most common dose measure is average daily dose (ADD), which is used to predict or assess the noncarcinogenic effects of a chemical: ADD =
C × IR × B BW × AT
where ADD is the potential average daily dose, BW is the body weight (kg), B is the bioavailability (unitless), AT is the time period over which the dose is averaged (days), C is the mean exposure concentration (e.g., mg/m3, mg/L or mg/g), and IR is the inhalation, ingestion, or skin absorption rate (e.g., m3/day, L/day or g/day). It is important to note that this equation is very general and may be modified to assess exposure via specific pathways (Paustenbach and Madl 2008; US EPA 2009). To assess a carcinogen, doses are usually averaged over a lifetime and are presented as lifetime average daily doses (LADDs), even though exposure does not
768
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
occur over the entire lifetime (US EPA 1992a). The biological response is usually described in terms of lifetime probabilities (e.g., the increased risk of developing cancer during a 70-year lifetime is 5 in 100,000). The potential LADD is calculated as follows, with all variables equivalent to those used to calculate ADD except for averaging time (AT), which is replaced with lifetime (LT) (days): LADD =
C × IR × B BW × LT
Although other measures of chronic dose may be more appropriate for predicting the hazard posed by specific chronic or carcinogenic toxicants, such as an areaunder-the-blood-concentration (AUC) curve or the peak target tissue concentration, the LADD is the most common dose metric used in carcinogen risk assessment (Paustenbach and Madl 2008).
30.3.2.
Estimating Risk
Once a dose metric is selected and estimated, a dose extrapolation model can be applied to estimate cancer risk. The choice of the model will be driven by the likely mechanism of action of the chemical or agent. For example, if the substance is a genotoxic material, such as radiation, a linear model would be used. A threshold model or nonlinear model might be used if the chemical or agent is not genotoxic (Paustenbach 2002; Williams and Paustenbach 2002). The general theory behind both models is discussed below. 30.3.2.1. Linear Regulatory Approach. Cancer risk estimates are most often based on the assumption that the dose–response relationship is linear at low doses (i.e., exposures corresponding to risks less than 1 in 100) (US EPA 1989). Under this simplistic assumption, risk is directly related to dose and can be estimated using the following equation: Risk = LADD × CSF where risk is the probability of an individual developing cancer as a result of the exposure of interest (unitless), LADD is the potential lifetime average daily dose (mg/kg-day), and CSF is the cancer slope factor (mg/kg-day)−1. Cancer slope factors are established by the U.S. EPA (or state health agencies) to reflect the relative toxicity of carcinogenic substances. The U.S. EPA values can be found on the U.S. EPA’s Integrated Risk Information System (IRIS) website. The U.S. EPA defines a slope factor as “an upper bound, approximating a 95% confidence limit, on the increased cancer risk from a lifetime exposure to an agent” (Integrated Risk Information System 2008). The slope factor therefore represents an “upper bound” or “plausible upper limit” value. The “true” cancer risk is not expected to exceed this value, and may, in fact, be substantially less or even zero (i.e., because the compound is not a human carcinogen or the dose–response relationship has an effec-
30.3. APPLICATION OF ESTIMATED HISTORICAL EXPOSURE VALUES TO CANCER RISK ESTIMATES
769
tive threshold above the level of exposure). According to the U.S. EPA (1989), use of the 95% upper confidence limit (UCL) allows regulators and decision-makers to be “reasonably confident that the ‘true risk’ will not exceed the risk estimate derived through use of this model and is likely to be less than that predicted” (US EPA 1989). The U.S. EPA reports acceptable cancer risks to fall in the range of one excess case in 10,000 people (1/10,000 or 10−4) to one excess case in 1,000,000 people (1/1,000,000 or 10−6) (US EPA 1989).
30.3.2.2. Nonlinear Approach. When there are sufficient data to support an assumption of nonlinearity for carcinogens, a margin of exposure (MOE) analysis is typically used to estimate cancer risk. MOEs are used either when a compound’s mode of action leads to a dose–response relationship that is nonlinear, or when the mode of action may theoretically have a threshold, although the U.S. EPA does not generally try to distinguish between these different scenarios (US EPA 1996b). The risk in this case is not extrapolated as a probability of an effect at low doses, but instead represents the toxicity point of departure (i.e., the beginning of the extrapolation), divided by the estimated dose: MOE =
POD LADD
where MOE is the margin of exposure (unitless), POD is the dose representing the point of departure (mg/kg-day) and LADD is the potential lifetime average daily dose (mg/kg-day). Common points of departure (POD) include the No Observed Adverse Effect Level (NOAEL) and the benchmark dose, or LED10 (i.e., lower 95% confidence limit on a dose associated with 10% extra risk). In general, the ideal point of departure is the dose in which the key events in tumor development would not occur in a heterogeneous human population, thus representing a “no effect level” (US EPA 1999a). The U.S. EPA therefore recommends that MOE analyses be based on “precursor responses” rather than on tumor incidences, since precursor events can often be detected with greater sensitivity (US EPA 1999a). An MOE analysis provides useful information on the distance between the exposure of interest and the range of observation in which cancer risk is inferred to be sublinear (US EPA 1999a). It is the role of the risk manager to decide whether a given margin of exposure is acceptable under applicable management policy criteria, although the risk characterization will provide supporting information to assist the decision-maker in making this determination (US EPA 1999a). As noted in the dioxin reassessment by the U.S. EPA: “Generally speaking, when considering either background exposures or incremental exposures plus background, MOEs in the range of 100–1000 are considered adequate to rule out the likelihood of significant effects occurring in humans based on sensitive animal responses or results from epidemiologic studies. The adequacy of the MOE to be protective of health must take into account the nature of the effect at the ‘point of departure,’ the slope of the dose–response curve, the adequacy of the overall database, interindividual variability in the human population, and other factors” (US EPA 2000c).
770
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
30.3.3. Use of Probabilistic Analysis to Refine Dose Estimates Probabilistic analysis, as discussed above for use in exposure reconstruction, can also be used to estimate risk, incorporating all of the available data (Beck and Cohen 1997; Cronin et al. 1995; Mertz et al. 1998; Simon 1997). In addition to exposure variables, data forming the basis of the toxicological criteria, such as cancer slope factors, are also amenable to Monte Carlo-style analysis where a robust database exists (Baird et al. 1996; Boyce 1998; Cox 1996; Crouch 1996a,b; Evans et al. 1994a,b; Hill and Hoover 1997; Shlyakhter et al. 1992; Sielken 1989; Sielken and Stevenson 1997; Sielken and Valdez-Flores 1996; Velazquez et al. 1994). As with exposure variables, the advantage to this approach is that it allows all data to be used (and weighted appropriately, where necessary), thus avoiding relying on a single experiment or endpoint.
30.4.
SUMMARY
This chapter presents a comprehensive review of the different types of exposure reconstruction methods that are available to industrial hygienists, epidemiologists, and risk assessors. There are a wide variety of possible techniques that can be used, many of which have advantages and disadvantages. The U.S. ATSDR has suggested considering two key issues when reconstructing exposure scenarios: first, whether the available data (either measured or modeled) are of sufficient quality and quantity to evaluate the necessary exposure pathways; and second, how the critical data gaps (if identified) should be filled (Agency for Toxic Substances and Disease Registry 2005). Therefore, the optimal technique for a specific situation depends on the type of data and the resources available (Gerin 1990). Quantitative combination techniques, such as Bayesian decision analysis, should be applied to historical exposure reconstruction, dose, and cancer risk estimation whenever appropriate to help to improve the quality of estimates that are generated. Techniques such as these allow for the systematic evaluation of multiple types of data, which, in turn, helps to ensure that critical data gaps in the reconstruction have been adequately addressed. Ramachandran (2001) demonstrated the application of a Bayesian decision analysis framework for retrospective exposure assessment of workers in a nickel smelter using existing exposure data, professional judgment, and Monte Carlo analysis together (Ramachandran 2001; Ramachandran and Vincent 1999). Furthermore, the use of rapidly developing techniques for both biological monitoring and PBPK modeling are allowing increasingly refined exposure and dose reconstruction efforts (Coble et al. 2005; Jonsson and Johanson 2002; Sohn et al. 2004; Yokley et al. 2006). Regardless of the chosen exposure reconstruction method, it is important to document the exposure estimation process, and, whenever possible, evaluate any uncertainties inherent in the selected methodology and the results obtained using Monte Carlo or a similar technique (Benke et al. 2001; Stewart et al. 1996, 2003). Uncertainty analysis techniques are increasingly common in exposure reconstruction, and can be critical to understanding the variability and uncertainty in the estimates generated.
30.4. SUMMARY
771
The use of a transparent, stepwise process to reconstruct exposure values is extremely important. The five steps outlined in this chapter, including identifying the goals of the reconstruction, organizing and ranking the available data, identifying key data gaps, selecting an appropriate methodology for the reconstruction, and conducting an uncertainty analysis, will allow for a well-organized and likely more robust reconstruction effort, which is particularly critical when the reconstructed values will be used quantitatively, such as in a cancer risk assessment, in order to minimize the effect of exposure misclassification. Many studies have evaluated the effect of exposure misclassification, which can have a significant impact on the final results of an epidemiology study or risk assessment. Careful selection of the data and methods to be used in the reconstruction can help to minimize this potentially important confounding factor (Blair et al. 2007; Grandjean et al. 2004; Blair and Stewart 1992). An increased interest and focus on appropriate methods for exposure reconstruction, such as the guidelines recently published by the American Industrial Hygiene Association (AIHA) (Viet et al. 2008), and the chapter on exposure reconstruction in the newest edition of the mathematical modeling publication from AIHA (Keil 2009), should also serve to improve the quality of exposure reconstruction efforts. Once historical exposures have been estimated, these values can be used to calculate potential dose, which, in turn, can be used to estimate the theoretical cancer risk associated with the exposure of interest. General dose equations have been presented, but it is important to note that these equations may incorporate many assumptions, which may lead to uncertainty in the estimations. Each of these assumptions should be carefully evaluated, and, depending on the anticipated use of the estimated values, an in-depth uncertainty analysis should be considered (Paustenbach and Madl 2008). In addition to uncertainty in exposure, dose, and cancer estimations, the possibility of multiple simultaneous exposures to different chemicals needs to be considered. Risks from simultaneous exposure to more than one carcinogenic substance are typically estimated by assuming that the individual risks are additive. This process assumes that intakes of individual substances are relatively small, that there are no synergistic or antagonistic chemical interactions, and that all chemicals produce the same toxic effect (US EPA 1989). However, because health risks from exposures to chemical mixtures are generally based on a combination of upper-bound risks calculated for individual compounds, these risk assessments tend to be overly conservative (Gaylor and Chen 1996; Hwang and Chen 1999; Kodell and Chen 1994). Cancer risks from multiple exposure pathways also need to be assessed. In general, multiple pathways are also assumed to be additive (US EPA 1989). However, the additive approach is only a valid assumption for a systemic chemical whose toxic effects are similar despite the route of exposure. For example, ingestion of trichloroethylene probably poses about the same cancer risk when it is inhaled or absorbed through the skin. In contrast, a chemical like hexavalent chromium is a known inhalation carcinogen because of its mutagenic effects in the lung, but poses little to no carcinogenic risk via ingestion or dermal contact because of the reductive capacity of the stomach and the organics in the skin (De Flora 2000; Paustenbach et al. 2003; Proctor et al. 2002). In today’s world, where the background risk of cancer in the United States is one in two for men and one in three for women (0.5 and 0.33, respectively) (Rodricks
772
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
1992; Ropeik and Gray 2002; Tran et al. 2000), it is important to understand the meaning of an estimated cancer risk from historical exposure. These risks are managed differently by different organizations, and a theoretical increased risk of cancer of 1 in 1000 may be acceptable in one circumstance, but may trigger a major remediation effort in another. For these reasons, it is necessary for all cancer risk estimations to be placed in context, and for all underlying assumptions and implications to be well understood and documented.
REFERENCES Abadie, M., Limam, J., Bouilly, J., and Genin, D. (2004). Particle pollution in the French high-speed train (TGV) smoker cars: Measurement and prediction of passengers exposure Atmospheric Environment 38, 2017–2027. Acquavella, J. F., Alexander, B. H., Mandel, J. S., Burns, C. J., and Gustin, C. (2006). Exposure misclassification in studies of agricultural pesticides: Insights from biomonitoring. Epidemiology 17, 69–74. Agency for Toxic Substances and Disease Registry (ATSDR) (2005). Public Health Assessment Guidance Manual, U.S. Department of Health and Human Services. American Conference of Industrial Hygienists (2008). 2008 TLVs and BEIs based on the documentation of the threshold limit values for chemical substances and physical agents and biological exposure indices. ACGIH, Cincinnati, OH. American Industrial Hygiene Association (2008). Draft documentation for the Biological Environmental Exposure Limits (BEELs). 4,4′-Methylene Dianiline and Creatinine, AIHA Press, Fairfax, VA. Andersen, M. E., Clewell, H., 3rd, and Krishnan, K. (1995). Tissue dosimetry, pharmacokinetic modeling, and interspecies scaling factors. Risk Anal 15, 533–537. Andersen, M. E., and Conolly, R. B. (1998). Mechanistic modeling of rodent liver tumor promotion at low levels of exposure: An example related to dose-response relationships for 2,3,7,8tetrachlorodibenzo-p-dioxin. Hum Exp Toxicol 17, 683–690. Andersen, M. E., MacNaughton, M. G., Clewell, H. J., and Paustenbach, D. J. (1987). Adjusting exposure limits for long and short exposure periods using a physiological pharmacokinetic model. Am Ind Hyg Assoc J 48, 335–343. Anderson, P. D., and Yuhas, A. L. (1996). Improving risk management by characterizing reality: A benefit of probabilistic risk assessment. Hum Ecol Risk Assess 2, 55–58. Armstrong, T. W., and Haas, C. N. (2007). A quantitative microbial risk assessment model for Legionnaires’ disease: Animal model selection and dose-response modeling. Risk Anal 27, 1581–1596. Armstrong, T. W., Pearlman, E. D., Schnatter, A. R., Bowes, S. M., 3rd, Murray, N., and Nicolich, M. J. (1996). Retrospective benzene and total hydrocarbon exposure assessment for a petroleum marketing and distribution worker epidemiology study. Am Ind Hyg Assoc J 57, 333–343. Aylward, L. L., Barton, H. A., and Hayes, S. M. (2009a). Biomonitoring equivalents (BE) dossier for toluene (CAS No. 108-88-3). Regul Toxicol Pharmacol 51, S27–S36. Aylward, L. L., Brunet, R. C., Starr, T. B., Carrier, G., Delzell, E., Cheng, H., and Beall, C. (2005). Exposure reconstruction for the TCDD-exposed NIOSH cohort using a concentration- and agedependent model of elimination. Risk Anal 25, 945–956. Aylward, L. L., and Hayes, S. M. (2009). Biomonitoring Equivalents (BE) dossier for 2,4dichlorophenoxyacetic acid (2,4-D) (CAS No. 94-75-7). Regul Toxicol Pharmacol 51, S37–S48. Aylward, L. L., Hays, S. M., Karch, N. J., and Paustenbach, D. J. (1996). Relative susceptibility of animals and humans to the cancer hazard posed by exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin using internal measures of dose. Environ Sci Technol 30, 3534–3543. Aylward, L. L., LaKind, J. S., and Hayes, S. M. (2009b). Biomonitoring equivalents (BE) dossier for trihalomethanes. Regul Toxicol Pharmacol 51, S68–S77. Baird, S. J. S., Cohen, J. T., Graham, J. D., Shlyakhter, A. I., and Evans, J. S. (1996). Noncancer risk assessment: A probabilistic alternative to current practice. Hum Ecol Risk Assess 2, 79–102.
REFERENCES
773
Beck, B. D., and Cohen, J. T. (1997). Risk assessment for criteria pollutants versus other noncarcinogens: The difference between implicit and explicit conservatism. Hum Ecol Risk Assess 3, 617–626. Benke, G., Sim, M., Forbes, A., and Salzberg, M. (1997). Retrospective assessment of occupational exposure to chemicals in community-based studies: Validity and repeatability of industrial hygiene panel ratings. Int J Epidemiol 26, 635–642. Benke, G., Sim, M., Fritschi, L., Aldred, G., Forbes, A., and Kauppinen, T. (2001). Comparison of occupational exposure using three different methods: Hygiene panel, job exposure matrix (JEM), and self reports. Appl Occup Environ Hyg 16, 84–91. Blair, A., Stewart, P., Lubin, J. H., and Forastiere, F. (2007). Methodological issues regarding confounding and exposure misclassification in epidemiological studies of occupational exposures. Am J Ind Med 50, 199–207. Blair, A., and Stewart, P. A. (1992). Do quantitative exposure assessments improve risk estimates in occupational studies of cancer? Am J Ind Med 21, 53–63. Bouyer, J., and Hemon, D. (1993). Retrospective evaluation of occupational exposures in populationbased case-control studies: General overview with special attention to job exposure matrices. Int J Epidemiol 22(Suppl 2), S57–S64. Boyce, C. P. (1998). Comparison of approaches for developing distributions for carcinogenic potency factors. Hum Ecol Risk Assess 4, 527–589. Bukowski, J., Korn, L. R., and Wartenberg, D. (1995). Correlated inputs in quantitative risk assessment: The effects of distributional shape. Risk Anal 15, 215–219. Burgess, G. L., Dippnall, W. M., Ravandi, M. R., and Cherry, N. M. (2004). Retrospective noise estimates for British nuclear workers using an alternative approach. Ann Occup Hyg 48, 117–127. Burmaster, D. E. (1998a). A lognormal distribution for time spent showering. Risk Anal 18, 33–36. Burmaster, D. E. (1998b). Lognormal distributions for skin area as a function of body weight. Risk Anal 18, 27–32. Burmaster, D. E., and Anderson, P. D. (1994). Principles of good practice for the use of Monte Carlo techniques in human health and ecological risk assessments. Risk Anal 14, 477–481. Burmaster, D. E., and von Stackelberg, K. (1991). Using Monte Carlo simulations in public health risk assessments: Estimating and presenting full distributions of risk. J Expo Anal Environ Epidemiol 1, 491–512. California Environmental Protection Agency (2007). Chemicals known to the state to cause cancer and reproductive toxicity. Office of Environmental Health Hazard Assessment, CalEPA, ed. Carrieri, M., Bonfiglio, E., Scapellato, M. L., Macca, I., Tranfo, G., Faranda, P., Paci, E., and Bartolucci, G. B. (2006). Comparison of exposure assessment methods in occupational exposure to benzene in gasoline filling-station attendants. Toxicol Lett 162, 146–152. Centers for Disease Control and Prevention (2005). Third National Report on Human Exposure to Environmental Chemicals. U.S. Department of Health and Human Services, National Center for Environmental Health, ed. Chang, H. Y., Tsai, C. Y., Lin, Y. Q., Shih, T. S., and Lin, Y. C. (2004). Urinary biomarkers of occupational N,N-dimethylformamide (DMF) exposure attributed to the dermal exposure. J Expo Anal Environ Epidemiol 14, 214–221. Checkoway, H., and Rice, C. H. (1992). Time-weighted averages, peaks, and other indices of exposure in occupation epidemiology. Am J Ind Med 21, 25–33. Chen, C. C., Wu, K. Y., and Chang, M. J. W. (2004). A statistical assessment on the stochastic relationship between biomarker concentrations and environmental exposures. Stoch Environ Res Risk Assess 18, 377–385. Cherrie, J. W. (1999). The effect of room size and general ventilation on the relationship between nearand far-field concentrations. Appl Occup Environ Hyg 14, 539–546. Cherrie, J. W., and Schneider, T. (1999). Validation of a new method for structured subjective assessment of past concentrations. Ann Occup Hyg 43, 235–245. Cherrie, J. W., Semple, S., Christopher, Y., Saleem, A., Hughson, G. W., and Philips, A. (2006). How important is inadvertent ingestion of hazardous substances at work? Ann Occup Hyg 50, 693–704. Chu, H., Wang, Z., Cole, S. R., and Greenland, S. (2006). Sensitivity analysis of misclassification: A graphical and a Bayesian approach. Ann Epidemiol 16, 834–841.
774
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
Clavel, J., Glass, D. C., Cordier, S., and Hemon, D. (1993). Standardization in the retrospective evaluation by experts of occupational exposure to organic solvents in a population-based case–control study. Int J Epidemiol 22(Suppl 2), S121–S126. Coble, J., Arbuckle, T., Lee, W., Alavanja, M., and Dosemeci, M. (2005). The validation of a pesticide exposure algorithm using biological monitoring results. J Occup Environ Hyg 2, 194–201. Cohen Hubal, E. A., Sheldon, L. S., Burke, J. M., McCurdy, T. R., Berry, M. R., Rigas, M. L., Zartarian, V. G., and Freeman, N. C. (2000). Children’s exposure assessment: A review of factors influencing Children’s exposure, and the data available to characterize and assess that exposure. Environ Health Perspect 108, 475–486. Cooper, J. A., Ferson, S., and Ginzberg, L. (1996). Hybrid processing of stochastic and subjective uncertainty data. Risk Anal 16, 785–792. Copeland, T. L., Holbrow, A. M., Otani, J. M., Connor, K. T., and Paustenbach, D. J. (1994). Use of probabilistic methods to understand the conservatism in California’s approach to assessing health risks posed by air contaminants. Air Waste 44, 1399–1413. Copeland, T. L., Paustenbach, D. J., Harris, M. A., and Otani, J. (1993). Comparing the results of a Monte Carlo analysis with EPA’s reasonable maximum exposed individual (RMEI): A case study of a former wood treatment site. Regul Toxicol Pharmacol 18, 275–312. Corn, M., and Esmen, N. A. (1979). Workplace exposure zones for classification of employee exposures to physical and chemical agents. Am Ind Hyg Assoc J 40, 47–57. Couch, J., Fleming, D., and Herman, A. (2005). A retrospective chemical exposure assessment of benzene and carbon tetrachloride for a United States naval shipyard. In American Industrial Hygiene Conference and Exposition, Anaheim, CA. Coughlin, S. S., and Chiazze, L., Jr. (1990). Job-exposure matrices in epidemiologic research and medical surveillance. Occup Med 5, 633–646. Cox, L. A. J. (1996). More accurate dose–response estimation using Monte Carlo uncertainty analysis: The data cube approach. Hum Ecol Risk Assess 2, 150–175. Cronin, W. J. t., Oswald, E. J., Shelley, M. L., Fisher, J. W., and Flemming, C. D. (1995). A trichloroethylene risk assessment using a Monte Carlo analysis of parameter uncertainty in conjunction with physiologically-based pharmacokinetic modeling. Risk Anal 15, 555–565. Crouch, E. A. (1996a). Uncertainty distributions for cancer potency factors: Combining epidemiological studies with laboratory bioassays—The example of acrylonitrile. Hum Ecol Risk Assess 2, 130–149. Crouch, E. A. (1996b). Uncertainty distributions for cancer potency factors: Laboratory animal carcinogenicity and interspecies extrapolation. Hum Ecol Risk Assess 2, 103–129. Crump, K. S. (1998). On summarizing group exposures in risk assessment: Is an arithmetic mean or a geometric mean more appropriate? Risk Anal 18, 293–297. Cullen, A. C., and Frey, H. C. (1999). Probabilistic Techniques in Exposure Assessment, Plenum Press, New York. Dahmann, D., Bauer, H. D., and Stoyke, G. (2008). Retrospective exposure assessment for respirable and inhalable dust, crystalline silica and arsenic in the former German uranium mines of SAG/SDAG Wismut. Int Arch Occup Environ Health 81, 949–958. De Flora, S. (2000). Threshold mechanisms and site specificity in chromium(VI) carcinogenesis. Carcinogenesis 21, 533–541. Deadman, J. E., Church, G., Bradley, C., Armstrong, B. G., and Theriault, G. (1997). Task-based estimation of past exposures to 60-hertz magnetic and electric fields at an electrical utility. Scand J Work Environ Health 23, 440–449. Dement, J. M., Harris, R. L., Symons, M. J., and Shy, C. M. (1983). Exposures and mortality among chrysotile asbestos workers. Part I: Exposure estimates. Am J Ind Med 4, 399–419. Dement, J. M., Loomis, D., Richardson, D., Wolf, S., and Myers, D. (2008). Estimates of historical exposures by phase contrast and transmission electron microscopy in North Carolina USA asbestos textile plants. Occup Environ Med 66, 574–583. Dosemeci, M., Chen, J. Q., Hearl, F., Chen, R. G., McCawley, M., Wu, Z., McLaughlin, J. K., Peng, K. L., Chen, A. L., Rexing, S. H., et al. (1993). Estimating historical exposure to silica among mine and pottery workers in the People’s Republic of China. Am J Ind Med 24, 55–66. Dosemeci, M., Rothman, N., Yin, S. N., Li, G. L., Linet, M., Wacholder, S., Chow, W. H., and Hayes, R. B. (1997). Validation of benzene exposure assessment. Ann NY Acad Sci 837, 114–121.
REFERENCES
775
Drummond, I., Murray, N., Armstrong, T., Schnatter, A. R., and Lewis, R. J. (2006). Exposure assessment methods for a study of mortality and cancer morbidity in relation to specific petroleum industry exposures. J Occup Environ Hyg 3, 513–520. Duan, N., and Mage, D. T. (1997). Combination of direct and indirect approaches for exposure assessment. J Expo Anal Environ Epidemiol 7, 439–470. Egeghy, P. P., Quackenboss, J. J., Catlin, S., and Ryan, P. B. (2005). Determinants of temporal variability in NHEXAS-Maryland environmental concentrations, exposures, and biomarkers. J Expo Anal Environ Epidemiol 15, 388–397. Eickmann, U., Liesche, W., and Wegscheider, W. (2007). Harmonization and further development of models to calculate airborne contaminant concentrations at the workplace. Gefahrstoffe–Reinhalt Luft 67, 127–132. Eisen, E. A., Smith, T. J., Wegman, D. H., Louis, T. A., and Froines, J. (1984). Estimation of long term dust exposures in the Vermont granite sheds. Am Ind Hyg Assoc J 45, 89–94. Esmen, N. (1979). Retrospective industrial hygiene surveys. Am Ind Hyg Assoc J 40, 58–65. Esmen, N. A., and Corn, M. (1998). Airborne fiber concentrations during splitting open and boxing bags of asbestos. Toxicol Ind Health 14, 843–856. Esmen, N. A., Hall, T. A., Phillips, M. L., Jones, E. P., Basara, H., Marsh, G. M., and Buchanich, J. M. (2007a). Chemical process-based reconstruction of exposures for an epidemiological study. Part II. Estimated exposures to chloroprene and vinyl chloride. Chem Biol Interact 166, 264–276. Esmen, N. A., Hall, T. A., Phillips, M. L., and Marsh, G. M. (2007b). Chemical process based reconstruction of exposures for an epidemiological study. I. Theoretical and methodological issues. Chem Biol Interact 166, 254–263. Esmen, N. A., Kennedy, K. J., Hall, T. A., Phillips, M. L., and Marsh, G. M. (2007c). Classification of worker exposures. Chem Biol Interact 166, 245–253. Evans, J. S., Graham, J. D., Gray, G. M., and Sielken, R. L., Jr. (1994a). A distributional approach to characterizing low-dose cancer risk. Risk Anal 14, 25–34. Evans, J. S., Gray, G. M., Sielken, R. L., Jr., Smith, A. E., Valdez-Flores, C., and Graham, J. D. (1994b). Use of probabilistic expert judgment in uncertainty analysis of carcinogenic potency. Regul Toxicol Pharmacol 20, 15–36. Fallentin, B., and Kamstrup, O. (1993). Simulation of past exposure in slag wool production. Ann Occup Hyg 37, 419–433. Fehrenbacher, M. C., and Hummel, A. A. (1996). Evaluation of the mass balance model used by the Environmental Protection Agency for estimating inhalation exposure to new chemical substances. Am Ind Hyg Assoc J 57, 526–536. Fenske, R. A., Kissel, J. C., Lu, C., Kalman, D. A., Simcox, N. J., Allen, E. H., and Keifer, M. C. (2000). Biologically based pesticide dose estimates for children in an agricultural community. Environ Health Perspect 108, 515–520. Finley, B. L., and Paustenbach, D. J. (1994). The benefits of probabilistic exposure assessment: Three case studies involving contaminated air, water, and soil. Risk Anal 14, 53–73. Finley, B. L., Proctor, D. M., Scott, P. K., Harrington, N., Paustenbach, D. J., and Price, P. (1994). Recommended distributions for exposure factors frequently used in health risk assessment. Risk Anal 14, 533–553. Frey, H. C., and Patil, S. R. (2002). Identification and review of sensitivity analysis methods. Risk Anal 22, 553–578. Friesen, M. C., Davies, H. W., Teschke, K., Marion, S., and Demers, P. A. (2005). Predicting historical dust and wood dust exposure in sawmills: Model development and validation. J Occup Environ Hyg 2, 650–658. Friesen, M. C., Davies, H. W., Teschke, K., Ostry, A. S., Hertzman, C., and Demers, P. A. (2007). Impact of the specificity of the exposure metric on exposure–response relationships. Epidemiology 18, 88–94. Friesen, M. C., Demers, P. A., Spinelli, J. J., and Le, N. D. (2003). Validation of a semiquantitative job exposure matrix at a Soderberg aluminum smelter. Ann Occup Hyg 47, 477–484. Friesen, M. C., Macnab, Y. C., Marion, S. A., Demers, P. A., Davies, H. W., and Teschke, K. (2006). Mixed models and empirical Bayes estimation for retrospective exposure assessment of dust exposures in Canadian sawmills. Ann Occup Hyg 50, 281–288.
776
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
Fritschi, L., Nadon, L., Benke, G., Lakhani, R., Latreille, B., Parent, M. E., and Siemiatycki, J. (2003). Validation of expert assessment of occupational exposures. Am J Ind Med 43, 519–522. Fritschi, L., Siemiatycki, J., and Richardson, L. (1996). Self-assessed versus expert-assessed occupational exposures. Am J Epidemiol 144(5), 521–527. Gaffney, S., Moody, E., McKinley, M., Knutsen, J., Madl, A., and Paustenbach, D. (2008). Worker exposure to methanol vapors during cleaning of semiconductor wafers in a manufacturing setting. J Occup Environ Hyg 5, 313–324. Gaffney, S., Panko, J. M., Unice, K. M., Burns, A. M., Kreides, M. L., Gelatt, R. H., Bocher, L. E., and Paustenbach, D. J. (2009). Occupational exposure to benzene at the ExxonMobil refinery in Baytown, TX. J Exp Sci Env Epidemiol. doi: 10.1038/jes.2009.53. Gargas, M. L., Finley, B. L., Paustenbach, D. J., and Long, T. F. (1999). Environmental health risk assessment: Theory and practice. In General and Applied Toxicology, Vol. 3, Ballantyne, B., Marrs, T., and Syverson, T., eds., Macmillan, London, pp. 1749–1809. Gaylor, D. W., and Chen, J. J. (1996). A simple upper limit for the sum of the risks of the components in a mixture. Risk Anal 16, 395–398. Gerin, M. (1990). Recent approaches to retrospective exposure assessment in occupational cancer epidemiology. Recent Results Cancer Res 120, 39–49. Gerin, M., Fletcher, A. C., Gray, C., Winkelmann, R., Boffetta, P., and Simonato, L. (1993). Development and use of a welding process exposure matrix in a historical prospective study of lung cancer risk in European welders. Int J Epidemiol 22(Suppl 2), S22–S28. Gitelman, M., and Dement, J. (1996). Simulated 1, 1, 1 Trichloroethane exposure during brake repair. Appl Occup Environ Hyg 11, 1177–1179. Glass, D. C., Adams, G. G., Manuell, R. W., and Bisby, J. A. (2000). Retrospective exposure assessment for benzene in the Australian petroleum industry. Ann Occup Hyg 44, 301–320. Glass, D. C., Spurgeon, A., Calvert, I. A., Clark, J. L., and Harrington, J. M. (1994). Retrospective assessment of solvent exposure in paint manufacturing. Occup Environ Med 51, 617–625. Goldberg, M., Kromhout, H., Guenel, P., Fletcher, A. C., Gerin, M., Glass, D. C., Heederik, D., Kauppinen, T., and Ponti, A. (1993). Job exposure matrices in industry. Int J Epidemiol 22(Suppl 2), S10–S15. Gosselin, N. H., Brunet, R. C., Carrier, G., Bouchard, M., and Feeley, M. (2006). Reconstruction of methylmercury intakes in indigenous populations from biomarker data. J Expo Sci Environ Epidemiol 16, 19–29. Grandjean, P., Budtz-Jorgensen, E., Keiding, N., and Weihe, P. (2004). Underestimation of risk due to exposure misclassification. Int J Occup Med Environ Health 17, 131–136. Greenland, S. (2001). Sensitivity analysis, Monte Carlo risk analysis, and Bayesian uncertainty assessment. Risk Anal 21, 579–583. Groah, W. J. (2006). Predicting indoor-air formaldehyde concentrations from emissions of ureaformaldehyde-bonded wood products using the Versar model. Forest Products J 55, 97–100. Gustafson, P., and Greenland, S. (2006). Curious phenomena in Bayesian adjustment for exposure misclassification. Stat Med 25, 87–103. Haas, C. N. (1997). Importance of the distributional form in characterizing inputs to Monte Carlo risk assessments. Risk Anal 17, 107–113. Hall, T. A., Esmen, N. A., Jones, E. P., Basara, H., Phillips, M. L., Marsh, G. M., Youk, A. O., Buchanich, J. M., and Leonard, R. C. (2007). Chemical process based reconstruction of exposures for an epidemiological study. III. Analysis of industrial hygiene samples. Chem Biol Interact 166, 277–284. Hallock, M. F., Smith, T. J., Woskie, S. R., and Hammond, S. K. (1994). Estimation of historical exposures to machining fluids in the automotive industry. Am J Ind Med 26, 621–634. Hamed, M. M., and Bedient, P. B. (1997). On the effect of probability distributions of input variables in public health risk assessment. Risk Anal 17, 97–105. Harber, P., Muranko, H., Shvartsblat, S., Solis, S., Torossian, A., and Oren, T. (2003). A triangulation approach to historical exposure assessment for the carbon black industry. J Occup Environ Med 45, 131–143. Hattis, D. B., and Burmaster, D. E. (1994). Assessment of variability and uncertainty distributions for practical risk analyses. Risk Anal 14, 713–730.
REFERENCES
777
Hawkins, N. C., and Evans, J. S. (1989). Subjective estimation of toluene exposures: A calibration study of industrial hygienists. Appl Ind Hyg 4, 61–68. Hayes, S. M., and Aylward, L. L. (2009). Biomonitoring equivalents (BE) dossier for acrylamide (AA) (CAS No. 79-06-1). Regul Toxicol Pharmacol 51, S57-S67. Hayes, S. M., Aylward, L. L., and LaKind, J. S. (2009a). Introduction to the biomonitoring equivalents Pilot Project: Development of guidelines for the derivation and communication of biomonitoring equivalents. Regul Toxicol Pharmacol 51, S1–S2. Hayes, S. M., Aylward, L. L., LaKind, J. S., Bartels, M. J., Barton, H. A., Boogaard, P. J., Brunk, C., DiZio, S., Dourson, M., Goldstein, D. A., Lipscomb, J., Kilpatrick, M. E., Krewskim, D., Krishnan, K., Nordberg, M., Okino, M., Tan, Y., Viau, C., and Yager, J. W. (2008). Guidelines for the derivation of biomonitoring equivalents: Report from the Biomonitoring Equivalents Expert Workshop. Regul Toxicol Pharmacol 51, S4–S15. Hayes, S. M., Nordberg, M., Yager, J. W., and Aylward, L. L. (2009b). Biomonitoring equivalents (BE) dossier for cadmium (Cd) (CAS No. 7440-43-9). Regul Toxicol Pharmacol 51, S49–S56. Hein, M. J., Waters, M. A., van Wijngaarden, E., Deddens, J. A., and Stewart, P. A. (2008). Issues when modeling benzene, toluene, and xylene exposures using a literature database. J Occup Environ Hyg 5, 36–47. Helton, J. C., and Davis, F. J. (2002). Illustration of sampling-based methods for uncertainty and sensitivity analysis. Risk Anal 22, 591–622. Henn, S. A., Utterback, D. F., Waters, K. M., Markey, A. M., and Tankersley, W. G. (2007). Task- and time-dependent weighting factors in a retrospective exposure assessment of chemical laboratory workers. J Occup Environ Hyg 4, 71–79. Herrick, R. F., Stewart, J. H., Blicharz, D., Beall, C., Bender, T., Cheng, H., Matthews, R., Sathiakumar, N., and Delzell, E. (2005). Exposure assessment for retrospective follow-up studies of semiconductorand storage device-manufacturing workers. J Occup Environ Med 47, 983–985. Hewett, P., Logan, P., Mulhausen, J., Ramachandran, G., and Banerjee, S. (2006). Rating exposure control using Bayesian decision analysis. J Occup Environ Hyg 3, 568–581. Hill, R. A., and Hoover, S. M. (1997). Importance of the dose–response model form in probabilistic risk assessment: A case study of health effects from methylmercury in fish. Hum Ecol Risk Assess 3, 465–481. Hoffman, F. O., and Hammonds, J. S. (1994). Propagation of uncertainty in risk assessments: The need to distinguish between uncertainty due to lack of knowledge and uncertainty due to variability. Risk Anal 14, 707–712. Hornung, R. W., Greife, A. L., Stayner, L. T., Steenland, N. K., Herrick, R. F., Elliott, L. J., Ringenburg, V. L., and Morawetz, J. (1994). Statistical model for prediction of retrospective exposure to ethylene oxide in an occupational mortality study. Am J Ind Med 25, 825–836. Hwang, J. S., and Chen, J. J. (1999). An evaluation of risk estimation procedures for mixtures of carcinogens. Risk Anal 19, 1071–1076. Ignacio, J. S., and Bullock, W. H. (2006). A Strategy for Assessing and Managing Occupational Exposures, American Industrial Hygeine Association, Fairfax, VA. Integrated Risk Information System (2008). IRIS Glossary/Glossary Archive/Acronyms and Abbreviations. Ivannikov, A. I., Gaillard-Lecanu, E., Trompier, F., Stepanenko, V. F., Skvortsov, V. G., Borysheva, N. B., Tikunov, D. D., and Petin, D. V. (2004). Dose reconstruction by EPR spectroscopy of tooth enamel: Application to the population of Zaborie village exposed to high radioactive contamination after the Chernobyl accident. Health Phys 86, 121–134. Johansen, K., Tinnerberg, H., and Lynge, E. (2005). Use of history science methods in exposure assessment for occupational health studies. Occup Environ Med 62, 434–441. Johnson, D. L., and Esmen, N. A. (2004). Method-induced misclassification for a respirable dust sampled using ISO/ACGIH/CEN criteria. Ann Occup Hyg 48, 13–20. Jonsson, F., and Johanson, G. (2002). Physiologically based modeling of the inhalation kinetics of styrene in humans using a Bayesian population approach. Toxicol Appl Pharmacol 179, 35–49. Kauppinen, T., Toikkanen, J., and Pukkala, E. (1998). From cross-tabulations to multipurpose exposure information systems: a new job-exposure matrix. Am J Ind Med 33, 409–417. Kauppinen, T. P., Pannett, B., Marlow, D. A., and Kogevinas, M. (1994). Retrospective assessment of exposure through modeling in a study on cancer risks among workers exposed to phenoxy herbicides, chlorophenols and dioxins. Scand J Work Environ Health 20, 262–271.
778
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
Kawasaki, T., Kono, K., Dote, T., Usuda, K., Shimizu, H., and Dote, E. (2004). Markers of cadmium exposure in workers in a cadmium pigment factory after changes in the exposure conditions. Toxicol Ind Health 20, 51–56. Keil, C. (2000). Mathematical Models for Estimating Occupational Exposure to Chemicals, American Industrial Hygiene Association Press, Fairfax, VA. Keil, C., and Murphy, R. (2006). An application of exposure modeling in exposure assessments for a university chemistry teaching laboratory. J Occup Environ Hyg 3, 99–106. Keil, C., Simmons, C., and Anthone, T. R. (2009). Mathematical Models for Estimating Occupational Exposure to Chemicals. Industrial Hygiene Association Press, Fairfax, VA. Keil, C. B. (1998). The development and evaluation of an emission factor for a toluene parts-washing process. Am Ind Hyg Assoc J 59, 14–19. Keil, C. B., Akbar-Khanzadeh, F., and Konecny, K. A. (2001). Characterizing formaldehyde emission rates in a gross anatomy laboratory. Appl Occup Environ Hyg 16, 967–972. Keil, C. B., and Nicas, M. (2003). Predicting room vapor concentrations due to spills of organic solvents. AIHA J (Fairfax, VA) 64, 445–454. Keyes, D. L., Chesson, J., Ewing, W. M., Faas, J. C., Hatfield, R. L., Hays, S. M., Longo, W. E., and Millette, J. R. (1991). Exposure to airborne asbestos associated with simulated cable installation above a suspended ceiling. Am Ind Hyg Assoc J 52, 479–484. Kleinerman, R. A., Romanyukha, A. A., Schauer, D. A., and Tucker, J. D. (2006). Retrospective assessment of radiation exposure using biological dosimetry: Chromosome painting, electron paramagnetic resonance and the glycophorin a mutation assay. Radiat Res 166, 287–302. Kodell, R. L., and Chen, J. J. (1994). Reducing conservatism in risk estimation for mixtures of carcinogens. Risk Anal 14, 327–332. Koo, H. J., and Lee, B. M. (2005). Human monitoring of phthalates and risk assessment. J Toxicol Environ Health A 68, 1379–1392. Kudla, I. (1997). Exposure to benzene-contaminated toluene and bone marrow disorders—A retrospective exposure assessment Appl Occup Environ Hyg 12, 11–14. Laden, F., Hart, J. E., Eschenroeder, A., Smith, T. J., and Garshick, E. (2006). Historical estimation of diesel exhaust exposure in a cohort study of U.S. railroad workers and lung cancer. Cancer Causes Control 17, 911–919. Lavoue, J., Begin, D., Beaudry, C., and Gerin, M. (2007). Monte Carlo simulation to reconstruct formaldehyde exposure levels from summary parameters reported in the literature. Ann Occup Hyg 51, 161–172. Leung, H. W., and Paustenbach, D. J. (1988). Application of pharmacokinetics to derive biological exposure indexes from threshold limit values. Am Ind Hyg Assoc J 49, 445–450. Lewis, S. J., Bell, G. M., Cordingley, N., Pearlman, E. D., and Rushton, L. (1997). Retrospective estimation of exposure to benzene in a leukaemia case–control study of petroleum marketing and distribution workers in the United Kingdom. Occup Environ Med 54, 167–175. London, L., and Myers, J. E. (1998). Use of a crop and job specific exposure matrix for retrospective assessment of long-term exposure in studies of chronic neurotoxic effects of agrichemicals. Occup Environ Med 55, 194–201. Macaluso, M., Larson, R., Delzell, E., Sathiakumar, N., Hovinga, M., Julian, J., Muir, D., and Cole, P. (1996). Leukemia and cumulative exposure to butadiene, styrene and benzene among workers in the synthetic rubber industry. Toxicology 113, 190–202. Macaluso, M., Larson, R., Lynch, J., Lipton, S., and Delzell, E. (2004). Historical estimation of exposure to 1,3-butadiene, styrene, and dimethyldithiocarbamate among synthetic rubber workers. J Occup Environ Hyg 1, 371–390. Madl, A. K., and Paustenbach, D. J. (2002a). Airborne concentrations of benzene and mineral spirits (Stoddard solvent) during cleaning of a locomotive generator and traction motor. J Toxicol Environ Health A 65, 1965–1979. Madl, A. K., and Paustenbach, D. J. (2002b). Airborne concentrations of benzene due to diesel locomotive exhaust in a roundhouse. J Toxicol Environ Health A 65, 1945–1964. Madl, A. K., Gaffney, S. H., Balzes, J. L., and Paustenbach, D. J. (2009). Airborne asbestos concentrations associated with heavy equipment brake removal. Ann Occup Hyg 53, 839–857.
REFERENCES
779
Madl, A. K., Unice, K. M., Brown, J. L., Kolanz, M. E., and Kent, M. S. (2007). Exposure-response analysis for beryllium sensitization and chronic beryllium disease among workers in a beryllium metal machining plant. J Occup Environ Hyg 4(6), 448–466. Mage, D. T., Allen, R. H., Gondy, G., Smith, W., Barr, D. B., and Needham, L. L. (2004). Estimating pesticide dose from urinary pesticide concentration data by creatinine correction in the Third National Health and Nutrition Examination Survey (NHANES-III). J Expo Anal Environ Epidemiol 14, 457–465. Mangold, C., Clark, K., Madl, A., and Paustenbach, D. (2006). An exposure study of bystanders and workers during the installation and removal of asbestos gaskets and packing. J Occup Environ Hyg 3, 87–98. Mannetje, A., Fevotte, J., Fletcher, T., Brennan, P., Legoza, J., Szeremi, M., Paldy, A., Brzeznicki, S., Gromiec, J., Ruxanda-Artenie, C., Stanescu-Dumitru, R., Ivanov, N., Shterengorz, R., Hettychova, L., Krizanova, D., Cassidy, A., van Tongeren, M., and Boffetta, P. (2003). Assessing exposure misclassification by expert assessment in multicenter occupational studies. Epidemiology 14, 585–592. Maslia, M. L., Sautner, J. B., Aral, M. M., Gillig, R. E., Reyes, J. J., and Williams, R. C. (2001). Historical reconstruction of the water-distribution system serving the Dover Township area, New Jersey: January 1962–December 1996. Matthiessen, R. C. (1986). Estimating chemical exposure levels in the workplace: Well-characterized monitoring data are needed to assess worker exposure levels during the manufacture, processing, and use of industrial chemicals. Chem Eng Prog, 30–34. McKinnery, W. N., Jr., and Moore, R. W. (1992). Evaluation of airborne asbestos fiber levels during removal and installation of valve gaskets and packing. Am Ind Hyg Assoc 53, 531–532. McNamee, C. (1996). Retrospective assessment of occupational exposure to hydrocarbons—Jobexposure matrices versus expert evaluation of questionnaires. Occup Hyg 3, 137–143. Meredith, S. K., Bugler, J., and Clark, R. L. (2000). Isocyanate exposure and occupational asthma: A case-referent study. Occup Environ Med 57, 830–836. Mertz, C. K., Slovic, P., and Purchase, I. F. (1998). Judgments of chemical risks: Comparisons among senior managers, toxicologists, and the public. Risk Anal 18, 391–404. Mokhtari, A., Christopher Frey, H., and Zheng, J. (2006). Evaluation and recommendation of sensitivity analysis methods for application to stochastic human exposure and dose simulation models. J Expo Sci Environ Epidemiol 16, 491–506. Monge, P., Partanen, T., Wesseling, C., Bravo, V., Ruepert, C., and Burstyn, I. (2005). Assessment of pesticide exposure in the agricultural population of Costa Rica. Ann Occup Hyg 49, 375–384. Morgan, M. D., and Henrion, M. (1990). Uncertainty: A Guide to Dealing with Uncertainty in Quantitative Risk and Policy Analysis, Cambridge University Press, Cambridge. Mowat, F., Weidling, R., and Sheehan, P. (2007). Simulation tests to assess occupational exposure to airborne asbestos from asphalt-based roofing products. Ann Occup Hyg 51, 451–462. National Institute for Occupational Safety and Health (NIOSH). (2002). Interactive RadioEpidemiological Program Final Report. NIOSH-IREP, Cincinnati, OH. National Institute for Occupational Safety and Health (NIOSH). (2007). External dose reconstruction implementation guideline. Office of Compensation Analysis and Support (OCAS). Cincinnati, OH. Nelson, D., Nelson, R., Hart, K., and Asal, N. (1995). Hydrocarbon exposure assessment methodology for an epidemiologic study of renal disease. Appl Occup Environ Hyg 10, 299–310. Nestorov, I. (2003). Whole body pharmacokinetic models. Clin Pharmacokinet 42, 883–908. Nestorov, I. A. (1999). Sensitivity analysis of pharmacokinetic and pharmacodynamic systems: I. A structural approach to sensitivity analysis of physiologically based pharmacokinetic models. J Pharmacokinet Biopharm 27, 577–596. Nicas, M. (2000). Two-zone model. In Mathematical Models for Estimating Occupational Exposure to Chemicals, Keil, C. ed., AIHA Press, Fairfax, VA. Nicas, M. (2003a). Estimating methyl bromide exposure due to offgassing from fumigated commodities. Appl Occup Environ Hyg 18, 200–210. Nicas, M. (2003b). Using mathematical models to estimate exposure to workplace air contaminants. Chem Health Saf January/February, 14–21.
780
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
Nicas, M., Plisko, M. J., and Spencer, J. W. (2006). Estimating benzene exposure at a solvent parts washer. J Occup Environ Hyg 3, 284–291. Nieuwenhuijsen, M. J., Grey, C. N., and Golding, J. (2005). Exposure misclassification of household pesticides and risk perception and behaviour. Ann Occup Hyg 49, 703–709. Olsen, G. W., Logan, P. W., Hansen, K. J., Simpson, C. A., Burris, J. M., Burlew, M. M., Vorarath, P. P., Venkateswarlu, P., Schumpert, J. C., and Mandel, J. H. (2003). An occupational exposure assessment of a perfluorooctanesulfonyl fluoride production site: Biomonitoring. AIHA J (Fairfax, VA) 64, 651–659. Owen, C. V., Acquavella, J. A., Lynch, J., and Bird, M. G. (1992). An industrial hygiene methodology developed in support of a retrospective morbidity case–control study. Am Ind Hyg Assoc J 53, 540–547. Panko, J. M., Gaffney, S. H., Burns, A. M., Unice, K. M., Kreider, M. L., Booher, L. E., Gelatt, R. H., Marshall, J. R., and Paustenbach, D. J. (2009). Occupational exposure to benzene at the ExxonMobil refinery at Baton Rouge, Louisiana (1977–2005). J Occup Environ Hyg 6, 517–529. Paustenbach, D., and Madl, A. (2008). The practice of exposure assessment. In Principles and Methods of Toxicology, 5th Edition, Hayes, A. W., ed., CRC Press, Andover, MA, pp. 475–534. Paustenbach, D. J. (2000). The practice of exposure assessment: A state-of-the-art review. J Toxicol Environ Health B Crit Rev 3, 179–291. Paustenbach, D. J. (2002). Human and Ecological Risk Assessment: Theory and Practice, John Wiley & Sons, Hoboken, NJ. Paustenbach, D. J., Finley, B. L., Mowat, F. S., and Kerger, B. D. (2003). Human health risk and exposure assessment of chromium (VI) in tap water. J Toxicol Environ Health A 66, 1295–1339. Paustenbach, D. J., Madl, A. K., Donovan, E., Clark, K., Fehling, K., and Lee, T. C. (2006). Chrysotile asbestos exposure associated with removal of automobile exhaust systems (ca. 1945–1975) by mechanics: Results of a simulation study. J Expo Sci Environ Epidemiol 16, 156–171. Paustenbach, D. J., Price, P. S., Ollison, W., Blank, C., Jernigan, J. D., Bass, R. D., and Peterson, H. D. (1992). Reevaluation of benzene exposure for the pliofilm (rubberworker) cohort (1936–1976). J Toxicol Environ Health 36, 177–231. Pleil, J. D., Kim, D., Prah, J. D., and Rappaport, S. M. (2007). Exposure reconstruction for reducing uncertainty in risk assessment: Example using MTBE biomarkers and a simple pharmacokinetic model. Biomarkers 12, 331–348. Proctor, D. M., Otani, J. M., Finley, B. L., Paustenbach, D. J., Bland, J. A., Speizer, N., and Sargent, E. V. (2002). Is hexavalent chromium carcinogenic via ingestion? A weight-of-evidence review. J Toxicol Environ Health A 65, 701–746. Proctor, D. M., Panko, J. P., Liebig, E. W., and Paustenbach, D. J. (2004). Estimating historical occupational exposure to airborne hexavalent chromium in a chromate production plant: 1940–1972. J Occup Environ Hyg 1, 752–767. Ramachandran, G. (2001). Retrospective exposure assessment using Bayesian methods. Ann Occup Hyg 45, 651–667. Ramachandran, G., and Vincent, J. H. (1999). A Bayesian approach to retrospective exposure assessment. Appl Occup Environ Hyg 14, 547–557. Rappaport, S. M. (1991). Selection of the measures of exposure for epidemiology studies. Appl Occup Environ Hyg 5, 448–457. Riley, D. M., Newby, C. A., Leal-Almeraz, T. O., and Thomas, V. M. (2001). Assessing elemental mercury vapor exposure from cultural and religious practices. Environ Health Perspect 109, 779–784. Ripple, S. R., Widner, T. E., and Mongan, T. R. (1996). Past radionuclide releases from routine operations at Rocky Flats. Health Phys Soc 71, 502–509. Rodricks, J. (1992). Calculated Risks: The Toxicity and Human Health Risks of Chemicals in Our Environment, Cambridge University Press, Cambridge. Romundstad, P., Haldorsen, T., and Ronneberg, A. (1999). Exposure to PAH and fluoride in aluminum reduction plants in Norway: Historical estimation of exposure using process parameters and industrial hygiene measurements. Am J Ind Med 35, 164–174. Romundstad, P. R., Ronneberg, A., Leira, H. L., and Bye, T. (1998). Health survey of former workers in a Norwegian coke plant: Part. 1. Estimation of historical exposures. Occup Environ Med 55, 616–621.
REFERENCES
781
Ropeik, D., and Gray, G. (2002). Risk: A Practical Guide for Deciding What’s Really Safe and What’s Really Dangerous in the World Around You, Houghton Mifflin Company, Boston. Sahmel, J., and Boeniger, M. (2006). Dermal exposure assessments. In A Strategy for Assessing and Managing Occupational Exposures, Ignacio, J. S., and Bullock, W. H., eds., American Industrial Hygiene Association, Fairfax, VA, pp. 137–161. Saltelli, A. (2002). Sensitivity analysis for importance assessment. Risk Anal 22, 579–590. Sanderson, W. T., Petersen, M. R., and Ward, E. M. (2001). Estimating historical exposures of workers in a beryllium manufacturing plant. Am J Ind Med 39, 145–157. Schultz, M., Cisek, J., and Wabeke, R. (1995). Simulated exposure of hospital emergency personnel to solvent vapors and respirable dust during decontamination of chemically exposed patients. Ann Emerg Med 26, 324–329. Sedman, R., Funk, L. M., and Fountain, R. (1998). Distribution of residence duration in owner occupied housing. J Expo Anal Environ Epidemiol 8, 51–58. Seel, E. A., Zaebst, D. D., Hein, M. J., Liu, J., Nowlin, S. J., and Chen, P. (2007). Inter-rater agreement for a retrospective exposure assessment of asbestos, chromium, nickel and welding fumes in a study of lung cancer and ionizing radiation. Ann Occup Hyg 51, 601–610. Seixas, N. S., and Checkoway, H. (1995). Exposure assessment in industry specific retrospective occupational epidemiology studies. Occup Environ Med 52, 625–633. Seixas, N. S., Robins, T. G., and Becker, M. (1993). A novel approach to the characterization of cumulative exposure for the study of chronic occupational disease. Am J Epidemiol 137, 463–471. Seixas, N. S., Robins, T. G., and Moulton, L. H. (1988). The use of geometric and arithmetic mean exposures in occupational epidemiology. Am J Ind Med 14, 465–477. Shade, W. D., and Jayjock, M. A. (1997). Monte Carlo uncertainty analysis of a diffusion model for the assessment of halogen gas exposure during dosing of brominators. Am Ind Hyg Assoc J 58, 418–424. Sheehan, P., Malzahn, D., Goswami, E., and Mandel, J. H. (2008). Simulation of benzene exposure during use of a mineral spirit solvent to clean elevator bearing housings. Hum Ecol Risk Assess 14, 421–432. Shlyakhter, A. I., Goodman, G., and Wilson, R. (1992). Monte Carlo simulation of rodent carcinogenicity bioassays. Risk Anal 12, 73–82. Sielken, R. L., Jr. (1989). Useful tools for evaluating and presenting more science in quantitative cancer risk assessment. Tox Subst J 9, 353–404. Sielken, R. L., Jr., and Stevenson, D. E. (1997). Opportunities to improve quantitative risk assessment. Hum Ecol Risk Assess 3, 479–490. Sielken, R. L., Jr., and Valdez-Flores, C. (1996). Comprehensive realism’s weight-of-evidence based distributional dose–response characterization. Hum Ecol Risk Assess 2, 175–193. Siemiatycki, J., Fritschi, L., Nadon, L., and Gerin, M. (1997). Reliability of an expert rating procedure for retrospective assessment of occupational exposures in community-based case–control studies. Am J Ind Med 31, 280–286. Simon, T. (1997). Combining physiologically based pharmacokinetics modeling with Monte Carlo simulation to derive an acute inhalation guidance value for trichloroethylene. Regul Toxicol Pharmacol 26, 257–270. Sivulka, D. J., and Seilkop, S. K. (2009). Reconstruction of historical exposures in the U.S. nickel alloy industry and the implications for carcinogenic hazard and risk assessments. Regul Toxicol Pharmacol 53, 174–185. Smith, A. E., Ryan, P. B., and Evans, J. S. (1992). The effects of neglecting correlations when propagating uncertainty and estimating population distribution of risk. Risk Anal 12, 467–474. Smith, R. L. (1994). Use of Monte Carlo simulation for human exposure assessment at a superfund site. Risk Anal 14, 433–439. Smith, R. W., Sahl, J. D., Kelsh, M. A., and Zalinski, J. (1997). Task-based exposure assessment: Analytical strategies for summarizing data by occupational groups. Am Ind Hyg Assoc J 58, 402–412. Smith, T., Hammond, S. K., Hallock, M., and Woskie, S. R. (1991). Exposure assessment for epidemiology: Characteristics of exposure. Appl Occup Environ Hyg 6, 441–447.
782
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
Smith, T. J., Hammond, S. K., Laidlaw, F., and Fine, S. (1984). Respiratory exposures associated with silicon carbide production: Estimation of cumulative exposures for an epidemiological study. Br J Ind Med 41, 100–108. Smith, T. J., Hammond, S. K., and Wong, O. (1993). Health effects of gasoline exposure: I: Exposure assessment for U.S. distribution workers. Environ Health Perspect 101, 13–21. Smith, T. J., Quinn, M. M., Marsh, G. M., Youk, A. O., Stone, R. A., Buchanich, J. M., and Gula, M. J. (2001). Historical cohort study of US man-made vitreous fiber production workers: VII. Overview of the exposure assessment. J Occup Environ Med 43, 809–823. Sohn, M. D., McKone, T. E., and Blancato, J. N. (2004). Reconstructing population exposures from dose biomarkers: Inhalation of trichloroethylene (TCE) as a case study. J Expo Anal Environ Epidemiol 14, 204–213. Spencer, J. W., and Plisko, M. J. (2007). A comparison study using a mathematical model and actual exposure monitoring for estimating solvent exposures during the disassembly of metal parts. J Occup Environ Hyg 4, 253–259. Stanek, E. J., and Calabrese, E. J. (1995). Soil ingestion estimates for use in site evaluation based on the best tracer method. Hum Ecol Risk Assess 1, 133–156. Stewart, A., Blair, A., Cubit, D., Bales, R., Kaplan, S., Ward, J., Gaffey, W., O’Berg, M., and Walrath, J. (1986). Estimating historical exposures to formaldehyde in a retrospective mortality study. Appl Ind Hyg 1, 34–41. Stewart, P., and Stenzel, M. (2000). Exposure assessment in the occupational setting. Appl Occup Environ Hyg 15, 435–444. Stewart, P. A., Carel, R., Schairer, C., and Blair, A. (2000). Comparison of industrial hygienists’ exposure evaluations for an epidemiologic study. Scand J Work Environ Health 26, 44–51. Stewart, P. A., Lees, P. S., Correa, A., Breysse, P., Gail, M., and Graubard, B. I. (2003). Evaluation of three retrospective exposure assessment methods. Ann Occup Hyg 47, 399–411. Stewart, P. A., Lees, P. S., and Francis, M. (1996). Quantification of historical exposures in occupational cohort studies. Scand J Work Environ Health 22, 405–414. Stewart, P. A., Zaebst, D., Zey, J. N., Herrick, R., Dosemeci, M., Hornung, R., Bloom, T., Pottern, L., Miller, B. A., and Blair, A. (1998). Exposure assessment for a study of workers exposed to acrylonitrile. Scand J Work Environ Health 24(Suppl 2), 42–53. Suh, H., Koutrakis, P., and Ebelt, S. (2004). Detailed characterization of indoor and personal particulate matter concentrations. Final Report. Contract No. 00-302. Prepared for the California Air Resources Board Research Division, Sacramento, CA, by the Harvard School of Public Health, Boston, MA. ten Berge, W. F. (2008). Homepage of Wil ten Berge, Vol. 2008. Tomerlin, J. R., Berry, M. R., Tran, N. L., Chew, S. B., Petersen, B. J., Tucker, K. D., and Fleming, K. H. (1997). Development of a dietary exposure potential model for evaluating dietary exposure to chemical residues in food. J Expo Anal Environ Epidemiol 7, 81–101. Tran, N. L., Locke, P. A., and Burke, T. A. (2000). Chemical and radiation environmental risk management: Differences, commonalities, and challenges. Risk Anal 20, 163–172. US Centers for Disease Control and Prevention, National Center for Environmental Health (US CDCP) (2008a). National Health and Nutrition Examination Survey (NHANES), Vol. 2008. US Centers for Disease Control and Prevention (US CDCP), National Institute for Occupational Safety and Health (2008b). NIOSH Program Area—OCAS—Technical Documents Used in Dose Reconstruction, Vol. 2008, p. 2008. US Environmental Protection Agency (US EPA) (1986). Final report: Formaldehyde exposure model— Description and demonstration. US Environmental Protection Agency (US EPA) (1989). Risk assessment guidance for Superfund. In Human Health Evaluation Manual (Part A), Vol. 1, Interim Final, Office of Emergency and Remedial Response, Washington, D.C. US Environmental Protection Agency (US EPA) (1992a). Guidelines for exposure assessment: Notice. Fed Reg 57, 22888–22938. US Environmental Protection Agency (US EPA) (1992b). Supplemental guidance to RAGS: Calculating the concentration term. US Environmental Protection Agency (USEPA), Office of Solid Waste and Emergency Response, Washington, D.C.
REFERENCES
783
US Environmental Protection Agency (US EPA) (1995). Guidance for risk characterization. Science Policy Council, Washington, D.C. US Environmental Protection Agency (US EPA) (1996a). Draft guidelines for carcinogenic risk assessment. Fed Reg 61, 17960–18011. US Environmental Protection Agency (US EPA) (1996b). Proposed guidelines for carcinogen risk assessment. Office of Research and Development, Washington, D.C. US Environmental Protection Agency (US EPA) (1997). Guiding principles for Monte Carlo analysis. Risk Assessment Forum, Office of Research and Development, Washington, D.C. US Environmental Protection Agency (US EPA) (1999a). Review of revised sections of the proposed guidelines for carcinogen risk assessment. Science Advisory Board, Washington, D.C. US Environmental Protection Agency (US EPA) (1999b). Risk assessment guidance (RAGS3A) for conducting probabilistic risk assessment. US Environmental Protection Agency (USEPA), Washington, D.C. US Environmental Protection Agency (US EPA) (2000a). Human health risk assessment for diazinon. US Environmental Protection Agency (USEPA), Office of Pesticide Programs Health Effects Division, Washington, D.C. US Environmental Protection Agency (US EPA) (2000b). Options for development of probability distributions for exposure factors. US Environmental Protection Agency (USEPA), Office of Research and Development, National Center for Environmental Assessment, Washington, D.C. US Environmental Protection Agency (US EPA) (2000c). Science Policy Council Handbook, Office of Science Policy, Office of Research and Development, Washington, D.C. US Environmental Protection Agency (US EPA) (2003a). Dietary exposure potential model: A model using extant food databases to estimate dietary exposure to chemical residues. Available from http:// www.epa.gov/nerlcwww/depm.htm. (accessed October 17, 2008). US Environmental Protection Agency (US EPA) (2003b). SHEDS-PM: A population exposure model for predicting distributions of PM exposure and dose from both outdoor and indoor sources. Available from http://cfpub.epa.gov/si/si_public_record_Report.cfm?dirEntryID=63055 (accessed October 17, 2008). US Environmental Protection Agency (US EPA) (2005). Ritualistic use of mercury simulation: A preliminary investigation of metallic mercury vapor fate and transport in a trailer. Office of Solid Waste and Emergency Response. Washington, DC, EPA/540/-04/006. US Environmental Protection Agency (US EPA) (2009). Risk assessment guidance for Superfund (RA65), volume 1: Human health evaluation manual (Part F: Supplemental guidance for inhalation risk assessment). Office of Research and Development, Washington, DC. van Tongeren, M. J., Barker, R. D., Gardiner, K., Harris, J. M., Venables, K. M., Harrington, J. M., and Newman Taylor, A. J. (1998). Retrospective exposure assessment for a cohort study into respiratory effects of acid anhydrides. Occup Environ Med 55, 692–696. van Veen, M. P. (2001). CONSEXPO 3.0 Consumer exposure and uptake models. National Insitute of Public Health and the Environment, Research for Man and Environment, Bilthoven, the Netherlands. van Wendel de Joode, B., van Hemmen, J. J., Meijster, T., Major, V., London, L., and Kromhout, H. (2005). Reliability of a semi-quantitative method for dermal exposure assessment (DREAM). J Expo Anal Environ Epidemiol 15, 111–120. Velazquez, S. F., McGinnis, P. M., Vater, S. T., Stiteler, W. S., Knauf, L. A., and Schoeny, R. S. (1994). Combination of cancer data in quantitative risk assessments: Case study using bromodichloromethane. Risk Anal 14, 285–292. Vermeulen, R., and Kromhout, H. (2005). Historical limitations of determinant based exposure groupings in the rubber manufacturing industry. Occup Environ Med 62, 793–799. Vermeulen, R., Li, G., Lan, Q., Dosemeci, M., Rappaport, S. M., Bohong, X., Smith, M. T., Zhang, L., Hayes, R. B., Linet, M., Mu, R., Wang, L., Xu, J., Yin, S., and Rothman, N. (2004). Detailed exposure assessment for a molecular epidemiology study of benzene in two shoe factories in China. Ann Occup Hyg 48, 105–116. Vernez, D., Bruzzi, R., Kupferschmidt, H., De-Batz, A., Droz, P., and Lazor, R. (2006). Acute respiratory syndrome after inhalation of waterproofing sprays: A posteriori exposure–response assessment in 102 cases. J Occup Environ Hyg 3, 250–261.
784
CHAPTER 30
EXPOSURE RECONSTRUCTION AND CANCER RISK ESTIMATE DERIVATION
Vernez, D. S., Droz, P. O., Lazor-Blanchet, C., and Jaques, S. (2004). Characterizing emission and breathing-zone concentrations following exposure cases to fluororesin-based waterproofing spray mists. J Occup Environ Hyg 1, 582–592. Viet, S. M., Stenzel, M., Rennix, C., Armstrong, T., and Couch, J., eds. (2008). Guideline on Occupational Exposure Reconstruction, AIHA, Fairfax, VA. von Grote, J., Hurlimann, C., Scheringer, M., and Hungerbuhler, K. (2003). Reduction of occupational exposure to perchloroethylene and trichloroethylene in metal degreasing over the last 30 years: influences of technology innovation and legislation. J Expo Anal Environ Epidemiol 13, 325–340. von Grote, J., Hurlimann, C., Scheringer, M., and Hungerbuhler, K. (2006). Assessing occupational exposure to perchloroethylene in dry cleaning. J Occup Environ Hyg 3, 606–619. Vose, D. (1996). Quantitative risk analysis: A Guide to Monte Carlo Simulation Modeling. John Wiley & Sons, New York. Walker, K. D., Catalano, P., Hammitt, J. K., and Evans, J. S. (2003). Use of expert judgment in exposure assessment: Part 2. Calibration of expert judgments about personal exposures to benzene. J Expo Anal Environ Epidemiol 13, 1–16. Walker, K. D., Evans, J. S., and MacIntosh, D. (2001). Use of expert judgment in exposure assessment. Part I. Characterization of personal exposure to benzene. J Expo Anal Environ Epidemiol 11, 308–322. Wartenberg, D., Uchrin, C., and Coogan, P. (1991). Estimating exposure using kriging: A simulation study. Environ Health Perspect 94, 75–82. Weisel, C., Yu, R., Roy, A., and Georgopoulos, P. (1996). Biomarkers of environmental benzene exposure. Environ Health Perspect 104(Suppl 6), 1141–1146. Williams, P. R., Panko, J. M., Unice, K., Brown, J. L., and Paustenbach, D. J. (2008). Occupational exposures associated with petroleum-derived products containing trace levels of benzene. J Occup Environ Hyg 5, 565–574. Williams, P. R., and Paustenbach, D. J. (2002). Risk characterization: Principles and practice. J Toxicol Environ Health B Crit Rev 5, 337–406. Williams, P. R., and Paustenbach, D. J. (2003). Reconstruction of benzene exposure for the Pliofilm cohort (1936–1976) using Monte Carlo techniques. J Toxicol Environ Health A 66, 677–781. Williams, P. R., and Paustenbach, D. J. (2005). Characterizing historical industrial hygiene data: A case study involving benzene exposures at a chemical manufacturing facility (1976–1987). J Occup Environ Hyg 2, 341–350. Williams, P. R. D., Holicky, K. C., and Paustenbach, D. J. (2003). Current methods for evaluating children’s exposures for use in health risk assessment. J Childrens Health 1, 41–98. Wilschut, A., ten Berge, W. F., Robinson, P. J., and McKone, T. E. (1995). Estimating skin permeation. The validation of five mathematical skin permeation models. Chemosphere 30, 1275–1296. World Health Organization (2005). Principles of Characterizing and Applying Human Exposure Models, World Health Organization, Geneva. Yokley, K., Tran, H. T., Pekari, K., Rappaport, S., Riihimaki, V., Rothman, N., Waidyanatha, S., and Schlosser, P. M. (2006). Physiologically-based pharmacokinetic modeling of benzene in humans: A Bayesian approach. Risk Anal 26, 925–943. Yu, R., Tan, W., Matthew, R., Andjelkovich, D. A., and Levine, R. (1990). A deterministic mathematical model for quantitative estimation of historical exposure. Am Ind Hyg Assoc 51, 194–201. Zartarian, V. G., Xue, J., Ozkaynak, H., Dang, W., Glen, G., Smith, L., and Stallings, C. (2006). A probabilistic arsenic exposure assessment for children who contact CCA-treated playsets and decks, Part 1: Model methodology, variability results, and model evaluation. Risk Anal 26, 515–531.
INDEX Note: Page numbers in italics refer to Figures; those in bold to Tables. absorption, distribution, metabolism, and excretion (ADME), of carcinogens, 558 absorption rate, and vehicle-dependent effects, 74 accelerator mass spectrometry, radiolabeling method associated with, for measuring DNA adducts, 317, 319, 321, 323 “acceptable” risk, in regulatory decisionmaking, 24 acetone, simulation studies of, 744 acetylaminofluorene (AAF), 423 dose-response study of, 197 metabolic activation of, 170, 171 acinar cell tumors, pancreatic, 440 acrylonitrile, carcinogenic classification of, 413 ACToR database, 606 acute apoptosis, after exposure to PPARα activators, 444. See also apoptosis acute lymphocytic leukemia (ALL), incidence of, 400, 401 acylating agents, SAR analysis of, 524 ADD. See average daily dose additives, 78. See also food additives “additivity to background” concept, 669 adenocarcinomas, mammary, 125 adenoma, in thyroid C-cell carcinomas, 701 adenomatous polyposis coli (APC) protein product, 140 administration time, in gene mutation assays, 340 adnexa, combining neoplasms of, 709 adrenal gland, classifying neoplasms of, 708
adverse effect, threshold for, 621n. adversity, reevaluation of, 390 aerosol particles, estimating exposures of, 754 aerosols, in animal inhalation studies, 72 aflatoxin, cancer associated with, 103 aflatoxin B1 chemical structure of, 425 linear dose-response of, 629 metabolic activation of, 172 age and apoptosis, 152 and cancer incidence, 126, 398–399, 399, 401 and chemical exposure, 414–415 age-dependent adjustment factor (ADAF), 568 Agency for Toxic Substances and Disease Registry (ATSDR), US, and reconstructing exposure scenarios, 740–741, 766, 770 age-period-cohort (APC) models, 643 aging, theories of, 153, 154 Agriculture, Dept. of (USDA), US, 77 Air Resources Board (ARB), California, 73 air standards for asbestos, 83 regulatory considerations, 72–73 scientific issues, 70–72 Air Toxics Hot Spots Program, 73 aldehydes, SAR analysis of, 523 aldrin/dieldrin, cancellation of, 4 alimentary tract tissue, combining neoplasms of, 709, 710 alkylating agents DNA reactive, 234 mutagenicity of, 627–628 alkyl esters, SAR analysis of, 522–523
Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright © 2010 John Wiley & Sons, Inc.
785
786
INDEX
α2U-globulin (α2U-g) hypothesis, 486 α2U-globulin (α2U-g) nephropathy and CPN, 493 human relevance of study of, 495–496 and RTT development, 487 α2U-globulin (α2U-g) protein, 482 function of, 483 and hyaline droplets, 484 and increased incidence of RTT, 493–495 in renal carcinogenesis, 485–486 and renal tubule tumors, 483 alpha-2u-nephropathy, SAR of, 533–534 altered foci cells (AFCs), 106 American Conference of Governmental Hygienists, 743 American Industrial Hygiene Association, 743 Ames test, 228, 233, 234 appearance of Ames plates, 275, 275 basis of, 274 in ICH guidelines, 244–245, 260 limitations of, 239 in REACH policy, 254 sensitivity of, 276 in transgenic animals, 263 4-aminobiphenyl, key events in carcinogenicity of, 372, 372, 374, 374 2-amino-3,8-dimethylimidazo, carcinogenicity bioassay of, 207 2-amino-3,8-dimethylimidazo[4,5-f] quinoxaline (MeIQx), low-dose carcinogenicity of, 209–214, 210, 210, 211, 212, 213, 214 2-amino-1-methyl-6 phenylimidazo[4,5-b] pyridine (PhIP) carcinogenicity bioassay of, 207 low-dose carcinogenicity of, 215–216, 216, 217 analogy and causal association, 409 and judging causation, 196 androgens, and hepatic tumor risk, 430 aneugenic effects, compared with clastogenic effects, 306 aneugenicity, REACH definition of, 253 aneugens detection of, 280 nondisjunction of chromosomes induced by, 283
aneuploidy, 253, 280 animal data, to estimate human potency, 70 animal inhalation studies, 71–72 animal models use of data from, 14–15 (see also specific animal models) in in vivo genotoxicity assays, 300–301, 302–303 animal studies, interpretation of, 716 animal testing, minimizing, 392 anoikis, in carcinogenesis, 121–122 anoxia, and in vivo testing, 263 antioxidant defense, 431 antioxidants, endogenous, 661–662 Apaf-1 gene, 112 apoptosis, 662 acute, 444 and age, 152 cellular, 108, 108 in human diseases, 109 and PPARα activator suppression of, 448–449 processes, 110, 110 and species differences in PPARα activation, 457–458 sum total of, 109 tipping point in, 111 apoptosis resistance of cancer cells, 111–112 and p53 gene, 112 apoptosome, formation of, 111 Aprt assay, 329, 331, 333 apurinic sites, in PAH-induced cancer process, 179–180 area-under-the-blood-concentration (AUC) curve, 768 arl hydrocarbon receptor (AhR), 666 aromatic amines, 234, 423 and bladder cancer, 170, 501 SAR of, 527–528 arsenic, 84 carcinogenic classification of, 412 exposure reconstruction for, 761 hormesis review criteria for, 196 Monte Carlo analysis of, 765 PHGs for, 76 simulation studies of, 744 arylating agents, SAR analysis of, 524 aryl hydrocarbon receptor (AhR), in rodent hepatocarcinogenesis, 428, 428–429
INDEX
asbestos carcinogenic classification of, 412 concentration-dependent effects of, 74 documenting exposure to, 70 expert exposure studies on, 756, 757 exposure reconstruction for, 763 regulation of, 83 simulation studies of, 744 asbestos fiber exposures, JEM for, 748, 749 ATSDR, US. See Agency for Toxic Substances and Disease Registry, US atypical tubule hyperplasia, 482 autoimmune disease, IDS-balance in, 124 autophagy, imperfect, 112 average daily dose (ADD), 767 AXIN1 gene protein product, 140 azinphosmethyl, exposure reconstruction for, 763 azo dyes, 423 azoxymethane (AOM), colon tumors treated with, 216, 217 bacteria, gene mutation testing in, 273–276, 275, 276 bacterial cystitis, 511 bacterial reverse mutation (Ames) test, in ICH guidelines, 260. See also Ames test base excision repair (BER), and UDS test, 324 basophils, differentiation of, 132–133, 133, 134 Bayesian data analysis, 756, 760 Bayesian decision analysis, 770 bay region, on PAHs, 173, 174, 175, 177 BBP, chemical structure of, 531 BCME. See bis(chloromethyl) ether BDII rats 4-DAST study, 53, 53–54 DENA study, 53, 54, 54 BDIII rats, 4-DAB study in, 51, 51 BD IX rats, EMI study in, 55, 57 benchmark dose (BMD), 671–672, 672 benchmark dose (BMD) approach, in carcinogenic risk assessment, 41 benchmark dose lower bound (BMDL), 671–672, 672 benefit-risk assessment, for pharmaceuticals, 259
787
benign categorization as, 702–703 defining term, 701 benzene, 83 biological monitoring of, 747 carcinogenic classification of, 412 documenting exposure to, 70 estimating exposures of, 754 expert exposure studies on, 756, 758 exposure reconstruction for, 761, 762, 763 as hazardous air pollutant, 82 JEM for, 749, 750 Monte Carlo analysis of, 765 NTP study of, 80 simulation studies of, 744 and in vivo testing, 264 benzidine, carcinogenic classification of, 412 benzo[a]pyrene (B[a]P) identification of, 169 metabolic activation of, 173, 175 benzo[a]pyrene diol epoxides (BPDE) deoxyguanosine adducts, 175, 176 and DNA stereoisomeric adducts, 174–175, 175 metabolic activation of B[a]P to, 173, 175, 182 BER. See base excision repair Berenblum, Isaack, 103 beryllium, carcinogenic classification of, 413 “best available technology” approaches, 24 β-catenin gene, 60 β-catenin protein, 114, 140–141 bezafibrate, chemical structure of, 531 Big Blue® mice, 631 Big BlueTM models, in vivo gene mutation assays in, 334, 336, 338, 341 Big Blue® Rats, GST-P positive foci in, 211 Billroth, Theodor, 1027 Biocidal Products Directive (BPD), EU, 49–50 biocides defined, 48 regulatory schemes for EU Biocidal Products Directive, 49–50 US Federal Insecticide, Fungicide, and Rodenticide Act, 48–49
788
INDEX
bioindicators, using, 369 bioinformatics, 84, 590, 606 Biological Environmental Exposure Limit (BEEL), 743 Biological Exposure Indices (BEIs), 743 biological gradients and causal association, 409 in judging causation, 195 biologically-based dose-response (BBDR), 576 biologically based dose-response (BBDR) modeling, 616, 666, 674–675 biological monitoring, in exposure reconstruction, 743, 745, 746–747 biological plausibility, and judging causation, 195 biology, systems, 606–609 biomarkers classification of, 406 distinguished from bioindicators, 369 in molecular epidemiology, 415 to monitor cancer progress, 155 myofibroblasts, 113 Biomonitoring Equivalents (BE) Pilot Project, 743 biotin-dUTP, used in SCE, 327 bis(chloromethyl) ether (BCME), carcinogenic classification of, 412 black carbon, estimating exposures of, 752 bladder, urinary, classifying neoplasms of, 707 bladder cancer and aromatic amines, 170 causative factors for, 501 and multistage carcinogenesis, 183–184 in rat, 672 risk assessment for, 511–512 and urinary solids, 505, 510 bladder tumors, and urinary solids, 502– 505, 503, 504 blood tests, in experimental studies, 381 BN rats, MeIQx carcinogenicity in, 213 body dose, 737 body weight, in experimental studies, 381 bone marrow cell differentiation in, 132–133, 133 mammalian erythrocyte micronucleus test with, 304–306 bone marrow chromosome aberration test, 308–310 bone tissue, combining neoplasms of, 709
“boosting” rules, in SAR analysis, 525 Boveri, Theordor Heinrich, 102 BRCA1 gene, 135–136 BrdU labeling index, 672 breast cancer genetic mutations associated with, 403 risk for, 402 breast cancer 1 gene (BRCA1), 135–136 Burnet, Mac, 124 “busting” rules, in SAR analysis, 525–526 1,3-butadiene and affected-animal count, 722, 723, 724 animal MOA for, 372, 372, 375 carcinogenic classification of, 412 carcinogenicity of, 719 carcinogenic potency estimates for, 726, 726 distribution-based methods for, 724, 725, 726, 726–730, 727–729, 731, 732 estimating combined potency for, 721 exposure reconstruction for, 762 JEM for, 750 multisite time-dependent analysis of tumor incidence in mice exposed to, 730, 731 potency estimates for exposure to, 732, 733 source data for, 721–722, 722, 723 survival of mice exposed to, 721, 723 total tumor incidences in mice exposed to, 722, 723 tumor incidences in mice exposed to, 721, 722, 724, 725 “butter yellow” (4-DAB), 50 butylated hydroxyl toluene, JEM for, 750 cachexia, and Warburg’s hypothesis of GLU mismanagement, 150–151 cadmium biological monitoring of, 747 carcinogenic classification of, 413 calcium and formation of urinary solids, 509 in urine, 506 calcium stearate, JEM for, 750 calculi, urinary, and bladder cancer, 510. See also urinary solids California, PHG for carcinogens in, 76 California Air Toxics program, 73 California Environmental Protection Agency, OEHHA of, 720
INDEX
California Safe Drinking Water and Toxic Enforcement Act (1986), 84 Canadian Environmental Protection Act (CEPA), 389, 391 Canadian National Dose Registry (CNDR), 650 cancer. See also bladder cancer apoptosis in, 111 biology of history, 99–103 inflammation in, 124–130 mechanisms, 103–105 deaths caused by, 397 epidemiology of, 397 experimental reproduction of, 103 growth, 138–139, 139 (see also carcinogenesis) incidence of, 152, 399 irritation theory of, 169 mechanistic data for, 387 nongenetically induced, 122 occupationally associated with, 397 prevalence of, 3 prevention of diet-related, 78 risk factors for, 398 stages of, 117, 118, 129–130 stem cell biology and, 130–139, 132–, 139 substances causally associated with, 411–414, 412–413 as two-stage process, 200 cancer-associated fibroblasts (CAFs), 113 cancer dose, ten percent of (CD10) defined, 684 extrapolation between species, 686, 687–690, 691, 691, 691–692, 694 extrapolation of intraspecies variation in, 692, 693, 693, 694 uses for, 693–694 and variation within species, 684–686, 685, 686, 694 cancer potency, procedure for calculating, 84. See also potency cancer-predictive tests, 228. See also genetic toxicity testing cancer process. See also carcinogenesis initiation, 420, 420–422, 421 progression in, 420, 421, 422–423 promotion in, 420, 421, 422 cancer registry data, analysis of, 643–344 cancer response, 558. See also extrapolation
789
cancer risk background of, 771–772 calculating, 87 lifetime, 683 potency estimates for 1,3-butadiene exposure in, 733 potential bladder, 512 proliferation of normal cells in, 200 cancer risk assessment. See also risk assessment defaults in, 67 defining steps in, 9–10 DMA, 672 dose-response relationships in, 50–60, 51, 53, 54, 56, 57, 57, 59, 60 historical exposure values for, 766–767, 770, 771 dose, 767–768 risk, 768–769 historical perspectives on, 4–9, 6–8 hormetic information in, 663 human relevancy in, 595 linearity assumption in, 668–669 and low-dose extrapolation, 669–670 mathematical model for, 60–61 overall aim of, 363 PBPK models in, 563, 563–565, 564 for chloroform, 576–577 for dichloromethane, 574, 574–576 for dioxane, 577–578 extrapolation from individual carcinogens, 571–574, 573 high-dose to low-dose and interspecies extrapolation, 565–568, 566, 567 intraspecies extrapolation, 568–570, 570 route-to-route extrapolation, 571 for trichloroethylene, 578 for vinyl chloride, 575–576 for volatile organic chemical mixtures, 578–579, 579 predictive modeling in, 609 principles and practices, 69 process of, 6–8, 557 (Q)SAR in, 544, 545, 546–547 quantitative, 66 traditional, 659–660 uncertainties in, 684 in US, 3 WOE in, 615
790
INDEX
cancer slope factor (CSF), 557, 768 cancer stem cell (CSC), 135. See also stem cells capillary electrophoresis, for measuring DNA adducts, 317, 319, 321, 323 captan, cancer evaluation of, 673 carbonaceous particulates, JEM for, 749 carbon black, JEM for, 749 carbon dioxide (CO2), in Warburg effect, 148, 149 carbon monoxide (CO), exposure reconstruction for, 763 carbon tetrachloride, JEM for, 749 Carcinogen Assessment Group (CAG), EPA, 18 carcinogen-DNA adduct formation, at level of DNA mutation, 208 carcinogenesis, 419. See also hepatocarcinogenesis changing views on, 676 clonal aspects of, 116–123, 118 clonal models of, 137, 138, 661 dynamic evolution model of, 138 EPA evaluation of, 4 events of, 142 experimental studies on, 393 4-aminobiphenyl-induced, 374, 374 without genetic positive findings, 145 genetic susceptibility to, 17 genomics of, 595 hormesis in, 663–664, 664 irritation theory of, 102 multiple mutations in, 676 multistage, 182–185, 183 nongenetic events in, 121 nonlinear hormesis, 663–664, 664 post-DNA damage mechanisms, 662–663 pre-DNA damage mechanisms, 661–662 parallel evolution model of, 137 “point of no return” in, 152 potential for multistage, 619 process of, 636 reversal of, 131 role for α2U-g nephropathy in, 485–486 stages of, 117, 118, 521 stem cell model of, 138, 142
telomeres in, 153 thyroid, 532–533 urinary solids in, 508–510 velocity of, 56 carcinogenesis, chemical biological initiation of, 147–152, 149, 150, 151 epigenetic, 172 and genetic toxicology testing, 225 history of, 168–169 metabolic activation theory of, 170 multistage, 185 stages of, 182 carcinogenic agents. See also specific agents effects on cancer risk of, 655 investigation of, 655 carcinogenicity CD10 measure of, 684 determination of, 5 experimental studies on, 378 genomics in, 85 genotoxicity tests for, 259–260 human, 15–16 IARC classification of, 39 non-threshold concept of genotoxic, 208 route-specific, 71 The Carcinogenicity Potency Database (Kirkland et al.), 241 carcinogenicity studies design of, 379, 379–380, 381 perinatal, 382 technical adequacy of, 384 carcinogenicity studies, rodent, 699 combining neoplasms in, 700 criteria for, 704–710, 705–709 rationale for, 701–702, 711 differentiating benign from malignant neoplasms in, 702–703 carcinogenicity testing, long-term, 209. See also specific tests Carcinogenic Potency Database (CPDB), 682, 684–685 carcinogen identification, determining statistical significance for, 718 carcinogens acceptable risk for, 3 ADME of, 558 bladder, 501
INDEX
chemical at initiation stage, 421 (Q)SAR studies of, 520 classification of, 38–40 classified into initiators, 636 complete, 183 direct-acting, 621 DNA-reactive, 664–666 dose-response relationships in, 50–60, 51, 53, 54, 56, 57, 57, 59, 60 in foods, 76–77 genotoxic, 78, 171, 171–172 inhibition of, 219 potassium bromate, 216, 218, 218–219 SAR analysis of, 522–528 thresholds for, 209 hepatic classification of, 423, 423 genotoxic agents, 423, 424, 424–425, 425 nongenotoxic, 425–433, 426, 428, 429, 431 HRF applied to, 373 human, 411, 412–413, 414 inhalation exposure to, 70 IRIS classification of, 411, 412–413, 414 limiting exposure to, 617 linear low-dose extrapolation for, 619 low-dose linearity of, 675 lung, expert exposure studies on, 757 mixtures of, 573–574 nongenotoxic, 58, 85, 233, 272 epidemiological data for, 643 MSCE model of, 640–642, 641 SAR of, 528–534 TSCR model, 637–640, 638 nonlinear extrapolation for, 677 nonmutagenic, and nonlinearity, 666–667 PHG for, 76 “potency” of, 682 public health impact of, 5–6 in rodent studies, 700 sites for, 716 threshold of toxicity for, 660 ultimate, 208, 219 viruses, 402 carcinogens, mutagens, or reproductive toxicants (CMRs), classification as, 44, 45
791
Carcinogens Assessment Group (CAG), 4 carcinomas, hepatatocellular. See hepatatocellular carcinoma cardiovascular system, classifying neoplasms of, 706 cartilage, combining neoplasms of, 709 case-control studies in epidemiology, 13 goal of, 405 case reports, in epidemiological studies, 13, 407 caspases activation of, 110 in apoptosis, 110 casual models, 191–192 causality criteria for, 410 determining, 408–410 cell biology, surveillance systems in, 126 cell biology of cancer chemical carcinogenesis, 147–152, 149, 150, 151 clonal aspects of, 116–123, 118 epigenetic, 142–147 facilitation of supporting cells in, 113–116 inflammation in, 124–130 programmed cell removal, 107–112, 108, 110 specific gene sets in, 139–140 notch signaling pathway, 141–142 polycomb group, 140 sonic hedgehog, 140 Wnt pathway, 140–141 stem cells, 130–139, 132–, 139 in vitro systems, 105–107 cell-cycle delay, 662 cell death, in comet assay, 285 cell proliferation after exposure to PPARα activators, 444 in CPN, 490 and folic acid supplementation, 653 and multiple mutations, 676 receptor-mediated, 528–529 SAR of, 528–529 and species differences in PPARα activation, 457–458 and tumorigenesis, 667 cell theory, origins of, 101–102
792
INDEX
cellular adhesion molecules (CAMs), 113, 114 cellular differentiation, 662 Center for Disease Control and Prevention (CDC), US, National Center for Environmental Health of, 743 cervix, classifying neoplasms of, 708 CFR. See Code of Federal Regulations chaotropic agents, in carcinogenesis, 149 Chemical Assessment and Management Program (ChAMP), US EPA’s, 45–46 Chemical Industries Institute of Toxicology (CIIT), α2U-g protein study of, 486 chemicals. See also carcinogens in drinking water exposure to, 75 lifetime cancer risks of, 76, 77 experimental tumor promoter, 119 without genetic positive findings, 145 genotoxic and nongenotoxic, 630–631 multiple MOAs of, 631 pre-market genetic toxicity testing of, 226 testing for carcinogenicity of, 67 toxic, and analytical capabilities, 66 chemicals, industrial regulatory schemes for REACH, 44–45 US Toxic Substances Control Act, 42–44 voluntary initiatives for evaluating Chemical Assessment and Management Program of US EPA, 45–46 High Production Volume Chemicals Challenge Program, 46 OECD’s investigation of HPV chemicals, 46–47 Voluntary Children’s Chemical Evaluation Program of US EPA, 47–48 Chemical Safety Assessment (CSA), under REACH, 254 Chemical Safety Report (CSR), under REACH, 44 chemical screening, 15 chemicals industry, European, 252 chemokines, role in inflammation of, 125 chemopreventive agent, folate as, 653
childhood, early-life exposures associated with, 568 children exposure regulations for, 80 exposure to soil of, 81 Child-Specific Exposure Factors Handbook (USEPA), 8 Chinese hamster cells gene mutation testing in, 277, 279 in micronucleus analysis, 281 Chinese tin miners, analysis of lung cancer in, 650, 651, 652 chlordane/heptachlor, cancellation of, 4 chlorinated byproducts in drinking water, US EPA regulation for, 28 chloroform cancer evaluation of, 673 EPA’s assessment of, 85 PBPK models for, 576–577 chloromethyl methyl ether (CMME), carcinogenic classification of, 412 chloroprene, exposure reconstruction for, 761 chromated copper arsenate (CCA), control of, 84 chromatin, in epigenetic biology, 143 chromium carcinogenic classification of, 412 expert exposure studies on, 756, 757 JEM for, 750 chromium, hexavalent, 75 cancer risk of, 771 documenting exposure to, 70 multiroute exposure to, 71 simulation studies of, 744 chromosomal damage (clastogenicity) mammalian erythrocyte micronucleus test for, 304 in vitro testing for, 279, 279–280 in in vivo genotoxicity assays, 294–295 chromosome aberration test bone marrow, 294, 296, 298, 300 interpretation of, 309 limitations of, 309–310 principle of, 308 purpose of, 308 regulatory acceptance of, 308 study design for, 308–309 in genetic toxicity testing, 231 in REACH policy, 254
INDEX
chromosome breakage (clastogenicity), 227 chromosomes discovery of, 102 micronucleus analysis of, 281 chromosome theory, Boveri-Sutton, 102 chronic progressive nephropathy (CPN), 482 early exacerbation of, 484 grading severity of, 490, 490 human relevance of study of, 495–496 impact of MTBE on, 494 incidence of, 489 and increased incidence of RTT, 493–495 neuropathy (α2U-g) associated with, 493 and renal carcinogenesis, 489–490 RTT associated with exacerbation of, 491 cigarette smoke particles, exposure reconstruction for, 762 cigarette smoking, and bladder cancer, 501. See also smoking cII lambda phage gene, in vivo gene mutation assays for, 335, 337, 339 ciprofibrate chemical structure of, 531 in MOA analysis after PPARα activation, 444 clastogenic effects, aneugenic effects compared with, 306 clastogenicity (chromosome breakage), 227. See also chromosomal damage Clean Air Act, US (CAA) (1970), 24, 72, 83 clofibrate chemical structure of, 531 in MOA analysis after PPARα activation, 444 clonal dominance model, of carcinogenesis, 138 clonal expansion, in PPARα activator MOA, 451 clonal model of carcinogenesis, 661 clonal selection model, of carcinogenesis, 137, 138 CMME. See chloromethyl methyl ether coal tar, in history of carcinogenesis, 168–169 Cochran-Armitage test, 382 Code of Federal Regulations (CFR), 49
793
coherence and causal association, 409 and judging causation, 195 cohort studies, 13, 403–405, 404, 651 coke oven emissions, carcinogenic classification of, 412 colon cancer genetic mutations associated with, 403 incidence of, 137 and infection, 124 and multistage carcinogenesis, 184 Colorado Plateau Uranium Miners (CPUM), analysis of lung cancer in, 650, 651 colorants, NTP study of, 80 colorectal cancer, folate and, 653, 654 colorectal tumors clonal composition of, 137 and DCC gene, 113–114 combined chronic/cancer bioassay, for regulatory risk assessment, 379 comet assay, 262, 294, 296, 298, 300 advantages of, 313 interpretation of, 312–313 limitations of, 314 principle of, 311 purpose of, 311–314 regulatory acceptance of, 311 study design for, 311–312 Commission for Investigation of Health Hazards of Chemical Compounds in the Work Area (MAK Commission), German, 670 Committee of Human Medicinal Products (CHMP), 257 Committee to Coordinate Toxicology and Related Programs, DHEW, 238 Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA) (1980), 81–82 computational biology, and risk assessment, 16–17 computational toxicology historical perspective on, 598–599 pharmacodynamic and dose-response modeling in, 602 pharmacokinetic variability and uncertainty, 601–602 quantitative risk assessment in, 599–601
794
INDEX
computational toxicology (cont’d) and risk assessments, 602–603 analysis, interpretation, and informatics, 606 high-throughput testing, 604–606 21st-Century, 603 systems biology, 606–609 conazoles, gene expression profile for, 592–593 conceptus, germinal layers in developing, 130 connective tissue, combining neoplasms of, 709 conservative assumptions, application of, 33–34 CONSEXPO model, 753, 755 consistency, and causal association, 408 constitutive androstane receptor (CAR) in rodent hepatocarcinogensis, 429, 429–430 Consumer Product Safety Commission (CPSC), US, on lifetime cancer risk, 683 consumer product standards regulatory considerations, 83–84 scientific issues, 82 Contaminants of Potential Concern (COPC), of World Trade Center Indoor Air Task Force Working Group, 8 control measures, development of, 383 Cori cycle, 151, 151–152 cost-benefit analyses and public health policy, 677 in regulatory decision-making, 24 courts and conservative assumptions, 33–34 new methodology in, 32–33 risk assessment controversy in, 29–30 COX enzymes, role in infection of, 127–128 COX genes, role in infection of, 127–128 creosote, carcinogenic classification of, 413 Crick, Francis, 146 cross-sectional (prevalence) studies, 406 croton oil, 104 Crystal Ball add-in software, 729 crystallization, of urine, 507 cyclohexane, estimating exposures of, 754
cyclooxygenase-2 (COX-2), overexpression of, 120 cyclopenta-ring oxidation mechanism, in metabolic activation of PAH, 181–182 cyclosporine, Monte Carlo analysis of, 765 cynomolgus monkey, response to PPARα activators of, 458, 459 CYPs, in in vitro testing, 273, 274 cystitis, bacterial, 511 cytochalasin B, for in vitro micronucleus assay, 246–247, 281, 282 cytochrome P450 enzymes genotoxic metabolites and, 68 reactive intermediates activated by, 662 cytokines, cancer inflammation propagated by, 127 cytotoxicants, hepatocyte, 433 cytotoxicity and MOA, 667 and rodent hepatocarcinogenesis, 433 and in vitro testing, 246 2,4-D, chemical structure of, 531 Dangerous Substances Directive, of EU, 38 Darwinian selection, limitations of, 146 data, in risk assessment, 30–33 data gaps, in exposure reconstruction, 740 Data Quality Act (DQA) (2000), 28 data quality requirements, in risk assessment, 28–30 Daubert decision, 32 Dbl-1 assay, 329, 331, 333 D.C. Circuit court, EPA regulation in, 25 decalin NTP testing of, 484 in renal carcinogenesis, 486, 488 renal tumor response by, 493 decision making environmental, 191 and scientific uncertainties, 12 decision-tree-based approach, of HRF, 434 default causal models, 192 default dose-response models, 192–193 default methods, for cancer risk assessment, 67 “default options,” in US EPA’s guidelines, 25–27 defect rates, acceleration of, 152 DEHA, chemical structure of, 531
INDEX
DEHP chemical structure of, 531 in MOA analysis after PPARα activation, 444 Delaney Clause, in FFDCA, 79, 80 deletions, in initiation stage of carcinogenesis, 421 de minimis (negligible risk), 79, 389 demographics of cancer age, 398–399, 399, 401 gender, 399, 400 race/ethnicity, 399, 400 denadieldrin, dose-response characteristics of, 59, 59–60, 60 dendritic cells (DC), in inflammation, 125 deoxyribonucleic acid. See DNA depurinating adducts, in PAH-induced cancer process, 179–180 DEREK, 547 dermal exposure estimating, 753, 755 to soil, 81 detail, in exposure reconstruction, 739 deterministic models, to estimate exposures, 751, 752, 753 dextran sulfate sodium (DSS), 128 dialkylnitrosamine, interaction with nucleic acids of, 55 1,4-dichlorobenzene (DCB) cancer risk from, 84 in renal carcinogenesis, 486 dichlorodiphenyltrichloroethane (DDT) cancellation of, 4 hormetic biphasic dose-responses in, 197 1,1-dichloroethane, Monte Carlo analysis of, 765 dichloromethane animal MOA for, 371, 371, 373 key events in humans for, 374 dichloromethane (DCM) (methylene chloride) IUR for, 575 PBPK modeling of, 599, 600 PBPK models in, 574, 574–576 dieldrin cancellation of, 4 and ROS, 432 diesel engine exhaust, carcinogenic classification of, 413
795
diet and cancer risk, 402 and renal carcinogenesis, 489 Dietary Exposure Evaluation Model (DEEM), 755 Dietary Exposure Potential Model (DEPM), 755 dietary studies, weight gain reductions in, 385 diethylnitrosamine (DENA), dose-response relationships for, 53, 53–54, 54 digestive system, classifying neoplasms of, 706 dihydrodial epoxide, reactive, 527 4-dimenthylaminoazobenzene (4-DAB), dose-response relationships in, 50–53, 51, 53 4-dimenthylaminostilbene (4-DAST), doseresponse relationships for, 53, 53–54 dimenthylhydrazine, metabolic activation of, 172 dimethyintroasamine, metabolic activation of, 172 dimethyl dicarbamate, 78 dimethyldithiocarbamate, exposure reconstruction for, 762 dimethylnitrosamine (DMA), 425, 672 DINP, chemical structure of, 531 diol epoxide, metabolic activation of PAH through, 174 dioxane, PBPK models for, 577–578 dioxins, 654 exposure reconstruction for, 745 nonlinear response to, 666 directed acyclic graph (DAG), 201 disease and exposure, 408–409 and genotoxic metabolites, 68–69 Distributed Structure-Searchable Toxicity (DSSTox) Database Network (EPA, US), 605 DMBA, early studies with, 103–104 DMF, biological monitoring of, 747 DNA (deoxyribonucleic acid) as carcinogen target, 55 chemically damaged, 289–290 in initiation stage of carcinogenesis, 421 and radiation damage, 618–619
796
INDEX
DNA (deoxyribonucleic acid) (cont’d) ROS damage to, 181 stem cell, 136 as target for chemical carcinogens, 52 DNA (deoxyribonucleic acid) adducts chemical structures of, 314 defined, 314 dose-response curves for, 629–630 measuring, 315 assay limitations for, 317, 320, 322–324 interpretation of, 316–317 methods used for, 316–323 principle of, 315–316, 318–319 purpose of, 314–315 regulatory acceptance of, 315 sensitivity of, 320 specificity of, 320, 321 strengths of, 322–323 DNA comets, formation of, 284, 284 DNA (deoxyribonucleic acid) damage and DNA repair, 662 in nonlinear carcinogenesis, 661 PPARα activator-induced, 446 processes limiting, 662 in in vivo genotoxicity assays, 294, 295 DNA (deoxyribonucleic acid) repair key mechanisms of, 662 in nonlinear dose response, 627 DNA synthesis, unscheduled, in vitro test for, 283–284 dose in cancer bioassays, 381 in in vivo genotoxicity assays, 293, 298–299 dose-cancer response, in cancer discovery, 154–155 dose metrics PBPK model in, 569 US ATSDR, 766–767 dose response biologically-based, 576 biology behind, 632 hormetic, 624, 663 logarithmic scale for, 621 nonlinear, 661 in risk assessment process, 370 dose-response assessment, 9–10, 37, 383, 736 combined incidence in, 717 for dichloromethane, 574, 574–575
for mixtures, 623 and MOA, 615–616 PBPK models in, 563–564, 564, 566, 566 for vinyl chloride, 575 dose-response curve broadening of, 623 concept of threshold in, 621 controversy over, 29–30 evaluation of shape of, 629 linear, 617, 617 in low-dose region, 674 nonlinearities in, 626–627 dose-response data to estimate cancer risk, 716 modeling of, 557 dose-response models cumulative impact in, 572 hormetic, 194, 202 J-shaped hormetic, 198, 199 LNT-based, 199 dose-response relationships in cancer progression, 119 characteristics of, 52–53, 53 in chemical carcinogenesis for 4-DAB, 50–53, 51, 53 for 4-DAST, 53, 53–54 for DENA, 53–54, 54, 54 for dieldrin, 59, 59–60, 60 for ENU, 55, 55, 56, 58 for dioxane, 577–578 and formation of urinary solids, 502 in risk characterization, 388–390 single, 716 drinking water chloroform in, 577 exposure to, 75 lifetime cancer risks of chemicals in, 76, 77 MCL for, 76 standards, 73 Drosophila melanogaster, and epigenetic processes, 146 drug development, hormesis review criteria for, 197 drugs, pre-market genetic toxicity testing for, 226. See also pharmaceuticals dust exposure reconstruction for, 761 simulation studies of, 744
INDEX
dynamic evolution model, of carcinogenesis, 138 ecologic studies, 69, 407 economics, of cancer prevention, 45 2-EH, chemical structure of, 531 electron spin resonance (ESR), 452 electrophilicity, 170 embryological rest theory, 137 EMT. See epithelial-to-mesenchymal cell transition endocrine disruptors, 80 endocrine system, classifying neoplasms of, 707 endometrial cancer genetic mutations associated with, 403 risk for, 402 endothelium classifying neoplasms of, 706 combining neoplasms of, 709 end-stage kidney disease, and CPN process, 491 environmental chemicals, guidelines for testing, 239–240 Environmental Mutagen Society, 238 environmental protection, (Q)SAR analysis in, 546 Environmental Protection Agency (EPA), US, 441, 520 Cancer Guidelines of, 669–670 carcinogenicity descriptors of, 39–40 carcinogenicity protocols of, 636–637 and classification of renal tumors, 485–486 creation of, 4 on default dose-response models, 192 dose-response evaluation of carcinogens used by, 40 draft guidelines of, 69 drinking water standards of, 75–76 Framework for Evaluating a Mutagenic MOA of, 631 framework for extrapolation proposed by, 364 Guidelines for Carcinogen Risk Assessment of, 363, 364, 365, 408, 411n, 660–661 initial carcinogen risk assessment guidelines of, 25–27 institutional transition for, 17
797
on lifetime cancer risk, 683 mode of action defined by, 587 NCEA of, 674 Office of Pesticide Programs of, 673 Oncologic Cancer Expert System of, 530, 549 original risk assessment guidelines of, 717 pesticides regulated by, 77 quantitative extrapolations of, 682 and rat kidney cancer controversy, 32 revised cancer risk guidelines of, 557–558 rewrite of carcinogenic risk assessment guidelines of, 27 risk assessment approach of, 5–6, 194 risk assessment data used by, 30–31 risk-benefit analysis of, 19 on risk management, 191–192 Toxic Air Programs of, 674 virtual tissue models of, 609 weight-of-evidence requirement of, 202 enzyme-altered foci (EAF), in rat liver, 653–655 enzyme-linked immunosorbant assays (ELISA), 588 enzymes. See also specific enzymes in metabolic activation process, 170 in nonlinear carcinogenesis, 662 for Phase I metabolism of PAHs, 176–177 eosinophils, differentiation of, 132–133, 133, 134 epidemiological data analysis of cancer registry data, 643 individual data, 644 likelihood function (Poisson Regression), 643–644 on human carcinogens, 683 epidemiologic studies, 208, 397 of cancer risk, 660 case reports, 407 demographics, 398–400 descriptive, 13 effects of exposure in, 414–415 evaluation of determining causal association, 408–410 quality of studies, 407–408
798
INDEX
epidemiologic studies (cont’d) of hypolipidemic pharmaceuticals, 467 infectious agents in, 415 meta-analysis, 14, 407 methodology case-control, 405 cohort studies, 403–405, 404 cross-sectional (prevalence), 406 ecologic analyses, 407 molecular epidemiology, 406 proportionate studies, 405 molecular epidemiology, 415 in risk assessment, 13–14 used by risk assessors, 13 well-conducted, 12–13 epidemiology defined, 12 in history of cancer, 101 and identification of key events, 375 premise of, 13 of urinary tract calculi, 510–511 epigenetic biology, and nuclear traffic, 142–147 epigenetic carcinogens, (Q)SAR analysis of, 528–529 epigenetics, of defects in phenotypic plasticity, 148 epithelial cells, in local chemical carcinogenesis, 122 epithelial-to-mesenchymal cell transition (EMT), 122 erythrocytes differentiation of, 132–133, 133, 134 regeneration of, 132 erythropoiesis, and in vivo testing, 263 Escherichia coli, testing for mutagenicity in, 227–228 estimation methods, to reconstruct exposures deterministic models, 751, 752, 753 stochastic (probabilistic) models, 753 estrogen α2U genes repressed by, 483–489 and hepatic tumor risk, 430 ethanol, and ROS, 432 ethnicity and cancer incidence, 399, 400 and leukemia incidence, 401 ethyl benzene, renal tumor associated with, 491, 492
ethyl carbamate (urethane), metabolic activation of, 172 ethyl ether, estimating exposures of, 754 ethyl hydroxyethylnitrosamine (EHEN), 485 ethylnitrosourea (ENU) alkylation of guanine by, 56 dose-response relationships for, 55, 57, 57 euchromatin, 144 Europe genotoxicity testing guidelines in, 240 in history of genotoxicity testing, 242 and ICH guidelines, 244 European Centre for Validation of Alternative Methods (ECVAM), 31, 242 European Chemicals Agency (ECHA), 44 European Food Safety Authority, 85 European Food Safety Authority/World Health Organization (EFSA/WHO) international conference (2005), 78 European Union (EU), 15 chemical testing guidelines of, 243 Dangerous Substances Directive of, 38 NOAEL uncertainty factor approach used by, 41 precautionary principle promoted by, 34 REACH policy of, 252–256 Expanded Simple Tandem Repeat (ESTR) assay, 252 experimental evidence, and judging causation, 195 experts, in exposure reconstruction process, 756, 759 exponent index, IWR in, 572 exposure air sampling for, 70 background, 622, 623, 624 chemical compared with radiation, 620 for Chinese tin miners, 651, 652 dermal, 753, 755 detailed individual information for, 644 and disease, 408–409 drinking water, 75 effect at different ages, 414–415 environmental vs. occupational, 410 human chemical, 102 ingestion, 753, 755 to irritants, 125
INDEX
limiting dietary, 77 minimization of, 617 multiroute, 71 and mutagensis, 107 occupational vs. environmental, 69 PBPK modeling of, 577 residential, 81 routes of, 736 soil, 81 in in vivo genotoxicity assays, 293, 298–299 exposure assessment, 10, 37, 736–737 exposure estimation, 383 Exposure Factors Handbook (US EPA), 7, 81, 751 exposure/potency index (EPI), 389, 625 exposure reconstruction approaches to, 737 combined methodologies for, 759–760, 761–763 goals of, 738 methodology for, 737, 740–741, 770 identifying data gaps, 740 organizing information, 738–740 qualitative estimation methods, 755–759, 757–758 quantitative methods, 741–743, 744, 745 semiquantitative methods, 745, 748, 749–750, 751, 752, 753, 754, 755 process of, 737–738, 771 uncertainty analysis for, 764, 765, 766 exposure-response relationship, and judging causation, 195 exposure studies dose selection in, 419 of hepatocarcinogenesis induced by PPARα activator, 453 expression time, in gene mutation assays, 340 extrapolation default paradigms for, 681 from experimental data, 625–631 high-dose to low-dose, 565–568, 566, 567 from individual carcinogens to mixtures, 571–574, 573 interspecies, 391, 565–568, 566, 567
799
linear of cancer incidence, 615 historical perspective on, 616–625, 617, 618 scientific basis for, 632 linear low-dose, 616 low-dose and cancer risk assessment, 669–670 outside US, 670 quantitative, 681 in risk assessment, 364, 393 of rodent data, 586–587 route-to-route, 571 extrapolation methods, to reconstruct exposures, 748, 751 factory workers, cancer development in, 208 false-positive results in genetic toxicity testing, 232, 234 role of culture conditions in, 286 feasibility, in regulatory decision-making, 24 federal appeals court, US EPA’s guidelines in, 27 Federal Food, Drug, and Cosmetic Act (FFDCA), US (1938), 4, 79 Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA), US, 19–20, 48–49, 83 fiberglass, JEM for, 749 fibers carcinogenicity of, 535 carcinogenicity/toxicity of characteristics related to, 537–539 mechanisms of, 534–537 SAR analysis of, 539–540, 540 fibroblast growth factor (FGF), 113 Fischer 344 rats, 494 fish, hormetic biphasic dose-responses in, 197 Fisher’s exact probability test, 700 fixation time, in gene mutation assays, 340 fjord region, on PAHs, 176, 177 flavoring agents, and TTC concept, 85 fluorescent in situ hybridization (FISH) aneuploidy detected by, 281–282, 283 comet assay combined with (COMET-FISH), 314 folate, and colorectal cancer, 653, 654
800
INDEX
follow-up testing, for in vivo genotoxicity assays, 294–295 food additives FFDCA provisions on, 79 guidelines for testing, 239–240 negligible risk concept for, 78 Food and Drug Administration (FDA), US, 77 AAF study of, 197 core testing battery of, 240 on lifetime cancer risk, 683 risk-benefit analysis of, 19 Food Quality Protection Act (FQPA), US (1996), 79, 80 food standards regulatory considerations, 77–80 scientific issues, 76–77 formaldehyde BBDR modeling of, 674 carcinogenic classification of, 413 estimating exposures of, 752 expert exposure studies on, 757 gene expression profile for, 591–592 JEM for, 749, 750 Monte Carlo analysis of, 765 nonlinearity of, 665 predicting emission rates for, 751 as site-of-contact carcinogen, 75 tumorigenicity of, 631 framework analysis, 364–365 for determining MOA, 375 human relevance in, 372–375, 374 human relevance of animal MOA in, 365–367, 366, 366 MOA in, 365 Frank-Starling Law of the Heart, 607 F344 MeIQx carcinogenicity in, 211, 211–212, 213 PhIP carcinogenicity in, 216, 216, 217 furans, exposure reconstruction for, 745 Galen, 99–100 gap junction connections (GJs), 113, 114–115 and cell-to-cell communication, 115 xenochemicals diluted by, 115–116 gardeners, exposure to soil of, 81 gas chromatography, for measuring DNA adducts, 317, 319, 321, 323
gas stations, abandoned, 82 gemfibrozil, chemical structure of, 531 gender and cancer incidence, 399, 400 and leukemia incidence, 401 in in vivo genotoxicity assays, 300–301, 302 gender differences, in lung cancer risk, 650 Gendron, Claude Deshais, 100 gene expression and cancer prophylaxis, 139 plasticity in, 133 gene expression profiles conazoles, 592–593 formaldehyde, 591–592 glutaraldehyde, 591–592 nongenotoxic hepatocarcinogens of, 594 perfluorooctanoic acid, 590–591 value of, 594–595 gene mutation assays, 294, 297, 299, 301 autosomal vs. nonautosomal, 342 in vivo, 328–329 animals per group, 343 endpoints for, 328–329 interpretation of, 329 limitations of, 329 mutant selection in, 332–333 principle of, 329, 330–331 purpose of, 328 regulatory acceptance for, 329 reporter gene, 330–331 species used in, 330–331 strengths of, 329 tissues targeted for, 332–333 in vivo transgenic endpoints for, 334–335 mutant selection in, 338–339, 340 principle of, 334–335 reporter gene, 336–337 species for, 336–337 tissues targeted by, 336–337 genetic background, and genotoxic metabolites, 68–69 genetic changes, in tumor-associated stroma, 130 genetic mutations, and cancer risk, 402, 403 genetics early history of, 102 in risk assessment, 17
INDEX
genetic toxicity testing, 225 approaches to, 229, 231–232 development of, 226, 227–228 predictive ability of, 235 types of tests, 228–229, 230 in vitro compared with in vivo, 233, 235 in vivo, 226 genetic toxicology assays. See also genotoxicity assays the 3Rs of, 303 in vivo, 344 endpoints used, 289–291, 290, 293, 294–295 in risk assessment, 291–292 genital system, classifying neoplasms of, 708 genomics, 84, 185, 390 case studies, 590–593, 591 in predictive toxicology, 593–594 and risk assessment paradigm, 589–590, 591 genotoxic agents, in hepatocarcinogenic compounds, 423, 424, 424–425, 425 genotoxicity mutagenicity compared with, 226–227 nonlinearity of, 245 protocols, 145 screening assays for, 392–393 structural alerts of, 525 genotoxicity assays, in vivo, 344 complementary or follow-up gene mutation assays, 328–329, 328–344 sister-chromatid exchange assay, 326–328 unscheduled DNA synthesis test, 324–326 complementary or follow-up tests, 310 comet assay, 311–314 DNA adducts, 314–324, 316–323 parameters and criteria for, 292–293, 294–301, 302–303 required, 303 bone marrow chromosome aberration test, 308–310 mammalian erythrocyte micronucleus test, 304–308 genotoxicity guidelines ICH, 243–248 OECD, 243
801
genotoxicity testing battery approach for, 247 origins of, 272–273 philosophic changes in, 240 recent changes to, 242 regulatory, 264 in vitro, 241, 241–242, 250–251 in bacteria, 273–275, 275, 276 cells used for, 273 comet assay, 284, 284–285, 285 strengths and limitations, 285–286 for unscheduled DNA synthesis, 283–284 in vivo, 251–252, 258–261 choice of in vivo test, 261–263 evaluation of results with, 262–264 genotoxicity testing guidelines bacterial test strains in, 275 for genotoxic impurities in pharmaceuticals, 256–258, 258 historical overview of, 238–243, 241 internationally relevant, 242 of IPCS, 249 IWGT on, 248–249 under REACH, 252–256 for in vivo testing, 252 genotoxic potential, endpoints for, 272 Germany, genotoxicity testing guidelines in, 240–241 germ cell mutagens, human, 225–226 germ cell testing, 251–252 GLOBAL software, 718, 724 glucose cancer cell metabolization of, 148 and Cori cycle, 151, 151–152 and Warburg Effect, 148–149, 149, 150 glutaraldehyde, gene expression profile for, 591, 592 glutathione S-transferase-P (GST-P) positive foci, 629 glycolysis, anaerobic, 151 Good Laboratory Practice (GLP) standards, 588 Gpt delta (spi), in vivo gene mutation assays in, 335, 337, 339 Gpt delta (6-thioguanine), in vivo gene mutation assays in, 335, 337, 339 granite dust, exposure reconstruction for, 763
802
INDEX
groundwater contaminants, exposure reconstruction for, 762 GST-P-positive foci, in rat hepatocarcinogenesis, 210, 210, 210, 213 guideline/GLP studies, 588 guidelines, risk assessment, 7, 8, 25–27. See also genotoxicity guidelines; specific guidelines guinea pig, response to PPARα activators of, 456, 457–458 Haber’s rule, 51–53, 59, 138 haloalkanes, SAR analysis of, 523 hamster, CD10 estimates in, 686, 688, 689, 691, 691, 691. See also Chinese hamster cells; Syrian hamster Harderian gland carcinomas, 728, 728, 730 Harderian gland neoplasms, 701 harm, and hormetic dose-response, 201 hazard characterization, 386–388 defined, 386 mechanistic data in, 387 hazard function, 637 hazard identification, 37, 383–386, 736 adequate high dose in, 384 combined incidence measures in, 718 defined, 9 genotoxicity testing for, 259 (Q)SAR in, 544, 545, 546–547 signs of treatment-related toxicity, 384–385 statistical vs. biological significance, 385–386 hazards environmental, 38 physiochemical, 38 toxicological, 38 hazard testing for regulatory risk assessment combined chronic/cancer bioassay, 379–380, 379–382 limited in vivo studies, 382–383 perinatal carcinogenicity studies, 382 and risk assessment, 391–393 Health and Safety Data Reporting (HaSDR) rule, of TSCA, 43–44 Health Education and Welfare Department (DHEW), US, Subcommittee on Environmental Mutagenesis of, 238
Health Professional’s Follow-Up Study, 649 health protection, (Q)SAR analysis in, 546 heat shock proteins, 153 Heiger, Ian, 169 Helicobacter pylori, 124 hemangiosarcomas, 702 hematopoiesis process, 132–133, 134 hematopoiesis stem cells (HSCs), 132, 134, 135 hematopoietic system, classifying neoplasms of, 706 hemocytoblasts (HSCs), 132, 133 hepatitis C virus (HCV), 455 hepatocarcinogenesis PPARα activator-induced, 444, 445–447 receptor-mediated processes in, 435 stages of, 420, 420, 421 hepatocarcinogenesis, rodent, 592 human relevance framework for, 434 initiation in, 420, 420–422, 421 progression in, 420, 421, 422–423 promotion in, 420, 421, 422 hepatocarcinogenic peroxisome proliferators, SAR of, 530–532 hepatocarcinogens, nongenotoxic, 594 hepatocellular carcinoma (HCC), 104, 113 hepatitis C virus-induced, 455–456 incidence of, 103 NF-κB in, 129 hepatocytes, 443 initiated, 421–422 micronucleus assay with, 307 in PPARα activator MOA, 451 hepatotoxicants, genomic signatures of, 593–594 herbicides, exposure reconstruction for, 762 heritable effects, genotoxicity tests for, 260 H1 histone, in epigenetic biology, 143 high content analysis (HCA), 597, 610 high content analysis (HCA) assays, 604 high-performance liquid chromatography, for measuring DNA adducts, 317, 319, 321, 323 High Production Volume (HPV) chemicals defined, 46 OECD’s investigation of, 46–47 High Production Volume (HPV) Chemicals Challenge Program, EPA, 46
INDEX
high production volume (HPV) existing substances, evaluation of, 42, 45 high-throughput molecular technologies, 588 high-throughput screening (HTS), 597, 604–606, 610 Hill, Sir Austin Bradford, 408–409 Hill criteria, in epidemiological studies, 13–14 Hippocrates, 99 histology, early field of, 101 histone acetyltransferases (HATs), in epigenetic biology, 144 The Histone Code (Jenuwein et al.), 143 histone deacetylases (HDACs), in epigenetic biology, 144 histone demethyltransferases (HDMTs), in epigenetic biology, 144 histopathology of hepatocarcinogenesis, 420 of α2U-g nephropathy, 485–489 hockey stick regression model, 213 homeodynamic equilibrium, 109 hormesis concept of, 196 defined, 664 J-shaped curved, 664 as principal regulatory model, 201 theory, 663 hormesis review criteria, 196 hormetic models arguments justifying, 198 J-shaped, 198, 199 MVK model, 199–200, 200 hormetic statement, validity of, 198 hormones and cancer risk, 402 in rodent hepatocarcinogenesis, 430 Howell-Jolly bodies, 304 Hprt assay, 328, 330, 332 H-ras gene, mutation of, 211 Human Genome Project (HGP), 17, 146, 185 human relevance framework (HRF), 366, 373–375, 374, 387–388 application of, 393 rodent hepatocarcinogenesis in, 434 humans and biological responses in animals, 68 cancer risk assessment in, 208
803
CD10 estimates in, 691, 691 and relevance of liver tumors induced by PPARα activation, 441–442, 442 and relevance of PPARα activatorinduced rodent liver tumor response, 467 response to PPARα activators of, 458, 459–460 substances carcinogenic to, 411, 412– 413, 415 TD25 estimates in, 690, 691 toxicokinetic variability among, 618 hyaline droplet nephropathy, 483–484 hydrocarbons, expert exposure studies on, 758 hydrocyanic acid (HCA), 51 hydroquinone, RTT associated with, 491–492 8-hydroxy-2′-deoxyguanosine (8-OH-dG), as marker for oxidative DNA damage, 211, 211 hyperparathyroidism, and urinary calcium levels, 504 hyperplasia and excess LACA, 150 progression from, 701 hyperthermia, and in vivo testing, 263 hypothermia, and in vivo testing, 263 hypoxia, in chemical carcinogenesis, 149, 150. See also oxidative stress I cells, 121 ICRP. See International Commission on Radiological Protection immune responses, in DNA repair, 662 immunoassays, for measuring DNA adducts, 316, 318, 320, 322 immunodefense system (IDS) and cellular repair, 125 mammalian, 124 immunological response, and heat shock proteins, 153 incidence, cancer, 398–399, 399 and age, 126 analysis of data for, 404–405 combined, 717 defined, 404n and diet, 402 and hormonal differences, 402 infectious agents and, 402
804
INDEX
incidence, cancer (cont’d) and inherited genetic alterations, 402, 403 linear extrapolation of, 615 multisite time-dependent analysis of, 730, 731 proportionate, 405 in rat carcinogenicity studies, 699 and smoking, 400, 401, 402 total, 732 incidence function, age-specific, 637 industrial hygiene measurements, in exposure reconstruction, 741–742 industrial hygienists, expert opinions of, 756, 759 infection, and cancer incidence, 124 infectious agents cancer associated with, 415 cancers due to, 402 inflammation, and cancer, 124–130 informatics, 606 Information Quality Act (2000), 28 In(GSD) estimates, 692, 693, 694 inhalation evaluation of carcinogenicity by, 72 of fibers, 538–539 inhalation exposures interspecies scaling of, 70 route-specific, 71 inhalation unit risk (IUR), 557 inhibition studies, PPARα activator MOA, 450, 451, 452–454 initiation in cancer process, 420, 420–422, 421 process of, 636 initiation, in carcinogenesis, 117, 118 experiments in, 182, 182–183 in mouse skin, 183, 184 use of term, 182 Initiation, Promotion, and Progression (I-P-P) stages, 123 Integrated Risk Information System (IRIS), US EPA’s, 6, 411, 412–413, 414, 668, 768 integumentary system, classifying neoplasms of, 705 interaction weighing ratio (IWR), determination of, 572 Interactive RadioEpidemiological Program (IREP), NIOSH’s, 760
Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM), 31 Interagency Regulatory Liaison Group (IRLG), US, 6 Interagency Testing Committee (ITC), TSCA, 43 “Interim Genomics Policy,” of US EPA, 32 International Agency for Research on Cancer (IARC), 15, 388, 398, 501 carcinogenic classification of, 414 carcinogenicity evaluations by, 39 and classification of renal tumors, 485–486 on DCB, 84 International Association on Environmental Mutagen Societies (IAEMS), 248 International Commission on Radiological Protection (ICRP), 665 International Conference on Harmonisation (ICH), 1991 Brussels conference of, 244 International Conference on Harmonisation (ICH) guidelines, 244–248 on genotoxicity testing of pharmaceuticals, 259–260 pharmaceutical, 242, 243–248 for qualification of impurities, 256–257 revised ICH S2, 260 S2R1, 264 in vivo testing, 259 International Conferences on Environmental Mutagens (ICEMs), 248 International Life Sciences Institute (ILSI), 242, 520 framework for extrapolation proposed by, 364 mode of action framework of, 587 RSI of, 367 International Life Sciences Risk Sciences Institute (ILSI-RSI), 387–388, 434 International Programme on Chemical Safety (IPCS), 242, 249, 434 framework for extrapolation proposed by, 364 and hazard characterization, 387 mode of action framework of, 587 International Workshop on Genotoxicity Tests (IWGT), 242, 248–249 intestines, classifying neoplasms of, 706
INDEX
intrinsic pathway, 111 Inventory Update Reporting (IUR), TSCA, 45 investigational new drugs (IND), in ICH guidelines, 247–248 in vitro assays, 378 in vitro micronucleus test in ICH guidelines, 245–248 OECD Guidelines for, 243 validation of, 283 in vitro models, in cell biology, 105–107 in vitro testing, 264 bacterial reverse mutation assay, 250 for chromosome damage, 279, 279–280 comet assay, 284, 284–285, 285 concentrations used in, 245–246 evaluation of results of, 250 follow-up to, 250–251 for gene mutation, 276–279, 278 in bacteria, 273–276, 275, 276 in mammalian cells, 276–279, 278 mammalian assay, 250 metabolic activation in, 273, 274 micronucleus test, 280–283, 282, 283 nonrelevant positive findings in, 246–247 strengths and limitations, 285–286 for unscheduled DNA synthesis, 283–284 in vivo assays, shorter-term, 383 in vivo micronucleus assay, basis of, 261 in vivo studies, limited, 382–383 in vivo testing, 264, 289. See also genetic toxicology agents compared with in vitro testing, 291 endpoints used for, 289–291, 290, 293, 294–295, 344 CHECK follow-up to, 251 in genetic toxicity testing, 231 parameters and criteria for, 292–293, 294–301, 302–303 and REACH, 254–255 for risk assessment, 258–261, 291–292 choice of in vivo test, 261–263 evaluation of results with, 263 strategy for germ cell testing, 251–252 ionizing radiation (IR) biological monitoring of, 746, 747 cancer estimates for, 365 dose reconstruction of, 759 as mutagenic agent, 664
805
I-P-P. See Initiation, Promotion, and Progression I-P protocols, 104–105, 107 IRIS. See Integrated Risk Information System IRLG. See Interagency Regulatory Liaison Group, US iron oxide, simulation studies of, 744 irritants, and carcinogenesis, 125 irritation theory, of cancer, 169 isocyanates exposure reconstruction for, 763 SAR analysis of, 524 I-stage. See also initiation in carcinogenesis, 104–105 evaluation of F-344 rat liver foci during, 105 GJ function in, 115–116 in progressive carcinogenesis, 117, 118, 154 Japan genotoxicity testing guidelines in, 240 and ICH guidelines, 244 job exposure matrix (JEM), 745, 748, 749–750 Joint FAO/WHO Expert Committee on Food Additives (JECFA), 85 J-shaped dose-response model, 193, 196, 198, 199, 203, 663, 664 judicial decisions, 30. See also courts; specific decisions Kaplan-Meier estimates, 724 Kaposi’s sarcoma, 124 keratinocytes, 134 key events for characterizing MOA, 368, 368–370 defined, 364 in MOA framework, 371, 371–372, 372 temporal sequence of, 371 in tumor development, 368, 368–369 kidney, classifying neoplasms of, 707 kidney cancer incidence of, 399, 399 in rat, 32 kidney tumor, in male rat, SAR of, 533–534 knockout mouse models, 382, 427 Koch, Robert, 102
806
INDEX
Kupffer cells inflammation altered by, 454 and PPARα activators, 449 LACA and Cori cycle, 151, 151–152 excess, 148 ionization of, 150 lacI gene, mutation of, 211, 212 lactic dehydrogenase A, 148 lactofen, chemical structure of, 531 LacZ bacterial gene, 336 LADDs. See lifetime average daily doses lead, in ceramic ware, 78 legionella, estimating exposures to, 754 leukemia benzene-induced, 69 incidence of ALL, 400, 401 leukemia deaths, in survivors of Hiroshima and Nagasaki, 618, 618 leukemogenesis, 57 leukocytes, as regulators of cancer development, 124 Leydig cell tumors, testicular, 440 life stage, and genotoxic metabolites, 68–69 Life Stage Study (LSS), of atomic bomb survivors, 365 lifestyle factors and cancer incidence, 400, 401, 402 and carcinogenesis, 116 in epidemiology, 12 and genotoxic metabolites, 68–69 lifetime average daily doses (LADDs), 767–768, 768, 769 ligand inductibility, in human response to PPARα activators, 462 likelihood function, analysis of, 643–344 d-limonene in renal carcinogenesis, 486, 488 renal tumor response by, 493 lindane, and ROS, 432 linear, no-threshold (LNT) models, 192, 193, 199 linearity EPA on, 669–670 low-dose, and cancer risk, 660 linearity assumption, for direct acting mutagens, 668
linearized multistage (LMS) model, 30, 668 linear no threshold (LNT) extrapolation, 367 linear no-threshold (LNT) model, 665 linear regulatory approach, to estimating risk, 768–769 Linnaeus, 100 lipid homeostasis, PPARα regulation of, 445 lipid metabolism, PPARα agonists in, 427 lipid 2,3,7,8-tetrachloro-dibenzo-dioxin (TCDD), biological monitoring of, 746 lipophilic chemicals in soil, 81 and water standards, 75 liquid chromatography, for measuring DNA adducts, 317, 319, 321, 323 Lister, Joseph, 100, 102 liver. See also rat liver as carcinogen target, 420 classifying neoplasms of, 707 for in vitro genotoxicity testing, 273, 274 liver cancer in humans vs. rodents, 587 incidence of, 420 MeIQx-induced, 214, 214 and multistage carcinogenesis, 184 liver carcinogens, hormetic hypotheses for, 197 liver cell, mammalian, UDS test in, 324–326 liver damage, MeIQx-induced, 212–213, 214 liver tumors after exposure to PPARα activators, 444 in PPARα activator MOA, 451 rodent and human data on, 466, 467 and species differences in PPARα activation, 457–458 liver weight, and species differences in PPARα activation, 457–458 LNT-based dose-response model, 192, 193, 199 local tissue array (LTA) and homeostasis, 124 improper cellular development in, 140 loss of heterozygosity (LOH), 121 lower limit on effective dose01 (LED01), 41
INDEX
lower limit on effective dose10 (LED10), 769 lowest feasible concentration (LFC), 73 lowest observed adverse effect level (LOAEL), for non-carcinogens, 10 lung, classifying neoplasms of, 705 lung cancer among coal miners, 102 radiation and, 650–651, 651 and radon levels, 665 smoking-related promotion in, 649–650 lung deposition, in inhalation exposure, 70 lung particle overload, 70 L5178Y mouse lymphoma assay, 277, 278, 279 lymphocytes, in micronucleus analysis, 281, 282, 283, 307 lymphoma benzene-induced, 69 genetic mutations associated with, 403 “lymphoreticular infiltrate,” 125 lymphoreticular tissue, combining neoplasms of, 709 machining fluids, exposure reconstruction for, 763 macrophages in inflammation, 125 tumor-associated, 126 magnesium and formation of urinary solids, 509 in urine, 506 malignancy, Initiation, Promotion, and Progression stages of, 123 malignant categorization as, 702–703 defining term, 701 malignant regrowth, after cancer surgery, 155 malondialdehyde (MDA), 433 mammalian cell mutagenicity test, in genetic toxicity testing, 231 mammalian erythrocyte micronucleus test. See also micronucleus test interpretation of, 306 limitations of, 306–308 principle of, 304 purpose of, 304
807
regulatory acceptance of, 304 sensitivity of, 306–307 study design for, 305–306 mammalian test systems for chromosome damage, 279, 279–280 for genotoxin detection, 276 mammary gland, classifying neoplasms of, 705 manifestation time, in gene mutation assays, 340 margin of exposure (MOE), 78 margin of exposure (MOE) analysis, 671, 769 margin of exposure (MOE) approach, 671–673, 672, 676–677 Markov Chain Monte Carlo (MCMC) simulations, PBPK modeling using, 569, 570 mass spectrometry, for measuring DNA adducts, 317, 319, 321, 323 mast cells (MCs), in inflammation, 125 matrix metalloproteases (MMPs), in tumor progression, 130 maximal sensitivity approach, to in vitro testing, 245 Maximum Available Control Technology (MACT) requirement, 24 maximum contaminant level goal (MCLG), establishing, 75–76 maximum contaminant level (MCL), establishing, 76 maximum likelihood estimate (MLE), in risk assessment, 30 maximum tolerated dose (MTD) controversy about, 385 and regulatory standards, 68 mechanism of action contrasted with MOA, 364 US EPA definition of, 203 Medium-production volume (MPV) chemicals, evaluation of, 45 meiosis, 130 memory, abnormal cellular, in carcinogenesis, 147 Mendel, Gregor Johann, 102 mercury biological monitoring of, 746 estimating exposures of, 752 mesothelioma, 74
808
INDEX
meta-analysis in cancer risk assessment, 14 in epidemiological studies, 407 metabolic activation theory, 169–172 metabolomics, 185, 390, 589, 589 metals biological monitoring of, 746 in soil, 81 metazoans, humans as, 119 methanol estimating exposures of, 752, 754 simulation studies of, 744 methyl bromide, estimating exposures of, 754 methyl chloroform, exposure reconstruction for, 762 methylclofenapate, chemical structure of, 531 methylene chloride, PBPK modeling of, 599, 600 methylene diphenyl diisocyanate (MDI), as “high risk” pollutant, 29 methyl isobutyl ketone (MIBK), short-term inhalation studies of, 494–495 methylnitrosourea, metabolic activation of, 172 methyl tertiary butyl ether (MTBE) and aα2U-g nephropathy, 493–494 exposure reconstruction for, 761 PBPK model for, 600–601 Michael Addition acceptors, SAR analysis of, 524 Michaelis-Menten equation, 626–627 microarrays, 590 microbial infection, and cancer incidence, 124 micronucleus assay in bone marrow or peripheral blood, 261, 294, 296, 298, 300 cytochalasin B in, 281, 282 with different mammalian cell types, 241 ICH guidelines for in vivo, 261 in peripheral blood, 262 REACH policy for in vitro, 254 use of cytochalasin B in, 246 in vitro, 280–283, 282, 283 microRNAs (miRNA), 450 Miller, Elizabeth C., 169–172 Miller, James A., 169–172 mineral spirits, simulation studies of, 744
de minimis level of risk, 79, 389 mitogen-activated protein (MAP) kinase, activation of, 433 mitogen-activated protein (MAP) kinase pathway, tumor suppressor activities of, 662 mitosis, and age, 152 mixtures risk assessment, 622–623 “Mobile Source” regulations, 83 mode of action (MOA), 84 complex, 454–455 cytotoxicity in, 666, 667 defined, 11, 364, 441, 661 DNA-reactive, 368, 368–369 and dose-response assessment, 615 establishing, 587–588 establishing key events in support of, 367–372, 368, 371, 372 examples of, 364 and hazard characterization, 388 human relevance of, 365 for induction of bladder cancer, 502 information on, 11 key events for characterizing, 368, 368–370 of liver tumor induction, 439 of liver tumors induced by PPARα activation, 441–442, 442 and low-dose linearity, 660–661 need for additional, 371 nonlinear, 670–671 and PBPK modeling, 599–600 in risk assessment, 9 in tumor development, 615 in US EPA cancer guidelines, 442 US EPA definition of, 587 mode of action (MOA), rodent and cell growth alterations, 449–450 and cell proliferation/apoptosis balance, 447–449 chemically-independent MOAs of PPARα activators, 455–456 complex MOAs of PPARα activators, 454–455 inhibition studies, 450, 451, 452–454 oxidative stress in, 444, 445–447 PPARα activation, 443, 444, 445 role of NF-κB in, 447 and species differences in PPARα MOA, 456–460, 457–458
INDEX
mode of action (MOA) analysis, 365 framework analysis for determining, 375 in US EPA risk assessment, 441 WOE in, 441 molecular biology, and risk assessment, 16–17. See also cell biology of cancer molecular epidemiology, 406 monkey, response to PPARα activators of, 458, 459 monoclonality, of certain tumors, 120–121 monocytes, differentiation of, 132–133, 133, 134 monotonic functions, 203 Monte Carlo analysis, 753, 760, 770 of formaldehyde, 765 procedure, 729, 733 of toxicological criteria, 770 Monte Carlo risk analysis modeling, 10 Monte Carlo simulations, 719, 720, 764 Moolgalvkar-Dewanjii-Venzon (MVK) model, 199–200, 200 mortality analysis of data for, 404–405 cancer, 397 in cohort studies, 404 proportionate, 405 retrospective studies, 404 smoking attributable, 400, 401, 402 mothballs, cancer risk from, 84 mouse. See also specific strains for bone marrow micronucleus test, 261 cancer risk assessment in, 208 as carcinogenesis model, 227 carcinogenicity studies in, 67 CD10 estimates in, 686, 687, 688, 691, 691 combined chronic/cancer bioassay in, 379, 381 response to PPARα activators of, 457 screening for chemical carcinogens in, 501 mouse (coat) spot assay, 328, 330, 332 mouse lymphoma assay, of gene mutation, 277, 278, 279 mouse lymphoma cells, in micronucleus analysis, 281 mouse lymphoma tk+/− assay (MLA), 247 mouse retinoblast (eye spot) assay, 328, 330, 332
809
mouse skin carcinogenesis in, 183–184 experiments on, 636 mouse urinary protein (MUP), 484–485 MSTAGE software, 719, 724 mucosal cells, intestinal, 134–135 MultiCase MC4PC, 547 multiple endocrine neoplasia type 2, genetic mutations associated with, 403 multistage clonal expansion (MSCE) model, 637 age-dependent parameters in, 642 constant model parameters for, 641–642 development of, 640–641, 641 dose-response in, 642–643 multistage models, use of, 655 muraglitazar, and formation of urinary solids, 509, 512 S-Mustards, SAR analysis of, 523 mutability, in tumor progression, 119 mutagenesis, 106 causes of, 122 dose-response for, 628 experimental studies on, 393 in history of genotoxicity testing, 239 and Warburg Effect, 148–149, 149 mutagenicity genotoxicity compared with, 226–227 as MOA, 616 REACH definition of, 253 mutagenicity testing methods. See also genotoxicity testing guidelines classical series of, 240 IPCS guidelines in, 249 mutagens classification of, 227 and genetic toxicology testing, 225 MutaTM Mouse, in vivo gene mutation assays in, 334, 336, 338, 341 mutant frequency, measurement of, 341–342 mutational events, in progression from benign to malignant neoplasms, 703 Mutational Theory on the Origin of Cancers (Bauer), 182 Mutation Research (Kirkland et al.), 249 mutations dose-response curves for, 629–630 and initiation process, 636 in initiation stage of carcinogenesis, 421
810
INDEX
mutations (cont’d) measurement of, 277 point, 227 mycotoxins, 423 myeloblasts, 132 myeloperoxidase, 126 myofibroblast biomarkers, 113 NADPH oxidase, 126 nafenopin, in MOA analysis after PPARα activation, 444 nafenopino, chemical structure of, 531 nanomaterials carcinogenicity of, 543 defined, 540 physicochemical properties of, 545 SAR analysis of, 543–544, 545 toxic effects of, 540–543, 543 nanotoxicology, 84 nasal cavity, classifying neoplasms of, 705 National Academy of Sciences (NAS), US, 9, 79 1977 report of, 619–620 1983 report of, 598 National Ambient Air Quality Standards, 72 National Cancer Institute, US, carcinogen testing program of, 234 National Cancer Institute/National Toxicology Program (NCI/NTP), US, 682 National Center for Environmental Assessment (NCEA), of US EPA, 674 National Center for Environmental Health (NCEH), of CDC, 743 National Death Index (NDI), 404n National Health and Nutrition Examination Survey, 743 National Institute for Occupational Safety and Health (NIOSH), US, on dose reconstruction, 759 National Institute for Public Health and the Environment (RIVM), of the Netherlands, 616, 670 National Institute of Occupational Safety and Health (NIOSH), Pocket Guide to Chemical Hazards of, 72–73 National Research Council (NRC), US on genomic data, 32 on nongenotoxic carcinogens, 666
reports of, 375–376 on risk assessment, 9 risk assessment protocol of, 587 on risk management, 191–192 on risks to infants and children, 80 Toxicity Testing in the 21st Century initiative of, 674 Toxicity Testing reports of, 379, 390, 392, 588 National Scale Air Toxics Assessment, 674–675 National Toxicology Program (NTP), US, 15, 39, 75, 420 decalin tested by, 484 experimental studies of, 383 on hyaline droplet nephropathy, 493 known carcinogen list of, 414 regulatory role of, 80 reports of, 717 RoC of, 40 standard cancer bioassays conducted by, 122–123 TBA study of, 494 on validation studies, 228 NCEA. See National Center for Environmental Assessment NCI. See National Cancer Institute neonatal mouse model, 382–383 neoplasia B[a]P-induced, 169 nongenetic epigenetic alterations in, 147 neoplasms benign, 718 from benign to malignant, 701 combined in rodent carcinogenicity studies, 699–700 criteria for, 704–710, 705–709 rationale for, 701–702, 711 formation of, 422–423 and IDS, 126 regression of, 703 subclassifying, 702–703 neoplasms, combining for common cell type in different tissues, 709, 710 criteria for, 704–710, 705–709 by organ and tissue, 704, 705–709 by site, 704, 709, 710 guidelines for, 700 rationale for, 701–702, 711 neoplastic disease, early description of, 101
INDEX
nephropathy, α2U-globulin, 483–484 NER. See nucleotide excision repair nervous system classifying neoplasms of, 708–709 combining neoplasms of, 709 net nuclear grain (NNG) count, in UDS test, 325 neurogenic malignancies, median induction period of, 55–57 neutrophils, differentiation of, 132–133, 133, 134 nickel expert exposure studies on, 756, 757 exposure reconstruction for, 762, 763 JEM for, 750 nickel refinery dust, carcinogenic classification of, 412 nickel subsulfide, carcinogenic classification of, 413 nitrate, estimating exposures to, 752 nitrosamines, 234 N-nitrosodiethylamine (DEN) carcinogenicity bioassay of, 207 low-dose hepatocarcinogenicity of, 215 N-nitrosodimethylamine (DMN) carcinogenicity bioassay of, 207 low-dose hepatocarcinogenicity of, 215 N-Mustards, SAR analysis of, 523 N-Nitrosamides, SAR analysis of, 523, 526 N-Nitrosoamines, 423 N-nitroso compounds, low-dose hepatocarcinogenicity of, 215 N-nitrosomorpholine (NNM), exposure to, 654 N-nitroso-N-ethylurea (ENU), single-dose experiments with, 55, 55, 56, 58 no-effect level, for MeIQx mutagenicity and carcinogenicity, 211 nonlinear approach, to risk assessment, 769 nonlinearity DNA-reactive carcinogens and, 664–666 explanation for, 661 nonmutagenic carcinogens and, 666–667 in practice, 670–671 BBDR modeling, 674–675 captan, 673 chloroform, 673 harmonization of cancer and noncancer risks, 675–676 with RfD or MOE approach, 671–673, 672
811
nonparenchymal cells (NPCs), 443 nonsteroidal anti-inflammatory drugs, 125 no observed adverse effect level (NOAEL), 41, 769 no observed effect level (NOEL), for noncarcinogens, 10 normochromatic erythrocytes (NCE), 304, 305 No Significant Risk Levels (NSRLs), 84 notch signaling pathway, 141–142 Notice of Commencement of Manufacture or Import, 43 nuclear factor kappa B (NF-κB) activation of, 433 and hepatocarcinogenesis, 439 in modulation of hepatocyte fate, 453 in PPARα activator MOA, 447 in tumorigenesis, 128–129 nuclear factor kappa B (NF-κB) activation after exposure to PPARα activators, 444 in PPARα activator MOA, 451 and species differences in PPARα activation, 457–458 nucleosomes, in epigenetic biology, 143 nucleotide excision repair (NER), and UDS test, 324 Nurses’ Health Study, 649 obesity, and cancer, 402 observed association, and judging causation, 195 occupational cancer, scrotal, 100–101 occupational diseases, historical perspective on, 100 Occupational Safety and Health Act (OSHA) (1970), US, 72, 742 Occupational Safety and Health Administration (OSHA), US benzene standard of, 33 cancer guidelines of, 26 formaldehyde risk assessment of, 30 and risk assessment, 24–25 risk-benefit analysis of, 19 octamer-binding transcription factor-4, 141–142 Oct-4 gene, 141–142 Office of Environmental Health Hazard Assessment (OEHHA), California’s, 84, 720
812
INDEX
Office of Management and Budget (OMB), US peer review mandated by, 28 Proposed Risk Assessment Bulletin of, 29 quality standards of, 28–29 Office of Pesticide Programs (OPP), of US EPA, 673 Office of Pesticide Programs (OPP), US EPA, 755 Office of Prevention, Pesticides and Toxic Substances (OPPTS), US EPA, 38, 225 Office of Science and Technology Policy (OSTP), US, 7 Office of Solid Waste and Emergency Response (OSWER), EPA, 7, 8 Ogg1-null mouse studies, 447 “omics” revolution in, 604 studies, 390 technologies, 345, 588–589 in toxicology, 139 Oncologic Cancer Expert System, EPA’s, 530 one-hit hypothesis, 617 one-hit model, 619, 624 One Substance, One Registration (OSOR) framework, 19 ontogenesis, normal, 120 OPP. See Office of Pesticide Programs oral cavity, classifying neoplasms of, 706 oral studies, 381 organic solvents expert exposure studies on, 758 JEM for, 750 Organisation for Economic Co-operation and Development (OECD) genotoxicity guidelines of, 243 HPV chemicals investigation of, 45–47 international harmonized test guidelines of, 225 test guidelines of, 242 validation requirements of, 31 Organisation for Economic Co-operation and Development (OECD) guidelines for mammalian erythrocyte micronucleus test, 304 for in vitro micronucleus test, 283
for in vitro SCE assay, 326 for in vivo testing, 252 organophosphate exposures JEM for, 748, 749 organ toxicity, 344 OSWER. See Office of Solid Waste and Emergency Response, EPA ovarian cancer genetic mutations associated with, 403 risk for, 402 ovary, classifying neoplasms of, 708 oxidative stress after exposure to PPARα activators, 444 and hepatic carcinogenesis, 431, 431–433 markers of, 452 in PPARα activator MOA, 451 and species differences in PPARα activation, 457–458 oxide activation, metabolic activation of PAH through, 174 oxygen metabolism, and tumor promotion, 150 packaging, chemicals absorbed from, 78 pancreas, classifying neoplasms of, 707 pancreatic cancer, survival rate for, 404, 404 pancreatic islets, classifying neoplasms of, 707 Paneth cells, 136 papillomas, squamous cell, 702 parallel evolution model, of carcinogenesis, 137 particle handling, interspecies differences in, 71 particles carcinogenicity of, 535 carcinogenicity/toxicity of characteristics related to, 537–539 mechanisms of, 534–537 SAR analysis of, 539–540, 540 Pasteur, Louis, 100, 102 Pasteur Effect, 148 PCDFs. See polychlorinated dibenzofurans peer review, US OMB mandate for, 28–29 n-pentane, estimating exposures of, 754 perchlorate, exposure to, 617 perchloroethylene (PERC), estimating exposures of, 754
INDEX
perfluorooctanesulfonic acid (PFOS), biological monitoring of, 747 perfluorooctanoic acid (PFOA), 590–591 biological monitoring of, 747 chemical structure of, 531 permissible exposure limit (PEL), for asbestos, 72 peroxisomal ACO, 446 peroxisome proliferator-activated receptor (PPAR), 426 peroxisome proliferator-activated receptor alpha (PPARα), 85 and hepatocarcinogenesis, 439 in human liver, 428 in rodent hepatocarcinogenesis, 441–442, 442 peroxisome proliferator-activated receptor alpha (PPARα) activator, 443 defined, 440 human response to, 460–461, 461 allelic variants of, 461–462 differences in ligand inductibility, 462 differences in transcriptional networks, 463–465 PPARα gene and protein, 463 truncated PPARα, 463 MOA analysis after exposure to, 443, 444, 445 MOA for rodent liver tumors induced by, 443 and cell proliferataion/apoptosis balance, 447–449 chemically independent MOA, 455–456 complex MOAs, 454–455 inhibition studies, 450, 451, 452–454 mechanisms of cell growth alterations, 449–450 NF-κB in, 447 oxidative stress in, 444, 445–447 and species differences, 456–460, 457–458 MOA of, 465, 465–466 in rodent hepatocarcinogenesis, 440 species-specific effects of, 443 weak, 445 peroxisome proliferator-activated receptor alpha (PPARα) agonists, in rodent hepatocrcinogenesis, 425–428, 426
813
peroxisome proliferator-activated receptor alpha (PPARα) gene, 463 peroxisome proliferator-activated receptor alpha (PPARα) protein, 463 peroxisome proliferator-activated response elements (PPRES), 426, 426–427 peroxisome proliferators, 440 chemical structure of, 531 SAR of, 530–532 peroxisomes, 440 Perrier® water, concentration of benzene in, 30 persistent bioaccumulative and toxic substances (PBTs), classification as, 44, 45 pesticides biological monitoring of, 746, 747 cancellation of, 4 control of cancer risk from, 84 defined, 48 dietary exposures to, 80 endocrine effects of, 80 FFDCA provisions on, 79 under FIFRA, 48–49 in household products, 83 residues, 79 US EPA’s regulation of, 19 pesticide safety standard, single risk-only, 80 p53-deficient cells, in gene mutation testing, 279 p53 tumor suppressor protein, in tumorassociated fibroblasts, 130 p53 tumor supressor gene, 112 PGCs. See primordial germ cells P-glycoprotein (Pgp), 662 pharmaceutical industry and HTS studies, 604 and ICH guidelines, 244 pharmaceuticals genotoxicity assessment of, 259 testing for genotoxic impurities in, 256– 258, 258 pharmacodynamic processes, adult-child differences in, 568–569 pharmacodynamics, systems models of, 608 pharmacokinetic profiles, and PBPK models, 560 pharmacokinetics, systems models of, 608
814
INDEX
phenobarbital exposure to CAR, 429, 429–430 and ROS, 432 phosmet, exposure reconstruction for, 763 phosphate and formation of urinary solids, 509 in urine, 506 phthalates, biological monitoring of, 747 physiologically based pharmacokinetic (PBPK) models, 608 in cancer risk assessment, 563, 563–565, 564 for chloroform, 576–577 for dichloromethane, 574, 574–576 for dioxane, 577–578 extrapolation from individual carcinogens, 571–574, 573 high-dose to low-dose and interspecies extrapolation, 565–568, 566, 567 intraspecies extrapolation, 568–570, 570 route-to-route extrapolation, 571 for trichloroethylene, 578 for vinyl chloride, 575–576 for volatile organic chemical mixtures, 578–579, 579 characteristics of, 558–562, 559, 561, 562 conceptual representation of, 559 development and application of, 599–601 development of, 558 equations employed in, 560, 561 evaluation of, 562 examples of, 564 parameters of, 561 in rat-human extrapolation, 567 uncertainty, sensitivity, and variability analyses in, 562 using, 579–580 validation of, 561–562 physiologically based pharmacokinetics (PBPK), in risk assessment, 9 pica, 81 Pig-a assay, 329, 331, 333 Pig-a gene mutations, 344 pituitary gland, classifying neoplasms of, 707 plasmid lacZ, in vivo gene mutation assays in, 334, 336, 338 Plasmid pKM101, 274
platelets, differentiation of, 132–133, 133, 134 plausibility, and causal association, 409 plutonium, exposure reconstruction for, 763 point mutation, 227 point of departure (POD), 769 in carcinogenic evaluation, 671 in carcinogenicity measure, 684 defined, 365 EPA cancer guidelines for, 626 identification of, 625 tumor incidence data for, 720 politics, and evaluation of carcinogenesis mechanisms, 85 polonium, cancer associated with, 103 polybrominated biphenyls (PBBs), 428, 428–429 polychlorinated biphenyls (PCBs), 428, 428–429, 654 exposure reconstruction for, 745 nonlinear response to, 666 SAR of, 529–530 tolerances for, 78 polychlorinated dibenzofurans (PCDFs), Monte Carlo analysis of, 765 polychlorinated dioxins (PCDDs) Monte Carlo analysis of, 765 nonlinear response to, 666 polychromatic erythrocytes (PCEs), 304, 305 polyclonal competition model, 122 polycomb group (PcG), functions of, 140 polycyclic aromatic hydrocarbons (PAHs), 234 bay- or fjord-region diol epoxides of, 173–178 biological monitoring of, 746 isolated from coal tar, 169 JEM for, 749 metabolic activation of cyclopenta-ring oxidation mechanism, 181–182 diol epoxide mechanism in, 173–178, 174–176, 182 o-quinone/reactive oxygen species mechanism, 180, 180–181, 182 radical cation mechanism, 178–180, 179, 182 SAR analysis of, 527 simulation studies of, 744
INDEX
population risk, estimation of, 69 Portsmouth Naval Shipyard, 756 position effect variegation (PEV), 145 potassium bromate (KBRO3) carcinogenicity bioassay of, 207–208 lose-dose carcinogenicity of, 216, 218, 218–219 potency, carcinogenic defined, 682 time-dependent analysis of, 730 potency for cancer risk determination of, 732–733, 733 development of combined, 717 Pott, Sir John Percival, 168, 397 PPARα-null mice, inhibition studies in, 452 32 p-postlabeling assay, for measuring DNA adducts, 316, 318, 320, 322, 322–323 precautionary principle, 34, 86 predictivity, for in vitro testing, 286 Preliminary Assessment and Information Reporting (PAIR) rule, of TSCA, 43 Preliminary Remediation Goals (PRGs), of US EPA, 82 premalignant lesions reversibility of, 154 TSCE modeling of, 644–649 preputial gland carcinomas, 728, 729 preputial neoplasms, classifying, 705 prevention, cancer economic benefits of, 45 toxicological, 148 primordial germ cells (PGCs), apoptosis of, 141–142 probabilistic analysis, 764. See also Monte Carlo analysis probabilistic statistical methods, to estimate exposures, 753 process integrity, defined, 29 product development, (Q)SAR in, 544 product liability litigation, 30 and new risk assessment methods, 32 risk assessment in, 31 progenitor cells, epigenetically disrupted, 147–148 progression in cancer process, 420, 421, 422–423 for carcinogenesis in mouse skin, 183, 184 in rat carcinogenicity studies, 699
815
proliferation, and carcinogenesis, 59 promotion in cancer process, 420, 421, 422 of carcinogenesis in mouse skin, 183, 184 experiments in, 182, 182–183 use of term, 182 prophylaxis, in chemical carcinogenesis, 139 proportionate incidence ratio (PIR), 405 proportionate mortality ratio (PMR), 405 propylene glycol monobutyl ether (PGMBE), 487–488 prostaglandins, cancer inflammation propagated by, 127 prostate cancer genetic mutations associated with, 403 incidence of, 399, 399 survival rate for, 404, 404 prostate gland, classifying neoplasms of, 708 proteomics, 185, 390, 588, 589 P-stage biomarkers for, 120 in carcinogenic process, 104–105 evaluation of F-344 rat liver foci during, 105 GJ function in, 115–116 in progressive carcinogenesis, 117, 118, 154 public health agencies exposure levels determined by, 3 mission of, 3 public health goals (PHGs), 76 Q-real time RT-PCR, 185 (Q)SAR, types of, 518 (Q)SAR analysis assessing validity of, 519 classification for, 518–519 Critical Evaluation and Improvement of, 548 of epigenetic carcinogens, 528–529 expanding use of, 517 mechanistic understanding in, 520–521 QSAR equations in, 547 software for, 546 TD50 in, 518 uses for, 548
816
INDEX
(Q)SAR studies of chemical carcinogens, 520 model development in, 519 scientific soundness of, 519–520 qualitative estimation methods, for reconstructing exposures expert/professional judgment, 756, 757–758 exposure determinants for, 755–756 quantitative approach, to risk assessment, 557 quantitative methods, of exposure reconstruction, 738 biological monitoring data for, 743, 745, 746–747 current data for, 742 historical data in, 741–743, 744, 745 simulated exposure data for, 742–743, 744 quantitative structure-activity relationships analysis, 517. See (Q)SAR analysis quinoid compounds, SAR analysis of, 524 o-quinone, metabolic activation of PAH through, 174 quinones, SAR analysis of, 524 [4,5-f] quinoxaline (MeIQx) carcinogenicity bioassay of, 207 and DNA adduct formation, 211, 211, 213, 214 race and cancer incidence, 399, 400 and leukemia incidence, 401 radiation. See also ionizing radiation low-linear-energy-transfer, 650 and lung cancer, 650–651, 651 radical cations in metabolic activation of PAHs, 178– 180, 179, 182 metabolic activation of PAH through, 174 radiolabeling coupled with liquid scintillation counting, for measuring DNA adducts, 316, 318, 320, 322 radium, cancer associated with, 103 radon, multiroute exposure to, 71 Ramazzini, Bernardino, 100 rat. See also specific strains for bone marrow micronucleus test, 261 cancer risk assessment in, 208
as carcinogenesis model, 227 carcinogenicity bioassay in, 207 carcinogenicity studies in, 67 CD10 estimates in, 686, 687, 689, 691, 691, 691 combined chronic/cancer bioassay in, 379, 381 CPN study in, 489 experiments in kidney tumorigenesis with, 218–219 human relevance of experiments on, 495 MeIQx hepatocarcinogenicity in, 212 MTBE study in, 493–494 PhIP colon carcinogenicity in, 216, 216, 217 response to PPARα activators of, 457 RTT studies in, 491–492 screening for chemical carcinogens in, 501 rat colon, carcinogenicity of PhIP in, 215– 215, 216, 217 rat kidney cancers, use of data from, 32 rat liver enzyme-altered foci in, 653–655 MeIQx in, 209–214, 210, 210, 211, 212, 213, 214 N-nitroso compound hepatocarcinogenicity of, 215 for in vitro genotoxicity testing, 273, 274 rat liver foci model, 382 reaction kinetics, 630 reactive nitrogen species (RNS) and altered cancer metabolism, 127 and GJ function, 115 reactive oxygen species (ROS) and altered cancer metabolism, 127 cellular production of, 431, 433 and GJ function, 115 and hepatic carcinogenesis, 431–432 and infection, 126 and liver cancer, 445 overproduction of, 432 particles in formation of, 539 in tumorigenic processes, 180, 180–181, 182 recommended exposure limits (RELs), 72 “Red Book” (NRC), 6, 23, 25, 240 redundancy, in cell biology, 123 reference dose (RfD), 616
INDEX
Registration, Evaluation, Authorization, and Restriction of Chemicals (REACH), EU, 44–45, 391 Ames test required by, 254 animal experiments of, 255 Chemical Safety Assessment of, 254 classification system of, 254–255 objective of, 252–256 regulation on, 19, 85 scope of, 86 stepwise strategy of, 256 testing of chemicals under, 253 toxicological information required by, 252–253 in vivo genotoxicity tests in, 254 regulation basis for, 699 for cancer risk, 193 of genetic toxicity tests, 228–229, 230 goal of, 609 (Q)SAR analysis in, 546 and testing strategies, 391 regulatory acceptance, for in vivo genotoxicity assays, 294–295 regulatory agencies. See also specific agencies genotoxicity testing responsibilities of, 238–239 guidelines of, 25 validation requirements of, 31 regulatory decision-making, risk assessment in, 24 regulatory science, US, cancer risk assessment in, 194–198 regulatory standards air standards, 70–73 and cancer risk assessments, 66–69 consumer product standards, 82–84 food standards, 76–80 soil standards, 81–82 water standards, 73–76 relative source contribution, concept of, 75 relative total growth (RTG), 247 reliability, of scientific data, 32 renal tubule tumors (RTTs), 482 and exacerbation of CPN, 491–492 increase in incidence of, 491–492 and MIBK study, 495 MOA (α2U-globulin) for chemicals increasing, 483–489
817
renal tumors, chemicals increasing incidence of, 489–490 reporter genes, used in vivo gene mutation assays, 336–337, 340 reporting requirements, for industrial chemicals, 43 Report on Carcinogens (RoC), of US NTP, 40 Reregistration Eligibility Documents (REDs), 83–84 Resource and Conservation Recovery Act (1976), 82 respiratory system, classifying neoplasms of, 705 respiratory tract tissue, combining neoplasms of, 709, 710 restriction fragment length polymorphisms (RFLPs), 137 retinoid X receptor-alpha (RXRα), 426, 426 reverse mutation assays, basis of, 274, 275, 276 ribonucleic acids (RNAs), selective production of, 145 risk causal vs. spurious, 195 communication strategies for, 20 overestimation of, 610 perception of, 20 risk analysis and management, causal network for, 193, 194 risk assessment. See also cancer risk assessment application in dose-response analyses, 388–390 hazard characterization, 386–388 hazard identification, 383–386 application of rodent toxicity in, 419–420 assumptions in, 33–34 chemical, 607 classic model for, 736 data quality requirements in, 28–30 data used in, 30–33 defined, 598 enhancements in quantitative pharmacodynamic and dose-response modeling, 602 pharmacokinetic variability and uncertainty, 601–602
818
INDEX
risk assessment (cont’d) physiologically based pharmacokinetic modeling, 599–601 epidemiologic data in, 411 evolution of testing strategies in, 390–391 experimental studies required in, 392 goal of, 609 harmonization of cancer and noncancer risks in, 675–676 and hazard testing, 391–393 historical perspective on, 598–599 judicial review of, 30 modeling in, 606–607 new paradigm for, 590, 591 non-cancer, 659–660 original purpose of, 21 policy inputs into, 23 and precautionary principle, 34 principles of classification of carcinogens, 38–40 current carcinogenic, 40–41 process of, 598 dose-response characteristics in, 370 human relevance in, 372–375, 374 quantitative, 609, 711 in regulatory decision-making, 24–25 role of experimental studies in, 392 role of guidelines in, 25–27 science of, 4 steps in, 37 in 21st century with advances in molecular and computation biology, 16–17 and genetic susceptibility, 17 uncertainties in, 393 Risk Assessment Forum, of US EPA, 766 Risk Assessment Guidance for Superfund (RAGS) manuals, EPA (1989), 82 Risk Assessment of Genotoxic Carcinogens in Food Task Force, 78 risk-based prioritizations (RBPs), of ChAMP, 45 risk-benefit analyses, 19–20 risk characterization, 10, 38, 383, 736 risk characterization process, 363–364 key events in, 367–372, 368, 371, 372 MOA in, 365–367, 366, 366
risk management applications in acceptance and communication, 20–21 international, 18–19 risk-benefit analysis, 19–20 U.S., 17–18 for cancer risk assessments, 191 causal defaults in, 193 defined, 10 science of, 18 risk management measures (RMMs), 38 risk predictions, validation of, 197 Risk Science Institute (RSI), of ILSI, 367 risk-specific dose, 615 RIVM. See National Institute for Public Health and the Environment (RIVM), the Netherlands Robust Summaries, in EU approval process, 50 ROCK 1, 112 rodent cancer bioassays, hormesis in, 663 rodent models, formation of urinary solids in, 504–505. See also specific models rodents. See also carcinogenicity studies, rodent; guinea pig; hampster; mouse; rat as carcinogenesis model, 227 carcinogenicity studies in, 67 chronic/cancer bioassays in, 391 hormetic biphasic dose-responses in, 197 in safety bioassays, 434 screening for chemical carcinogens in, 501 as testing surrogates for humans, 419 toxicity studies with, 586–587 urinary solid carcinogenesis in, 508–510 urinary tract calculi in, 510–511, 512 RSI. See Risk Science Institute S-adenosylmethionine (SAM), 144 Safety Working Party (SWP), 257 Safe Water Drinking Act (1974), 76 safrole, 423, 425 salmonella testing, 239 genetic, 228 mutation test, 231 reverse mutation test, 241 standard genotoxicity assays, 68
INDEX
Salmonella typhimurium strains, in gene mutation tests, 274–275 SAM. See smoking attributable mortality sampling time, in in vivo genotoxicity assays, 293, 298–299 Sanitary and Phytosanitary (SPS) agreement, of WTO, 18–19 science, and risk assessment, 31–32 Science Advisory Board, US EPA, 32 science policy international differences in, 61 and risk assessment, 23–24, 34 scientific method, and science policy, 34 screening tests, 406 scrotum, cancer of, 168, 397 “seed and soil hypothesis,” 117 semiquantitative reconstructive methods for dermal or ingestion exposures, 753, 755 estimation methods, 751, 752, 753, 754, 755 exposure data matrices, 745, 748, 749–750 extrapolation methods, 748, 751 senescence, cell, 153 Setubal principles, 519 sex, and genotoxic metabolites, 68–69 Shapiro-Wilk p values, 685, 686, 686 Shh. See supersonic hedgehog siblings, and epigenetic control, 145–146 silica exposure reconstruction for, 761 JEM for, 750 simulation studies of, 744 silica exposures, JEMs for, 748 simulation studies, in exposure reconstruction, 742–743, 744 single-cell evolution cancer, 123 single-cell gel electrophoresis, 262, 284, 284–285, 285, 311. See also comet assay single-walled carbon nanotubes (SWCNT), 542 single-zone mixing factor model, 751, 752, 753 sister-chromatid exchange (SCE) assay, 228, 294, 297, 299, 301 interpretation of, 327–328 limitations of, 327–328 principle of, 326
819
purpose of, 326–328 regulatory acceptance of, 326 study design for, 326–327 skeletal muscle tissue, combining neoplasms of, 709 skeletal system, classifying neoplasms of, 709 skin classifying neoplasms of, 705 combining neoplasms of, 709 skin cancer, early study of, 102 skin tumors, initiation of, 179. See also mouse skin slimicides, NTP study of, 80 Small Business Liability Relief and Brownfields Revitalization Act (2002), 82 smoking and cancer incidence, 400, 401, 402 and lung cancer, 649 Smoking and Health Report, DHEW, 410 smoking attributable mortality (SAM), 400, 401, 402 smooth muscle tissue, combining neoplasms of, 709, 710 Society for Risk Analysis, 19 Society of Toxicology (SOT), US, 197 sodium salts, and formation of urinary solids, 509–510 soil, Monte Carlo analysis of, 765 soil standards regulatory considerations, 81–82 scientific issues, 81 solubilization, of urine, 507 sonic hedgehog (Shh), synthesis of, 140 SOT. See Society of Toxicology, US species, in in vivo genotoxicity assays, 300–301, 302 species differences molecular basis of, 460 in PPARα MOA, 456–460, 457–458 specificity, and causal association, 409 spot-test method, for screening mutagens, 228 spray emissions, estimating exposures of, 754 squamous cell carcinomas, in vitro production of, 107 staghorn calculi, 511
820
INDEX
state governments, waste site policies of, 82 statistical analysis in experimental studies, 382 for exposure reconstruction, 737 statistical inference models, 609 statistically significant response, and biological significance, 385 stem cell biology, and cancer, 130–139, 132–, 139 stem cell model, of carcinogenesis, 138 stem cells compartment, 131–132, 132, 134 differentiation of, 132–133, 134 epigenetically disrupted, 147–148 hierarchical compartment for, 136 homeodynamic equilibrium maintained by, 135 polyclonal disruption of, 147 re-genesis of, 135 steroids, and hepatic tumor risk, 430 Stochastic Human Exposure and Dose Simulation (SHEDS) model, US EPA, 753 stochastic techniques, to estimate exposures, 753 Stoddard solvent IIC (SS IIC), 487–488 stomach, classifying neoplasms of, 706 stomach cancer, and infection, 124 strained ring systems, SAR analysis of, 522 Strategic Plan for Evaluating Toxicity of Chemicals, USEPA, 8 stress, and carcinogenesis, 153 structural alerts (SA), 525 structure-activity relationship (SAR) of fibers and particles, 534, 534–540, 540 of nanomaterials, 540–544, 543 qualitative or semiquantitative, 518 structure-activity relationship (SAR) analysis expansion of database/knowledge base for, 548 expansion of integrative approaches for, 549 of genotoxic carcinogens direct-acting electrophilic functional groups, 522–526 requiring metabolic activation, 526–528
of nongenotoxic carcinogens, 528–534, 531 principles of mechanism-based, 521–522 styrene expert exposure studies on, 757 exposure reconstruction for, 762 JEM for, 750 PBPK modeling of, 599 subcutis, classifying neoplasms of, 705 subpopulations, susceptible, in U.S. EPA’s guidelines, 27 sulfate esters, cyclopenta-ring diols conjugated to, 182 sulfotransferases, PAPS, 170, 171 “Sunset Date,” 45 Superfund Amendments and Reauthorization Act (SARA) (1986), 82 Superfund program, 81–82 Superfund sites, risk assessment procedures for, 82 supersonic hedgehog (Shh), role of, 140 Supreme Court, US Daubert decision, 32 on risk assessment, 24 risk assessment controversy in, 29 Surveillance Epidemiology and End Results (SEER), NCI, 3 survival rates, five-year, 404, 404 susceptibility, adult-child differences in, 568 symmetrical mitotic division, 135 syncytium, defined, 113 Syrian hamster, response to PPARα activators of, 456, 457, 459 systems biology, 147, 606–609 systems modeling, 607–609 2,4,5-T, chemical structure of, 531 TCA, chemical structure of, 531 TD. See tolerable dose TD50 of carcinogenic potency, 682 experimental variation in, 683 Technical Guidance Document (TGD), in EU approval process, 50 telomeres, and aging, 153 temporality, and causal association, 409 teratocarcinoma, development of, 131 teratoma, in ectopic pregnancy, 131
INDEX
tert butyl alcohol (TBA), hyaline droplet accumulation associated with, 494 test batteries, in genetic toxicity testing, 232 testicle, classifying neoplasms of, 708 tests, for regulatory approval of commercial chemicals, 228–229, 230. See also toxicity testing 2,3,7,8-tetrachloro-dibenzo-dioxin (TCDD) exposure reconstruction for, 761 nonlinear response to, 666 tetrachlorodibenzo-p-dioxin (TCDD), 428 promotional effect of, 654–655 and ROS, 432 tetrachlorodibenzo-p-dioxin (TCDD)-related compounds, SAR analysis of, 529 tetrachloroethylene Monte Carlo analysis of, 765 NTP study of, 80 tetradecanoyl phorbol acetate (TPA), 104 threshold of toxicity definitions of, 627, 628 identification of, 659 minimum detection, 646 Threshold of Toxicological Concern (TTC), 85, 623 concept of, 257 generic staged values for, 258 regulatory policy on, 258 thymidine, tritiated, in UDS testing, 325 thyroid carcinogenesis, SAR of rodent, 532–533 thyroid gland, classifying neoplasms of, 707 tiered-testing approach, of US EPA, 47–48 time-to-tumor data, 732 time-weighted average (TWA) calculation of, 742 in exposure reconstruction, 738, 739 tissue damage, concentration-related, 74 tissues. See also specific tissues in PBPK model, 560 in in vivo genotoxicity assays, 300–301, 302 Tk assay, 329, 331, 333 TNF-related apoptosis-inducing ligand (TRAIL), 110, 110, 111 tobacco, lung cancer initiation, 649–650
821
toluene estimating exposures of, 754 expert exposure studies on, 757, 758 top dose, in in vivo genotoxicity assays, 292–293, 296–297 ToxCastTM, 605, 606, 609 Toxic Air Programs, of EPA, 674 “toxic equivalency” approach, 623 toxicity high-dose, 659 route-specific, 73–74 of urinary tract solids, 510 toxicity pathway, risk assessment based on, 16–17 Toxicity Prediction by Komputer-Assisted Technology (TOPKAT), 547 toxicity studies, conventional rodent, 586 toxicity testing, 605 basis of, 391 evolution of, 390 genetic, 225 (see also genetic toxicity testing) mandates for, 378 potential human adverse effects predicted by, 419 Toxicity Testing in the 21st Century initiative, 674–675, 676 toxicogenomics, 390, 593–594 defined, 589, 589 at transcriptional level, 594 toxicokinetic models, in risk assessment, 9 toxicokinetics, nonlinearities in, 627 toxicological methods, developing better, 32 toxicology computational, 597 goal of, 61 toxicology studies, 609 toxic response, dose-dependent transitions in, 620 Toxics Release Inventory (TRI), 86 Toxic Substances Control Act (1976), US, 42–44, 45, 72, 238 toxic tort, 30 and new risk assessment methods, 32 risk assessment in, 31 ToxRisk software, 719 ToxTree, 547 transcriptomics, 588, 589
822
INDEX
transformation stage, in carcinogenesis, 117, 118 transgenic animal models, 382 development of, 341 sources of oxidative stress in, 452 trans-muconic acid, biological monitoring of, 746 transplants, bone marrow stem-cell transfusions, 133 transporter proteins, in nonlinear carcinogenesis, 662 trend tests, 382 trichloroethylene (TCE) cancer risk of, 771 estimating exposures of, 754 exposure reconstruction for, 762 PBPK models for, 578 TSCA. See Toxic Substances Control Act, US tumor development key events for, 368, 368–369 role of inflammation in, 124 tumor dose-response curve concave, 626 shape of, 624 tumor enhancing properties, 383 tumorigenesis and host defense, 127 metabolic activation of PAH and, 173–182 NF-κB in, 128–129 tumorigenicity, dose-response curve for, 630 tumor incidence data, 67–68 tumor necrosis factor alpha (TNFα), 110, 110, 111 in HCC, 129 and PPARα activators, 449, 450 as tumor promotor, 129 tumor progression biology of, 109 role of stem cells in, 131–132 tumor promoters, 422 tumor promotion in carcinogenesis, 117, 118 and inflammatory mediators, 129 and oxygen metabolism, 150 tumors, 114. See also specific tumors and excessive cell production, 109 multicellular involvements in, 125
summing of related types of, 717–718 summing of unrelated types of addition of independent potency values, 718–719 affected-animal count, 718 distribution-based methods, 719–721, 725, 726, 726–730, 727–729, 731, 732 tumor-suppressor genes, genetic variation in, 58 twin studies, epigenetic mechanisms in, 145 two-stage clonal expansion (TSCE) model age-dependent model parameters for, 639 analysis of premalignant lesions using, 644–645 constant parameters in, 645–646 distribution of premalignant cells in, 648–649 general time-dependent parameters, 646–647 minimum detection threshold, 646 piecewise constant parameters in, 647 development of, 637–638, 638 dose-response in, 642–643 EAFs in, 654 interaction between agents studied with, 651 joint analysis of premalignant and malignant lesions under, 649 lag time in, 640 for lung cancer incidence, 649–652, 651, 652 quantitative reconstruction methods with, 638–639 two-zone model, for estimating exposure, 753, 754 umuD activity, deficiency in, 274 uncertainties accounting for scientific, 10–11 in cancer risk assessment, 10, 684 and conservative assumptions, 33 and conservative decision making, 193 in exposure assessment process, 736–737 in exposure reconstruction, 766 and extrapolation, 389 in risk assessment, 393 in tumor incidence data, 68
INDEX
uncertainty analysis, Monte Carlo technique, 764, 765, 766 United Kingdom, Committee of Mutagens of, 243 United States cancer in, 3 and ICH guidelines, 244 risk management in, 17–18 unit risk factor, for DCM, 575 unscheduled DNA synthesis assay, 294, 297, 299, 301 interpretation of, 325 limitations of, 325–326 principle of, 324 purpose of, 324 regulatory acceptance of, 324 study design for, 325 upper confidence limit (UCL), in risk assessment, 30 uranium cancer associated with, 103, 650, 651 exposure reconstruction for, 763 urethane, and in vivo testing, 264 urinalysis, 381, 512 urinary solids collection of urine for detection of, 507–508 formation of, 502–505, 503, 504 direct and indirect, 502–505, 504 and formation of tumors, 508–509 urinary factors influencing, 505–507 types of, 502–503 urinary system, classifying neoplasms of, 707 urinary tract solids, assessment of, 512 urine acidification of, 510 collection of, 507–508 composition of, 505 interspecies comparison, 508 pH of, 506 solubilization vs. crystallization in, 506–507 urothelium tissue, combining neoplasms of, 709 U-shaped dose-response, 663, 664 uterus, classifying neoplasms of, 708 validation, of new test methods, 31–32
823
validation studies, for genetic toxicity testing, 234 variability, in exposure reconstruction, 766 vascular endothelial growth factor (VEGF), 130 VCCEP. See Voluntary Children’s Chemical Evaluation Program VEGF. See vascular endothelial growth factor very persistent and very bioaccumulative substances (vPvBs), classification as, 44, 45 vinyl chloride (VC) angiosarcoma induced by, 565, 566 carcinogenic classification of, 413 carcinogenic threshold for, 665 exposure reconstruction for, 761 PBPK model for, 575–576 risk assessment of, 6 Virchow, Rudolf, 101–102, 124, 125 virtually safe dose (VSD), 671 calculation of, 576 defined, 683 determination of, 41 virtual tissue models, 609 viruses carcinogenic, 402 interaction with chemical agents of, 415 volatile organic chemical (VOC) mixtures, PBPK models for, 578–579, 579 volatile organic compound (VOC), in cancer risk assessment, 569 Voluntary Children’s Chemical Evaluation Program (VCCEP), of US EPA, 47–48 Warburg Glycolytic Effect, 148–149, 149, 150, 151 waste sites, abandoned hazardous, 81–82 water (H2O), insufficient oxidation, 148, 149 water standards. See also drinking water regulatory consideration, 75–76, 77 scientific issues, 73–75 Weibull model, for dose-response functions in carcinogenesis, 60–61, 724 weight of evidence (WOE), 3 in cancer assessment, 615 for combining neoplasms, 711 descriptors, 386
824
INDEX
weight of evidence (WOE) (cont’d) determination of, 5 for determining carcinogenicity animal models, 14–15 descriptors for, 15–16 epidemiologic studies, 12–14 to establish MOA in animals, 367–368 of PPARα activators, 465, 465–466 welding fumes, JEM for, 750 Western immunoblotting, 588 Wilder’s law of initial value, 56 wild-type mice, inhibition studies in, 452 Wistar rats, kidney tumorigenesis experiments in, 218–219 Wnt/β-catenin pathway, 127 Wnt pathway, 140–141 wood dust, exposure reconstruction for, 761 worker variability, in exposure reconstruction, 742 working groups, in IWGT process, 248 workshop, on MOE approach (2008), 78–79 World Health Organization (WHO)
IPCS of, 242, 249 NOAEL uncertainty factor approach used by, 41 World Trade Center Indoor Environment Assessment, 8 World Trade Organization (WTO), agreement on Sanitary and Phytosanitary of, 18–19 wounds, early tumors as, 130 WY-14,643, 452 chemical structure of, 531 in MOA analysis after PPARα activation, 444 xylene, JEM for, 750 p-xylene, simulation studies of, 744 zero risk tolerance, social and political influences on, 21 zero-tolerance policy failure of, 5 limitations of, 4–5 zinc oxide, simulation studies of, 744 zygote, 130