Principles of Anticancer Drug Development (Cancer Drug Discovery and Development)

Cancer Drug Discovery and Development Series Editor: Beverly A. Teicher, Genzyme Corporation Framingham, MA, USA

For other titles published in this series, go to www.springer.com/series/7625

Principles of Anticancer Drug Development Edited by

Manuel Hidalgo, MD, PhD Universidad CEU San Pablo, Spain

S. Gail Eckhardt, MD University of Colorado at Denver, USA

Elizabeth Garrett-Mayer, PhD Medical University of South Carolina, South Carolina, USA

Neil J. Clendeninn, MD, PhD CANAID, Inc., Hanalei, USA

Editors Manuel Hidalgo, MD, PhD Department of Oncology School of Medicine CEU San Pablo University Madrid, Spain and Centro Integral Oncológico Clara Campal (CIOCC) Madrid, Spain and Gastrointestinal Clinical Research Unit Centro Integral Investigaciones Oncológicas (CNIO) C/ Melchor Fernández Almagro 3 Madrid, Spain [email protected]

S. Gail Eckhardt, MD Professor and Division Head, Medical Oncology Stapp Harlow Chair in Cancer Research University of Colorado at Denver Aurora, CO 80045 USA [email protected] Elizabeth Garrett-Mayer, PhD Hollings Cancer Center Medical University of South Carolina Charleston, SC 29425 USA [email protected] Neil J. Clendeninn, MD, PhD CANAID, Inc Drug Development Consultant 96714 Hanalei USA [email protected]

ISBN 978-1-4419-7357-3 e-ISBN 978-1-4419-7358-0 DOI 10.1007/978-1-4419-7358-0 Springer New York Dordrecht Heidelberg London © Springer Science+Business Media, LLC 2011 All rights reserved. This work may not be translated or copied in whole or in part without the written permission of the publisher (Springer Science+Business Media, LLC, 233 Spring Street, New York, NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed is forbidden. The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights. While the advice and information in this book are believed to be true and accurate at the date of going to press, neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be made. The publisher makes no warranty, express or implied, with respect to the material contained herein. Printed on acid-free paper Springer is part of Springer Science+Business Media (www.springer.com)

Blurb A practical guide to the design, conduction, analysis and reporting of clinical trials with anticancer drugs.

v

Preface The development of cancer drugs, from the preclinical studies to final randomized clinical trial is a science per se. The process involves multiples distinct steps and requires the participation of multiple individuals with unique expertise such as toxicologist, pharmacologists, pathologists, statisticians, clinicians and ethics and regulatory experts. In addition, it involves several different organizations such as academic center, pharmaceutical industry and governmental organizations. Teaching and learning this process is not a simple task. The editors are seasoned clinical investigators with different backgrounds who spend considerable time teaching student and junior colleagues the nuts and bolts of drug development both at their own institutions but also through active participations in workshops and seminars on the topic. It became apparent to us that there are no sources in which the basis and principles of drug development are concisely summarize. This book has been written to fill that gap and to provide a guide for the beginners of drug development as well as a consultation manual for more advance drug developers. It is intended to provide a practical tool for the design, conduction, analysis and reporting of a clinical trial as well as to establish a developmental plan for a new agent. The book is organized into five parts – all of them written by experts and renowned authors who have done a great job putting the chapters together. Part I summarizes basic concepts in biostatistics and in clinical and analytical pharmacology that are needed to understand the clinical drug development process. Part II provides a comprehensive summary of preclinical studies that are required before a medical agent can be tested in humans. Part III deals with clinical trial design from phase I to phase III as well as with correlative studies in clinical trials including the more classic pharmacokinetics and the newer molecular imaging and tissue biomarkers. Part IV is an important section that outlines the FDA requirement for testing and approving a drug for cancer treatment. Part V focuses on more specific descriptions of developmental strategies for the different classes of anticancer agents ranging from conventional cytotoxic agents to molecularly targeted agents. The final section outlines the resources and perspective of the National Cancer Institute. We expect this book to be a night table manual and guide for those interested in the complex but rewarding field of anticancer drug development and the place to get started when training in this field. We also hope that this text book would be useful to our peer teachers in drug development. Manuel Hidalgo S. Gail Eckhardt Elizabeth Garrett-Mayer Neil J. Clendeninn

vii

Contents Part I   1 Basic Biostatistics for the Clinical Trialist............................................. Elizabeth G. Hill and Elizabeth Garrett-Mayer

3

  2 Fundamental Concepts in Clinical Pharmacology............................... Daniel L. Gustafson and Erica L. Bradshaw-Pierce

37

  3 Bioanalytical Methods in Clinical Drug Development......................... Walter J. Loos, Peter de Bruijn, and Alex Sparreboom

63

Part II   4 Preclinical Models for Anticancer Drug Development......................... Edward A. Sausville

89

Part III   5 Phase I Clinical Trials with Anticancer Agents.................................... 117 Stephen Leong, Justin Call, Alex Adjei, and Wells Messersmith   6 Phase II Trials with Anticancer Agents................................................. 141 Hui K. Gan, J. Jack Lee, and Lillian L. Siu   7 Phase III Clinical Trials with Anticancer Agents................................. 163 Wendy R. Parulekar and Daniel J. Sargent   8 Pharmacokinetic Studies in Early Anticancer Drug Development................................................................................... 189 Alex Sparreboom and Sharyn D. Baker   9 Pharmacodynamic Studies in Early Phase Drug Development........... 215 D. Ross Camidge, Robert C. Doebele, and Antonio Jimeno 10 Prediction of Antitumor Response......................................................... 257 Fred R. Hirsch and Yu Shyr 11 Imaging Studies in Anticancer Drug Development.............................. 275 David A. Mankoff ix

x

Contents

Part IV 12 Role of the US Food and Drug Administration in Cancer Drug Development................................................................. 305 Ann T. Farrell, Ramzi N. Dagher, and Richard Pazdur Part V 13 Early Clinical Trials with Cytotoxic Agents.......................................... 335 M.J.A. de Jonge and Jaap Verweij 14 Challenges and Successes in Developing Effective Anti-angiogenic Agents............................................................................ 347 Laura Q.M. Chow and S. Gail Eckhardt 15 Targeted Therapeutics in Cancer Treatment........................................ 403 Colin D. Weekes and Manuel Hidalgo 16 Cancer Chemoprevention........................................................................ 463 Christopher H. Lieu, William N. William Jr, and Scott M. Lippman 17 Combined Modality Therapy in Cancer Management........................ 483 David Raben and Kyle Rusthoven 18 Cancer Vaccines....................................................................................... 519 Daniel Laheru 19 Optimising the Development of Antibodies as Treatment for Cancer.......................................................................... 535 Craig P. Carden, Hendrik-Tobias Arkenau, and Johann S. de Bono 20 Oligonucleotide Therapeutics................................................................. 569 Cy A. Stein, Britta Hoehn, and John Rossi 21 Anticancer Drug Development in Pediatric Patients............................ 589 Lia Gore and Margaret Macy 22 Clinical Trials in Special Populations..................................................... 603 S. Percy Ivy, Merrill J. Egorin, Chris H. Takimoto, and Jeannette Y. Wick Part VI 23 NCI-Sponsored Clinical Trials................................................................ 631 Andriana Papaconstantinou and Janet E. Dancey Index.................................................................................................................. 659

Contributors Alex A. Adjei, MD, PhD, FACP Department of Medicine, Roswell Park Cancer Institute, Elm & Carlton Streets, Buffalo, NY 14263, USA [email protected] Dr. Hendrik-Tobias Arkenau, MD, PhD The Medical Professorial Unit, Prince of Wales Medical School, University of New South Wales, Level 1, South Wing, Edmund Blacket Building, Avoca Street, Sydney, NSW, 2031, Australia [email protected] Sharyn D. Baker, PharmD, PhD Department of Pharmaceutical Sciences, St. Jude Children’s Research Hospital, 262 Danny Thomas Place, CCC Room I5308, Mail Stop 313, Memphis, TN, 38105-3678, USA [email protected] Erica L. Bradshaw-Pierce, PhD Department of Pharmacokinetics, Dynamics and Metabolism, Pfizer Global Research and Development, 10646 Science Center Drive, San Diego, CA 92121 [email protected] Justin Call, MD Medical Oncology, Developmental Therapeutics Program/GI Malignancies, University of Colorado Cancer Center, Mail Stop 8117, PO Box 6511, Aurora, CO, 80045, USA [email protected] D. Ross Camidge, MD, PhD Developmental Therapeutics and Thoracic Oncology Programs, Clinical Thoracic Oncology Program, University of Colorado Cancer Center, Aurora, CO, USA [email protected] Craig P. Carden, MBBS, FRACP Drug Development Unit, Section of Medicine, The Institute of Cancer Research, The Royal Marsden Hospital NHS Trust, Sutton, London, SM2 5PT, UK [email protected]

xi

xii

Contributors

Laura Q.M. Chow, MD, FRCPC Division of Medical Oncology, Department of Medicine, University of Washington, 825 East lake Avenue East (SCCA) MS: G4-940, Campus Box 358081, Seattle, Washington, USA 98109-1023 [email protected]; [email protected]; [email protected] Ramzi N. Dagher, MD Worldwide Regulatory Strategy, Global Regulatory, Pfizer Inc., 50 Pequot Avenue MS6025-C5141, New London, CT, 06320, USA; Oncology Business Unit, Pfizer Inc., 50 Pequot Avenue MS6025-C5141, New London, CT, 06320, USA [email protected] Janet E. Dancey, MD, FRCPC Investigational Drug Branch, Cancer Therapy Evaluation Program, Division of Cancer Treatment and Diagnosis, National Cancer Institute, 6130 Executive Blvd, EPN 7131, Rockville, MD 20892, USA [email protected] Peter de Bruijn, BSc Laboratory of Translational Pharmacology, Department of Medical Oncology, Erasmus University Medical Center, Groene Hilledijk 301, PO Box 5201, 3008 AE, Rotterdam, The Netherlands [email protected] Dr. Johann S. de Bono, MD, FRCP, MSc, PhD Section of Medicine and Drug Development Unit, Institute of Cancer Research, Royal Marsden Hospital, Downs Road, Sutton, Surrey, SM2 5PT, UK [email protected] M.J.A. de Jonge, MD, PhD Department of Medical Oncology, Erasmus University Medical Center (Daniel den Hoed Kliniek), P.O. Box 5201, 3008 AE, Rotterdam, The Netherlands [email protected] Robert C. Doebele, MD Division of Medical Oncology, Department of Medicine, University of Colorado Cancer Center, University of Colorado at Denver Anschutz Medical Campus, Aurora, CO, USA [email protected] S. Gail Eckhardt, MD Professor and Division Head, Medical Oncology Stapp Harlow Chair in Cancer Research University of Colorado at Denve, Aurora, CO 80045, USA [email protected]

Contributors

xiii

Merrill J. Egorin, MD, FACP University of Pittsburgh Cancer Institute, Room G27E, Hillman Research Pavilion 5117 Centre Avenue, Pittsburgh, PA, 15213-1863, USA [email protected] Ann T. Farrell, MD Division of Hematology Products, Office of Oncology Drug Products (OODP), Center for Drug Evaluation and Research, US Food and Drug Administration, Silver Spring, MD, 20993-0002, USA [email protected] Hui K. Gan, MBBS, FRACP, PhD Drug Development Programme, University Avenue, Room 5-224, Princess Margaret Hospital 610, Toronto, ON, M5G 2M9, Canada [email protected] Elizabeth Garrett-Mayer, PhD Hollings Cancer Center, Medical University of South Carolina, Charleston, SC 29425, USA [email protected] Lia Gore, MD, FAAP The Children’s Hospital, Center for Cancer and Blood Disorders and Developmental Therapeutics Program, University of Colorado Cancer Center, Box B115 13123 East 16th Avenue, Aurora, CO, 80045, USA [email protected] Daniel L. Gustafson, Ph.D Department of Clinical Sciences, Colorado State University, ACC226, Veterinary Teaching Hospital, 300 West Drake Road, Fort Collins, CO, 80523-1620, USA; Pharmacology Core, CU Cancer Center, Colorado State University, ACC226, Veterinary Teaching Hospital, 300 West Drake Road, Fort Collins, CO, 80523-1620, USA; CSU Animal Cancer Center, Colorado State University, ACC226, Veterinary Teaching Hospital, 300 West Drake Road, Fort Collins, CO, 80523-1620, USA [email protected] Manuel Hidalgo, MD, PhD Department of Oncology, School of Medicine, CEU San Pablo University, Madrid, Spain; Centro Integral Oncológico Clara Campal (CIOCC), Madrid, Spain; Gastrointestinal Clinical Research Unit, Centro Integral Investigaciones Oncológicas (CNIO), C/ Melchor Fernández Almagro 3, 28029, Madrid, Spain [email protected] Elizabeth G. Hill, PhD Biostatistics Core, Hollings Cancer Center, Medical University of South Carolina, 86 Jonathan Lucas Street, Room 118D, MSC 955, Charleston, SC, 29425 – 9550, USA [email protected]

xiv

Contributors

Fred R. Hirsch, MD, PhD Department of Medical Oncology, University of Colorado Cancer Center, Aurora, CO 80045, USA [email protected] Britta Hoehn, PhD City of Hope National Medical Center, 1450 E. Duarte Road, Duarte, CA, 91010, USA [email protected] S. Percy Ivy, MD Investigational Drug Branch, Cancer Therapy Evaluation Program, Division of Cancer Treatment and Diagnosis, National Cancer Institute, 6130 Executive Blvd, Suite 7131, Rockville, MD 20852, USA [email protected] Antonio Jimeno, MD, PhD Medical Oncology, University of Colorado Cancer Center, Mail Stop 8117, PO Box 6511, Aurora, CO, 80045, USA;Developmental Therapeutics/ Pharmacodynamic Laboratory, Developmental Therapeutics, Head and Neck Cancer and Stem Cell Programs, University of Colorado Cancer Center, Mail Stop 8117, PO Box 6511, Aurora, CO 80045, USA [email protected] Dan Laheru, MD Skip Viragh Center for Pancreas Cancer Clinical Research and Patient Care, The Sol Goldman Pancreatic Cancer Research Center, The Sidney Kimmel Comprehensive Cancer Center Bunting-Blaustein, The Johns Hopkins University School of Medicine, CRB Room G89, 1650 Orleans Street, Baltimore, MD 21231, USA [email protected] J. Jack Lee, PhD, MS, DDS Division of Quantitative Sciences, Department of Biostatistics, The University of Texas M.D. Anderson Cancer Center, 1400 Pressler Street, Unit Number: 1411 Room Number: FCT4.6012, Houston, TX, 77030, USA [email protected] Stephen Leong, MD Medical Oncology, Developmental Therapeutics Program/GI Malignancies, University of Colorado Cancer Center, Mail Stop 8117, PO Box 6511, Aurora, CO, 80045, USA [email protected] Christopher H. Lieu, MD Department of Thoracic/Head and Neck Medical Oncology, Unit 432, The University of Texas MD Anderson Cancer Center, Houston, TX, 77030-4009, USA

Contributors

xv

Scott M. Lippman, MD Department of Thoracic/Head and Neck Medical Oncology, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Blvd., Unit 432, Houston, TX 77030-4009, USA [email protected] Walter J. Loos, Ph.D Laboratory of Translational Pharmacology, Department of Medical Oncology, Erasmus University Medical Center, Groene Hilledijk 301, PO Box 5201, 3008 AE, Rotterdam, The Netherlands [email protected] Margaret Macy, MD The Children’s Hospital, Center for Cancer and Blood Disorders, 13123 East 16th Avenue, Aurora, CO 80045, USA [email protected] David A. Mankoff, MD, PhD Department of Radiology, Seattle Cancer Care Alliance, G2-600, 825 Eastlake Avenue East, Seattle, WA 98102, USA [email protected] Wells Messersmith, MD, FACP GI Medical Oncology Program, University of Colorado Cancer Center, Mail Stop 8117, 12801 East 17th Avenuem, Aurora, CO 80045, USA; Division of Medical Oncology, University of Colorado Cancer Center, Mail Stop 8117, 12801 East 17th Avenue, Aurora, CO, 80045, USA [email protected] Andriana Papaconstantinou, PhD Technical Resources International Inc., 6500 Rock Spring Drive, Suite 650, Bethesda, MD, 20817, USA [email protected] Wendy R. Parulekar, MD, FRCPC NCIC Clinical Trials Group, Queen’s University, Kingston, ON, Canada; Department of Oncology, Queen’s University, Kingston, ON, Canada [email protected] Richard Pazdur, MD Office of Oncology Drug Products, Center for Drug Evaluation and Research, U.S. Food and Drug Administration, Silver Spring, MD, 20993-0002, USA [email protected]

xvi

Contributors

David Raben, MD Department of Radiation Oncology, University of Colorado Denver, Anschutz Cancer Pavilion, MS F-706, 1665 N. Ursula St., Suite 1032, Aurora, CO 80045-0510, USA [email protected] John Rossi, PhD Department of Molecular and Cellular Biology, Dean, Irell and Manella Graduate School of Biological Sciences, Beckman Research Int. of City of Hope, 1500 East Duarte Road, Duarte, CA, 91010, USA [email protected] Kyle Rusthoven, MD Department of Radiation Oncology, University of Colorado Health Sciences Center, 1665 N. Ursula St., Suite 1032, Denver, CO, 80045-0508, USA [email protected] Daniel J. Sargent, PhD Division of Biomedical Statistics and Informatics, Mayo Clinic, Rochester, MN, 55905, USA [email protected] Edward A. Sausville, MD, PhD Marlene and Stewart Greenebaum Cancer Center, University of Maryland, 22 S. Greene Street, Room S9D07, Baltimore, MD, 21201, USA [email protected] Yu Shyr, Ph.D Cancer Research, Vanderbilt University School of Medicine, Nashville, TN, USA; Division of Cancer Biostatistics, Vanderbilt University School of Medicine, Nashville, TN, USA; Cancer Biostatistics Center, Vanderbilt-Ingram Cancer Center, 2220 Pierce Avenue, 571 Preston Building, Nashville, TN 37232-6848, USA [email protected] Lillian L. Siu, MD Division of Medical Oncology and Hematology, Princess Margaret Hospital, University of Toronto, 610 University Avenue, Suite 5-718, Toronto, ON, Canada, M5G 2M9 [email protected] Alex Sparreboom, PhD Department of Medical Oncology, Erasmus MC, Rotterdam, The Netherlands and Department of Pharmaceutical Sciences, St. Jude Children’s Research Hospital, Memphis, TN, USA [email protected]

Contributors

xvii

Cy A. Stein, MD, PhD Department of Oncology, Albert Einstein-Montefiore Cancer Center, Montefiore Medical Center, 111 E. 210 Street, Bronx, NY 10467, USA [email protected] Chris H. Takimoto, MD, PhD Translational Medicine, Ortho Biotech Oncology R&D, 145 King of Prussia Road, Mail Stop RA-2-2, Radnork, PA, 19087, USA [email protected] Colin D. Weekes, MD, PhD Developmental Therapeutics Program/GI Oncology, University of Colorado Health Science Center, Mail Stop 8117 RC1 South, Rm 8123 12801 E. 17th AvenueAuroraP.O. Box 6511, CO, 80045, USA; Shana Spears, Denver, CO, USA [email protected]. Jeannette Y. Wick Pharmaceutical Management Branch, Cancer Therapy Evaluation Program, National Cancer Institute, Rockville, MD, USA [email protected] William N. William Jr, MD Department of Thoracic/Head and Neck Medical Oncology, Unit 432, The University of Texas MD Anderson Cancer Center, Houston, TX, 77030-4009, USA Jaap Verweij, MD, PhD Dept. of Medical Oncology, Erasmus University Medical Center, PO Box 2040, 3000 CA, Rotterdam, The Netherlands; Gravendijkwal 230, 3015 CE, Rotterdam, The Netherlands [email protected]

Part I

Chapter 1

Basic Biostatistics for the Clinical Trialist Elizabeth G. Hill and Elizabeth Garrett-Mayer

1.1 Introduction The purpose of this chapter is to acquaint the reader with some typically used biostatistical principles and methods in anticancer drug development for summarizing and analyzing data. Understanding and properly interpreting statistics is critically important for drug development. Each stage of development, ranging from preclinical studies to phase III clinical trials, utilizes some form of statistical analysis whether it is as simple as the calculation of a mean or as complex as a longitudinal model with a complicated correlation structure. Proper statistical design and analysis will be critical for making valid inferences and moving to the next phase of research. Most cancer centers and pharmaceutical companies will have a biostatistics group or division. The role of these biostatisticians is to support cancer research by assisting with study design, statistical analysis, and presentation of results. It is highly recommended that, in addition to understanding the basic statistical principles utilized in oncology, oncology drug developers utilize the biostatisticians in their institution or company and consider and treat them as part of the research team. It is well-known that drug development cannot be done independently and requires a host of experts: biostatistical expertise is critical to valid and efficient research, from basic science to preclinical research to clinical trials.

1.2 Example The most effective way to demonstrate statistical methods used in drug development is by example. In the sections that follow, we present selected results from a phase II study of the farnesyltransferase inhibitor tipifarnib in patients with acute E. Garrett-Mayer (*) Division of Biostatistics and Epidemiology, Hollings Cancer Center, Medical University of South Carolina, 86 Jonathan Lucas Street, Room 118G, Charleston, SC 29425, USA e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_1, © Springer Science+Business Media, LLC 2011

3

4

E.G. Hill and E. Garrett-Mayer

myelogenous leukemia (AML) [3]. Briefly, farnesyltransferase inhibitors (FTIs) are potent and selective inhibitors of intracellular farnesyltransferase (FTase) which is an enzyme that catalyzes the transfer of farnesyl moiety to the cysteine terminal residue of a substrate protein. A number of intracellular proteins are substrates for prenylation via FTase (including Ras). Interruption of prenylation may prevent substrates from maturation which may result in inhibition of cellular events that depend on the function of those substrates. FTIs have been shown to be “unselective,” targeting proteins involved in different pathways, despite the initial presumption that FTase inhibition would specifically target posttranslational processing of Ras. The use of FTIs is expanding in the treatment of hematological malignancies, especially in AML but also in other leukemias, myelodysplastic syndromes, and myeloproliferative disorders in large part due to their oral bioavailability and acceptable toxicity profile. The FTI tipifarnib (Zarnestra, R115777) had been shown to have in vitro activity against a wide range of malignancies. Based on promising preliminary results in phase I testing, a multicenter phase II study was developed in poorrisk previously untreated AML patients. One-hundred fifty-eight patients with poor-risk AML were enrolled between March 2001 and December 2005 and followed in this single-arm study. “Poor-risk” was defined as having at least one of the following (1) age ³ 65; (2) adverse cytogenetic profile; (3) AML arising from an antecedent hematologic disorder; or (4) therapy-related AML. Patients received 600 mg oral tipifarnib twice daily for 21 days followed by a rest period of up to 42 days to allow for peripheral blood count recovery. Additional cycles were administered (up to a total of four) if stable disease or ­clinical response were observed. Clinical endpoints of interest included complete and partial remission, duration of complete remission (CR), overall survival, and tolerability. The authors also explored correlative endpoints including farnesylation of the surrogate protein HDJ-2 (measured at baseline and day 8) and normalized baseline expression levels of ERK and AKT phosphorylation and their relation to clinical outcomes. Note that only 15% (24 of 158) of AML patients had evaluable data for ERK and AKT expression. Additional information relating to the rationale, study design, and measurement of correlative endpoints can be found in the original article by Lancet et al. [3].

1.3 Aims, Endpoints, and Data Analysis The distinction between a study’s aims and endpoints is often unclear to cancer researchers, and subsequently these terms are commonly confused with one another. In most cancer research, a plan is developed where specific aims or goals are stated. In the AML example described below, the study’s primary aim was to define the antileukemic activity of tipifarnib (the investigational agent) in adults with poor-risk, previously untreated AML [3]. Secondary study aims described evaluating expression levels of ERK and AKT, and examining their ability to predict response to agents such as tipifarnib. For each aim, there is a corresponding

1  Basic Biostatistics for the Clinical Trialist

5

endpoint (or outcome) used for that aim’s quantitative assessment. An endpoint should be measurable in the sense that it can be observed and recorded for each individual in the study, whether the individual be mouse, patient, or cell line. For example, a measure of tipifarnib’s efficacy is its ability to induce a CR, and so, in the AML example, the investigators evaluate patients’ CR status (achieved or not) as the primary study endpoint. Whether or not a patient has a CR is determined by clinical criteria outlined in the study’s design, and CR status is recorded for each study subject. The endpoint is summarized across subjects using data analysis, resulting in both an estimated CR rate, as well as a measure of that estimate’s uncertainty. Data analysis facilitates the averaging of information across subjects, resulting in measures (here, estimated CR rate and its associated uncertainty) that provide an objective assessment of a study’s primary aim. To summarize, for any research project, the aims need to be clearly stated. For each aim, an endpoint of interest is identified which provides for a quantitative assessment of the aim. At the completion of the study, inference is made regarding the study’s aim via formal statistical analysis methods, some of which are described in this chapter.

1.4 Variable Types Biostatisticians in drug development generally refer to their data as comprising a set of “variables” because in most instances, the numerical observable data varies across subjects, where a subject may be a cell line, a mouse, a patient, etc. It is important to distinguish between several different kinds of variables and to specify, when defining a variable, how it is measured. As an example, we are often interested in gene expression. However, it is not always clear how gene expression is measured: in a given study, the researcher could be using two categories of expression (expressed vs. not expressed), or she could be using a numeric value of expression that can take any value within a specified range. These two different types of variables are treated differently in statistical analyses.

1.4.1 Continuous Variables Continuous variables can take any value within a given (and potentially wide) range of values. In our tipifarnib AML study, an example of a continuous variable is expression of phosphorylated AKT. Although it is does not have a wide distribution, the values of baseline AKT in our study range from 0.03 to 2.11, as shown in Fig. 1.1a. Another commonly utilized continuous variable is age. Note that age can be measured in fine increments, such as weeks, days, or even minutes. Age measured in days or weeks is commonly used in animal studies; however, for clinical applications age expressed in years is generally the preferred metric.

6

E.G. Hill and E. Garrett-Mayer

Fig. 1.1  (a) Density plot of day 8 AKT expression values. Tick marks along x-axis indicate observed data points. Mean and median values are indicated by vertical solid and dashed gray lines, respectively. (b) Density plot of day 8 AKT expression values on the log scale. Mean and median are the same and indicated by vertical solid gray line. Horizontal black line at height of 0.1 indicates the 95% confidence interval for the true day 8 log AKT expression value. (c) Density plot of difference between day 0 and day 8 log AKT expression. Horizontal black line at 0.1 shows 95% confidence interval for the difference. Vertical gray line is plotted at 0, indicating no difference in values

1  Basic Biostatistics for the Clinical Trialist

7

1.4.2 Categorical Variables Categorical variables have several categories to which an individual may belong. Categorical variables with only two categories are called binary or dichotomous and examples could include gender (with categories of male and female), mutant (with categories of mutant vs. wild-type), or clinical response (with categories of nonresponders and responders). More than two categories are possible as well: a variable with three categories could be genotype, with levels defined as homozygous dominant, homozygous recessive, and heterozygous. Note that there is no specific ordering to genotype: it could be coded numerically with 1 = homozygous dominant, 2 = heterozygous, and 3 = homozygous recessive. Or, the numeric assignments could be transposed without any loss of interpretation. This implies that genotype is a nominal categorical variable. Another common example of a nominal categorical variable is race, which can take a number of categories, but the numeric values assigned are irrelevant. Another class of categorical variables is ordinal variables, where there are a fixed (and relatively small) number of categories, but the ordering is meaningful. Common examples of ordinal variables in clinical cancer research are cancer stage, performance status, or grade. For example, there are discrete values assigned to cancer stage and the ordering of the categories is meaningful: stage 2 is higher than stage 1, and stage 3 is higher than stage 2.

1.4.3 Time-to-Event Variables The predominant clinical outcomes in cancer research are time-to-event variables: overall survival (time to death), progression-free survival (time to progression or death), and disease-free survival [time to relapse or recurrence (or death)]. Time-toevent variables are defined by the occurrence of an event. In a clinical trial, the time from enrollment until death is used to measure overall survival. At first glance, this may seem to be a continuous variable because it is a time that can be measured in very small increments. However, most time-to-event variables in cancer research have the additional characteristic that they can be censored, meaning that some of the individuals under study may not experience the event during the time course of the experiment or trial. In the example of overall survival, patients who do not die by the end of the study are considered censored at the time at which they were last known to be alive.

1.4.4 Variable Transformation In many cases, variables will naturally take one form, but be transformed to another for convenience. For example, age is a continuous variable, but for the purposes of

8

E.G. Hill and E. Garrett-Mayer

analysis and interpretation, it may make more sense to create a new variable with three age categories, such as <40 years, 40–65 years, and >65 years. This may have some utility, as mentioned, for interpretation, but some information about age is lost. Specifically, when using the categorical example of age above, two patients whose ages are 66 and 91 are considered equivalent with respect to age. Other common transformations are applied to continuous variables to reduce skewness or asymmetry in their distributions. Skewness can create problems in data analysis by allowing a few data values that are relatively extreme (i.e., outliers) to have substantial influence on inferences. An example of skewness can be seen in Fig. 1.1a where a density plot of AKT at day 8 is shown. Notice that most of the data lies close to the left side of the plot toward 0; however, there are a few points scattered to the right. This distribution is called right skewed (or positively skewed) because it has a long right tail. To symmetrize the distribution of day 8 AKT, we can apply a logarithmic transformation, shown in Fig. 1.1b, that results in “pulling in” the right tail and making the distribution look more bell shaped. Notice how the points that may have been considered outliers in Fig.  1.1a would no longer be labeled as outliers after this transformation.

1.5 Data Description and Displays In statistical practice, there is an important distinction between a parameter and a statistic. A parameter is a quantity whose true value is unknown and is the measure we are trying to estimate. For example, in theory there is a true CR rate to tipifarnib in poor-risk AML patients. This could be determined by treating every poor-risk AML patient with tipifarnib and observing their response. This approach is, of course, impractical. Instead, we collect data on a sample of patients from the identified population, and construct a statistic (or estimate) as our best guess of the true parameter’s value. Statisticians are also concerned with an estimated parameter’s uncertainty – how much faith do I have that the estimate represents the truth? – and so accompanying each estimated statistic is a measure (usually an interval) describing a range of values consistent with the data within which the true parameter could lie. Thus, the parameter is the unknown value we are trying to make inferences about, and the statistic and its associated uncertainty are quantities calculated based on sample data.

1.5.1 Continuous Variables Continuous variables have a number of summary statistics used to describe their distributions which generally fall into two common types: statistics to summarize the center of the distribution, and those to describe the data’s variability or spread. Statistics used to summarize the center are usually the mean and the median.

1  Basic Biostatistics for the Clinical Trialist

9

The mean is simply the arithmetic average, calculated by adding up all the observed values of the variable and dividing by the number of values. The median is the middle observation (or 50th percentile) and can be found by sorting the data from lowest to highest and identifying the value in the middle of the sorted list. In the case where there is an even number of values, the median is the average of the middle two data points. In our AML example, the mean AKT expression on day 8 is 0.51 and the median is 0.33, as shown in Fig. 1.1a. In the case of skewed data, this is a common result: the median and the mean are different. The mean will be quite sensitive to skewness and extreme values, while the median will not be sensitive. In Fig. 1.1a, notice that the median is closer to the bulk of the data points while the mean tends to be displaced in the direction of the outliers. When describing the center of skewed data, the median is often preferred. Now consider Fig. 1.1b where a log transform of AKT on day 8 has been applied. Because the data is symmetric (i.e., it is not skewed), the mean and the median are almost the same (in this example they are the same to four decimal places). Variability is most commonly measured by the range and the standard deviation (SD). The range is the difference between the largest and smallest values, but instead it is common to report the minimum and maximum values for a particular variable. The SD is a one-number summary that describes how far the data tend to deviate from the mean. In the AKT example in Fig. 1.1a, the range is 0.03–2.11 and the SD is 0.51. The standard error is a related measure of variability and will be described later when confidence intervals are discussed. Another measure of spread of the data is the interquartile range (IQR). Recall that the median is the middle data point, or the 50th percentile. Using the same approach of sorting the data, we can identify the 25th and the 75th percentiles of the data. The IQR is defined as the 75th percentile minus the 25th percentile. For expression of AKT at day 8, the 25th and 75th percentiles are 0.20 and 0.53, resulting in an IQR of 0.53 − 0.20 = 0.33. As is the case with the range, it is common practice to report the 25th and 75th percentiles rather than their difference. At least as important as the summary statistics used to quantify the center and spread of continuous variables are data displays that show the overall distribution. There are various ways to display the distribution of a variable, one of which (a density plot) is shown in Fig.  1.1. Other common plots are boxplots, histograms and, dotplots. Figure 1.2 demonstrates each of these plots for the distribution of age in the AML clinical trial example. The boxplot (for age, shown in Fig.  1.2a), also known as a box and whisker plot, emphasize quartiles of the distribution and its skewness. The lower and upper limits that define the box are the 25th and 75th percentiles. The line crossing the middle of the box indicates the location of the median (i.e., the 50th percentile). As noted above, the IQR is the distance between the 25th and 75th percentile. In Fig. 1.2, the 25th, 50th, and 75th percentiles are 69, 74, and 78, and the IQR is 78 − 69 = 9. The upper whisker is the line drawn from the box out to the smallest data point within 1.5 times the IQR from the 75th percentile. In the age ­distribution in Fig.  1.2a, the 75th percentile is 78 and 1.5 times the IQR is 9 × 1.5 = 13.5. Hence, the whisker could be drawn as far as 78 + 13.5 = 91.5. However, the largest

10

E.G. Hill and E. Garrett-Mayer

Fig. 1.2  Graphical displays of age for 158 high-risk AML patients treated with tipifarnib. (a) Boxplot. (b) Histogram. (c) Dotplot

observed age in the study is 85 so the whisker stops at 85. The lower whisker is defined in an analogous way and here, the 25th percentile minus 1.5 times the IQR is 69 − 13.5 = 55.5, and so the whisker is drawn to 56, the largest observed value

1  Basic Biostatistics for the Clinical Trialist

11

above 55.5. There are, however, two outliers: age values of 34 and 46. These are indicated using ­individual points in the plot. That is, any points that fall outside the allowed limits for the whiskers are plotted using individual points. To interpret a boxplot, we look at the location of the median in relation to the upper and lower quartiles and the length of the whiskers. These comparisons provide information about the relative symmetry vs. skewness of the data, and also provide a range where the bulk of the data lie: we know the middle 50% of the data lies within the extent of the box, and that the remaining 50% are above and below the box. The whiskers and outliers provide information about the tails of the distribution. A boxplot with one whisker that is significantly longer than the other implies skewness. A histogram is another popular data display tool for continuous variables. It bins the data into a number of categories and then plots the number of observations in each bin vs. each category. This is not the same as a bar chart which is more general. The y-axis of a histogram provides either the frequency or proportion of observations in a bin. (The same is not true of a bar chart.) In Fig. 1.2b, age is plotted in bins with widths of 5 years. Like the boxplot, this figure provides information about the skewness, range, and center of the data. For age, we see that there is evidence of left-skewness (due to the left tail) and there are two outliers with values below 50 years. Note that the size of the bins can alter the interpretation of histograms. Most statistical software packages have a default algorithm for determining the width and number of bins. However, this varies across packages and altering the bin width can lead to different inference. The dotplot is a very simple tool to see all of the raw data of a variable. It is most useful in situations where the sample size is relatively small, and there is interest in looking at a particular continuous variable across subgroups. Figure 1.2c shows the dotplot for age. As in Fig. 1.2a, b, two outliers are notable and some left-skewness is seen. Notice also that the data are “jittered” horizontally, facilitating visualization of overlapping data points. Failure to add noise to the plot makes it is impossible to tell how many data points are represented by a single symbol. In our example of age, although there are 158 patients in the study, there are only 28 unique values of age so that jittering the points is imperative to displaying all the data in a figure such as Fig. 1.2c.

1.5.2 Categorical Variables Categorical variables are summarized using tabulations of counts and proportions or percentages. In the case of a binary variable, such as gender, the proportion of male individuals provides all the information necessary to summarize its distribution. For categorical variables with three or more categories, proportions and counts per category are used. Table  1.1 shows a tabulation of counts and percentage of AML patients in each of three response categories.

12 Table 1.1  Distribution of response in the tipifarnib AML study

E.G. Hill and E. Garrett-Mayer

Complete remission Partial remission/hematologic improvement Nonresponse a Total

N 22 15

% 13.9 9.4

121 158

76.6 100

a  Includes stable disease, progressive disease, and not evaluable for response

1.5.3 Time-to-Event Variables Recall that time-to-event variables are characterized by censoring when some of the individuals under study do not experience the event of interest by the end of the study, or are lost to follow-up before the event has occurred. As a result, means, SDs, and other summary statistics appropriate for continuous variables are not valid. However, the median applies in situations where a large fraction of the individuals have incurred the event. Other time-to-event summary statistics used include the estimated survival fraction at a given time point. For example, in the AML example, the median survival among patients who did not respond to treatment was only 3.6 months, while the median survival among patients who had a CR was 14.4 months. The estimated fraction of patients alive at 12 months are 13% and 66% in nonresponders and responders, respectively. Note that the fraction surviving and median survival are not calculated using the methods described in previous sections for continuous or categorical variables. Censoring needs to be accounted for and the most common approach for estimating these summary statistics is by using the product-limit estimator, also known as the Kaplan–Meier estimator. Without providing great detail, the fraction of patients without the event is estimated at each time point, accounting for how many patients are still at risk of experiencing the event at that time point (called the risk set). Patients are removed from the risk set when they have had an event or are censored. This approach is very commonly accepted and explained in greater detail in [5, 6]. Figure 1.3 demonstrates overall survival in our AML study, where patients are defined by three categories (1) complete remission (CR), (2) partial remission and hematologic improvement (PR/HI), and (3) nonresponse (NR). The display is called a Kaplan–Meier plot because the Kaplan–Meier estimates of overall survival are shown. Notice that the curve for each group is a step-function relating time, t, to the fraction of individuals alive at time t, denoted S(t). Each step represents a time at which one or more patients has had the event of interest, and so the curve steps down, indicating a lower survival fraction at that point. On the survival curves, in addition to the steps indicating when events occurred, there are tick marks indicating the times at which patients who do not experience the event are censored. This provides information as to what time the patient left the risk set, and is important for understanding the censoring patterns and the number of patients still under study at any given time.

1  Basic Biostatistics for the Clinical Trialist

13

Fig. 1.3  Kaplan–Meier curve of overall survival for patient experiencing CR (black solid line), partial remission/hematologic improvement (gray solid line), and nonresponse (light gray solid line). Median survival is indicated by dotted black lines. Vertical tick marks on survival curves show censoring times

The survival curve always begins at S(t) = 1 for t = 0, indicating everyone under study is at risk of the event at time 0. The median survival for a group can be found by drawing a horizontal line at S(t) = 0.5 and evaluating the time where it intersects the survival curve. In Fig. 1.3, the horizontal line at 0.5 intersects the curves at 3.6, 12.5, and 14.4 months for the CR, PR/HI, and NR groups, respectively. Other estimates, such as the 12-month survival, can be found be drawing vertical lines up from a particular time of interest. Twelve-month survival estimates for these three groups are 66, 51, and 13% for the CR, PR/HI, and NR groups, respectively.

1.5.4 Confidence Intervals Confidence intervals provide information about a likely range of values for a given parameter of interest, such as a mean expression level or a true response rate. Confidence intervals are created for many parameters of interest and provide a measure of the estimated parameter’s precision. We most often see 95% confidence intervals, but 90 and 99% confidence intervals are also fairly common. A 95% confidence interval is an interval which we are 95% certain contains the true value of the parameter. For example, the mean of log AKT at day 8 is −1.1 and the range of log AKT at day 8 is −3.6 to 0.8. The estimated 95% confidence interval for log AKT at day 8 is (−1.5, −0.7). This means that we are 95% confident that the true mean of log AKT at day 8 lies somewhere between −1.5 and −0.7. Note that the

14

E.G. Hill and E. Garrett-Mayer

95% confidence interval provides inference about the mean: it does not provide information about a likely range of values that we might observe for individuals in the population. This can be noted by looking at the 95% confidence interval in Fig. 1.1b, which is shown as a small horizontal black line at a height of 0.1. The confidence interval is relatively narrow compared to the range of the data. The width of the confidence interval depends on three things (1) the variability of the data in the sample (i.e., patient heterogeneity with regard to the variable of interest), (2) the level of confidence desired (e.g., 95%), and (3) the sample size. The variability of the data in the sample will depend on the patient population you choose. For example, the variability in PSA (prostate-specific antigen) values in a sample of healthy male volunteers will be much smaller than the variability of PSA in a sample of men with relapsed prostate cancer. We expect that men without prostate cancer will all have PSA values in the range of 0–4 ng/ml, while men with refractory prostate cancer will have PSA values ranging anywhere from 4  ng/ml into the tens of thousands. This latter group will have much greater variability which will affect our confidence in a mean estimate. As noted previously, the level of confidence chosen is most often 95%, but in some cases, a 90 or 99% confidence level is justified when we are satisfied with less confidence while gaining a narrower range, or require greater confidence at the expense of a wider interval. The width of the confidence interval also depends directly on the sample size: as the sample size increases, the width of the confidence interval decreases. This is intuitive: the more information we collect, the more certainty we have in our estimate. For formulas for construction of confidence intervals, see [1, 5, 6].

1.5.5 Confidence Intervals for Means and Differences in Means In the previous section, the 95% confidence interval for mean day 8 log AKT was presented and is also shown in Fig. 1.1b. In addition to the mean value at day 8, we may also be interested in the mean of the difference between log AKT at day 0 and day 8. By calculating a confidence interval for this difference, we obtain both a range of reasonable values for the difference as well as evidence to support or refute the hypothesis that the difference differs from zero. A common use of the confidence interval is to test whether a difference in means is equal to zero: if 0 is not within the 95% confidence interval we conclude that the difference differs meaningfully from 0. This is an example of the duality between confidence intervals and hypothesis testing (hypothesis testing is described in Sect. 1.6 of this chapter). Using the difference in log AKT between days 0 and 8 as an example, we construct a 95% confidence interval by taking the difference between the day 0 and the day 8 log AKT values resulting in a single calculated difference per patient. The data for this is shown in Fig.  1.1c, where a density plot is shown in addition to the observed ­differences along the bottom of the figure. The estimated mean difference is −0.11 and the 95% confidence interval for the mean difference is (−0.68, 0.46), indicated by the horizontal black line at a height of 0.10. This implies that we are 95% confident that

1  Basic Biostatistics for the Clinical Trialist

15

the true average difference lies somewhere between −0.68 and 0.46. The vertical gray line indicates a difference of 0 and is the location at which there is no difference between the day 0 and day 8 values. Notice that the 95% confidence interval overlaps this vertical line suggesting that the mean difference between the day 0 and day 8 log AKT values does not differ from 0.

1.5.6 Confidence Intervals for Proportions and Comparisons of Proportions The interpretation of confidence intervals remains generally the same, regardless of the parameter of interest. In the case of proportions, we use a different method for estimating the confidence interval, but nonetheless it has the same meaning. In the AML tipifarnib example, 22 of 158 patients, or 14%, had a CR. The report of this statistic will be better understood by providing a 95% confidence interval which will convey, in addition to our observed remission rate, a range of likely true remission rates if this treatment approach were applied in general to the high-risk AML population (consistent with those patients in our trial). As described above, the confidence interval width depends on the sample size and our level of confidence. It also depends on the variability of remissions in the population but, in the case of proportions, the variability is determined by the true remission proportion. For example, if the true proportion is near 1 (or 0) most of the subjects will (or will not) experience a remission, and therefore the variability in remissions is low. Conversely, variability in the event is highest for true proportions near 0.5. The 95% confidence interval for the true CR rate to tipifarnib in AML patients is (0.09, 0.20). The confidence interval for the true CR rate is somewhat asymmetric. We may have expected the observed remission rate of 0.14 to lie in the middle of the interval, but this is not the case, and the asymmetry is not due to rounding. As estimates of proportions get close to 0 or 1, the corresponding confidence intervals become increasingly asymmetric. For example, only three patients experienced a partial remission, yielding an observed partial remission rate of 0.02 with a 95% confidence interval of (0.004, 0.05). Here, the distance between the estimated remission rate and the interval’s upper bound is roughly twice the distance from the estimated rate to the lower bound. The second thing to notice about the 95% confidence interval for the CR rate to tipifarnib is its fairly narrow width of 0.11 (0.11 = 0.20 − 0.09). As mentioned above, the width of the interval depends on the sample size. If our sample size had been only 50, an observed CR rate of 0.14 would have a 95% confidence interval of (0.06, 0.27), for a width of 0.21. And, if we had a much larger sample size of 400, the width would be only 0.07. Often confidence intervals for proportions are created using approximate approaches. These approximations work very well under two conditions (1) the sample size is reasonably large, and (2) the proportion is not close to 0 or 1. It is difficult to provide rules determining reasonably large and not close to 0 or 1 because they depend on each other. For example, a proportion of 0.90 is not very

16

E.G. Hill and E. Garrett-Mayer

close to 1 if the sample size is 1,000 but it would be considered close to 1 if the sample size were only 20. But, in general, almost all statistical software packages can generate exact confidence intervals, so reliance on approximations is not necessary, although it is still very commonly seen. There are other parameters that we use to compare proportions in different subpopulations. For example, odds ratios or relative risks are often used for quantifying the risk or benefit associated with an exposure or treatment. Of the AML patients treated with tipifarnib who had at least three cycles of treatment 41.2% (14/34) had a CR, compared to only 6.5% (8/124) among patients who did not complete three or more cycles. We can use an odds ratio to represent this difference in CR by cycles. Specifically, the odds ratio for CR in this example is 10.15: the odds of experiencing a CR for patients who are able to complete three cycles of treatment is ten times that of the odds for patients who had fewer than three cycles. We calculate this by taking the odds of a CR in patients with three or more cycles (41.2/58.8% = 0.700) and dividing it by the odds of CR in patients with fewer than three cycles (6.5/93.5% = 0.069). The 95% confidence interval for the odds ratio is then used to determine if there is an association between the exposure (i.e., three or more cycles) and the outcome (CR). In this case, the 95% confidence interval for the odds ratio for CR in patients with and without at least three cycles of tipifarnib is (3.4, 31.2). Recall that an odds ratio of 1 indicates no association between CR and three or more cycles of treatment. Because this confidence interval does not contain one, we can conclude that there appears to be a significant association between receiving three or more cycles of tipifarnib and CR. However, note that this is a rather simplistic analysis where we have not adjusted for additional confounders that may play a role in a patient’s ability to receive three or more cycles. It is likely that patients least able to tolerate treatment also have other factors making them less likely to respond to treatment.

1.5.7 Confidence Intervals for Time-to-Event Parameters The interpretation of the confidence interval for time-to-event parameters, such as median survival or 12-month survival, is the same as for other types of parameters. Recall that the estimated median survival for AML patients who had a CR to tipifranib was 14.4 months and the median survival in patients who were nonresponders was 3.6 months. The associated 95% confidence intervals for median survival (in months) are (9.7, Inf) and (2.9, 5.2), where “Inf ” represents a bound of infinity. This is not uncommon: in cases where there are relatively few patients on study when the median survival is achieved and the survival does not drop dramatically below 0.5 by the end of the study, the upper limit of the 95% confidence interval may be infinite. To interpret this, we would state that we are 95% confident that the true median survival in AML patients who achieved a CR is greater than 9.7 months. Similarly, we are 95% confident that the true median survival in nonresponders lies somewhere between 2.9 and 5.2 months.

1  Basic Biostatistics for the Clinical Trialist

17

1.6 Hypothesis Testing 1.6.1 From Research Question to Statistical Hypothesis In the Lancet study, 10 of 75 (13%) poor-risk AML subjects with an unfavorable cytogenic profile achieved a CR. A natural question to ask is how this rate compares to CR rates in the same patient population receiving standard treatment. Is the tipifarnib rate better? Worse? Different? Because subjects in the Lancet study received only tipifarnib, there is no internal comparative arm. However, investigators frequently use published or historical rates in comparable patient populations to compare treatments in single-arm studies. For example, Leith et al. [4] conducted a study of elderly AML patients in which they investigated the association between patient cytogenics and response. They report a 21% CR rate among elderly poor-risk AML patients in response to standard chemotherapy (standard-dose cytosine arabinoside and daunomycin + rhG-CSF). For illustrative purposes, we use the CR rate reported by Leith et al. [4] as the historical CR rate with which to compare the tipifarnib rate. To quantify the relationship between the tipifarnib and chemotherapy CR rates, we use hypothesis testing, a statistical approach that allows us to draw conclusions from sample data and infer to the entire population. Hypothesis testing begins with a statement of “no effect,” appropriately called the null hypothesis (H0). For the current example, our null hypothesis states that the tipifarnib CR rate is equal to the historical chemotherapy rate. Specifically, we write H0: ptipifarnib = 0.21, where ptipifarnib is the true CR rate among elderly poor-risk tipifarnib-treated AML patients with unfavorable cytogenics. A second statement, called the alternative hypothesis (H1 or sometimes HA), summarizes the research question of interest and is phrased in contrast to H0. Here, a reasonable alternative hypothesis states the tipifarnib rate is different from the historical chemotherapy rate and is written as H1: ptipifarnib ¹ 0.21. The latter hypothesis is called a two-sided alternative and captures in a single statement two one-sided alternatives, specifically (1) the tipifarnib rate is better than the chemotherapy rate (ptipifarnib > 0.21) and (2) the tipifarnib rate is worse than the chemotherapy rate (ptipifarnib < 0.21). A twosided alternative is appropriate when there is no reason to assume a priori that the effect of the new treatment will be better or worse than that of the standard treatment.

1.6.2 Evaluating Evidence Through p-values A 13% CR rate is smaller than the 21% published chemotherapy rate, but this ­difference may be a chance occurrence, that is, an observation not attributable to tipifarnib treatment. Is there sufficient evidence in the data to allow us to rule out random variation as an explanation for the observed tipifarnib rate? Stated another way, if the true CR rate among this subgroup of elderly poor-risk tipifarnib-treated AML patients is the same as the historical chemotherapy rate of 21%, how unusual is an observation of 10 CRs in 75 subjects?

18

E.G. Hill and E. Garrett-Mayer

To answer this question, we need an understanding of the distribution of the f­ requency of observed CRs under conditions specific to this study. These studyspecific conditions refer to the composition of the study population (elderly poorrisk tipifarnib-treated AML patients with unfavorable karyotype); the sample size of 75 subjects; and the null-hypothesized tipifarnib CR rate of 21%. Under these conditions, how many CRs in 75 subjects should we expect to observe? How ­variable is the number of CRs in 75 subjects? Addressing these questions requires repeating the study many times under identical conditions and observing the ­number of CRs for each repetition. A more practical approach is to conduct a computer-based simulation in which a virtual “coin,” with probability of a head equal to 0.21, is tossed 75 times and the number of heads observed. Tossing a coin and recording heads or tails is like observing a patient from our study population following treatment with tipifarnib and deciding if the subject has or has not experienced a CR. Using a computer, the simulation can be repeated thousands of times in a matter of seconds, rapidly providing information pertaining to variability needed to address our hypothesis. Figure 1.4a shows a bar chart of the number of CRs observed in 100 repetitions of our simulated study. From this graph we note that 10 CRs were recorded in four of the 100 repetitions, or 4% of the simulations. If we could conduct our simulation an infinite number of times, the bars would “smooth out” and we would observe a bar chart like the one shown in Fig.  1.4b. Figure  1.4b shows the exact sampling distribution of the number of CRs in 75 subjects, where the true CR rate is 21%. Here the bar heights are probabilities, where the probability of an outcome is loosely defined as the long-term proportion of simulations in which that outcome is observed. The probability of 10 CRs in 75 subjects is 0.031. To summarize how unusual an observation is, statisticians generally sum the probabilities of all outcomes at least as extreme as the one observed, where the “extremeness” of an event is measured by how probable it is relative to the observed outcome. In this case, nine or fewer CRs are extreme events since each is less probable than the observed outcome of 10 CRs in 75 subjects. For the same reason, 22 or more CRs are considered extreme. The bars corresponding to extreme events are shaded dark gray in Fig.  1.4b, and each bar’s height (probability) is no greater than the height of the bar corresponding to the observed outcome of 10. Therefore, the probability of observing 10 CRs in 75 subjects, or any observation at least as extreme, is found by summing the heights of the dark gray bars in Fig. 1.4b and is equal to Pr(10 CRs) 

probability of observed event

+ Pr(9 CRs) + Pr(8 CRs) +  + Pr(1 CR) + Pr(0 CRs)  probability of events at least asextreme as, and to the “ left ”of the observed event

+ Pr(22 CRs) + Pr(23 CRs) +  + Pr(74 CRs) + Pr(75 CRs) = 0.12.    probability of events at least as extreme as and to the “ right”of the observed event

This probability is an example of a p-value. A p-value is always calculated assuming that the null hypothesis is true – in this case that the true tipifarnib CR rate is 21% – and represents the probability that the observed result or one more extreme is a ­random event. If a p-value is small – usually less than 0.05 – we ­eliminate random chance as an explanation for the observed results and reject the null hypothesis. A finding for

a

19

12

1  Basic Biostatistics for the Clinical Trialist

0.06

Probability 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Number of CRs in 75 subjects

0.00

0.02

4 2 0

0.04

Frequency 6

0.08

8

0.10

10

b

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 Number of CRs in 75 subjects

0.00

0.02

Density 0.04 0.06

0.08

0.10

c

0

5 10 15 20 25 Number of CRs in 75 subjects

30

Fig. 1.4  (a) Bar chart of the frequency of the number of CRs in 75 subjects for 100 simulations assuming the true CR rate is 21%. (b) Exact sampling distribution of the number of CRs in 75 subjects, assuming the true CR rate is 21%. Each bar’s height is the probability of observing the number of CRs indicated on the horizontal axis. (c) Normal approximation to the exact sampling distribution of the number of CRs in 75 subjects, assuming the true CR rate is 21%

which the p-value is smaller than 0.05 is said to be statistically significant or simply significant. A p-value greater than or equal to 0.05 indicates that, under the null hypothesis, the observed result is not so unlikely – the event could occur by chance 5% of the time or more – and we fail to reject the null hypothesis. Such findings are called nonsignificant. For our example, the p-value is 0.12, which is greater than 0.05. We therefore fail to reject the null hypothesis and conclude that the data provide ­insufficient evidence to claim that the true CR rate for elderly poor-risk tipifarnib-treated AML patients with unfavorable karyotype differs meaningfully from 21%.

20

E.G. Hill and E. Garrett-Mayer

On a final note, statisticians often compute p-values using an approximation to the exact sampling distribution of a statistic. This alleviates the need to construct exact sampling distributions that change as conditions vary from one problem to the next. This is similar to the approximation discussed in Sect. 1.4.2 for confidence intervals and is shown in Fig. 1.4c, the normal approximation density curve.

1.6.3 Types of Errors With each decision one can make concerning the null hypothesis – reject or fail to reject – there is a corresponding potential error or mistake. If the null hypothesis is rejected when in fact it is true, this is called a type I error. On the other hand, failing to reject the null hypothesis when in fact it is false is called a type II error. The probability of a type I error is represented by the Greek letter alpha (a) and the probability of a type II error is represented by the Greek letter beta (b). The layout in Table 1.2 displays the interpretations of type I and type II errors in the context of the hypothesis test of the tipifarnib CR rate. In this example, a type I error occurs if we conclude that the tipifarnib CR rate differs from the historical chemotherapy rate when it really is not different; a type II error occurs if we conclude that the tipifarnib rate is the same as the chemotherapy rate when it really is different. In drug discovery, we are typically more concerned with type I errors since rejecting the null hypothesis in error means a nonefficacious drug may advance to larger, more expensive (e.g., randomized) trials and, more importantly, patients will receive a drug that offers no additional clinical benefit. Phase II trials are often designed with a = b = 0.10 and phase III trials with a £ 0.05 and b £ 0.20.

1.7 Common One- and Two-Sample Tests 1.7.1 Comparing Proportions The hypothesis test highlighted in Sect. 1.6 is called a one-sample test of a proportion. This test is appropriate when interest surrounds relating a true but unknown proportion to a reference value. The test accounts for sampling variability inherent in

Table 1.2  Type I and type II errors in the context of the hypothesis test of the tipifarnib CR rate Truth Decision Tipifarnib CR rate Tipifarnib CR rate

Same as chemotherapy CR rate

Different from chemotherapy CR rate

Same as chemotherapy CR rate Different from chemotherapy CR rate

No error (1 − a) Type I error (a)

Type II error (b ) No error (1 − b )

1  Basic Biostatistics for the Clinical Trialist

21

estimating the unknown proportion but treats the reference value as a constant. For our example, the historical chemotherapy rate was actually constructed from an observation of 11 CRs in 52 subjects. Treating the 21% reference rate as “truth” ignores the fact that it was estimated from sample data, and therefore subject to sampling variability. An alternative test that accounts for the sampling variability in both the estimated tipifarnib and chemotherapy CR rates is a two-sample test of proportions whichis most commonly seen in randomized studies. Here the null hypothesis is H0:ptipifarnib = pchemotherapy, where ptipifarnib is defined in Sect. 1.5.1, and pchemotherapy is the true but unknown CR rate for elderly poor-risk chemotherapy-treated AML patients. The corresponding two-sided alternative is H0: ptipifarnib ¹ pchemotherapy. The test is conducted based on binary data from the two sets of patients under comparison. In this example, we construct the test based on a comparison of 10 CRs in 75 subjects with 11 CRs in 52 subjects. The p-value for this test is 0.17. This is slightly larger than the p-value for the corresponding one-sample test because the twosample test incorporates the uncertainty associated with estimating the chemotherapy CR rate in addition to the tipifarnib rate. The interpretation of the two-sample test is also slightly different. Our conclusion for the one-sample test was the true tipifarnib CR rate did not differ significantly from 21%. Here we conclude that the true tipifarnib CR rate does not differ significantly from the true chemotherapy CR rate, whatever that rate may be.

1.7.2 Comparing Means Comparisons between groups of continuous data are commonly constructed based on the relative locations of the data distributions’ centers, and the most common measure of central tendency is the mean. For example, to compare baseline levels of ERK phosphorylation (p-ERK) between responders and nonresponders, appropriate null and alternative hypotheses are H0: mresp = mnonresp and H1: mresp ¹ mnonresp, where mresp is the true but unknown average baseline level of p-ERK among responders, and mnonresp is the true but unknown average baseline level of p-ERK among nonresponders. Here responders are defined as subjects achieving either a complete or partial remission, or a hematologic improvement. Both the null and alternative hypotheses can be expressed equivalently based on a difference in means – that is, H0: mresp − mnonresp = 0 vs. H1: mresp − mnonresp ¹ 0 – and the test is carried out in a manner similar to that described in Sect. 1.5.2. Specifically, the test is based on estimates of the average baseline p-ERK levels among the eight responders and 21 nonresponders for whom biologic correlative data are available. The mean (SD) p-ERK levels for responders and nonresponders are 0.55 (0.75) and 0.36 (0.27), respectively. Additionally, it is important to understand characteristics of the distribution of the difference in sample means. Under assumptions that p-ERK baseline levels are normally distributed and independently sampled from both groups, and that the variance of p-ERK levels is unknown, the

22

E.G. Hill and E. Garrett-Mayer

shape of the distribution of the difference in sample means is unimodal and symmetric – much like a normal ­distribution. However, the distribution of the difference has “heavier” tails than a normal distribution, which is to say that extreme differences are more likely to occur than would be the case had the distribution been normal. The distribution of the difference in sample means under the stated assumptions is known as the t-distribution, and the corresponding hypothesis test is called a two sample t-test. (A one-sample t-test also exists and is appropriate when testing a true but unknown mean against a reference value.) In conducting any hypothesis testing, it is important to first evaluate how well underlying assumptions are satisfied. When assumptions are violated, the resulting inference is potentially compromised. As illustrated in Fig.  1.1a, day 8 p-ERK levels are positively skewed. Baseline p-ERK levels similarly violate the normality assumption (figure not shown). The mean and SD for baseline p-ERK levels provide additional evidence that approximate normality is not satisfied. A normally distributed variable has the property that 68% of its values fall within 1 SD of the mean, 95% within 2 SDs, and 99% within 3. Notice that for responders, 1 SD to the left of the mean (0.55 − 0.75), and for nonresponders 2 SDs to the left of the mean (0.36 − 2 × 0.27), results in implausible values – p-ERK expression levels cannot be negative. What can be done? As illustrated in Fig. 1.1b, a logarithmic transformation of day 8 p-ERK levels alleviates the distribution’s skewness resulting in a more symmetric, approximately normal shape. A logarithmic transformation of baseline p-ERK levels induces the same approximate normality (figure not shown). We therefore conduct a two-sample t-test on the log-transformed baseline p-ERK values. The null and alternative hypotheses are similarly stated, but mresp and mnonresp now represent the true but unknown average log baseline p-ERK levels among responders and nonresponders, respectively. The p-value associated with this test is 0.72 leading to the conclusion that average log baseline p-ERK levels do not differ significantly between responders and nonresponders. Using properties of logarithms,1 an equivalent conclusion is the average baseline p-ERK expression ratio comparing responders to nonresponders is not significantly different from 1. When transformation fails to induce normality, a test based on the ranked data is an alternative to the two-sample t-test. If a variable’s distribution for one group is centered at a larger value relative to a second group, data sampled from the first group will likely have larger ranks than data sampled from the second group. This is the idea behind the Wilcoxon rank-sum test, an example of a nonparametric test. A nonparametric test is one that makes no assumption about the form of the sample data’s distribution. The Wilcoxon rank-sum test is the nonparametric equivalent to the two-sample t-test. If we use the Wilcoxon rank-sum test to assess the association between baseline p-ERK level and response status, the p-value is 0.65, and our

1  Since log(A) − log(B) = log(A/B), then the equation log(A) − log(B) = 0 is equivalent to log(A/B) = 0. Exponentiation of both sides of the latter leads to the equivalent expression, A/B = 1.

1  Basic Biostatistics for the Clinical Trialist

23

conclusion is the same – baseline p-ERK levels do not differ significantly between responders and nonresponders.

1.7.3 The Chi-Square Test Tipifarnib is a FTase inhibitor so an important component of the Lancet study was the assessment of FTase inhibition. In addition to exploring FTase inhibition in AML isolates, Lancet study investigators examined inhibition in buccal (cheek) mucosa samples to determine if inhibition could be detected in normal tissue. The investigators report FTase inhibition failure in AML isolates from 14 of 57 (25%) subject samples but in only four of 49 (4%) normal tissue samples. In their discussion, the authors postulate that this difference potentially indicates a patient subpopulation with FTase posttranslational modification or possibly an alteration in drug accumulation, and may identify a patient cohort unlikely to benefit from tipifarnib. Table 1.3a shows a two-by-two table of the observed distribution of FTase inhibition status (yes or no) by sample type (AML isolate or normal tissue from buccal mucosa). This table is an example of a contingency table and is used to display the joint distribution of categorical variables. Usually, interest surrounds understanding the association (if any) between the row and column variables. Consistent with hypothesis testing strategies already presented, we construct a test of association under a null condition; we assume sample type and FTase inhibition status are independent in the sense that a sample’s origin – AML isolate or normal tissue – does not influence FTase inhibition.

Table 1.3  Observed (A) and expected (B) frequencies of FTase inhibition status (yes or no) by sample type (AML isolate or buccal mucosa) Farnesyltransferase inhibition status Sample type Yes No Total A: Observed AML isolate 43 14 57 Buccal mucosa 45 4 49 Total 88 18 106 B: Expected AML isolate 47.3 9.7 57 Buccal mucosa 40.7 8.3 49 Total 88 18 106 Expected frequencies are derived based on probability laws that assume independence between row and column variables

24

E.G. Hill and E. Garrett-Mayer

Using laws of probability, we derive a table of the frequencies we would expect to see if the variables under consideration really were independent.2 Table  1.3b shows the expected cell frequencies under an assumption of independence between sample type and FTase inhibition status. We test the variables’ independence based on how far the observed table deviates from that expected under independence. Such a test is called a chi-square test. Its name derives from the property that the statistic used to measure the discrepancy between the observed and expected tables has a distribution that can be approximated by a chi-square distribution, provided the sample size is large. The chi-square test based on Table 1.3a, b has a p-value of 0.025, indicating FTase inhibition status differs significantly by sample type.3 A chi-square test for a two-by-two contingency table is equivalent to the twosample test of proportions discussed in Sect. 1.6.1. However, chi-square tests apply more generally to tests of association between categorical variables with any number of levels. For example, we may be interested in knowing if response differs meaningfully across levels of Eastern Cooperative Oncology Group (ECOG) performance status (PS). We define response status (response or nonresponse) as in Sect. 1.6.2 – responders achieved complete or partial remission, or a hematologic improvement, while nonresponders had progression, stable disease, or were inevaluable. ECOG PS has three levels (0, 1, and 2) based on patient eligibility requirements. Table  1.4a shows the two-by-three contingency table for the joint distribution of response status and ECOG PS. The expected frequencies under independence are displayed in Table 1.4b. The corresponding chi-square test has a Table 1.4  Observed (A) and expected (B) frequencies of response status (response vs. nonresponse) by ECOG PS (0, 1, or 2) ECOG performance status Response status 0 1 2 Total A: Observed Response 12 23  2   37 Nonresponse 25 68 11 104 Total 37 91 13 141 B: Expected Response   9.7 23.9   3.4   37 Nonresponse 27.3 67.1   9.6 104 Total 37 91 13 141 Expected frequencies are derived based on probability laws that assume independence between row and column variables

2  For independent events, A and B, Pr(A and B) = Pr(A) × Pr(B). For example, if we assume FTase inhibition and sample type are independent, then from Table 1.3a the probability a sample comprises AML isolates and exhibits FTase inhibition is (57/106) × (88/106) » 0.446. Therefore, out of 106 total samples, we expect 0.446 ´ 106 » 47.3 to be AML isolates exhibiting FTase inhibition – assuming independence. The remaining cells in Table 1.3b are derived in a similar manner. 3  In this example, we assume FTase inhibition levels in AML isolates and buccal samples from the same patient are uncorrelated.

1  Basic Biostatistics for the Clinical Trialist

25

p-value of 0.46, and we conclude there is no significant association between response status and ECOG PS.

1.7.4 Fisher’s Exact Test The approximate chi-square distribution of the statistic measuring the discrepancy between the observed and expected frequencies is based on large-sample asymptotic. When sample sizes are small, an alternative test of independence is Fisher’s exact test. The test of independence between sample type and FTase inhibition status has a p-value of 0.037 based on Fisher’s exact test. For response status and ECOG PS, Fisher’s exact test yields a p-value of 0.53. Both examples result in equivalent inference compared to their corresponding chi-square tests discussed in Sect.  1.7.3. Had the tests conflicted, the more conservative finding (i.e., the one least in support of rejecting independence) would be reported or some would argue to report the finding of the exact test.

1.7.5 Testing Paired Data The hypothesis tests discussed thus far rely on an assumption that the data are independently sampled. Examples of data that violate this assumption are as ­follows: measures sampled from the same subject over time, for example, serum cytokine concentrations measured at baseline, week 1, week 4, and week 8 of a study; cluster-correlated measures, for example, standardized test scores of school-aged children from classrooms sampled from selected elementary schools in a state; and repeated measures, for example, visual acuity measures from the left and right eyes of the same subject. In the Lancet study, p-ERK levels were measured at baseline and at day 8. Is there a meaningful change in p-ERK levels from baseline? Although it may seem natural to assess the significance of the change in p-ERK using a two-sample t-test, the baseline and day 8 measures from the same subject do not represent independently sampled values. One remedy to this violation is to construct differences from the paired observations, resulting in a collection of independent measures of change. We construct the difference, d, from log-transformed p-ERK values, with each subject contributing a single value d = log(p-ERKday 8) − log(p-ERKbaseline). If the differences are meaningfully different from zero, we conclude change from baseline to day 8 in log p-ERK levels is significant. The corresponding null and alternative hypotheses are H0: D = 0 vs. H1: D ¹ 0, where D is the true mean difference in day 8 and baseline log p-ERK values. A test of the null hypothesis is accomplished using the one-sample t-test described in Sect. 1.7.2. Here, the p-value is 0.10, and we conclude there is not a significant change in log p-ERK from baseline to day 8. As described in Sect. 1.7.2, the conclusion of no significant difference

26

E.G. Hill and E. Garrett-Mayer

in log p-ERK levels is equivalent to a conclusion that the ratio of day 8 to baseline p-ERK levels does not differ significantly from 1.

1.7.6 Comparing Survival Times

1.0

In Sect. 1.4.3, we described survival endpoints as the predominant clinical outcome in cancer trials. The most common test to compare survival experiences between groups is the log-rank test. We consider a two-group comparison here, but the test easily extends to multiple groups. Consider two groups with corresponding survival functions S1(t) and S2(t). The log-rank test tests the null hypothesis H0: S1(t) = S2(t) for all times, t, vs. the alternative H1: S1(t) ¹ S2(t) for at least one time, t. The test is constructed from differences in the observed and expected number of deaths at each failure (death) time, under the null hypothesis that survival is the same in each group. The resulting test statistic has an approximate chi-square distribution. Figure  1.5 shows Kaplan–Meier estimates of overall survival for subjects younger than 75 years and subjects 75 years or more. The log-rank test yields a highly significant result with a p-value of 0.000022. (Most publications print very small p-values as being less than some threshold, typically p < 0.001, as indicated

0.6 0.4 0.0

0.2

Probability of Survival

0.8

Younger than 75 y 75 y or older

p<0.001 0

5

10

15 20 Time (in months)

25

30

Fig. 1.5  Kaplan–Meier estimates of the probability of survival comparing subjects younger than 75 years to those 75 years and older. The p-value corresponds to the log-rank test of the null hypothesis that the survival curves are equal

1  Basic Biostatistics for the Clinical Trialist

27

in Fig.  1.5.) We conclude overall survival for subjects under the age of 75 years differs significantly from survival for those 75 years of age or older. While there are many ways for survival functions to differ, the log-rank test is most useful at detecting differences in survival of the form S1(t) = S2(t)c, where c is a positive constant. Graphically, this alternative implies that the survival curves do not intersect. Therefore, when using the log-rank test to investigate differences in survival between groups, one should verify the survival curves are nonintersecting.

1.8 How Many Subjects Do I Need? The number of subjects needed to adequately conduct a study depends on the study’s purpose. Typically in cancer trials, the study’s quantitative goal can be described statistically as belonging to one of two categories (1) to estimate an endpoint with a desired level of precision or (2) to compare an endpoint between two or more groups. We discuss below sample size calculations in these contexts.

1.8.1 Precision-Based Calculations The goal of most phase II cancer trials is to obtain precise estimates of clinical endpoints so subsequent trials can be designed correctly. For the Lancet study, median progression-free survival and overall response are two clinical endpoints for which precise estimates may be desired. Recall from Sect.  1.5 a confidence interval measures the uncertainty associated with an estimate – the wider the interval, the lower our certainty in the estimate. Alternatively, a narrow interval indicates increased certainty, or precision, in our estimate. The half-width of a confidence interval is called the margin of error, and precision-based sample size calculations are based on its specification. As outlined in Sect. 1.5, factors influencing the size of the margin of error are (1) the data’s variability; (2) the confidence level (e.g., 90, 95, or 99%); and (3) the sample size. We illustrate the relationships among these factors based on sample size calculations to estimate a proportion from binary (0/1) data. Figure 1.6a shows the margin of error as a function of sample size for a fixed 95% confidence level and increasing variability in the binary data.4 Notice that as the data become less variable, the margin of error decreases (precision increases) for all sample sizes. Figure 1.6b shows the relationship between the margin of error and sample size for a fixed ­variance (0.25) and increasing confidence: the margin of error decreases with decreasing confidence level for all sample sizes. Finally, all curves shown in Fig. 1.6a, b illustrate that, for a given variance or confidence level, the margin of error decreases as sample size increases. These properties hold true across all endpoints, whether they are constructed from continuous, categorical, or event-time data.  For a binary variable with success probability p, the variance is equal to p(1 − p).

4

28

E.G. Hill and E. Garrett-Mayer

b

0.3

0.4

Confidence level 99% 95% 90%

0.1

0.2

0.3 0.2 0.0

0.0

0.1

Margin of error

0.4

Variance 0.25 0.24 0.21 0.16 0.09

Margin of error

a

10

20

30

40

50

10

20

Sample size

30 Sample size

40

50

Fig. 1.6  Margin of error as a function of sample size based on a confidence interval for a proportion derived from binary data with success probability, p. (a) Confidence level is fixed at 95% and variance increases from 0.09 (p = 0.1) to 0.25 (p = 0.5). (b) Variance is fixed at 0.25 and confidence increases from 90 to 99%

1.8.2 Test-Based Calculations When a study is designed to test a hypothesis (e.g., tipifarnib improves overall survival relative to standard chemotherapy in elderly poor-risk AML patients), factors influencing the required number of subjects are (1) the data’s variability; (2) the magnitude of the difference you want to detect, or effect size; (3) your willingness to tolerate erroneously rejecting the null hypothesis of no difference; and (4) the certainty with which you want to declare a difference when one actually exists. Items 3 and 4 in our list can be described in terms of probabilities shown in Table 1.1 Specifically, the chance of incorrectly rejecting the null hypothesis is the type I error rate, a. The probability of declaring a difference when one exists is called power, and is 1 minus the type II error rate, or 1 − b. A more succinct list of the factors influencing hypothesis testing-based sample size calculations is (1) variance; (2) effect size; (3) type I error rate, or a; and (4) power, or 1 − b. Figure  1.7 shows power as a function of sample size for differing variances, effect size, and type I error rate, based on a two-sample t-test comparing means. For a given sample size, we observe an increase in power with • Decreasing SD, for fixed effect size and a (Fig. 1.7a) • Increasing effect size, for fixed SD and a (Fig. 1.7b) • Increasing a, for fixed SD and effect size (Fig. 1.7c) These relationships are consistent with our intuition. For a fixed sample size, there is increased uncertainty in statistics estimated from data that are inherently more variable, which translates into decreased ability to find important differences.

1  Basic Biostatistics for the Clinical Trialist

29

b

0.4

0.4

Power

Power

0.6

0.6

0.8

0.8

1.0

1.0

a

10

20

30 Sample size

40

0.2

Effect size 0.5 1 1.5 2 2.5 3

0.0

0.0

0.2

Standard Deviation 1 1.5 2 2.5 3 3.5 50

10

20

30 Sample size

40

50

0.4

Power

0.6

0.8

1.0

c

0.0

0.2

Type I error rate 0.01 0.05 0.1

10

20

30 Sample size

40

50

Fig.  1.7  Power as a function of sample size based on a two-sample t-test for a difference in means. (a) Effect size = 2, a = 0.05, and SD increasing from 1 to 3.5. (b) SD = 1, a = 0.05, and effect size increases from 0.5 to 3. (c) Effect size = 2, SD = 2, and a increases from 0.01 to 0.1

For effect sizes, it is logical that the sample size requirements to detect large, obvious differences are smaller than when attempting to detect small, less noticeable changes. Finally, we note that a larger type I error rate makes it is easier to reject the null hypothesis in favor of the alternative, thereby increasing the chance of finding a difference that actually exists.

30

E.G. Hill and E. Garrett-Mayer

1.9 Multivariable Regression Analyses A multivariable5 regression analysis is a statistical modeling approach to assess the relative contributions of many variables to a single outcome. There are a number of reasons for performing such an analysis. Each of the following are examples of research questions for which a multivariable regression model is a suitable analytic approach: (1) a cancer researcher wants to develop a predictive model to quantify a subject’s risk of metastasis given their genetic, clinical, and demographic profile; (2) from a large group of variables, a cancer epidemiologist wants to identify those most associated with breast cancer recurrence; and (3) a clinical investigator wants an accurate measure of the risk of death for colon-cancer patients receiving experimental treatment relative to those on standard therapy, while adjusting for additional factors known to be associated with death (e.g., age or other comorbidities). In this section, we focus on the application of multivariable regression models to problems like the one presented in the third example, and begin with a simple illustration drawn from [3]. In that study, patients with unfavorable karyotypes had a 1.40-fold increase in the risk of death relative to those with favorable karyotypes. However, the 95% CI for this risk measure is 0.94–2.10 and contains the null value of 1. Further, the corresponding p-value is 0.10. We conclude patient karyotype is not significantly associated with death. This finding seems counterintuitive – certain cytogenic profiles are known to be “adverse” among AML patients. But we also know older subjects are more susceptible to death. It turns out that the cohort of patients with unfavorable karyotypes was younger than the favorable group. The median age of patients with unfavorable karyotypes was 72 years with a range of 46–85 years, while those with favorable karyotypes had a median age of 75 years with a range of 56–85 years. This difference may seem slight, but its influence is not inconsequential. If we control for the differences in age between the groups, the estimate of risk increases to 1.58, the corresponding 95% CI is 1.05–2.38 and the p-value is 0.027. Our “age-adjusted” conclusion is that karyotype and death are significantly associated. This example illustrates the effects of confounding, that is, the “masking” of the effect of an exposure (karyotype) on an outcome (death) by an extraneous variable (age). Here, age is said to confound the relationship between karyotype and death, and age is called a confounder. Estimates obtained from methods failing to account for the effects of confounding variables are said to be crude or unadjusted; those derived from methods that do account for their influence are said to be adjusted. The choice of method for a regression analysis is determined by the data-type of the outcome: continuous, approximately normally distributed endpoints are modeled using linear regression models; binary endpoints require the use of logistic

 The terms “multivariable” and “multivariate” are often used interchangeably. Strictly speaking however, a multivariable analysis refers to a statistical model of a single outcome as a function of multiple variables. In contrast, a multivariate analysis refers to the joint modeling of multiple outcomes simultaneously.

5

1  Basic Biostatistics for the Clinical Trialist

31

regression models (or other regression models for binary enpoints); and event-time endpoints are analyzed using hazard regression models, most commonly the Cox proportional hazards model. Since categorical and time-to-event endpoints dominate cancer trials, we restrict our discussion to logistic and hazard regression models, and direct the interested reader to [2] for a clinically oriented discussion of linear regression models.

1.9.1 Logistic Regression Response is a common categorical endpoint in cancer research. We dichotomize response as in Sect.  1.7.3 with responders defined as subjects achieving either a complete or partial remission, or a hematologic improvement. Suppose we are interested in quantifying the relationship between response status (responder vs. nonresponder) and karyotype (favorable vs. unfavorable), while controlling for the potentially confounding effects of subjects’ age, race, and sex. The multivariable logistic model relating the probability of response, p, to subject’s karyotype, controlling for the effects of age, race, and sex, is given by log (p / (1 - p ))= a + b ´ karyotype + g 1 ´ age + g 2 ´ race + g 3 ´ sex. Note that the left side of the model is stated in terms of the log odds of response, that is, log(p/(1 − p)). Had we attempted to model on the scale of the probability of response, we would need to constrain the right side of the model to be between 0 and 1. Modeling in terms of the log odds alleviates that constraint. The variables on the right side of the model (here, karyotype, age, race, and sex) collectively are referred to as covariates, predictors, or independent variables. The intercept and covariate coefficients (a, b, g1, g2, and g3) are called model parameters, and they quantify the effect of each covariate on the log odds of response. (The use of a and b in this context is distinct from the notation for type I and type II error rates introduced in Sect. 1.6.3.) Assuming the form of the model is correct, the parameters are estimated such that the fitted model – that is, the model with parameters replaced by their estimates – is best supported by the observed data. The model covariates are defined as follows: “karyotype” equals 0 for favorable and 1 for unfavorable; “race” equals 0 for white and 1 for nonwhite; and “sex” equals 0 for male and 1 for female. Using “hat” notation (^) to indicate estimated values, the parameter estimates for this model are aˆ = -0.0068 , bˆ = -0.36 , gˆ 1 = -0.0082 , gˆ 2 = 0.38 , and gˆ 3 = -0.69 . The fitted model is log (p / (1 - p ))= -0.0068 - 0.36 ´ karyotype - 0.0082 ´ age + 0.38 ´ race - 0.69 ´ sex. How can we use this fitted model to say something clinically relevant? Consider a 50-year-old white female subject with unfavorable karyotype. For this subject, the

32

E.G. Hill and E. Garrett-Mayer

values of the variables are karyotype = 1, age = 50, race = 0, and sex = 1. Substituting these values in the fitted model, we obtaint log (p / (1 - p))= -0.0068 - 0.36 ´ 1 - 0.0082 ´ 50 + 0.38 ´ 0 - 0.69 ´ 1 = -1.4668. In words, the log odds of response for a 50-year-old white female with unfavorable karyotype is −1.4668 – not an especially useful statement, as not many of us think of risk in terms of “log odds.” We obtain a more useful interpretation by transforming to the scale of odds; the odds of response for this individual is e−1.4668 = 0.23. Even better, we can transform to the probability scale6 and obtain a clinically interpretable ­statement of risk; the 50-year-old white female with unfavorable karyotype has a 0.19 probability of response. One of the most useful properties of the logistic model relates to comparing individuals equivalent in all respects except one. To illustrate, consider a second 50-year-old white female subject with favorable (rather than unfavorable) karyotype. The variables’ values for this subject are karyotype = 0, age = 50, race = 0, and sex = 1. Substituting into the fitted model, we obtain log (p / (1 - p))= -0.0068 - 0.36 ´ 0 - 0.0082 ´ 50 + 0.38 ´ 0 - 0.69 ´ 1 = -1.1068. Constructing the difference in the log odds of response for the subjects, we have log odds of response for subject 1 – log odds of response for subject 2 = −1.4668 −  (−1.1068) = −0.36. This value is exactly equal to bˆ , the estimated effect of karyotype on the log odds of response. This is not coincidental; terms in the difference relating to age, race, or sex cancel since the subjects are equivalent with respect to all factors except karyotype. By properties of logarithms described in Sect. 1.6.2, we can show the difference in the log odds is equivalent to the log odds ratio. Furthermore, we established that the difference in the log odds is equal to bˆ . Then log(odds of response for subject1) - log(odds of response for subject 2) æ odds of response for subject 1 ö = log ç = bˆ = - 0.36. è odds of response for subject 2 ÷ø  = 0.70 . = OR Finally, transforming to the scale of an odds ratio we obtain e The interpretation is that the subjects with an unfavorable karotype have 0.70 times the odds of a CR as subjects with a favorable karotype, adjusting for age, race, and sex. A more useful way to interpret this is that subjects with an unfavorable karyotype have a 30% reduction in the odds of response relative to those with favorable karyotypes, after controlling for the effects of age, race, and sex. -0.36

 Let log(p/(1 – p)) = x. Then p = 1/(1 + e–x). For the example in the text, log(p/(1 − p)) = −1.4668 so that p = 1/(1 + e1.4668) » 0.19.

6

1  Basic Biostatistics for the Clinical Trialist

33

In general, if bˆ is the estimated parameter corresponding to binary variable X ˆ from a multivariable logistic regression model, e b estimates the odds ratio of the outcome comparing subjects with X = 1 to those with X = 0, while controlling for the ˆ confounding effects of all other model covariates. For continuous X, e b estimates the odds ratio of the outcome for a 1 unit increase in X, while controlling for the confounding effects of all other model covariates. In addition to point estimates of model parameters, a fitted multivariable logistic regression model yields measures of uncertainty associated with estimated parameters in the form of estimated standard errors and 95% CIs, and a test for each parameter of the null hypothesis that the parameter equals 0. By exponentiating parameter estimates and the endpoints of associated 95% CIs, we obtain odds ratio and corresponding interval estimates. Additionally, the parameter’s significance test is equivalent to a test of the null hypothesis that the odds ratio is equal to 1.  = -0.36 , the associated 95% CI is −1.14 to 0.42, and the In our example, bˆ = log OR p-value of the corresponding hypothesis test is 0.40. Exponentiating, we obtain ˆ  = e -0.36 = 0.70 with a 95% of e−1.14 = 0.32 to e0.42 = 1.53. The latter CI contains e b = OR 1 which is consistent with a null finding and supported by the nonsignificant p-value. We conclude that the odds ratio of response comparing unfavorable to favorable karyotype does not differ significantly from 1, after adjusting for the confounding effects of age, race, and sex. Karyotype is not significantly­associated with response.

( )

1.9.2 Cox Proportional Hazards Regression Following a cancer diagnosis, the survival function provides a measure of the chance the patient will still be alive at time t. Another function, the hazard function, h(t), quantifies a cancer patient’s risk of death at time t. Informally, h(t) is the instantaneous death rate at time t. Suppose we are interested in quantifying the effect of karyotype on the hazard of death, controlling for the effects of age, race, and sex. For event-time endpoints, a common multivariable modeling approach is Cox proportional hazards regression. This approach models the hazard function as the product of a baseline hazard function, h0(t), common to all risk groups and a multiplicative term that modifies the baseline hazard based on specific covariate values. For the stated example, the Cox proportional hazards model is h(t ) = h (t ) ´ exp{b ´ karyotype + g 1 ´ age + g 2 ´ race + g 3 ´ sex}, 0

where the notation exp{x} means e x. The regression coefficients can be exponentiated and then represent hazard ratios. Parameter estimates from a fitted Cox proportional hazards model have a similar interpretation as presented for multivariable logistic regression models. Specifically, if bˆ  is the estimated parameter corresponding to binary variable X from a multivariable

34

E.G. Hill and E. Garrett-Mayer ˆ

Cox proportional hazards model, e b estimates the hazard ratio (HR) of the event comparing subjects with X = 1 to those with X = 0, while controlling for the conˆ founding effects of all other model covariates. For continuous X, e b estimates the hazard ratio of the event for a 1 unit increase in X, while controlling for the confounding effects of all other model covariates. As in multivariable logistic regression, the fit of a Cox model provides 95% CIs and hypothesis tests for model parameters. Continuing with our example, bˆ =  = 0.49, the corresponding 95% CI is 0.083–0.90, and the test of the null log HR hypothesis that b = 0 yields a p-value of 0.019. Upon exponentiation, we obtain ˆ  = e0.49 = 1.63 , and the corresponding 95% CI is e0.083 = 1.09 to e0.90 = 2.46. eb = HR The interval estimate does not contain the null value of 1 which is consistent with the significant p-value of 0.019. We conclude that karyotype is significantly asso­ ciated with death. Specifically, there is a 63% increase in the hazard of death comparing subjects with unfavorable karyotypes to those with favorable karyotypes, after adjusting for the confounding effects of age, race, and sex. Another way to interpret this is to say that the risk of death in those with unfavorable karyotypes is 1.63 times higher than in those with favorable karyotypes at all points in time. The proportionality assumption requires us to make the assumption that the ratio of risk is maintained over time. In Cox proportional hazards regression, the hazard is modeled such that hazard functions for different covariate patterns are proportional, that is, their ratio is a constant. Thus a simple assessment of the proportional hazards assumption is accomplished by constructing Kaplan–Meier estimates of survival; intersecting curves indicate a proportional hazards assumption violation.

( )

References 1. Daniel W. Biostatistics: A Foundation for Analysis in the Health Sciences. Wiley: New York; 8th edition (2004). 2. Katz MH. Multivariable Analysis: A Practical Guide for Clinicians. Cambridge University Press: Cambridge; 2nd edition (2006). 3. Lancet JE, Gojo I, Gotlib J, Feldman EJ, Greer J, Liesveld JL, Bruzek LM, Morris L, Park Y, Adjei AA, Kaufmann SH, Garrett-Mayer E, Greenberg PL, Wright JJ, and Karp JE. A phase 2 study of the farnesyltransferase inhibitor tipifarnib in poor-risk and elderly patients with previously untreated acute myelogenous leukemia. Blood. 2007;109(4):1387–94. 4. Leith CP, Kopecky KJ, Godwin J, McConnell T, Slovak ML, Chen IM, Head DR, Appelbaum FR, and Willman CL. Acute myeloid leukemia in the elderly: assessment of multidrug resistance (MDR1) and cytogenetics distinguishes biologic subgroups with remarkably distinct responses to standard chemotherapy. A Southwest Oncology Group Study. Blood. 1997;89(9): 3323–9. 5. Rosner B. Fundamentals of Biostatistics. Duxbury Press: Pacific Grove, CA; 6th edition (2005). 6. Van Belle G, Heagerty P, Fisher L, Lumley T. Biostatistics: A Methodology for the Health Sciences. Wiley-Interscience: New York; 2nd edition (2004).

1  Basic Biostatistics for the Clinical Trialist

Suggested Statistical Software 11. 12. 13. 14.

Stata SPSS Minitab SAS

35

Chapter 2

Fundamental Concepts in Clinical Pharmacology Daniel L. Gustafson and Erica L. Bradshaw-Pierce

2.1 Introduction Clinical pharmacology is the study of drugs in healthy volunteers and patients and defining the relationships between dose, drug exposure, and response in populations. Drug dose refers to an amount of drug administered via a particular dose route (e.g., intravenous, oral, subcutaneous). Drug exposure is a function of the concentration of drug in the body, and usually levels in the blood/plasma/serum serve as a surrogate, with respect to time. Response is a measure of effect and can relate to both advantageous (efficacy) and untoward (toxic) reactions. The dose– response relationship for a drug is based on the relationship between dose, dose route, and drug exposure that is defined by the pharmacokinetics (ADME) of the drug (Fig. 2.1). Pharmacokinetics (PK) is defined loosely as what the body does to the drug and is comprised of the absorption, distribution, metabolism, and elimination (ADME) profile. A key principal in clinical pharmacology is studying how drug PK differs amongst populations. Initial PK studies in humans for most drugs are done in healthy, volunteer populations and the results extrapolated to and refined in patient populations. Oncology drug development differs in this regard in that cancer drug PK is generally defined in phase I trial patient populations. The other major component of the dose–response relationship is pharmacodynamics (PD), which can be casually defined as what the drug does to the body. PD responses can be divided into efficacy and toxicity assessment and subdivided into direct or surrogate measures for each. The advent of molecularly targeted anticancer agents has increased the awareness and utility of PD endpoints in drug development and shifted the focus from models of drug toxicity [1] (e.g., neutropenia) to surrogates of target response [2].

E.L. Bradshaw-Pierce (*) Department of Pharmacokinetics, Dynamics and Metabolism, Pfizer Global Research and Development, 10646 Science Center Drive, San Diego, CA 92121 e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_2, © Springer Science+Business Media, LLC 2011

37

38

D.L. Gustafson and E.L. Bradshaw-Pierce

Fig. 2.1  Relationship of pharmacokinetics to dose–response

Pharmacokinetics

Dose

Absorption Distribution Metabolism Elimination

Response

2.2 Glossary Terms that are commonly used in PK and PD modeling and data analysis are listed and defined below along with the common abbreviation used to represent them.

2.2.1 Pharmacokinetic Terms Area under the curve (AUC) – A measure of drug exposure that is calculated as the product of plasma drug concentration and time. Calculation of the AUC can be done by geometric calculation or by integration of fitted equations. Bioavailability (F) – The fraction of an administered dose that reaches the ­systemic circulation. Calculated as a ratio of AUC following dosing via a given dose route to the AUC of the same dose given intravenously. Clearance (CL) – A factor that relates drug elimination to the plasma drug ­concentration. Calculated using the dose administered divided by the subsequent measured AUC. Elimination rate constant (kel) – The slope of the terminal phase of the plasma drug concentration vs. time curve. Elimination half-life (t1/2) – The time it takes plasma drug concentration to decrease by half during the terminal elimination phase. Maximum concentration (Cmax) – The maximum concentration of drug obtained in the plasma. Time to maximum concentration (Tmax) – The time it takes post dosing to reach a maximum plasma concentration. Volume of distribution – A factor that relates the amount of drug in the body to the concentration of drug in the plasma. This term can be calculated as an initial volume (VD), during the terminal phase (V), or under steady-state ­conditions (Vss).

2  Fundamental Concepts in Clinical Pharmacology

39

2.2.2 Pharmacodynamic Terms Affinity – The strength of the interaction between a receptor and a ligand. This interaction is described mathematically in terms of an association or dissociation constant. Efficacy – The capacity of a drug to induce a therapeutic effect. Intrinsic activity – The ability of a drug to induce a response at the level of a specific molecular target. Maximum response (Emax) – The largest response that a given drug can elicit at optimal concentrations. Potency – The concentration of a given drug that elicits a specific level of response. This term is used as a means of comparing the ability of compounds to induce a given response. Response (E) – A measured physiologic effect induced by a given concentration or dose of a drug. Often the measured response is a function of penetration of drug to the site of action as well as the molecular interactions and cascades that proceed.

2.2.3 Modeling Terms Compartmental modeling – Dividing the body into nonspecific body region groupings (compartments) that share similar rates of drug movement in and out for the purpose of describing plasma concentration vs. time data. Linear pharmacokinetics – Dose proportionality in plasma drug levels and exposure (AUC) such that clearance and volume of distribution do not change with dose. Noncompartmental modeling – The use of geometric and algebraic relationships to calculate pharmacokinetic parameters from plasma concentration vs. time data. Although the basis of noncompartmental modeling is classical statistical moment theory, current use of this approach in pharmacokinetics is generally limited to calculation of pharmacokinetic parameters using model-independent approaches and a series of equations for calculating common pharmacokinetic parameters. Nonlinear pharmacokinetics – The use of saturable (capacity-limited) equations to describe plasma concentration vs. time data in cases where drug absorption and/or elimination does not follow a first-order rate. In these cases, plasma drug concentration and drug exposure are not dose proportional. Population pharmacokinetic modeling – The incorporation of population ­characteristics into pharmacokinetic models in order to understand the sources of variability associated with drug exposure within a population. Physiologically based pharmacokinetics (PBPK) – A mathematical modeling technique incorporating physiological and physio-chemical properties to describe the absorption, distribution, metabolism, and elimination of drugs.

40

D.L. Gustafson and E.L. Bradshaw-Pierce

Unlike compartmental modeling, PBPK models are comprised of compartments that represent organ systems and attribute appropriate physiological and metabolic properties to that compartment.

2.3 Sampling Schedule and Study Design The sampling schedule and design of PK studies is dependent on the goals of the study but some basic principles apply for using specific modeling methodologies and the generation of valid PK parameters. Defining the PK of a drug, and the utility of the results, are dependent on how this drug is going to be dosed both in terms of route and frequency. The timing of sample collection needs to be based on components of the plasma concentration vs. time curve. For example, if a drug is dosed via an intravenous bolus, samples need to be collected early and fairly frequently to define the distribution from the plasma into tissues, and then samples collected at longer time intervals later to define the elimination component and the ­elimination half-life illustrated in Fig. 2.2. For the accurate collection of the terminal ­elimination half-life (t1/2), it is generally considered appropriate to collect time points out to at least three times the estimated length of the t1/2 (i.e., if the t1/2 is 6 h, samples should be collected out to at least 18 h). When drug dosing is via an extravascular route, sampling strategies need to take into account the estimation of Tmax and thus Cmax which may lead to more intense sampling around the range of Tmax. This scenario presupposes some knowledge of the drug’s PK and in the situation where prior data is not available the sampling schema should be altered in an iterative manner with initial measurements.

[Drug] plasma

1000

100

Distribution Elimination

10

1 Time →

Fig.  2.2  Bi-exponential relationship between the log drug plasma concentration vs. time f­ ollowing an intravenous dose

2  Fundamental Concepts in Clinical Pharmacology

41

Single drug PK studies are intended to define PK parameters for the given agent within a population or populations. Usually, these PK studies are carried out with dose escalation in oncology drug development and thus the linearity of drug elimination and distribution can be evaluated in these studies. Linear or nonlinear elimination and distribution are important characteristics of any given agent in that it allows or disallows for dose extrapolation to predict exposure. As illustrated in Fig. 2.3, linearity of drug elimination and distribution can be estimated based on the dose-dependence of AUC and Cmax, respectively. The determination of PK parameters also allows for the estimation of exposure scenarios based on altering dosing parameters (amount and frequency). PK studies with drug combinations are often intended to determine the effect of drug X on the PK of drug Y and vice versa. In oncology drug development this often involves the addition of a novel agent to a standard treatment protocol that may involve single or multiple agents. These types of PK studies are inherently complicated and may involve the generation of single agent PK in individual patients followed by combined agent PK to determine if an interaction takes place. This type of study design minimizes interpatient variability as single agent and combined agent PK are being compared in the same individual(s). However, with multiple agents involved, the single agent PK studies become difficult to coordinate and other strategies may be employed to determine any drug interaction. With multiple agent PK studies, the factors discussed for the single agent PK studies are also relevant with regards to sampling times for the proper estimation of various PK parameters. This means that for agents with disparate PK parameters, sampling may have to be individualized for each agent being tested. For example, if drug X has a t1/2 of 3  h and drug Y has a t1/2 of 24  h, sampling at 48 and 72  h may be exclusively for drug Y whereas earlier time points (4, 6, and 8 h) may provide useful data for both agents.

Fig. 2.3  Linear (closed circles, solid line) and nonlinear relationships (open circles and dashed line, closed triangles and dashed line) between drug exposure (AUC) and dose (a) and maximal plasma concentration (Cmax) and dose (b)

42

D.L. Gustafson and E.L. Bradshaw-Pierce

PK studies are often also carried out in special populations with the intent of determining if specific population characteristics influence drug elimination or distribution. Examples of population characteristics often studied include organ function and metabolic polymorphisms. These types of studies are dependent on the availability of individuals within the population of interest and what changes in PK are anticipated. For example, if a population is deficient in a specific metabolic pathway and hepatic drug elimination is expected to be decreased, this can influence the t1/2 as well as the Tmax and can change the sampling schedule that should be utilized. These considerations will be drug and population characteristic dependent and re-emphasize the importance of taking into account PK characteristics in study and sampling design. Another consideration when determining the sampling schedule is the analytical sensitivity of the assay that will be used to measure drug levels in samples. Utilization of more sensitive methods may allow for drug measurements at later time points and this can influence the calculated PK parameters [3] and the ability to sample out to three times the length of the elimination phase. Assay sensitivity, t1/2 and the dose being given influence what time points can be measured and should be considered when a sampling schema is being designed.

2.4 Patient Numbers and Sampling Intensity The number of patients required for a given study in clinical pharmacology is dependent on the question being addressed. For a typical phase I dose-escalation study, the purpose of the PK component is to determine drug exposure in the context of the toxicity being assessed at a given dose level. The number of samples per dose level obtained in phase I studies is dependent on the dose-escalation schedule being used and generally ranges from 1 to 6 patients. These dose-escalation studies can provide valuable information regarding the PK of a given agent, as shown in Fig. 2.3 for linear vs. nonlinear, regardless of the small number of patients within each dose level. Clinical pharmacology studies addressing differences between populations or to determine the presence or absence of a drug interaction would require sample numbers sufficient to either support or reject the null hypothesis based on data variability, degree of change in a parameter, and the level of statistical significance required. Power calculations can be performed to help determine the necessary sample size for these types of comparisons. However, most phase I study sample sizes are based on determining the maximum tolerated dose and not based on pharmacokinetic comparisons. Sampling intensity is a function of both the needs of the study and the reality of clinical sampling. In the outpatient setting, collecting 12 and/or 16  h samples is technically limiting and sampling between 8 and 24  h postdosing is not feasible except under special circumstances. Sampling intensity can also be influenced by the data analysis methods intended to be used. For example, if the data is to be analyzed by a three-compartment model (discussed below), the number of unknown variables

2  Fundamental Concepts in Clinical Pharmacology

a

Distribution Phase 3-4 sampling times

b Distribution Phase 3-4 sampling times

1000

[Drug]plasma

1000 [Drug]plasma

43

100

10

Elimination Phase 3-4 sampling times

1 0

6

12 Time

18

24

Equilibration Phase 3-4 sampling times

100

Elimination Phase 3-4 sampling times

10

1

0

6

12 18 24 30 36 42 48 Time

Fig. 2.4  Phases and recommended sampling density within each phase for a bi-exponential twocompartment model (a) and a tri-exponential three-compartment (b)

to be solved is 6 and thus the minimum number of samples required is 7 and they need to be distributed within the three specific phases of drug distribution, equilibration, and elimination for the accurate estimation of model parameters (Fig. 2.4). The intensity of sampling is also dependent on the time-frame within which each of these phases occurs which is compound dependent. Generally, the distribution phase is rapid, meaning frequent, more intense sampling during this early phase and more protracted sampling for the slower subsequent phases. Shown in Fig. 2.4 is sampling within the context of (a) two- and (b) three-compartment modeling to allow for proper curve resolution and accurate parameter estimation. As stated earlier, optimal estimation of the elimination phase includes sampling out to three times the elimination half-life. Sampling intensity is also a function of the amount of sample collected as the amount of blood collected is limited by patient size and other characteristics.

2.5 What is the Goal of the PK Study? It is critically important to determine the goals of PK studies a priori? The goals of the study will dictate the design of the study, including the frequency and intensity of sampling. If the purpose is to define the PK parameters and drug exposure of a drug in humans, then proper sampling strategies, as discussed earlier, should be put in place to assure that parameters can be accurately estimated. If the goal is to determine if limited sampling strategies can be utilized to determine a PK parameter that correlates with drug response or toxicity and employ this strategy in therapeutic drug monitoring, then the study must be designed in the context of determining what PK parameter correlates to a specific drug response and how best to estimate that PK parameter with regards to patient sampling. If a study is

44

D.L. Gustafson and E.L. Bradshaw-Pierce

intended to determine if a drug interaction is taking place (i.e., drug X alone vs. drug X with drug Y), then it must be determined if the PK will be done with drug X alone and with drug X and Y or whether just the combination of drug X and Y will be measured and the values compared to historic or literature values. An important component in planning these types of studies is the quality and type of historical data that will be used if that avenue of study design is chosen. In order to compare newly generated PK data with data from a previous study, it is important to note that similar dose ranges should be compared unless it is clear that the PK of the agent being studied is linear with dose, that the same modeling methodology should be used for data analysis, and that the sampling schedule should be similar to that employed in the earlier study.

2.6 Pharmacokinetics The pharmacokinetics of drugs are usually defined by its absorption, distribution, metabolism, and elimination, commonly referred to as ADME (Fig.  2.1). Drug absorption is a critical component to how a drug is dosed and generally determines the route of drug delivery. Whether given by oral or parenteral routes, a drug must pass through biological membranes to get into the blood stream and be distributed throughout the body. In the case of intravenous or intraarterial drug delivery, the crossing of the membranes is done by mechanical means. For oral, subcutaneous, intramuscular, and transdermal dosing, drug absorption is dependent on the concentration of drug in solution at the site of delivery, permeability, and the concentration gradient across membranes and this shown in (2.1). This equation shows that the rate of drug absorption,

Rate of absorbtion (ka ) = [drug]solution × permeability × concentration gradient,

(2.1)

is a function of drug dissolution if not given in solution (drug concentration), the ionization state and lipophilicity of the drug molecule (permeability) and the rate of perfusion at the site of administration such that absorbed drug is quickly removed and a diffusion driving concentration gradient maintained. Since permeability and the maintenance of a concentration gradient are relatively consistent within a local drug depot, drug absorption from extravascular sites often occurs via a first-order rate, which can be described by an absorption rate constant (ka) as long as the amount of drug in solution is not limiting. In the case of oral drug dosing, absorption is complicated by transit time through the gastrointestinal tract (GI), competing reactions with GI contents, metabolism by GI tissue, active transport from GI epithelium toward the gut lumen, and metabolism of absorbed drug by the liver due to portal blood outflow from the GI going through the liver prior to systemic re-circulation. All of these factors can limit the exposure to a given drug when it is orally dosed. Therefore, in the case of oral drug dosing, the actual amount of drug traversing the gastrointestinal epithelium and reaching the systemic

2  Fundamental Concepts in Clinical Pharmacology

45

circulation is tempered by these competing processes whereas drug absorption from parenteral sites is often simply a function of permeability across cellular membranes and access to the circulation. The measure of drug absorption and exposure from extravascular sites is termed bioavailability (F) and is determined by (2.2) and represents simply the drug exposure (AUC) following the extravascular dose in comparison to drug exposure following intravenous (IV) dosing.

F=

AUCextravascular × Dose IV . AUCIV × Doseextravascular

(2.2)

Drug distribution refers to the movement of a drug from the blood and is often described in terms of perfusion or diffusion-limited tissue uptake. Perfusion-limited drug movement into tissues describes drugs that essentially do not see biological membranes as a barrier and thus drug delivery to tissues is dependent only on the rate of delivery (i.e., blood flow). Diffusion-limited drug movement into tissues is dependent on the movement of drug molecules from the blood into the tissue with factors of facilitated transport and tissue properties effecting the rate and extent of drug uptake. Another component of drug distribution is the binding of drug molecules to proteins in the blood as well as in tissues. Drug binding within the blood compartment can include binding to blood cellular components as well as proteins and lipoproteins within the plasma component. The major plasma proteins responsible for drug binding are albumin and alpha 1-acid glycoprotein whose characteristics are summarized in Table 2.1. Protein binding within the blood can have a major impact on drug distribution, metabolism, and elimination depending on the degree and extent of the drug–protein interaction and can be a major conduit for drug–drug interactions. Since drug plasma concentrations are generally reported as total drug (protein bound and unbound), protein binding may be a confounding factor in the variability associated with plasma drug levels and drug effects. Simply stated, drug that is protein bound in the plasma will not distribute to tissues and thus cannot elicit a response, and will not be metabolized or eliminated. Therefore, the dynamics of drug movement from a protein bound to unbound state and how rapidly free drug is distributed to tissues, binds to effector sites, and is metabolized or transported will dictate drug action (Fig. 2.5). The pharmacokinetic descriptor of drug distribution is the volume of distribution (VD) which describes the relationship between the amount of drug in the body and the concentration in the plasma. The definition and calculation of this pharmacokinetic parameter is described elsewhere in this chapter. Drug metabolism involves the conversion of a drug molecule into another molecular entity. Drug metabolism can result in the formation of inactive, toxic, Table 2.1  Major plasma proteins that bind drugs Protein

Molecular weight (D)

Albumin a1-Acid glycoprotein Lipoproteins

65,000 44,000 200,000–3,400,000

Normal range of concentrations g/L mol/L 35–50 0.4–1.0

5–7.5 × 10−4 0.9–2.2 × 10−5 Variable

46

D.L. Gustafson and E.L. Bradshaw-Pierce

BLOOD PLASMA PROTEIN

DRUG

DRUG

TISSUE

TISSUE PROTEIN

DRUG

DRUG

METABOLISM TRANSPORT MET

Fig. 2.5  Dynamics of drug distribution between blood and tissues and the role of protein binding, metabolism, and drug transport in influencing drug movement from the blood to tissues

and active metabolites which can further complicate the pharmacology of given agents. The utilization of metabolizing systems within specific organs or cell types has led to the development of pro-drugs which are given in an inactive form with the specific intent of metabolism leading to the formation of an active form. The major metabolizing organ is the liver with contributions to metabolism from other tissues depending on the nature of the metabolizing system and its physiologic distribution. Metabolism is generally thought of as a mechanism of drug loss from the body, and this process generally follows saturation (Michaelis–Menten) kinetics and can be described by (2.3) where Vmax is the maximal rate of metabolism and Km is the Michaelis constant representing the apparent dissociation constant for the drug and metabolizing enzyme.

Rate of metabolism =

Vmax × [Drug] . K m + [Drug]

(2.3)

A majority of drug metabolism falls into a range of drug concentrations lower than the Km for metabolizing enzymes and thus the rate of metabolism is proportional to

2  Fundamental Concepts in Clinical Pharmacology

47

drug concentration and follows a first-order rate as described by (2.4). The term Vmax/Km represents a constant

Rate of metabolism =

Vmax × [Drug], Km

(2.4)

term with a rate (concentration/time) divided by concentration, resulting in units of time−1, which are the units of a first-order rate constant. In rare cases where the concentration of a drug is closer to or greatly exceeds the Km of a metabolizing system, the rate of drug metabolism will not be dose proportional and will result in either zero-order (2.5) or

Rate of metabolism = Vmax ,

(2.5)

saturation characteristics (2.3). In cases where drug-metabolizing systems are saturated and metabolism fails to be dose proportional, the relationships between dose and drug exposure become nonlinear and difficult to predict. Many drugs are metabolized by multiple metabolic pathways. Assuming that there is no interaction between the metabolic pathways, the total drug metabolism is simply the sum of the individual pathways. Drug metabolism is a major point of drug interaction due to a number of factors including competition for the active site of a metabolizing enzyme leading to competitive inhibition based on the affinity of each substrate. Other points of drug interaction can include the induction of drugmetabolizing enzymes which can lead to a proportional increase in the rate of drug metabolism. As stated earlier, drug metabolism is not necessarily synonymous with drug inactivation due to many drug metabolites being active and potentially toxic. Therefore, the role of metabolism in inactivation or activation of a given drug in terms of both efficacy and toxicity must be considered. Drug elimination is a catch-all phrase that describes the loss of drug from the plasma/serum. The major organs of drug elimination are the liver and the kidney, but drugs or drug metabolites can leave the body through any number of routes including urine, feces, exhaled air, sweat, saliva, and breast milk. However, for a majority of compounds the urine and feces are the major routes of elimination. Hepatic elimination is comprised of metabolism as well as transport of drugs into the bile and thus both metabolizing enzymes (i.e., P450, glucoronyl transferases, etc.) and drug transporters (i.e., ABC, OAT, OCT, etc.) are involved in hepatic elimination. Since active drug transporters are also abundant in the gastrointestinal epithelium it is reasonable to assume that drugs can also be eliminated directly from the GI into the feces. Shown in Fig. 2.6 is a schematic demonstration of the multiple mechanisms of drug elimination within the hepatobiliary and gastrointestinal circulation that results in drug elimination via the feces, and the potential for drug cycling (enterohepatic cycling) within these processes. The other major route of drug elimination is renal. Xenobiotics can be transported into the urine by glomerular filtration and active transport from the population of transporters that occur in the proximal tubule. Thus, drug ­accumulation in the

48

D.L. Gustafson and E.L. Bradshaw-Pierce

PORTAL VEIN

METABOLITE

DRUG

LIVER

METABOLITE

DRUG

BILE

DRUG

DRUG

INTESTINE METABOLITE TISSUE

LUMEN

FECES

Fig. 2.6  Potential fate of drug and drug metabolites within the hepatobiliary and gastrointestinal circulation which results in drug elimination via the feces or enterohepatic cycling

urine is from both passive filtering at the glomerulus as well as active transport primarily in the proximal tubule. Drug accumulation in the urine is tempered by reabsorption into the blood for drugs with high permeability of biological membranes and thus the renal elimination of highly lipophilic compounds is essentially limited to the urine concentration being equal to the plasma concentration. This problem is overcome by drug metabolism where functional groups may be added and parent drug or metabolite conjugated to hydrophilic compounds such as glucoronic acid, sulfate, or amino acids. Therefore, for many compounds significant renal elimination is limited to their conjugated (glucoronide, sulfate, glutathione, etc.) forms. Net renal elimination is summarized in (2.6).

Renal elimination = glomerular filtration + active secretion − reabsorption.

(2.6)

Routes of drug elimination besides renal and hepatic are dependent on the ­physio-chemical properties of the agent and its potential accumulation in various tissues and their secretions. For example, agents that are volatile will equilibrate between the liquid and the gas phase when present in the blood that is passing through the lung and thus be eliminated from the body through exhaled air. Drugs that are present in tissues that contain fluid-expressing glands would be expected to equilibrate between the fluid and the surrounding tissue and thus drug can be

2  Fundamental Concepts in Clinical Pharmacology

49

eliminated via any excreted fluid including sweat, tears, and milk depending on the presence of the drug within the tissue around the secreting gland, solubility within the secreted fluid and permeability across membranes to traverse through cell layers [4].

2.7 Pharmacokinetic Models Pharmacokinetic models utilize mathematical equations to describe drug concentrations measured in the body as a function of time. These mathematical models can be used to generate pharmacokinetic parameters that describe the processes of absorption, distribution, and elimination of a substance in the body. Models may also be used to predict plasma, and in some cases tissue, concentrations under different dosing schemes.

2.7.1 Compartmental Modeling Compartmental modeling is the mainstay of pharmacokinetic modeling. These models can be used to describe concentration data, estimate pharmacokinetic parameters and predict data. Compartmental models treat the body as though it is divided into distinct units which can be clearly and individually characterized. The compartments do not carry any anatomic or physiologic meaning, but can be considered as a tissue or group of tissues with similar blood flow, binding, and elimination characteristics. Compartmental modeling is performed on pharmacokinetic data sets. This means that models are fit to drug plasma concentration time-course data. Different models are fit until a “best-fit” model is identified which adequately describes the trends in the data. These models are linear differential equations which are used to describe the dynamic process of drug movement into and out of compartments. Drug enters and leaves a tissue compartment from a central or plasma compartment and is considered to be instantaneously and evenly distributed within the compartment. Since information on tissue drug concentrations, blood flow, or binding characteristics are not necessary, compartmental modeling can be quite useful when little information is available. Additionally, compartmental models are generally far less complex than physiologically based models. Although compartmental modeling can be used for prediction of data, this is generally limited to prediction of plasma concentrations and is not suitable for extrapolation between species. Another drawback to compartmental models is the potential inability to use the same model structure across different patients within a single study and between studies thereby limiting comparison of pharmacokinetic parameters within and between studies utilizing the same drug.

50

D.L. Gustafson and E.L. Bradshaw-Pierce

2.7.1.1 One-Compartment Model The most basic mathematical description of drug distribution and elimination is the one-compartment model. One-compartment models for the distribution and elimination of drug in the body are illustrated in Fig.  2.7. One-compartment models utilize a single central compartment and assume that equilibrium is rapidly/instantaneously achieved with tissues. Although the drug in the body, as a single “compartment,” cannot be measured, an assumption is made that the changes in blood (or other sampling compartment) reflect a proportional change in tissue drug concentrations. The rate of drug elimination from the central compartment in a one-compartment model is a first-order process. The mathematical description of linear decay of a drug from the compartment as a function of time is presented in (2.7) as the change in drug concentration (dC) over the change in time (dt).

−



dC = kC. dt

(2.7)

This can be rearranged and integrated to yield (2.8):

C (t ) = C0 e− kelt ,



(2.8)

where C(t) is the concentration in the central compartment at time t, C0 is the initial plasma concentration immediately following administration (at time 0), t is time and k is the first-order elimination rate constant (kel). Figure 2.8 shows a semilog plot of (2.8) yields a straight line. The slope of this line represents the firstorder elimination rate constant, kel. The initial plasma concentration, C0, can be obtained by extrapolation of the regression line back to the y-axis. Other important pharmacokinetic parameters can be obtained from drug plasma concentration time data that fit one-compartment models and the relationships shown in (2.9–2.13) are only valid for data that can be described adequately by the one-compartment description.

Dose (D) Bolus Extravascular (ka ) Infusion (k0 )

Central Compartment (VD, C)

kel

Fig. 2.7  Schematic representation of a one-compartment model. VD = apparent volume of distribution, C = concentration of drug in the compartment, kel = rate of drug elimination from the compartment, ka = rate of absorption for an extravascular dose, and k0 = rate of infusion for a drug delivered via a constant rate

2  Fundamental Concepts in Clinical Pharmacology 100

51

Co

Slope =

−kel 2.3

[Drug] plasma

10

1

0.1 6 Time

0

12

Fig. 2.8  Graphical representation of first-order elimination of a drug that fits a one-compartment model. The first-order elimination constant is represented by the slope of the log plasma concentration vs. time line

The apparent volume of distribution (VD) is the theoretical volume, which does not carry any physiologic meaning, in the body that the drug is distributed in. It is a constant used to relate the concentration of drug in the blood to the amount of drug in the body and is calculated by:

VD =

Dose . C0

(2.9)

Since this is a theoretical volume, it is possible for VD to be larger than the volume of the body which can be explained by extensive tissue binding of drug. The elimination half-life, t1/2, is the time it takes for the concentration to reach half of its initial value. This means that the concentration in the central compartment when t equals t½ is half of the initial concentration (C = ½C0). These are substituted into (2.8) and rearranged to yield (2.10):

t1/ 2 =

0.693 . kel

(2.10)

Clearance (CL) is the intrinsic ability of the body to eliminate drug from the blood. This can include distribution to tissues, or elimination from the body by metabolism and/or excretion. In the case of a one-compartment model, this process is linear as described above, regardless of the mechanism of clearance. Therefore, the

52

D.L. Gustafson and E.L. Bradshaw-Pierce

rate at which drug is eliminated from the body is directly proportional to the ability to clear the blood, kel  ¥ CL, and can be described by (2.11):

CL = VD × kel

(2.11)

The area under the drug concentration–time curve (AUC), calculated by integration from t = 0 to either t = ¥ or until the last sampling point, t, is used to describe total drug exposure. The AUC can also be estimated by (2.12):

C0 kel

(2.12)

Dose . CL

(2.13)

AUC =

or (2.13):

AUC =

Although the AUC is a model-independent parameter, it provides a quantitative measure of total systemic drug exposure. The AUC has been correlated to therapeutic efficacy or toxicity for several different chemotherapeutic agents. The graphical representation in Fig. 2.8 is only valid for IV bolus administration where the absorption processes is assumed to be instantaneous. However, onecompartment models can also be generated for drugs that are administered by IV infusion or other routes of administration. For administration other than IV bolus, an absorption/delivery phase is incorporated and can utilize first-order or zero-order dose input. While many of the rate processes in the body are first-order processes, there are few drugs that fit a one-compartment description. Therefore, the need for more complex mathematical descriptions is necessary. 2.7.1.2 Multicompartment Models Multicompartment models are used to describe drug concentration–time data that does not decay linearly as a single, first-order rate process. For drugs that are distributed to tissues (or “compartments”) or eliminated by different processes at different rates, ­multicompartment models are required to describe the data. In multicompartmental models, the transfer of a drug from the central compartment to the peripheral compartment(s), and vice versa, are represented by first-order rate constants (Fig. 2.9). Input and elimination of drug typically occur from the central compartment; however, the model may be modified to account from loss from the peripheral compartment(s) as well. In contrast to the one-compartment model, drug does not equilibrate or distribute instantaneously between the blood and tissues in a two-compartment model. Figure 2.2 is an example of the bi-exponential plasma concentration–time profile for drug that follows two-compartment pharmacokinetics. Following an IV bolus dose, drug equilibrates instantaneously within the central compartment, then ­distributes to the other tissues (the peripheral compartment) represented by the

2  Fundamental Concepts in Clinical Pharmacology

a Dose (D)

53

Bolus Extravascular (ka ) Infusion (k0 )

Central Compartment

k 12

Peripheral Compartment

k 21 k 10

b Dose (D) Peripheral Compartment

k 12

k 21

Bolus Extravascular (ka ) Infusion (k0 ) k 13

Central Com partment p

Peripheral Compartment

k 31

k 10

Fig. 2.9  Schematic representation of a (a) two- and (b) three-compartment models. kxy = rate of drug transport from one compartment to another, k10 = rate of elimination from the central compartment. Most models assume elimination from the central compartment; however, models can be modified for elimination from a peripheral compartment or multiple compartments

initial­rapid decline of drug. This is referred to as the distribution phase of the curve. Following the initial distribution, equilibrium between the central compartment and peripheral compartment is attained, which means that drug movement between the peripheral and central compartments is bidirectional. This is represented by a reduction in the decline of the plasma concentration–time curve and is referred to as the terminal elimination phase. Three-compartment models require the addition of a second peripheral or tissue compartment (Fig. 2.9b). The plasma concentration–time curves that fit three-compartment models exhibit tri-exponential behavior. In a three-compartment model, just as in the two-compartment model, we can assume that upon drug delivery, equilibrium is instantaneously achieved in the central compartment. However, in a threecompartment model drug distribution can be thought to occur in two phases; initial distribution to rapidly perfused tissues then more slowly to more poorly perfused tissues. Then, once equilibrium is achieved between peripheral compartments and the central compartment, a slower decline representing the terminal elimination phase occurs. Multicompartment models can be described by the sum of first-order processes. The following equations represent drugs that behave according to two- (2.14) and three-compartment models (2.15) (bi- and tri-exponential decay):

C (t ) = Ae−αt + Be−βt ,

(2.14)



C (t ) = Ae−αt + Be−βt + Ce− γt ,

(2.15)

54

D.L. Gustafson and E.L. Bradshaw-Pierce

where C(t) is the concentration in the central compartment at time t, A, B, and C represent the y-intercepts of each of the distribution and elimination phases of the plasma concentration–time curve, t is time and a, b, and g are the first-order elimination rate constant for each of the phases.

2.7.2 Nonlinear Pharmacokinetics Linear models assume that pharmacokinetic parameters do not change or change proportionally with altered or multiple doses. However, when a drug is given at a higher dose, or when multiple doses are administered, the pharmacokinetic behavior of that drug may deviate from the linearity observed with a lower or single dose. This disproportionate change in pharmacokinetics is termed nonlinear or dose-dependent pharmacokinetics. Nonlinear pharmacokinetics is determined by administering a drug at different dose levels and obtaining plasma concentration vs. time curves at each dose. From these plots there are a couple of methods to establish linearity: (1) Evaluate the lines (slope) of the log concentration–time plots for each dose. If the lines are parallel (i.e., slopes are equal) the drug follows linear pharmacokinetics for the given concentration range. (2) Plot the AUC, Cmax, or Css vs. dose. If the linear regression yields a straight line of the data, the drug follows linear pharmacokinetics for the given concentration range (Fig. 2.3). Sources of nonlinear pharmacokinetics are most often associated with the processes of drug absorption, distribution and elimination. Nonlinear absorption can arise from; saturation of carrier-mediated absorption, poor aqueous solubility or slow release of dosage forms, saturation of first-pass metabolic effects or doserelated changes in blood flow, gastric emptying and intestinal transit time [5]. In some cases, dividing a dose into multiple doses can improve absorption and reverse the dose-dependent effects. Saturable plasma protein and tissue binding are the major causes of nonlinear distribution. Elimination pathways such as metabolism and renal excretion can become saturated, induced or inhibited, resulting in nonlinear behavior. The majority of processes involved in the absorption, distribution, metabolism, and elimination of drugs involve protein-mediated reactions, which at high concentrations can become saturated. These protein-mediated processes can also be induced or inhibited by repeated dosing or by coadministration with other drugs, which can lead to nonlinear effects. The saturation leading to nonlinearity can be mathematically described by the Michaelis–Menton equation (2.3). A drug will exhibit nonlinear pharmacokinetics when concentrations are close to, and above, the Km value. At C >> Km the elimination rate is represented by (2.4), where drug elimination is proportional to the constant (Vmax/Km) and is first order and linear. It is also worth noting that pathologic alterations can give rise to nonlinear pharmacokinetics­. This is important in oncology where disease states may lead to alterations in liver and kidney function or where surgery can result in pathologic alterations.

2  Fundamental Concepts in Clinical Pharmacology

55

2.7.3 Noncompartmental Pharmacokinetics Noncompartmental pharmacokinetic analysis is a useful tool for calculating pharmacokinetic data. With compartmental analysis, the model used to describe pharmacokinetic data is the one that fits the data best. This can be problematic when comparing different doses or even different patients given the same drug at the same dose. The use of compartmental analysis can also be further complicated by sparse or inconsistent sample collection between patients. The use of noncompartmental analysis alleviates many of these problems associated with compartmental approaches. Despite sometimes being referred to as “model-independent” or “model-free approaches,” noncompartmental analysis obeys a definite model structure as can be seen in Fig. 2.10 [6, 7]. Noncompartmental analysis takes a “black box” approach to computation of pharmacokinetic parameters; where the focus is placed on the central measurement pool. As the term noncompartmental implies, this method of analysis allows for any number of exchanges to occur between any number of tissues without requiring the identification of any physiologic structures or the need to assign a number of compartments or number of exponentials to describe the plasma concentration–time data. This reduces the number of assumptions needed compared to compartmental analysis; however, it is important to stress that it is a reduction, not an elimination of assumptions. Noncompartmental analysis is an approach used to generate pharmacokinetic data that focuses on the output, pharmacokinetic parameters, rather than the mechanism. The limitation of noncompartmental analysis is that behavioral properties of drugs are emphasized rather than the mechanistic properties. Nevertheless, noncompartmental analysis is useful for addressing specific questions and making comparisons within and between studies. Noncompartmental calculations are based on the area under the plasma concentration–time curve, which can be solved for geometrically by the trapezoidal rule. Additional summary parameters, such as clearance, volume of distribution, and

3 2

4

1

n Central Measurement Pool

Input

Elimination (k10 )

Fig. 2.10  General “model” for noncompartmental analysis representing sampling from a central pool and distribution to peripheral compartments via unrepresented rates

56

D.L. Gustafson and E.L. Bradshaw-Pierce

mean residence time can then be calculated from this. Detailed explanations and derivation of equations for noncompartmental calculations can be found elsewhere [6].

2.7.4 Physiologically Based Pharmacokinetic Models Physiologically based pharmacokinetic (PBPK) models are more sophisticated pharmacokinetic models that mathematically incorporate principles of physiology, biochemistry, and chemical engineering to model the body as a chemical plant. The fundamental objective of PBPK modeling is to identify the principal organs or tissues involved in the disposition of the compound of interest, and to correlate absorption, distribution, and elimination within and among these organs and tissues in an integrated and biologically plausible manner. Compartments in PBPK modeling, in contrast to classical compartmental modeling, represent specific organs or tissue groups, requiring PBPK models to utilize a large body of physiologic and physio-chemical data. Strengths of PBPK models, compared to classical compartmental approaches, include the ability to extrapolate between doses, routes of administration and species, and the capability of a priori prediction of plasma and tissue distribution [8]. Due to the complexity of biological systems, several assumptions are imposed in the development of PBPK models, either to simplify the model or as a result of limited data. During the initial development of the model these assumptions include the following: (1) The model is flow limited. This means that organs are well-mixed systems that reach equilibrium with drug concentrations immediately. If tissue uptake studies indicate that the model is not flow limited, diffusion-limited approaches can be incorporated. (2) The concentration of drug in any given compartment is homogeneous. For well-perfused organs, this is likely to be an accurate assumption. However, for slowly perfused organs such as fat and muscle, this assumption is only a first-order approximation. The utility of PBPK modeling lies in its ability to describe concentration–time profiles for individual organs in addition to plasma. PBPK model development usually begins in preclinical animal models in order to obtain tissue concentrations, which for obvious reasons, is not feasible or is limited in humans. This allows for estimation of certain model parameters and for investigation of how various model parameters affect tissue concentrations by fitting a model to actual plasma and tissue data. This is critical in PBPK model development since alteration in model parameters can sometimes have no effect on the plasma PK or time course but can greatly affect specific tissue concentrations and PK. In developing a PBPK model, first the principal organs involved in the disposition of the drug need to be identified. The identification of these organs involves determination of tissue-specific metabolism, tissue partitioning, and tissue-specific excretion. A scheme is then usually formed where the normal physiology is followed in a graphical manner. Figure 2.11 is an example of a schematic representation of a PBPK model developed to describe

2  Fundamental Concepts in Clinical Pharmacology

CVSP

CVRP

CVK

QRP, CA

VRP QK, CA

Kidney PK Liver PL

Urinary Excretion

VSP

Rapidly Perfused PRP

VmaxAS KmAS

QSP, CA

Slowly Perfused PSP

CVL

57

VL

VK

Gut

QG, CVG

PG

VmaxB KmB VmaxLM KmLM

QL, CA QG, CA

VG

VmaxI KmI

Fecal Excretion Metabolism to t-OH-butyl-docetaxel

Fig. 2.11  Schematic representation of a PBPK model for docetaxel disposition (reprinted from [8])

plasma and tissue distribution of docetaxel [8]. Within the boundary of an identified compartment (e.g., an organ or tissue or a group of organs or tissues), whatever comes in must be accounted for via leaving, accumulation, or elimination from the compartment. The resulting “mass balance” is expressed as a mathematical equation with appropriate parameters carrying biological significance. An example of mass balance within a tissue compartment is given by (2.16):

C Vi × dCi = Qi × (Ca − vi ) − X , dt Pi

(2.16)

where, Vi = the volume of the compartment (organ or group of organs), Ci = concentration of drug in the compartment, Qi = blood flow to the compartment, Ca = concentration of drug in arterial blood, Cvi = drug concentration in venous blood leaving compartment i, Pi = tissue:blood partitioning of the drug for that compartment, and X = clearance term (metabolism, excretion). The differential mass balance equations for all compartments are solved simultaneously, which in turn is used for computer simulations predicting the time course for any given

58

D.L. Gustafson and E.L. Bradshaw-Pierce

parameter. In principle, if all the parameters that can affect the drug disposition are accounted for, the model should be capable of predicting what takes place in  vivo. Figure  2.12 illustrates the ability of PBPK models to accurately Mouse 20 mg/kg

5 mg/kg 10000

100000

Plasma Docetaxel [nmol/L]

Docetaxel [nmol/L]

Plasma 10000 1000 100

1000

100

10

10 Liver

Liver Docetaxel [nmol/L]

Docetaxel [nmol/L]

100000

10000

1000

1000

100

100 Intestine

Intestine Docetaxel [nmol/L]

100000 Docetaxel [nmol/L]

10000

10000

1000

100

10000

1000

100 0

4

8

12 16 Time [h]

20

24

0

4

Time [h]

8

12

Docetaxel [nmol/L]

Human 1000

1000

1000

100

100

100

10

10

10

1

0

10

20 30 Time (h)

40

50

1

0

10

20 30 Time (h)

40

50

1

0

10

20 30 Time (h)

40

50

Fig. 2.12  PBPK model developed for prediction of docetaxel plasma and tissue distribution in mice and humans. This PBPK model was able to accurately predict the plasma and tissue distribution in mice following IV bolus administration at two different doses (20 and 5 mg/kg). The model was then scaled to human organ, blood flow, tissue binding, metabolic and excretory parameters, and docetaxel plasma levels accurately predicted (36 mg/m2 by IV infusion). Symbols represent actual data points and the solid lines represent the PBPK model simulations (figure modified from [8])

2  Fundamental Concepts in Clinical Pharmacology 9 mg/kg IV

3mg/kg IV

59 6 mg/kg IP

1 mg/kg IP

Docetaxel [nM]

1000

100

10

1

0.1

0

1

2

3

4

5 6 Time (days)

7

8

9

10

Fig. 2.13  Examples of PBPK model simulations of docetaxel plasma concentrations in mice with differing doses, routes of administration, and schedules

predict plasma and tissue concentrations at different doses and the ability to extrapolate between species. It is important to note that the PBPK model simulations, represented by solid lines, exist without the data. In other words, the lines do not represent a “fit” to the data, rather data itself generated by the model. Figure  2.13 shows plasma concentration–time curves for docetaxel administered at different doses though different routes of administration predicted by a PBPK model. This illustrates the potential utility of PBPK models to aid in the advancement of “model-directed” experimental design of combination therapies and/or alternate dosing schedules. The value of PBPK models expands beyond preclinical research. In fact, PBPK models have been developed for several clinically used chemotherapeutic agents such as: docetaxel [8], adriamycin [9], ara-C [10], cisplatin [11], methotrexate [12], and capecitabine/5-FU [13]. PBPK models have the ability to be coupled to Monte Carlo simulation, which can then account for variability across model parameters such as induction of CYP3A activity or reduction of CYP3A activity due to impaired liver function or competitive inhibition by a coadministered drug. In fact, a PBPK model of doxorubicin was coupled to Monte Carlo simulation to predict the interaction of paclitaxel on doxorubicin pharmacokinetics [14]. This study showed that paclitaxel did not affect plasma pharmacokinetic of doxorubicin but did affect tissue pharmacokinetics. This information would not have been revealed by simple compartmental or noncompartmental modeling of the doxorubicin plasma concentration–time data.

60

D.L. Gustafson and E.L. Bradshaw-Pierce

2.7.5 Population Pharmacokinetics Population pharmacokinetics is the process by which individual characteristics within a population are utilized to try and identify sources of variability that can potentially be accounted for in drug dosing decisions. Thus, it is a useful tool for identifying the sources of pharmacokinetic variability and can aid in the design of alternative dosing regimens to enhance drug efficacy and safety. The foundations of population PK modeling were laid in the 1970s by Sheiner et  al. [15, 16], who showed that population PK modeling can estimate the average values of PK parameters and the interindividual variances of those parameters in a patient population. In addition to measuring interindividual variances, population PK can also account for some of this variability in terms of patient differences in genetic, physiological, pathological, and/or environmental factors. Thus, population-based methods facilitate the development of individualized dosing regimens based on patient-specific covariates. For example, should a drug be dosed based on the body weight of the patient (per kg) or per body surface area (BSA)? The question can be answered by taking body weight and body surface area into consideration when analyzing the pharmacokinetic data from a population, and determine the degree of dependence of specific pharmacokinetic parameters to these factors. There are various approaches to estimating population PK models with respect to the mathematical foundation, statistical aspects, software programs for implementation, and underlying assumptions; these include the naïve average data approach, the naïve pooled data analysis, the two-stage approach, and the nonlinear mixed-effect model approach [17]. The computer program NONMEM (nonlinear mixed-effect modeling), developed by Beal and Sheiner, is the most widely used method for the analysis of population PK data [18]. This program utilizes the least squares method, originally developed for individual subject curve fitting, for population data analysis. Thus, NONMEM treats the population study sample, rather than the individual, as a unit of analysis and generally requires fewer data points per individual (but many more individuals) than are normally required for PK analysis. In this way, a much more representative sample of the target population can be obtained and quantitative relationships between PK parameters and patient covariates can be investigated in a single step. This is illustrated in Fig. 2.14 where pharmacokinetic data for a population (panel A) can be analyzed via NONMEM with respect to a given population parameter (BSA) to establish a relationship that allows for estimation of pharmacokinetic parameters taking the population parameter into account to describe interindividual variability (panel B) as well as estimating intraindividual variability based on prediction and observation (panel C). Therefore, NONMEM-based population pharmacokinetics allows for the prediction of both interindividual (parameter-dependent) variability and intraindividual (random) variability. Population pharmacokinetic approaches are an important aspect of pharmacokinetic modeling but require collection of both pharmacokinetic and demographic data and require large numbers of patients to determine significant relationships between many population parameters.

2  Fundamental Concepts in Clinical Pharmacology

61

Fig.  2.14  Nonlinear mixed effects modeling (NONMEM) simultaneous estimation of parameters relating fixed effects and random effects to observed data for population pharmacokinetic modeling

References 1. Perry S. Reduction of toxicity in cancer chemotherapy. Cancer Res 1969; 29: 2319–25. 2. Thomas SM, Grandis JR. Pharmacokinetic and pharmacodynamic properties of EGFR inhibitors under clinical investigation. Cancer Treat Rev 2004; 30: 255–68. 3. Gustafson DL, Long ME, Zirrolli JA, et al. Analysis of docetaxel pharmacokinetics in humans with the inclusion of later sampling time points afforded by the use of a sensitive tandem LCMS assay. Cancer Chemother Pharmacol 2003; 52: 159–66. 4. Stowe CM, Plaa GL. Extrarenal excretion of drugs and chemicals. Annu Rev Pharmacol 1968; 8: 337–56. 5. Ludden TM. Nonlinear pharmacokinetics: clinical implications. Clin Pharmacokinet 1991; 20: 429–46. 6. Wagner JG. Noncompartmental and System Analysis. Pharmacokinetics for the Pharmaceutical Scientist. Lancaster, PA: Technomic Publishing Company; 1993. p. 83–99. 7. DiStefano JJ, III. Noncompartmental vs. compartmental analysis: some bases for choice. Am J Physiol Regul Integr Comp Physiol 1982; 243: R1–6. 8. Bradshaw-Pierce EL, Eckhardt SG, Gustafson DL. A physiologically-based phar­ macokinetic model of docetaxel disposition: from mouse to man. Clin Cancer Res 2007; 13: 2768–76. 9. Gustafson DL, Rastatter JC, Colombo T, Long ME. Doxorubicin pharmacokinetics: macromolecule binding, metabolism and elimination in the context of a physiological model. J Pharm Sci 2002; 91: 1488–501.

62

D.L. Gustafson and E.L. Bradshaw-Pierce

10. Dedrick RL, Forrester DD, Cannon JN, El Dareer SM, Mellett LB. Pharmacokinetics of 1-d-arabinofuranosylcytosine (Ara-C) deamination in several species. Biochem Pharmacol 1973; 22: 2405–17. 11. Farris FF, King FG, Dedrick RL, Litterst CL. Physiological model for the pharmacokinetics of cis-dichlorodiammineplatinum (II) (DDP) in the tumored rat. J Pharmacokinet Biopharm 1985; 13: 13–39. 12. Bischoff KB, Dedrick RL, Zaharko DS, Longstreth JA. Methotrexate pharmacokinetics. J Pharm Sci 1971; 60: 1128–33. 13. Tsukamoto Y, Kato Y, Ura M, Horii I, Ishikawa T, Ishitsuka H, Sugiyama Y. Investigation of 5-FU disposition after oral administration of capecitabine, a triple-prodrug of 5-FU, using a physiologically based pharmacokinetic model in a human cancer xenograft model: comparison of the simulated 5-FU exposures in the tumour tissue between human and xenograft model. Biopharm Drug Dispos 2001; 22: 1–14. 14. Gustafson DL. Use of physiologically-based pharmacokinetic modeling coupled to Monte Carlo simulation to predict the pharmacokinetic interactions between doxorubicin and taxanes in human populations. Proc Am Assoc Cancer Res 2002; 43: 208. 15. Sheiner LB, Rosenberg B, Marathe VV. Estimation of population characteristics of pharmacokinetic parameters from routine clinical data. J Pharmacokinet Biopharm 1977; 5: 445–79. 16. Sheiner LB, Rosenberg B, Melmon KL. Modelling of individual pharmacokinetics for computeraided drug dosage. Comput Biomed Res 1972; 5: 411–59. 17. Ette EI, Williams PJ. Population pharmacokinetics II: estimation methods. Ann Pharmacother 2004; 38: 1907–15. 18. Aarons L. Population pharmacokinetics: theory and practice. Br J Clin Pharmacol 1991; 32: 669–70.

Chapter 3

Bioanalytical Methods in Clinical Drug Development Walter J. Loos, Peter de Bruijn, and Alex Sparreboom

Abbreviations HPLC GC AAS ICP-MS UV F EC MS–MS

high-performance liquid chromatography gas chromatography atomic absorption spectrometry inductively coupled plasma mass spectrometry ultraviolet–visible detection fluorescence detection electrochemical detection tandem mass spectrometric detection.

3.1 Introduction Pharmacokinetic parameters determine to a large extent the pharmacological responses of individual patients. Moreover, differences in drug disposition account for a major share of interindividual differences in drug response. Drug doses required to achieve the same response in different patients may vary by more than one order of magnitude. It is for these reasons that incorporation of pharmacokinetic analyses in clinical cancer pharmacology has assumed its present importance in drug development. Detailed investigations on the absorption and disposition of a new drug are now required before it can be applied to human clinical trials. Pharmacologic research often utilizes drug level measurements to study the mechanism of drug action, including the contribution of metabolites to the observed drug effects.

A. Sparreboom (*) Department of Medical Oncology, Erasmus MC, Rotterdam, The Netherlands and Department of Pharmaceutical Sciences, St. Jude Children’s Research Hospital, Memphis, TN, USA e-mail: [email protected]

M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_3, © Springer Science+Business Media, LLC 2011

63

64

W.J. Loos et al.

Furthermore, bioavailability studies depend on drug level measurements; new drug formulations have to be tested for their bioequivalence in terms of established standard preparations. However, the most rapid growth at present occurs in the application of drug level measurements as a guide to optimize individual drug dosage regimens. Drug level measurements are based on analytical techniques suitable for the quantitative determination of drugs and their metabolites in biological samples. The most readily accessible body fluids are blood, saliva, and urine. While all of these fluids are utilized for drug assays, plasma or serum measurements may yield a better correlation between drug concentrations and effects. The analysis of drugs in whole blood should not be encouraged, since the erythrocyte/plasma concentration ratio is dependent on a number of variables that may impede the pharmacologic interpretation of the results. In addition, from a bioanalytical perspective, analysis of drug in whole blood samples is more challenging compared with plasma or serum. Therefore, blood samples should be centrifuged to obtain either plasma or serum, if the coagulation reaction has proceeded. There appears to be very little drug binding to the protein that forms the blood clot, i.e., fibrin, and assay results using plasma or serum are usually identical. It should be pointed out that the choice of the anticoagulant may affect assay results. Similarly, blood collection tubes may release substances such as plasticizers that can also interfere with the assay depending on the drug involved. The use of saliva for pharmacokinetic studies has been advocated for several anticancer drugs on the basis of its ready availability and the notion that saliva concentrations may reflect the concentration of unbound drug in plasma. However, the various salivary glands secrete fluids of different and varying composition, and thus, saliva concentrations may vary relative to unbound plasma concentrations, and only a few drugs show constant saliva/plasma concentration ratios. Urine drug analysis has normally a different set of objectives, which include mass balance studies, determination of renal clearance, and/or monitoring individual compliance with a prescribed dosage regimen. Because of the large fluctuations in urinary drug concentrations under different diuresis conditions, drug level measurements in spot urine samples are typically not sufficiently informative to estimate renal clearance or amount of drug excreted. For reasons outlined above, most laboratories now use and focus on plasma (or serum) for drug analysis in the context of anticancer drug development or subsequently, for routine implementation of approved agents. A literature survey of the various analytical techniques currently in use for the quantitative determination of anticancer drug concentrations in human plasma or serum is given in Table 3.1. With few exceptions, the methods employ chromatographic separation techniques in conjunction with a variety of detection procedures. The current chapter provides an introduction to the terminology and scope of bioanalytical methods used in pharmacokinetic studies of small-molecule anticancer agents, including methods for sample preparation and separation, as well as guidance for bioanalytical method validation.

3  Bioanalytical Methods in Clinical Drug Development

65

Table 3.1  Survey of bioanalytical techniques for small-molecule anticancer drugs HPLC Drug UV F EC MS–MS GC AAS/ICP-MS Actinomycin + + Aminopterin + + + Arsenic trioxide + + Azacitidine + + Bleomycin + + + Bortezomib + Busulfan + + + + Capacitabine + + Carboplatin + + + Carmustine + + + + Chlorambucil + + + Cisplatin + + + + Cladribine + + + Clofarabine + + Cyclophosphamide + + + Cytarabine + + + Dacarbazine + + + Dasatinib + Daunorubicin + + + + + Decitabine + Dexamethasone + + + + Docetaxel + + Doxorubicin + + + + + Epirubicin + + Erlotinib + + Estramustine + + + Etoposide + + + + Everolimus + + Floxuridine + + Fludarabine + + Fluorouracil + + + + Gefitinib + Gemcitabine + + Hydroxyurea + + Idarubicin + + Ifosfamide + + + Imatinib + + Irinotecan + + + Ixabepilone + Lapatinib + Lomustine + + + Mechlorethamine + + Melphalan + + + + + Mesna + + + Mercaptupurine + + + + (continued)

66

W.J. Loos et al. Table 3.1  (continued) Drug Methotrexate Mithramycin Mitomycin Mitoxantrone Nilotinib Oxaliplatin Paclitaxel Pemetrexed Pentostatin Procarbazine Raltitrexed Sorafenib Sunitinib Tamoxifen Tegafur Temozolomide Temsirolimus Teniposide Thiotepa Tioguanine Topotecan Trabectedin Vinblastine Vincristine Vindesine Vinorelbine Vorinistat

HPLC UV

F

EC

MS–MS

GC

+

+

+

+

+ +

+ +

+

+

+ + + + + + + + + + + + + + + + + + + +

+ + + +

+

+ + +

+ + +

+ + + + + + + + + + + + + + +

AAS/ICP-MS

+

+

+

+ +

+ + + + + +

3.2 Methods for Sample Preparation Suitable preparation of biological specimens is decisive in the successful application of an analytical technique. The preparation should be as simple as possible yet allow the specific assay of a drug in the presence of numerous biological (endogenous) compounds. The extent of sample workup is, therefore, largely determined by the selectivity of the analytical technique. Potentially interfering endogenous components need to be removed prior to analysis. A second objective of the preparation of biological specimens is to protect the analytical equipment from contamination by lipids, proteins, and undissolved particles. Biological sample preparation has to vary according to the technical demands of the various analytical instruments. Since the advent of highly selective methods that combined chromatographic separation and detection in a single unit, the second objective has gained in importance.

3  Bioanalytical Methods in Clinical Drug Development

67

Most analytical methods for the quantitation of anticancer agents in plasma determine the total drug concentrations, i.e., the drug bound to plasma proteins as well as free drug dissolved in plasma water. Since the free fraction of drug dissolved in the plasma water is able to cross cell membranes, it is considered as the biological active fraction. As a consequence, the relationship between total drug concentration in plasma and toxicity and/or efficacy will only be strong if the degree of plasma protein binding is constant within and between patients, or if the degree of drug binding to proteins is negligible such that alterations in binding make insignificant changes in unbound concentrations. Surprisingly, this holds true for most anticancer agents. There are, however, circumstances in which the quantification of unbound drug concentrations is essential, such as in case of irreversibly to nearly covalent protein-binding (e.g., platinum-containing drugs), proteinconcentration-dependent binding (e.g., epipodophyllotoxins), agents that undergo metabolic interconversion (e.g., campothotecin analogs), or in case protein binding is influenced by formulation vehicles (e.g., taxanes) [10].

3.2.1 Total Drug Measurements Adequate sample preparation is a key step in bioanalysis, especially during highthroughput analysis, in which sample preparation is becoming the rate-limiting step in the generation of pharmacokinetic data. Traditionally, sample preparation has been performed (manually) using protein precipitation, solid phase extraction, or liquid–liquid extraction, and all of these procedures can be transferred to automated multiwell plate formats and thus be adopted in high-throughput analysis [15, 18]. 3.2.1.1 Protein Precipitation Protein precipitation is a widely used, inexpensive, simple technique, involving the addition of ionic salts, organic solvents, or a combination of ionic salts and an organic solvent to plasma samples. After mixing and centrifugation, the clear supernatant can either be injected directly into an analytical system or first be concentrated by evaporation of the supernatant. Protein precipitation is a fast sample preparation technique, which can easily be automated. As there is no extraction involved, such as in the case of solid phase or liquid–liquid extraction, many salts and endogenous compounds are still present in the sample which might interfere with the analysis [4]. 3.2.1.2 Solid Phase Extraction With solid phase extraction, the analytes of interest are extracted from plasma by partitioning of the analyte between a solid phase and a liquid, thereby removing

68

W.J. Loos et al.

salts and endogenous compounds, which might interfere with the analysis, while concentrating the analytes. The solid phase extraction technique consists of four critical steps: conditioning of the solid phase matrix, sample loading, washing of the solid phase matrix while retaining the analytes of interests, and elution of the analytes of interest from the solid phase matrix. Several types of solid phase matrixes are available. The appropriate solid phase adsorbent material as well as a suitable washing solvent and eluting solvent should be chosen based on the physical and chemical properties of the analytes. Solid phase extraction, however, has several disadvantages such as the relative poor reproducibility caused by differences between batches of solid phase adsorbent material and difficulty in standardizing the use of the vacuum required in all steps of the procedure. Although solid phase extraction is expensive, it can be automated off-line as well as on-line, thereby improving the reproducibility while reducing the time-consuming and laborious work of an analyst [15]. 3.2.1.3 Liquid–Liquid Extraction Liquid–liquid extraction, also known as solvent extraction is, frequently used, mainly in studies involving relatively small numbers of samples, in which case automated solid phase extraction is typically not the method of choice due to its relatively time-consuming and expensive nature. Liquid–liquid extraction is relatively easy to develop, although it is usually more difficult to automate. During liquid–liquid extraction, an organic solvent, or a mixture of different organic solvents, immiscible with water, is added to plasma. Subsequently, the samples are vortex-mixed, during which an exchange of analytes into the organic phase will take place, the rate of which depends on the composition of the organic solvent used and the affinity and solubility of the analytes of interest in the organic phase relative to the water phase. The organic solvent is subsequently evaporated and the residue dissolved in an aqueous solution, which subsequently will be analyzed. Plasma proteins and the majority of salts are retained in the water phase. Compared to solid-phase extraction, liquid–liquid extraction is less selective and requires high-purity organic solvents [15, 18].

3.2.2 Unbound Drug Measurements For the measurement of unbound drug concentrations, various procedures can be applied, including equilibrium dialysis, ultrafiltration, ultracentrifugation, and protein precipitation, which will be briefly discussed below [14]. Besides these techniques, microdialysis can be applied to study the pharmacokinetics of unbound fractions of anticancer agents in plasma as well, and microdialysis probes for intravenous use have recently become commercially available [5, 8].

3  Bioanalytical Methods in Clinical Drug Development

69

3.2.2.1 Equilibrium Dialysis Equilibrium dialysis is regarded as the standard sample preparation technique for the analysis of unbound drug concentrations. Equilibrium dialysis is based on the establishment of an equilibrium of the analyte of interest between plasma, containing the analyte bound to proteins and free circulating in the plasma water, and a buffer solution. Both compartments are separated by a semipermeable membrane, through which small molecules will diffuse depending on their gradient. After a certain period of incubation, the nonprotein-bound fraction of the drug in the plasma compartment is in equilibrium with the fraction in the buffer solution. The buffer solution is subsequently analyzed, and the observed concentration is considered the unbound concentration in plasma. In contrast to ultracentrifugation and ultrafiltration, equilibrium dialysis is amenable to automation, allowing large numbers of samples to be analyzed. However, the technique faces several shortcomings such as the potential nonspecific absorption of analytes to the dialysis devices and membranes, volume shifts in the different compartments, and the relative long time required to establish the equilibrium. 3.2.2.2 Ultrafiltration Ultrafiltration is based on physical separation of free drug molecules in the plasma water by filtering plasma samples through a semipermeable membrane under pressure, generated by centrifugation. The ultrafiltrate can be analyzed, and the observed concentration is considered as the unbound concentration in plasma. The key advantage of ultrafiltration compared with equilibrium dialysis is the short analysis time, the lack of dilution effects caused by volume shifts, and its relative ease of use. As with equilibrium dialysis, potential nonspecific adsorption to the devices might be a limiting factor using the technique. In addition, the unbound fraction might alter during the centrifugation process as the protein and lipid concentration in the upper plasma compartment will increase in time. 3.2.2.3 Ultracentrifugation Ultracentrifugation can be an alternative to equilibrium dialysis and ultrafiltration, as it lacks some of the problems associated with these techniques. During ultracentrifugation, large macromolecules, such as proteins are spun down at high speed. The supernatant contains the free fraction of drugs dissolved in plasma water and are subsequently analyzed. Potential disadvantages of ultracentrifugation include the possibility that the unbound fraction is influenced by physical phenomena such as sedimentation, back diffusion, viscosity, and binding of analytes to lipoproteins present in the supernatant following centrifugation at high speed.

70

W.J. Loos et al.

3.2.2.4 Protein Precipitation In case of irreversible or nearly covalent binding of anticancer drugs, such as in the case of the currently clinically approved platinum-containing anticancer agents cisplatin, carboplatin, and oxaliplatin, unbound concentrations can be measured following a simple protein precipitation step, avoiding all pitfalls of equilibrium dialysis, ultrafiltration, and ultracentrifugation. In addition, no specific equipment or devices are required, and protein precipitation can be conducted in almost every laboratory and even at the bedside of patients. Compared to the other techniques, deproteinization is inexpensive and very rapid by its instantaneous precipitation of proteins, leaving the unbound fraction, after centrifugation, in the protein-free supernatant. Proteins can be precipitated by the addition of ionic salts or organic solvents to plasma samples. 3.2.2.5 Microdialysis (Extracellular Fluid) Microdialysis is a minimally invasive sampling method based on the diffusion of analytes from the interstitial compartment through a semipermeable membrane and enables direct assessment of tissue disposition and penetration of the free fraction of small molecules (Fig. 3.1). The concept of microdialysis has been applied extensively in neurological research where it has been used to monitor neurotransmitter concentrations in brain tissue. Microdialysis has also been applied in oncology; clinical drug disposition studies using microdialysis have been performed with a variety of anticancer drugs, including 5-fluorouracil, capecitabine, cisplatin, carboplatin, dacarbazine, and methotrexate [9]. Microdialysis involves inserting into tissue or fluid a probe that consists of an inlet through which an isotonic fluid, the so-called perfusate, matching the extracellular fluid (ECF) is infused. During flow of the perfusate along a semipermeable membrane, exchange of small molecules from the ECF into the perfusate will take place. The solution that exits the probe, the dialysate, can be collected for analysis (Fig. 3.1). Since the microdialysis procedure is not performed under equilibrium conditions, the concentration of the analyte under investigation in the dialysate will be different from that in the ECF. The relationship between the dialysate concentration and the ECF concentration is referred to as the relative recovery. Adequate calibration of the probes is one of the most important issues in using the microdialysis technique for quantitative analysis of unbound fractions of drugs because it allows the conversion of the microdialysate concentrations into extracellular concentrations (3.1).

Concentration in ECF = (Concentration in dialysate / Recovery) × 100

(3.1)

As the rates of diffusion for most analytes differ between aqueous solutions and tissue, adequate validation of the in vivo recovery is required to obtain a reliable estimate of the ECF concentrations. If several microdialysis probes are inserted, even into the same tissue, all probes should be calibrated as the physiological

3  Bioanalytical Methods in Clinical Drug Development

71 Dialysate Perfusate

Tissue cells

Analytes

Blood vessel

Microdialysis probe

Fig. 3.1  Principles of microdialysis

properties in (tumor) tissue may differ from region to region. Calibration of microdialysis probes based solely on in  vitro method is inappropriate, although such procedure can be useful to evaluate potential adsorption of analytes to microdialysis probes. The most frequently applied and accepted method to determine the relative in vitro recovery of an analyte is known as retrodialysis (3.2), in which the perfusion solution is spiked with a known concentration of the analyte. Retrodialysis is based on the assumption that the recovery is independent of the route of diffusion, as analyte molecules in the presence of a concentration gradient diffuse across the membrane in both directions. Recovery = (Concentration in perfusate − Concentration in dialysate) / Concentration in perfusate × 100% (3.2)

Because the presence of the analyte in the ECF will result in an overestimation of the relative recovery and thus an overestimation of ECF concentrations during the pharmacokinetic study, retrodialysis should be performed before the first drug administration. To avoid this problem, a radiolabelled form of the analyte might be added as internal standard to the perfusion liquid.

72

W.J. Loos et al.

3.3 Methods for Sample Separation Practical aspects for conducting pharmacokinetic studies include the development and validation of bioanalytical methods for drug measurement and construction of concentration-time profiles for calculation of pharmacokinetic parameters. The most commonly applied techniques for the quantitation of anticancer agents are high-performance liquid chromatography (HPLC) coupled to ultraviolet (UV) absorbance or fluorescence emission detectors, HPLC coupled to a tandem mass spectrometric detector (LC–MS–MS), atomic absorption spectrometry (AAS), and induction-coupled plasma mass spectrometry (ICP-MS).

3.3.1 Liquid Chromatography HPLC has become the most powerful tool in analytical chemistry. It has the ability to separate, identify, and quantitate compounds of interest that are present in any sample that can be dissolved in a liquid (Fig. 3.2). Currently, compounds in trace concentrations as low as parts per trillion may easily be identified. Furthermore, HPLC has been applied to just about any sample, such as pharmaceuticals, food, nutraceuticals, cosmetics, environmental matrices, forensic samples, and industrial chemicals [7]. The introduction of small particle sizes for the column packing material with pressured column systems, to shorten analysis time, has contributed to an explosion of HPLC drug assays in biological samples and has confirmed the versatility, efficacy, precision, and speed of this technique. Furthermore, new column packing materials, highly sensitive detectors, and reliable, quantitative injection systems have all contributed to the current utility of HPLC in pharmacokinetic studies during both clinical drug development as well as in routine drug level monitoring. The four basic types of chromatography are liquid–solid, partitioning, ionexchange, and exclusion chromatography. Liquid–solid chromatography or adsorption chromatography mainly uses silica gel particles as the stationary phase. Polar compounds are retained longer than lipophilic material on a silica gel column.

Sample injector Solvent delivery system

Fig. 3.2  Principles of HPLC

Detection UV/VIS Fluorescence ECD

3  Bioanalytical Methods in Clinical Drug Development

73

Partition chromatography is now mainly performed on columns containing a stationary phase chemically bonded to inert support material. It is called “normal phase” if the stationary phase is more polar than the mobile phase. In “reverse phase” partition chromatography, the stationary phase is more hydrophobic than the mobile phase, and this method is the most widely used mode of chromatography today. This dominance of reverse phase HPLC is based on the high hydrophobic nature of drugs relative to potentially interfering endogenous compounds, the rapid elution of polar endogenous compounds, and the possibility to analyze highly polar ionic and nonionic drugs in conjunction with ion-pair chromatography, including 6-mercaptopurine and methotrexate. Ion-exchange chromatography depends upon the exchange of ions between the mobile phase and the ionic site of the packing, for example, sulfonic acids and quaternary ammonium groups. The actual mode of separation may not depend exclusively on the ionic strength of the drug, with hydrophobicity also contributing to the chromatographic behavior. Finally, exclusion chromatography is based on the molecular size of the solute. It is potentially useful for the separation not only of macromolecules but also of small (<1,000 Da)molecular-weight substances. Further advances in instrumentation and column technology were made in the past 5 years to achieve very significant increases in resolution, speed, and sensitivity in liquid chromatography. Columns particles as small as 1.7 mm and instrumentation with specialized capabilities designed to deliver mobile phase at 15,000  psi (about 1,000  bar) were developed to achieve a new level of performance. These systems are currently referred to as ultraperformance liquid chromatography (UPLC), ultrahigh performance liquid chromatography (U-HPLC), or Hres fast liquid chromatography [19]. In the next section, the various detection techniques available in liquid chromatography are briefly discussed. 3.3.1.1 Ultraviolet and Visible Spectroscopy Ultraviolet and Visible Spectroscopy (UV/VIS) detection is one of the most commonly applied detection methods in conjunction with HPLC. It measures the absorption of monochromatic light by the solute according to the Lambert–Beer law. The concentration of an analyte in solution can be determined by measuring the absorbance at some wavelength in the range of 190–650 nm. The technique is capable of detecting traces of many drug-related compounds. However, if by-products and other impurities do not have a suitable chromophore, then the technique is not useful. Modern instruments can change the wavelengths during the run, so that each peak can be detected at its optimum wavelengths or can scan over a range of wavelengths in the case of photodiode-array detection. The selectivity of UV absorbance detection strongly depends on the wavelength used. Close to the minimum wavelength of 190 nm, most analytes or solvents used in a mobile phase absorb light. With an increase in the wavelength, the selectivity of the technique will increase. The maximum sensitivity that can be reached is approximately 10−8 or 10−9 mg/mL [7].

74

W.J. Loos et al.

3.3.1.2 Fluorescence Fluorescence detection is based on the property of certain molecules to emit light after excitation by UV radiation. It is one of the most sensitive detection techniques and therefore often used for trace analysis. Although the detector is very sensitive, its response is only linear over a relatively small concentration range, typically two orders of magnitude. The maximum sensitivity that can be achieved with fluorescence detection is approximately 10−9–10−11 mg/mL [7]. 3.3.1.3 Electrical Conductivity Measurements based on electrical conductivity detection (ECD) are frequently used for the analysis of inorganic acids, bases, and salts. The use of ECD is only possible for substances that are able to ionize and require a conductive mobile phase, and this technique is therefore almost exclusively used in reverse-phase HPLC. Electrochemical detection is based on the transfer of electrons between oxidizable or reducible molecules in solution and a solid conductor. The ECD detects this electrical current, and the electrode potential is set relative to a reference electrode, which has a constant potential difference with the solution. Electrochemical detection is a very sensitive method for use with high-performance liquid chromatography, and capable of detecting in the femtomolar range. The technique has found its way in many laboratories in the field of pharmaceutical chemistry, food chemistry, and environmental chemistry and is used for many applications such as microdialysis [7]. 3.3.1.4 Mass Spectrometry Mass spectrometry (MS) is the most sensitive detection technique of molecular analysis and has become the most important quantitative detection technique in analytical cancer pharmacology [17]. It has the ability to yield information about the relative molecular mass and the structure of the analyte. The principle of mass spectrometry is the production of ions that are subsequently separated according to their mass-to-charge ratio (m/z) and detected (Fig.  3.3). The selection of sample inlet depends upon the sample and the matrix. The two most common techniques for sample inlet into the mass spectrometer are gas chromatography and liquid chromatography. Many ionization methods are available in MS techniques, each with its own advantage and disadvantage depending on the type of sample and analyte under investigation. Electron ionization (EI) and chemical ionization (CI) are only suitable for gas-phase ionization, whereas fast atom bombardment (FAB), electrospray ionization (ESI), and matrix-assisted laser desorption (MALDI) are used to ionize condensed-phase samples. The most common ionization technique used for MS is the EI. It causes extensive fragmentation so that the molecular ion is not observed for many compounds, and therefore, it has its limitations. Fragmentation is minimized in CI by reducing the

3  Bioanalytical Methods in Clinical Drug Development

a

Single quadrupole Source

b

Quadrupole 1

Selection ion

Detection

Source

Quadrupole 1

Quadrupole 2

Ionization

Selection Parent ion

Collission

Ionization

75

Triple quadrupole Quadrupole 3

Selection Daughter ion

Detection

Fig. 3.3  Principles mass spectrometry

amount of energy produced by the reaction and is very useful for molecular mass determination. ESI is one of the atmospheric pressure ionization techniques in which the ionization takes place at atmospheric pressure and then transferred to the mass spectrometer. Atmospheric pressure chemical ionization (APCI) is the second atmospheric ionization technique. ESI and APCI differ in how ions are generated and in analyte compatibility. In particular, APCI is used for less polar and smaller compounds that have some volatility. In contrast, MALDI is based on the bombardment of sample molecules with laser light to bring sample ionization and is used for thermolabile, nonvolatile organic compounds especially those of high molecular mass and is used in biochemical areas for the analysis of proteins, peptides, and glycoproteins. After ions are formed in the source region, they are accelerated into the mass analyzer, and the ions are separated according to their m/z value. Analyzers are typically described as either continuous or pulsed. Continuous analyzers include quadrupole filters and magnetic sectors. They transmit a selected m/z to the detector, and the mass spectrum is obtained by scanning the analyzer so that different mass-to-charge ratio ions are detected. Pulsed mass analyzers are the other major class of mass analyzers, but they are less common. These instruments collect an entire mass spectrum from a single pulse of ions. Pulsed analyzers include timeof-flight and quadrupole ion trap mass spectrometers [17].

3.3.2 Gas Chromatography There are two basic modes of gas chromatography (GC), namely, gas–liquid partitioning and gas–solid adsorption. The latter is mainly applied to gases and highly volatile compounds such as ethanol. In analytical cancer pharmacology, the predominant GC

76

W.J. Loos et al.

method is by partitioning. Partitioning GC is performed either on columns packed with an inert, microparticulate support material that is coated with the stationary liquid phase or on capillary long tubes coated on the inside wall with the liquid phase. Capillary GC columns are highly efficient, since turbulence of the carrier gas is minimized, and these columns offer the greatest separation potential in GC. The major limitations of GC applications to organic material are the requirements for a sufficiently high vapour pressure and the thermostability at the GC temperatures. The most commonly applied GC detection techniques include thermoconductivity, flame ionization, nitrogen (phosphorus)-sensitive flame ionization, electron capture, flame photometry, and mass spectrometry. Although a large number of drugs can be and have been analyzed by GC-based procedures, this technique has been clearly overshadowed by recent HPLC developments, in particular HPLC coupled with MS detection, due to improved speed, sensitivity, precision, and specificity [6].

3.3.3 Atomic Spectroscopy Atomic spectroscopy, the interaction of light with gaseous atoms, is widely used for the quantification of approximately 70 elements. In oncology, the technique is mainly used for the quantification of platinum-containing drugs, such as cisplatin, carboplatin, oxaliplatin, and satraplatin, with sensitivities of the methods in the nanogram per milliliter and even picogram per milliliter range [1]. A requirement for atomic spectroscopy is the existence of gaseous atoms. A critical step in all atomic spectroscopic techniques is thus the atomization, the process in which a sample is volatilized and decomposed to produce free gaseous atoms. The efficiency and reproducibility of the atomization for a large part determines the sensitivity, accuracy, and precision of the method applied. Atomic spectroscopy can be divided into atomic absorption, atomic fluorescence, and atomic emission techniques and are categorized on the basis of the manner the samples are atomized. The most commonly applied atomic spectroscopic techniques for the quantitative determination of platinum-based anticancer agents are flame and graphite furnace atomic absorption. The atomic-spectroscopyaffiliated technique referred to as inductively coupled plasma mass spectrometry is now increasingly used and will likely replace the absorption-based techniques in pharmacokinetic studies. Which technique is most suitable in a specific setting depends on several factors, including sensitivity (Fig.  3.4) and throughput, and these issues are discussed below.

3.3.3.1 Atomic Absorption Spectroscopy Atomic absorption occurs when a ground state atom absorbs energy, i.e., the transfer of energy of a photon to an atom to promote an electron from the ground state to the excited state. The amount of light absorbed, at a wavelength specific for the

3  Bioanalytical Methods in Clinical Drug Development

77

Flame AAS Graphite furnace AAS ICP-MS

100

10 1 0.1 0.01 Detection limits (ug/L)

0.001

Fig. 3.4  Relative sensitivity

atom, is a measurement of the concentration of the atom in the sample. Atomic spectra have, in contrast to molecular absorption spectra, very narrow spectral lines, which make the technique highly specific. Originally, a flame was commonly applied to produce gaseous atoms. In the flame atomization, a spray of an aqueous solution is dispersed and mixed with gaseous fuel and oxidant and brought into the flame. At the base of the flame, located just above the tip of the flame, the solvent is evaporated, and the resulting solid particles are subsequently atomized in the center of the flame, the warmest part of the flame. Flame atomization, unfortunately, is not an efficient method. As the velocity of the sample–fuel mixture in the flame is high, only a small fraction of the samples is finally atomized and analyzed, and secondly, the residence time of atoms in the optical path, during which the absorption takes place, is relatively short (milliseconds). As a consequence, flame Atomic Absorption Spectroscopy (AAS) is relatively insensitive (Fig.  3.4), and flame atomizers are not used frequently in pharmacokinetic studies of platinum-based anticancer agents. Enhanced sensitivity has been achieved by using electrothermal vaporization of the sample on graphite furnaces. During graphite furnace AAS (also referred to as flameless AAS), a liquid sample is injected into the graphite furnace. The sample is first evaporated at a low temperature (100–150°C) to remove solvents. Subsequently, the remaining solid particles are pyrolyzed at higher temperatures (300–1,500°C), followed by an almost instant increase of the temperature to 1,600–3,000°C (~2,400°C for platinum), during which the atomization takes place. The absorption is, in contrast to flame AAS, measured during a few seconds as the atoms are retained in the graphite furnace. Graphite furnace AAS is thus more efficient, and as a result a more sensitive method, compared to flame AAS since all of the analyte in the injected sample is atomized, and moreover, analysis takes place during a prolonged period. On the other hand, the graphite furnace technique takes longer time than the flame technique, and fewer elements can be determined. Absorption of light by atoms occurs over a very narrow range of wavelengths (0.01–0.05 nm). In contrast to, for example, UV detectors coupled to HPLC, in which light sources are applied that emit over wide wavelength ranges, light sources for AAS in most cases only emit light of a single wavelength. In the light source, atoms of the same analyte are promoted to excited electronic states, subsequently emitting a single spectral line of light, making the technique of atomic absorption highly specific [1].

78

W.J. Loos et al.

3.3.3.2 Inductively Coupled Plasma Mass Spectrometry An alternative to AAS for platinum analysis is based on inductively coupled plasma spectrometry (ICP). Compared to graphite furnace AAS, this method can reach a higher level of sensitivity and productivity. The ICP setup comprises various components (e.g., a single quadrupole mass spectrometer) that are also used with other types of mass spectrometry, including HPLC-MS. In contrast to HPLC-MS, in which relative low temperature ion sources are used, the plasma ion source in ICP-MS uses an extreme high temperature of >5,000°C to break down the molecules. ICP-MS detects elemental ions instead of ions of intact molecules, and therefore, ICP-MS is more related to AAS than HPLC-MS. As in the single quadrupole mass analyzer, ions with a specific mass are selected and detected, and ICP-MS is capable of analyzing up to 40 elements in a single sample in less than 1 min [2].

3.4 Validation Requirements Bioanalytical methods play a pivotal role in pharmacokinetic, toxicokinetic, and bioequivalence studies and should be thoroughly validated to yield valid results. Especially in early clinical drug development, such as Phase 0/microdosing and Phase I studies, accurate and precise analysis at nano- or picomolar concentrations is needed, requiring the use of cutting-edge techniques. Bioanalytical method validation includes all of the procedures that demonstrate that a particular method used for quantitative measurement of analytes in a biological matrix is reliable and reproducible [12, 13, 16]. There might, however, be circumstances, such as during clinical phase I studies, in which a method is used over a relatively long period. In such cases, it is essential to demonstrate, by analysis of a calibration curve and quality control samples prior to analysis of study samples, that the method provides similar results as during the initial validation. In case study samples are analyzed at different laboratories, even using the same initially validated method, evidence should be provided that identical results will be obtained. The method should be adequately validated in each laboratory. Furthermore, interlaboratory reliability should be established by a crossvalidation using standards spiked with drug in the relevant biological matrix, as well as by using incurred samples. In case different techniques are used for the bioanalysis in one study (e.g., HPLC-UV and HPL-MS), a cross-validation of the different methods is required (Fig. 3.5). Besides full validation of analytical methods, a partial validation of an existing method might be necessary. The extent of the partial validation differs from situation to situation and might range from only one validation item to almost a full validation, depending on the changes made. Changes requiring a partial validation include, but are not limited to, changes in system components (e.g., a new detector), changes in sample processing procedures (e.g., from liquid–liquid extraction to

3  Bioanalytical Methods in Clinical Drug Development Fig. 3.5  Cross-validation

79

LC-MS/MS (ng/mL)

4000

3000

2000

1000

0 0

1000

2000 3000 HPLC (ng/mL)

4000

solid phase extraction) and sample volumes (e.g., for implementation of the method in samples from a pediatric study), changes in used anticoagulant during blood collection (e.g., from heparin to EDTA), changes of matrix within species (e.g., from mice to rat), and changes in concentration ranges (e.g., during the conduct of a clinical Phase I study). The critical parameters for bioanalytical method validation include range of reliable response and linearity, selectivity or specificity, sensitivity, accuracy, precision, and reproducibility and stability, and each of these are briefly discussed below [12, 13, 16].

3.4.1 Range of Reliable Response, Linearity, and Calibration Curves The range of reliable response is the range of concentrations over which the response of the detector, in general, is linear. In rare cases, the response curve for chemical assays might be nonlinear, which is acceptable if the nonlinearity is shown to be reproducible. The range of reliable response should be established prior to the bioanalytical method validation and determines the range of the calibration curve. Linear concentration versus response relationships in most cases can be established over approximately 3–4 logs, depending on the technique applied. The calibration curve standards should be prepared in exactly the same matrix as the study samples (i.e., potassium EDTA-derived human plasma). In case it is almost impossible to obtain the matrix (e.g., cerebrospinal fluid) in sufficient amounts, calibration curves might be prepared in a surrogate matrix. Equivalency in responses between the matrices should, however, be shown. Calibration curves

80

W.J. Loos et al.

should be constructed by spiking known concentrations of the analyte of interest to blank matrix, taking into account that the matrix might be diluted by a maximum of 5% (vol/vol). Calibration curves should consist of six to eight nonzero samples covering the expected range and acknowledging the established range of linear response. Besides the nonzero samples, a double blank and blank (e.g., matrix spiked with internal standard if applicable) should be processed to demonstrate selectivity and specificity during each run. The calibration curve standard with the lowest concentration reflects the lower limit of quantitation (LLQ or LLOQ), while the calibration curve standard with the highest concentration reflects the upper limit of quantitation (ULQ or ULOQ) of the bioanalytical method. It is generally accepted that from the processed calibration curve standards, 75% or a minimum of six standards should fall within 15% (20% at the LLQ) of their spiked (i.e., nominal) concentration, to be considered acceptable. Processed calibration curve standards with values falling outside the specified criteria might be excluded from the calibration curve. A weighting factor (i.e., 1/concentration) might be applied and should especially be considered.

3.4.2 Selectivity or Specificity Selectivity or specificity is the ability of a bioanalytical method to differentiate and quantify the analyte in the presence of other components in the sample, including potential metabolites and endogenous matrix components. The term selectivity is mostly used, as specificity implies that the analytical method is so selective that the detector only responds to a single compound. Selectivity should be demonstrated for each analyte of interest by processing at least six different sources of the specific matrix as double blank samples and spiked at the concentration of the LLQ. The responses in at least 80% of the double blank processed samples might be a maximum of 20% of the response of the analyte at LLQ and 5% of the response of the internal standard (if applicable). The backcalculated concentrations of at least 80% of the samples processed at the concentration of the LLQ should fall within 20% of the nominal concentration. Potential interference from exogenous components (e.g., concomitant medication and food supplements) should be tested to ensure that there is no interference. Pools of the matrix spiked with the analyte(s) at a concentration at the lower part (e.g., three times the LLQ) and higher part (e.g., 75% of the ULQ) and with the potential interfering substance should be processed. Back-calculated concentrations of the analyte(s) should be within 15% of the nominal concentration to show lack of interference.

3.4.3 Sensitivity The sensitivity of an analytical method is often expressed as the limit of detection (LOD) or lower limit of quantitation (LLQ or LLOQ). The LOD is defined as the

3  Bioanalytical Methods in Clinical Drug Development

81

lowest analyte concentration from which the response can be distinguished from the background noise. Several methods for the determination of the LOD exist, from which the signal-to-noise approach is most frequently applied. A signalto-noise ratio between 2 and 5 is generally accepted as being the LOD. As the LOD depends on several factors (e.g., purity of reagents, matrix differences), which might be highly variable between runs and are not easy to control, the LOD might differ between analytical runs and is therefore a less useful parameter for pharmacokinetic analyses. The LLQ, although formally not being synonymous with sensitivity of the analytical method, is a more appropriate parameter to establish and report during the validation. The LLQ is the lowest amount or concentration of an analyte that can be quantitated with an acceptable degree of accuracy and within-run and between-run precisions. The LLQ can be validated by several methodologies, from which the method based on back calculation of known spiked concentrations is most frequently used. Back-calculated concentrations of the analyte(s) in 80% of the processed samples should be within 20% of the nominal concentration (i.e., accuracy of 80–120%), while the within-run and between-run precisions, as well as the relative standard deviation of a set of processed samples within a run, should be £20% to be considered acceptable. In addition, the response of the detector at the concentration of the LLQ should be at least five times the response in blank processed matrix.

3.4.4 Accuracy The accuracy of an analytical method describes how close the mean test results obtained by the method are to the nominal spiked concentration of the analyte and is determined by replicate analysis of samples containing known amounts of the analyte (Fig. 3.6). The accuracy of a method is most commonly determined by replicate analysis of spiked pools (i.e., quality control samples; QC) of analyte in blank matrix at the concentration of the LLQ, two to three times the LLQ (i.e., QC Low), the midpoint of the calibration curve (i.e., QC Middle), and at 75–80% of the ULQ (i.e., QC High). A fourth QC sample with a concentration above the ULQ should be validated in case the concentration in study samples exceeds the ULQ (i.e., QC Diluted). This fourth QC sample should be appropriately diluted in blank matrix prior to processing in such a way that the concentration in the diluted sample falls within the range of the calibration curve. The LLQ and QC samples should be processed in quintuplicate in at least three separate runs, in which the calibration standards are prepared freshly and independently in each run. In general, the accuracy (3.3) is considered acceptable, in case the backcalculated concentrations of the analyte(s) in 80% of the processed samples falls within 15% (20% at LLQ) of the nominal concentration (i.e., accuracy of 85–115%; 80–120% at LLQ).

Accuracy = mean observed concentration / nominal concentration × 100% (3.3)

82

W.J. Loos et al.

Calibration Curve Standards

Detector response

precision: passed accuracy: passed precision: failed accuracy: passed precision: passed accuracy: failed

Nominal concentration

Fig. 3.6  Precision and accuracy

Precision describes how close the individual measures of an analyte are when the procedure is applied repeatedly to multiple aliquots of a single homogeneous volume of biological matrix. That is, it is a measure of the random error associated with the measurement process (Fig. 3.6). Precision of a sample is often expressed in the within-run precision and between-run precision. The within-run precision, also known as the repeatability, describes the precision of an analytical method at the tested concentrations within one analytical run, in which analytical circumstances are considered constant. The between-run precision, also known as the reproducibility, describes the precision of an analytical method over multiple analytical runs over a relative short period of time in the same laboratory, most commonly under the same analytical conditions and performed by a single analyst. The within-run and between-run precisions of an analytical method are most commonly determined using the observed concentration data of the processed pools of QC-samples as described for the accuracy. A widely accepted approach for the calculation of the within-run (WRP) and between-run precisions (BRP) is the analysis of variance [ANOVA; (3.4) and (3.5)]. In general, the WRP and BRP are considered acceptable, in case the values do not exceed 15% (20% at LLQ).

WRP (%) = [(wgMS)0.5/ mean observed concentration ] × 100%



(3.4)

BRP (%) = [[(bgMS − wgMS) / n] /mean observed concentration] × 100 (3.5) 0.5

In which, wgMS is the within-group Mean Square, bgMS is the betweengroup Mean Square, and n is the number of replicates within each day. In addition, the relative standard deviation, or coefficient of variation, of multiple

3  Bioanalytical Methods in Clinical Drug Development

83

(e.g., 5) replicates in each separate analytical run should not exceed 15% (20% at LLQ) to be considered acceptable (3.6).

CV(%) = standard deviation / mean observed concentration × 100% (3.6)

3.4.5 Reproducibility Besides establishing reproducibility as described above for the between-run precision, the reproducibility of an analytical method also might be evaluated by varying analytical settings (e.g., different batches of the column, column temperature, and/or mobile phase composition), performing the analysis at different analytical systems, by different analysts or at different laboratories. Those types of reproducibility tests are also known as the ruggedness and robustness of the analytical method.

3.4.6 Stability The stability of an analyte in fluid is a function of the matrix in which the analyte is dissolved, of the container systems in which the fluid is collected and stored, and of the storage condition of the containers. Stability of an analyte should, thus, be tested under the relevant conditions reflecting situations during sample analysis of the study samples. Conditions include the stability of analyte in the stock solution, in the biological matrix during short-term and long-term storage, in the matrix during multiple freeze–thaw cycles, and as processed sample in the autosampler of the analytical system. In addition, stability data of analyte in incurred samples (i.e., samples derived from study subjects) is of high interest. The stock solution stability should be performed at ambient temperature for at least 6 h. Hereafter, the amount of analyte in the solutions should be evaluated by comparison of the detector responses to those of a freshly prepared solution. The short-term storage stability of an analyte in the specific matrix and container should be evaluated at the concentration of QC Low and QC High in triplicate, by incubation of the samples for at least 4  h at ambient temperature. The minimum time of incubation at ambient temperature should be defined based on the expected time the study samples will be stored at ambient temperature before they will be frozen or processed. Other incubation temperatures (i.e., 4 or 37°C) might be tested as well. Observed concentrations in the incubated samples ideally should be compared to observed concentrations in nonincubated samples, processed in the same analytical run. Depending on the number of replicate observations, a twotailed t-test might be applied to demonstrate stability (i.e., no significant difference), while also several other approaches are considered appropriate (e.g., <10% difference in observed concentration with nonincubated sample). The long-term storage stability of an analyte in the specific matrix, container and storage temperature should also be evaluated at the concentration of QC Low and

84

W.J. Loos et al.

QC High, and should cover the period during which study samples are intended to be stored during the conduct of a study. Long-term storage stability might be evaluated during the conduct of the study, providing that study samples are not stored for a longer period than the currently validated long-term storage stability. The most common approach for validation of the long-term storage stability is the analysis of the QC samples in triplicate at predefined time points (e.g., 1, 3, 6, and 12 months) on a freshly prepared calibration curve. General acceptance criteria for accuracy are used to demonstrate stability. A more accurate and precise method is by comparing the observed concentrations of analyte in stored samples to those in the same pools stored at −130°C or lower (i.e., liquid nitrogen). As analytes in all matrices are considered to be stable at −130°C or lower, even when unstable at higher temperatures, analytical and processing errors are not influencing the observations. The influence of repeated freezing and thawing of samples, which might occur during the conduct of the study, should also be tested at the concentration of QC Low and QC High. It is recommended to perform the test up to at least three freeze–thaw cycles, during which triplicate aliquots of each concentration will be completely thawed at ambient temperature, after which they will be refrozen for at least 12 h. This cycle should be repeated at least two times, after which the samples will be analyzed, during which the observed concentrations ideally should be compared to nonfrozen samples or freshly prepared samples in the same analytical run. The stability of analyte in processed samples should be tested over a time period covering the expected time to complete the analysis during the conduct of the study samples, while it is desirable to test a longer time period to cover potential delays during the analysis (e.g., instrument failure). Processed sample stability should be tested under exactly the same conditions as processed samples (e.g., light protected, identical temperature) and therefore can best be performed stored in the auto­ sampler of the analytical system. Ideally, the processed sample stability will be ­performed by reinjections of the same samples at different time intervals. In cases where reinjection is not possible, multiple samples of the same pool should be processed and injected at different time-intervals. The latter method is less accurate and precise, as sample processing errors might influence the stability test. In most validated analytical methods, precision and accuracy are described for pools of blank matrix spiked with the analyte of interest. Incurred samples (i.e., study samples), however, differ in many ways from the spiked pools of QC samples. Incurred samples, for example, in most cases contain also (metabolic) degradation products of the analyte, which might lead to irreproducible results in case the metabolite reverts to the parent molecule ex vivo. As during the validation the metabolite might not have been added to the matrix, this potential lack of reproducibility will not be recognized. During the conduct of a study, especially during preclinical toxicokinetic studies in several different species and during early clinical studies, incurred sample reanalysis should be included to demonstrate the reproducibility of the method for clinical samples. Differences between incurred sample reanalysis results might be systematic (i.e., biased higher or lower results in repeated analysis) or might be random (i.e., results do not agree; however, there is no systematic bias). The number of samples subject to reanalysis depends on the precision of the method and whether

3  Bioanalytical Methods in Clinical Drug Development

85

or not there is a random or systematic difference. In the latter case, fewer samples are required. Samples for reanalysis should be selected in the range of observed concentration, near the peak concentration and during the elimination phase, during which it can be assumed metabolites will be present at higher levels. Besides the presence of metabolites, the performance of the analytical methodology might be different in incurred samples due to differences in protein binding, issues related to the recovery of an analyte during sample preparation, and/or matrix effects during the ionization process in mass-spectrometric-based methods [11].

3.4.7 Matrix Effect in LC–MS/MS Based Methods Liquid chromatography–tandem mass spectrometry (LC–MS/MS) is considered to be very selective; however, lack of selectivity might be observed as a result of ion suppression or enhancement by coeluting matrix components and/or metabolites, which in turn affects accuracy and precision of the method. Especially in case an unlabeled internal standard is used in the method, evaluation of the matrix effect is essential, as the potentially altered response of the analyte is not directly corrected by the response of the internal standard. In case an unlabeled internal standard is used, it is recommended to assess the accuracy of the method in ten different lots of the same matrix, spiked with the analyte at the same concentration. The absolute matrix effect (3.7) can be evaluated by comparing the absolute responses after injection of the analyte in neat solution with the response of analyte spiked to blank matrix extracts. The absolute matrix effect should be determined in methods with a labeled internal standard as well as in methods with nonlabeled internal standards [3]. Matrix effect = (response in matrix extract) / (response in neat solution) × 100% (3.7)

3.4.8 Recovery In most cases, biological matrices cannot be injected into the analytical system without being extracted. Recovery is a measurement of the extraction efficiency of an analytical method. This parameter (3.8) should be determined at different concentrations over the entire range of the method and will be determined comparing the absolute responses after injection of the analyte spiked to and extracted from the matrix and the response of the analyte spiked in a neat solution or blank extract. Recovery = (response spiked analyte) / (response extracted analyte) × 100%

(3.8) Although a near 100% recovery is desirable, it is not required. However, the recovery should be consistent, precise, and reproducible. In addition, the recovery of the internal standard should be assessed independently at the concentration as used in the analytical method.

86

W.J. Loos et al.

References 1. Balcerzak M (1997) Analytical methods for the determination of platinum in biological and environmental materials. A review. Analyst 122:67R–74R. 2. Brouwers EE, Tibben M, Rosing H, Schellens JH and Beijnen JH (2008) The application of inductively coupled plasma mass spectrometry in clinical pharmacological oncology research. Mass Spectrom Rev 27:67–100. 3. Chambers E, Wagrowski-Diehl DM, Lu Z and Mazzeo JR (2007) Systematic and comprehensive strategy for reducing matrix effects in LC/MS/MS analyses. J Chromatogr B Analyt Technol Biomed Life Sci 852:22–34. 4. Flanagan RJ, Morgan PE, Spencer EP and Whelpton R (2006) Micro-extraction techniques in analytical toxicology: short review. Biomed Chromatogr 20:530–538. 5. Hocht C, Opezzo JA, Bramuglia GF and Taira CA (2006) Application of microdialysis in clinical pharmacology. Curr Clin Pharmacol 1:163–183. 6. Hyotylainen T and Riekkola ML (2005) Solid-phase extraction or liquid chromatography coupled on-line with gas chromatography in the analysis of biological samples. J Chromatogr B Analyt Technol Biomed Life Sci 817:13–21. 7. Kazakevich Y and Lobrutto R (2007) HPLC for Pharmaceutical Scientists. John Wiley & Sons, Inc., Hoboken, New Jersey. 8. Kitzen JJ, Verweij J, Wiemer EA and Loos WJ (2006) The relevance of microdialysis for clinical oncology. Curr Clin Pharmacol 1:255–263. 9. Konings IR, Engels FK, Sleijfer S, Verweij J, Wiemer EA and Loos WJ (2009) Application of prolonged microdialysis sampling in carboplatin-treated cancer patients. Cancer Chemother Pharmacol 64:509–516. 10. Oravcova J, Bohs B and Lindner W (1996) Drug-protein binding sites. New trends in analytical and experimental methodology. J Chromatogr B Biomed Appl 677:1–28. 11. Rocci ML, Jr., Devanarayan V, Haughey DB and Jardieu P (2007) Confirmatory reanalysis of incurred bioanalytical samples. AAPS J 9:E336–E343. 12. Rosing H, Man WY, Doyle E, Bult A and Beijenen JH (2000) Bioanalytical liquid chromatographic method validation. A review of current practices and procedures. J Liq Chromatogr Relat Technol 23:329–354. 13. Shah VP, Midha KK, Findlay JW, Hill HM, Hulse JD, McGilveray IJ, McKay G, Miller KJ, Patnaik RN, Powell ML, Tonelli A, Viswanathan CT and Yacobi A (2000) Bioanalytical method validation – a revisit with a decade of progress. Pharm Res 17:1551–1557. 14. Sparreboom A, Nooter K, Loos WJ and Verweij J (2001) The (ir)relevance of plasma protein binding of anticancer drugs. Neth J Med 59:196–207. 15. Venn RF, Merson J, Cole S and Macrae P (2005) 96-Well solid-phase extraction: a brief history of its development. J Chromatogr B Analyt Technol Biomed Life Sci 817:77–80. 16. Viswanathan CT, Bansal S, Booth B, DeStefano AJ, Rose MJ, Sailstad J, Shah VP, Skelly JP, Swann PG and Weiner R (2007) Quantitative bioanalytical methods validation and implementation: best practices for chromatographic and ligand binding assays. Pharm Res 24:1962–1973. 17. Vogeser M and Seger C (2008) A decade of HPLC–MS/MS in the routine clinical laboratory – goals for further developments. Clin Biochem 41:649–662. 18. Wille SM and Lambert WE (2007) Recent developments in extraction procedures relevant to analytical toxicology. Anal Bioanal Chem 388:1381–1391. 19. Xu RN, Fan L, Rieser MJ and El-Shourbagy TA (2007) Recent advances in high-throughput quantitative bioanalysis by LC–MS/MS. J Pharm Biomed Anal 44:342–355.

Part II

Chapter 4

Preclinical Models for Anticancer Drug Development Edward A. Sausville

4.1 Introduction New molecules under consideration as novel cancer therapeutics face a number of challenges in their development path. Most prominent among these is that in contrast to therapeutic areas such as infectious diseases or inflammation, where a variety of model systems are predictive of clinical success assuming that pharmaceutical features of the candidate molecules can be properly designed or modified, such models are lacking in oncology. Thus, an overriding goal of a cancer drug development pathway should be to “fail fast” structures that have low likelihood of success, and “advance smartly” structures that will achieve the desired goal, which is ultimately the evolution of a successful registration strategy. This chapter will touch on the various types of yardsticks that can be applied throughout a drug’s development path to achieve these goals. Anticancer agents may be selected for advancement to clinical trial on the basis of empirically observed antiproliferative activity (rarely in modern times without at least some information on mechanism of action), or as the result of conscientious selection of a molecule which “targets” a process important for cancer cell proliferation. As it is exceptionally unusual for a molecule’s development path to be completely straightforward, the discovery and development process usefully is a mix of rationally designed strategies to enhance the likelihood of a successful outcome with planned openness for serendipitous observations to allow major useful refinements. Figure 4.1 illustrates key steps in the process of discovery and early development for a “targeted” agent, while Fig. 4.2 outlines usual issues for empirically recognized antiproliferative molecules.

E.A. Sausville (*) Marlene and Stewart Greenebaum Cancer Center, University of Maryland, 22 S. Greene Street, Room S9D07, Baltimore, MD 21201, USA e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_4, © Springer Science+Business Media, LLC 2011

89

90

E.A. Sausville

Compound library

"Counterscreen" vs.

"Targeted Screen"

Irrelevant target OR Non-target bearing cells

Biochemical / Biophysical

or

Cell-based

Hits

"On target"Activity in cell culture: Time / concentration for desired effect in vitro

"Lead Structure(s)" CHEMISTRY "Optimized Lead Structure(s) with on target in vitro activity

Pharmacology in test species / strain for in vivo efficacy

Desired Exposure time / concentration feasible? NO Target Dependent In vivo tumor efficacy model: activity? Advance to Toxicology / Pharmaceutic Development Fig. 4.1  Task flow for targeted cancer drug discovery campaign. Solid arrows define tasks that progress to the definition of a development candidate. Dashed arrows define tasks that reflect failure to meet criteria to progress leads. A compound library provides candidate structures to enter into either a biochemical or cell-based screen for activity that is dependent on the desired target’s function. Counter screens assure elements of selectivity and confirm nonredundancy against related but nondesired target(s), or serve as “controls” with the presence of an entity, e.g., a tumor suppressor gene, in whose absence the drug candidate would be active. “Hit” compounds are further refined by chemistry into lead structures. Leads if defined by a physical biochemical assay will have at least submicromolar potency against the target in a cellular context, with the ability to cross membranes. “Hit: compounds defined by a cell-based assay will likewise have at least analogous potency and will be demonstrated to directly affect the purified target or a process critical to target function. A critical qualifying property for further study is the evolution of a structure with a concentration × time exposure in vitro for the desired endpoint (e.g., cell death), which in pilot pharmacology studies in a test species is achievable. This allows use of a target-dependent animal model to qualify the molecule for advanced toxicology evaluation

4.1.1 Molecular and Chemical Descriptors of Successful Drugs Useful cancer therapeutics are of two types. One type comprises agents that have the properties of transient presence in the circulation, localization to tumor sites, modification of a cancer-relevant target, and facile systemic elimination to minimize damage to host tissues while effecting tumor cell death, which are all features that characterize the useful classical cytotoxic chemotherapeutic agents, including the nitrogen mustard derivatives, antimetabolites, platins, and antitumor antibiotics.

4  Preclinical Models for Anticancer Drug Development

91

Compound Library Empirical antiproliferative effect In vitro or in vivo

IN VITRO? Deconvolute mechanism or correlate with drug action in vitro: Novel? Better than Standard?

Determine time/concentration Necessary for drug effect

Determine Maximum Tolerated Dose (MTD) on schedule mimicking in vitro activity exposure

IN VIVO? Role of host metabolism to generate active drug?

Activity (T/C < 40%) in >33% of models tested in syngeneic or allogeneic xenografts or analogous criterion

Consider for Advanced Toxicology and Formulation Development

Fig. 4.2  Task flow for empirical drug discovery campaign. Following definition of an antitumor effect, either in vitro or in vivo, a key action is to define the mechanism of drug action so that a realistic assessment of its novelty can be understood. If empirical activity is observed in an animal model in  vivo, clarification as to whether the agent of interest acts directly on tumor cells or requires metabolism is an allied activity. Following detailed analysis of activity in a series of in vivo models, compounds with activity in at least 33% of models tested might be a candidate for advanced development, recognizing that only ~50% of those empirically selected molecules will ultimately have preliminary evidence of useful antineoplastic effect [60]

These molecules have in many cases reactive features, e.g., affinity of basic moieties or thiols for platinum compounds, or which lead to the generation of reactive intermediates such as carbonium ions (e.g., mustards including cyclophosphamide or melphalan). Microtubule-directed classical chemotherapeutic agents, while not chemically reactive in an analogous sense, have in their successful examples the intrinsic capacity to form protracted retention bound to their targets in tumor tissues (e.g., taxanes) [1], while clearing from the systemic circulation. In contrast a number of useful cancer therapeutics have the capacity to modify target or receptor function in an ongoing way and have value predicated on continuous drug exposure. Classically, this behavior is represented by the hormone receptor-directed agents, and more recently this feature is prominently illustrated by monoclonal antibodies and many of the oncogene product directed kinase antagonists such as gefitinib, erlotinib, imatinib, dasatinib, and sorafenib. A key strategic question therefore is to ascertain early in the discovery process whether the nature of the intended target calls for continuous or intermittent modulation

92

E.A. Sausville

in relation to the target’s presence or function in normal tissues. For example, while the oncogene-directed protein kinase inhibitors described above do seem to allow chronic dosing (e.g., for erlotinib) [2], likely owing to the absence of prominent or persistent corresponding oncogenic kinase action in normal cellular function, cell cycle regulatory kinase-directed agents have not historically been able to be dosed on continuous administration schedules (e.g., cyclin-dependent kinase inhibitors) [3]. Additional questions are whether there are features of target structure that might allow for the design or screening of tight binding or covalently bound modulators [4] or whether endogenous clearance mechanisms for the target allow strategies to modify function by enhancing target degradation (e.g., faslodex) [5] or by decreasing transcription of the target’s mRNA in a specific way or its translation (e.g., rapamycin analogs and “cap-dependent” mRNAs) [6]. Three general types of molecules are the starting points for cancer drug discovery campaigns. Completely synthetic structures are typically designed around a scaffold structure that is potentially proprietary either directly or after pharmaceutically directed chemical modification and which can be allied to a structural feature of the drug’s intended target. Useful properties of such synthetic structures to be included in design of structures frequently use the “rule of five,” an awareness tool for discovery chemists [7]. Compounds with two or more of the following characteristics are flagged as likely to have poor oral absorption: more than 5 H-bond donors; molecular weight >500; an oil/water distribution ratio (c log P) > 5; and sum of Ns and Os (a rough measure of H-bond acceptors) > 10. It is interesting that the classical chemotherapeutic agents that are actually quite useful do not actually conform to the “rule of five,” reflecting their reactive and therefore for the most part (although not always) necessary parenteral administration. A second origin for currently utilized cancer therapeutics is from “natural products” [8]. These usually connote extracts of plant (terrestrial or marine), microbial (bacterial or fungal), or marine (usually invertebrates) origin. Historically, a variety of useful anticancer drugs have emerged from these sources, including as diverse agents as cytosine arabinoside, the vincas, taxanes, anthracyclines, and camptothecins. Generally, the antiproliferative actions of these agents are a basis for host defense by the originating organism directed against competitors for a shared ecological niche. For example, taxanes in the bark of yew trees diminish susceptibility to invasive fungi. It should be noted that while the use of natural product extracts as actual anticancer-directed therapeutics has historical interest (e.g., in traditional Chinese medicines), most current use of such extracts in cancer drug discovery and development focuses on their serving as a source of drug discovery leads, from which pure compounds are isolated. Issues related to standardization of extracts with respect to potency and purity encourage the use of pure compounds derived from natural product extracts rather than the extracts themselves. While it is beyond the scope of this chapter to consider the intricacies inherent in natural product screening [9], points to consider in utilizing natural product extracts as a basis for drug discovery or development include the following. Is the identity of extract source (including location and date or season of collection) clear? Can repeated collections reliably generate extracts that can be benchmarked as similar until the

4  Preclinical Models for Anticancer Drug Development

93

actual basis for anticancer activity is identified? Is it clear whether a contaminating species (such a microbe or parasite) could be the real source of any observed biological or biochemical activity? Is there access to a bioassay or biochemical assay that can serve as basis for purification efforts? A third source of useful cancer drugs includes bioengineering, with the derivation of entities such as monoclonal antibodies, recombinant proteins, and peptides or peptide fragments, e.g., for use as vaccines. The evolution of monoclonal antibody technology from initial relatively crude utilization of wholly mousederived antibodies (with predictably limited clinical utility) through current availability of humanized antibodies has opened new vistas for design and implementation of new strategies particularly directed against cell surface receptors or secreted molecules [10, 11]. Factors affecting success in clinical application of antibodies, in addition to the derivation of humanized molecules with suitable anticipated production efficiencies; include affinity for the target (in this regard, solublized receptors for their cognate ligands may be advantageous); whether the antibody cross reacts with nontumor-related targets and the physiological consequences of this for the host; whether it fixes or does not fix complement or evoke other immunological effectors subsequent to antigen binding [12]; and whether the naked antibody is suitable for therapeutic use as opposed to the need for further conjugation with additional small or large molecules. Recombinant proteins such as chimeric fusion protein toxins may be limited by the degree to which they evoke a host immune response [13], so as humanized as possible molecules of this type are desirable to produce. In addition, while antibodies are generally perceived as readily adapted to folding and secretion in their producer cell types, recombinant proteins particularly those expressed in bacteria may experience varying folding and renaturation efficiencies affecting ultimately their clinical utility [14]. Whole bacterial, cellular or “gene therapy” constructs represent allied concerns (beyond the scope of the current chapter, but covered elsewhere in greater detail) [15] to those with bioengineered proteins, and in addition, there is the regulatory uncertainty stemming from the fact that no currently approved gene therapy can provide a reliable “road map” for development milestones. Potentially practical gene therapy vector systems must have a completely defined gene sequence, not contain extraneous DNA, and in their production not utilize certain selection markers, e.g., penicillin (in relation to allergy) or chloramphenicol (in relation to selection of new resistant strains) [16]. Systemic administration of foreign proteins must be considered a relative “negative” in developing certain of these constructs, although local administration into a tumor of a virus – propagated gene therapy may obviate this issue to a degree.

4.1.2 Empirical Versus Rational Discovery and Development Strategies Historically modern cancer chemotherapy extended in large part from the empirical observation of severe bone marrow suppression resulting from derivatives of

94

E.A. Sausville

mustard gas during and following World War I [17]. This suggested the potential value of analogous agents in the treatment of leukemias and lymphomas, which in turn led to the development of mass screening programs to assess the value of various chemicals and natural product extracts as possessing anticancer activity in animal and later cultured cell models. This “empirical” approach, i.e., defining drug candidates by the phenomenon of observable antiproliferative activity in experimental cancers or their derived cells, while inefficient, has produced many of the useful antineoplastic agents in current use ranging from the DNA-directed alkylating agents and the antitumor antibiotics such as anthracyclines and the microtubule-directed vincas, taxanes, and epothilones. A discussion of the implementation and management of the output of such empirical screening programs in whole animals with tumors, and in vitro in cell lines has been presented elsewhere [18, 19]. In contrast, “rational” drug development strategies embark on a discovery and development campaign in relation to altering molecular targets of known biological importance to the progression of the cancer. Such targets have been defined by defining the genetic abnormalities found in tumors (e.g., the p210bcr–abl or HER2neu oncoproteins in chronic myelogenous leukemia and breast cancer, respectively) [20, 21]; capitalizing on the normal role of a potential target in the originating tissue of a tumor (e.g., hormone receptors in breast and prostate cancer) [22, 23]; understanding the detailed biochemistry of cell replication (e.g., antimetabolites used for a wide variety of neoplasms); defining the binding partners of empirically derived antiproliferative agents, and then modifying molecules around defined structural requirements (e.g. mammalian target of rapamycin inhibitors) [24]; or modifying molecules to convey useful pharmacologic features in tumor as opposed to host tissues (e.g. capecitabine as a rationally constructed delivery system for enhanced 5-fluorouracil delivery to the tumor milieu) [25]. Predicting which targets will be most valuable to pursue is heavily influenced by whether one can define differential expression of a target in tumor as opposed to host tissues or “mutational activation” of a target in the tumor milieu. A truly differential expression of a target in tumor tissues really occurs if a protein derives from a translocation or mutation uniquely present in the tumor, or is present in a restricted, nonessential, at least transiently, host tissue. Mutational activation refers to a set of mechanisms including activating point mutations, deletions, insertions, use of certain single nucleotide polymorphisms that enhance transcription or stability of mRNA, or amplification of a gene leading to its product’s overexpression. These genetically originated changes result in the critical activation of pathways driving a tumor’s pleiotropic capacities for proliferation, avoidance of cell death mechanisms, limitless replication, and capacity to invade, provoke angiogenesis, and avoid immunologic defense mechanisms. The result is a state of “addiction” on the part of the tumor to the continued presence of the molecule in maintaining its existence, a concept first proposed with respect to the products of transforming oncogenes [26]. Proteins creating a state of addiction in tumors have consi­ derable potential value as targets for drug discovery, exemplified by the p210bcr–abl or HER2-neu cited above. Both of these targets demonstrate by mutational activation (translocation and amplification in many cases, respectively), the intense selection

4  Preclinical Models for Anticancer Drug Development

95

pressure in favor of their function in the tumor cells. The phenomenon of apparent oncogene addiction has given rise to the point of view that mutational analysis for individual tumors would provide evidence in the mutated target proteins a basis for drug discovery strategies and also allow “personalization” of therapy [27]. Recent examples of this type of strategy have yielded the observation that mutated and activated c-raf oncoproteins can be defined frequently in melanoma and colon cancers[28] and mutated and activated c-kit can be defined in certain melanomas [29], thus suggesting the potential value of inhibitors of those targets in those diseases. Once a suitable target has been defined from an assembly of biological, ­epidemiological, and clinical data, it can be considered “credentialed” as fit for a discovery campaign. The goal of the drug discovery process is to yield investigational agents that can then be entered into clinical trials with the hope of “validating­” the target by the emergence of a safe and effective therapeutic agent directed at the target from later stage clinical testing. A credentialed target can then be approached from a number of different directions to define suitable drug leads, as will be described below. Knowledge of the structure of the target from X-ray crystallographic or NMR spectroscopic or analogous techniques can add enormous value to this effort by suggesting candidate binding structures, which in turn can serve as the basis for scaffolds that can be “decorated” with various functional groups through “combinatorial” chemistry approaches, where many leads are generated often by parallel synthesis [30–32]. It is beyond the scope of this chapter to describe recent progress in chemical approaches to generating diverse molecules for use in screening­approaches, but a ready supply of candidate structures must be assured to fuel the screening strategies to be described below.

4.2 Methods for Large Volume Screening 4.2.1 Historical Perspective As recounted by Schepartz [33], empirical observations of toxicity of nitrogen mustard derivatives for the bone marrow compartment and later of Hitchings and Elion [34] that analogs of normal nucleic acid precursors could also effect marrow toxicity in the host and influence the course of model hematopoietic tumors resulted in the appearance of cancer drug screening programs at a number of institutions, e.g., use of the mouse sarcoma SA-180 at Memorial Sloan Kettering. In 1955, the National Cancer Institute, responding to a congressional directive, initiated a screening program through its Cancer Chemotherapy National Service Center. Compounds were collected or synthesized endogenously, and numbered according to the date of accession by the familiar “NSC” designation maintained to the present [35]. The initial screen utilized three tumors: L1210 leukemia, SA-180, and mouse mammary adenocarcinoma 755. Shortly thereafter, modifications were made so that

96

E.A. Sausville

Table 4.1  In vivo models for cancer drug development: pros and cons Model Advantages Disadvantages Survival models  Intraperitoneal Can perform in vivo Not measure tumor; therefore inference on drug action indirect  Biologic (e.g. engineered Can precisely define genetic Variable penetrance of Animal) origin of tumor phenotype; long latency; require much drug Tumor mass models  Endogenous; chemically Mimic human occurrence Variable incidence or latency induced  Syngeneic mouse tumor Intact host immune system and Not assess human cells stroma Potential “High Throughput” Multicompartment evaluation possible  Xenografts Can assess human tumors Immunocompromised entail artificial stroma Vary initial treatment characteristics (early/advanced) systematically  Orthotopic unless marked Mimic natural organ of origin Hard to score tumor growth (e.g. luminescent cells)

L1210 plus P388 leukemias plusB16 melanoma or Lewis lung. Both L1210 and P388 were chemically induced and passed in syngeneic mice. Screening by various routes of tumor inoculation (intraperitoneal, intravenous, or intracompartmental, e.g., intracranial) allowed the capacity of a test compound to cause increased life span (ILS) (Table 4.1). The predictive value of activity for solid tumors after activity in the mouse leukemia screens is limited. For example, Waud [18] reports that 1–2% of agents active in the murine leukemias were active against solid tumor murine in vivo models. Additional developments, recounted by Corbett et al. [36], have encouraged an end to primary in  vivo screening programs. These include the desire to limit needless use of ­animals, as only ~0.1% of compounds tested were active; prescreens in in vitro or cell culture systems allow increase of activity to ~5% of random materials and >25% of analogs; and the amount of material needed for such prescreens is considerably lower than the amount necessary for in vivo primary screening. These considerations prompted numerous organizations to move away from primary screening in vivo. In the case of the National Cancer Institute, successive generations of screens used an in  vivo leukemia prescreen, followed by detailed testing in solid tumor in vivo models, and a pilot program to evaluate explants of human tumors in a variant of the human tumor stem cell clonogenic assay. Since 1985 in a pilot mode, and since 1990, the National Cancer Institute has utilized a primary in  vitro cell line screen, with the use of in  vivo models only for more detailed testing [37].

4  Preclinical Models for Anticancer Drug Development

97

4.2.2 Ancillary Needs in Developing a Screening Program In addition to a source of compounds of suitable chemical diversity, modern empirical screening programs require “real time” information handling support to continuously allow input of data and ideally allow continuing refinement and curation of the information. As exemplified by the COMPARE algorithm (see Fig. 4.3) of the National Cancer Institute and its successors, including neural networks and “cluster analysis” [38–41], an appropriately constructed matrix of information handling algorithms can turn into a discovery engine itself, when the pattern of screening activity for a compound is aligned against, for example the expression of molecular targets in the test cell lines [42]. In a “rational” drug discovery

Percentage Growth

a

• Goal: COMPARE degree of similarity of a new compound to standard agents • Calculate mean GI50, TGI or LC50 • Display behavior of particular cell line as deflection from mean

50

0 −50

resistant

−100 −9

c

b

All Cell Lines

100

−8 −7 −6 −5 Log10 of Sample Concentration (Molar)

−4

mean

sensitive

• Calculate Pearson correlation coefficient: 1 = identity ; 0 = no correlation

d

Leukemia NSCLC Small Cell Lung Colon

Control

Jasplakinolide 120nM

Cucurbitacin E 50nM

CNS Melanoma Ovarian Renal Taxol

Halichondrin B Daunorubicin

Tubulin

Topoisomerase II

Fig. 4.3  The COMPARE algorithm for information mining from the NCI 60 cell drug screen. The method described in ref [38] is illustrated. Patterns of drug activity in vitro as concentration effect curves for growth inhibition (a) can be transformed (b) into a numeric representation by defining the particular behavior of a cell line with respect to a mean effect, such as 50% growth inhibition (GI50) or cell killing (LC50), and which can then be defined as numerically similar or different to other molecules by, e.g., a Pearson correlation coefficient. (c) Molecules with related mechanism of action, in this case tubulin interacting agents, have a distinct pattern of inhibition from the anthracylcine daunorubicin. This information allows the definition of agents with novel mechanism of action such as cucurbitacins and jasplakinolide. These natural products have a pattern of action distinct from the classical chemotherapy agents, yet have a related pattern of action (Pearson coefficient = 0.59) and both as shown in (d) act by disruption of the actin cytoskeleton, shown by disruption of rhodamine–phalloidin binding in cucurbitacin- or jasplakinolide-treated cells

98

E.A. Sausville

program directed against particular targets, correlation of target structure with features of compound activity can allow the design of subsequent iterations of a lead structure for testing. This may center on retention of a core scaffold for target directed activity while enhancing functionalities imparting pharmacological or pharmaceutic properties, e.g., a recent series of cdc25-directed inhibitors affecting cell cycle progression [43]. All screens need continuous “quality control” to assure validity that the screening behavior is constant from generation to generation of compounds. This can be accomplished by containing in each screening run a constant set of comparator compounds (both “positive” and “somewhat positive”) against whose behavior the screen can be continuously judged. In the case of cell-based screens, the passage number in comparison to a master cell bank, the serum lots used, the nutritional state of the cells including degree of confluence are all variables that must be carefully controlled to assure reproducibility. A quantitative means of assessing assay reproducibility is known as the Z¢ statistic, whose use has recently been considered in detail [44] and should be developed for each screen.

4.2.3 Types of “Large Volume” Screens “In silico” screening refers to various ways of querying databases of various sorts to discern potential lead structures directed against a target. “Computer Assisted Drug Design” generally starts with a collection of molecular descriptors of a collection of small molecules, reduced to bond lengths or other attributes, such as varying combinations of dipole charge moments. These may then be aligned with structural features of intended targets, to discern structures with anticipated favorable binding properties (exemplified in ref. [45]). Various “docking” algorithms [46, 47] can then refine leads obtained in this way to improve anticipated binding or interaction parameters with the target [48]. Ultimately, the results of this exercise are “optimized” (in a computational sense) lead structures, which then must be synthesized and utilized in actual assays that interact with the proposed target. “In vitro pure chemical screens” utilize collections of test small molecule structures in a biochemical (e.g., ATPase assay) [49] or biophysical perturbation (fluorescence resonance energy transfer) [50] dependent on enzymatic activity (e.g., protease activity acting on a substrate). Numerous strategies can be devised depending on the biochemical properties of the process to be modified. The major questions in designing such screens center on the availability of reagents (purified proteins can be difficult to produce in quantity and vary in activity from batch to batch); how faithful the biochemical activity in the screen is outside its biological context; and how to select test substances for use in the screen that will be useful in elucidating practical leads for further chemical modification. Traditional medicinal chemical approaches utilize structural features of substrates, products, or anticipated intermediary states. Where one does not have antecedent clues as to potential active small molecule structures, e.g., a small molecule that would

4  Preclinical Models for Anticancer Drug Development

99

abrogate protein–protein interactions, utilizing structural information is essential to focus the lead search. One approach that marries structural and biophysical data is the “SAR by NMR” approach which successfully utilized NMR to define small molecules that could bind with relatively low affinity to purified target proteins of the bcl2 family of antiapoptotic proteins, and then produced more complex structures that multiplied the apparent affinity to yield inhibitors with nanomolar potency [51]. The advantages of such molecularly defined biochemical or biophysical screens are the emergence of structures that have a clearly defined target around which optimization strategies can continuously be applied. The disadvantage is that unless the collection of small molecule scaffolds has been preselected to be compatible with physiologic barriers such as the traverse of cellular membranes, plasma protein binding, susceptibility to metabolism in the circulation or by hepatic disposition mechanisms, significant work must be directed to achieve those latter essential attributes. In addition, if the target in such pure biochemical screens has homology with other classes of target, e.g., kinases, phosphatases, proteases, etc., simultaneously applied “counter screens” must be in place to build enthusiasm that the positive leads emerging from the main screen has the requisite selectivity for target function. Ideally the “product” of a pure in vitro screening campaign is a family of closely related in terms of potency and selectivity small molecules that have likely pharmaceutical tractability that can then be entered into biological studies in tumor bearing and tumor-free animals. A distinct strategy to pure in vitro biochemical testing is the use of whole cells as the initial screening vehicle. Typically, cells grown in mass culture are exposed to the compounds in the screening set and at various times after drug exposure some measure of viability (e.g., the capacity of viable cell mitochondria to reduce tetrazolium dyes to various colored substances, protein mass, ATP content, etc.) in a relatively high throughput, plate-based methodology using colorimetric, luminescent, or fluorescent reporter systems. The “NCI 60” [37] cell line panel is prototypic of such screening efforts. The screen utilizes collections of cancer cell lines of defined passage number in relation to a master cell bank and catalogs susceptibility to test compounds in terms of cell kill and varying degrees of growth inhibition. Utilizing various computational algorithms [38, 41], the resulting patterns of activity can define rapidly chemotypes that are both directed at familiar mechanisms (e.g., microtubule, topoisomerase inhibition) and novel mechanisms (proteosome inhibitor, cell cycle regulatory kinase antagonist, etc.). The development and utilization of this screening system is described extensively elsewhere and is available through http://www.dtp.nci.nih.gov. The NCI 60 is exemplary of eukaryotic, nonmodified cell screen where the output (decreased cell proliferation) relates to compound antiproliferative actions rather than the action on any particular molecular target. Cell-based screens can be created with various modifications to create particular contexts in which screen leads would be active against particular targets. For example, in a most straightforward application, isogenic cell types for the over or under expression of a particular target can be created through transfection technologies (e.g. to define molecules

100

E.A. Sausville

affecting hypoxia-inducible factor) [52]. Screened small molecules with differential behavior in such screens can be allied then to some effect on the over- or underexpressed gene. An alternative strategy increasingly adopted includes the generation of specifically modified cells through expressed siRNA arising from various vector systems, with appropriate controls, to produce a pattern of activity that reflects the effects of down regulating a particular target [53]. Compounds that act as “synthetic lethals” with the diminished gene products can then be defined. The term derives from yeast genetics, where mutations can be defined which connote lethality only in the context of another coexisting mutation. Kaelin and colleagues have postulated that cells with specifically deleted target genes can be instrumental in defining molecules that would be useful in the context of mutated tumor suppressor genes [54]. The problem with all cell-based screens is that ultimately the screening leads must be “deconvoluted” to assure that the intended mechanism or target actually still ties to the actions of the positive molecules in the screen. For example, an over expressed kinase could detect not only a specific kinase antagonist, but also molecules such as heat shock protein antagonists [55] known to affect the stability and function of oncogenic kinases. Alternatively, if pathways downstream of a kinase target are mutationally activated in the recipient cell line of the target kinase, leads may be missed because the cell line is independent of the kinase action. Thus cell lines chosen for use in screening exercise that modify endogenous targets ideally would be well as well characterized as possible for up- and downstream modulators of target function. Non-eukaryotic cell models have been designed for use in cancer drug screening efforts. Yeast has figured prominently in efforts to define modulators of DNA damage checkpoint responses, given the extensive annotation of yeast cell cycle checkpoint regulatory molecules [56, 57]. While the information has in some cases served to illuminate novel chemotypes affecting these pathways, translation to human use has not been straightforward, in part because the biology of those targets in human cancer is incompletely defined. Likewise, zebrafish and nematodes have been proposed as a screening model to detect either process (angiogenesis) or pathway (DNA damage response and cancer susceptibility) modulators [58, 59], but in each of these cases, similar issues of facile deconvolution of the screened leads to particular targets and translation to higher organisms has not been easy to define, and the use of such screens must be considered exploratory.

4.2.4 Managing “Positive” and “Negative” Screening Results At the end of a cell-based initial screening effort, the product will be a collection of molecules meeting preset benchmarks of potency and selectivity of action for different cellular contexts. In addition to the generic issues discussed above of the need to deconvolute the compounds’ mechanism(s) of action, additional studies can help define initial strategies for in vivo evaluation. How potent must such molecules be to warrant further study? In the NCI 60 experience, compounds with IC50s

4  Preclinical Models for Anticancer Drug Development

101

<1 mM had a higher probability of activity in subsequent in vivo models than those that did not, stepping away from the class of alkylating agents as a group [60]. In vitro “area under the concentration × time” curve leading to cell growth inhibition to various degrees can be important in distinguishing compounds that require relatively brief exposures vs. those requiring protracted exposure. If protracted exposure is required, suitability of the molecule for oral formulation must be considered, as well as whether protracted inhibition of the target is compatible with anticipated large organism toxicology features. In addition, such “in  vitro” AUC definition provides an initial basis for interpreting initial pharmacological features of the molecules: compounds with rapid clearance in the organism but requiring protracted exposure [61] for efficacy can be put aside. Clonogenic assays refer to the ability of tumor cells to form colonies from single initiating cells either on plastic or in various semisolid media. Originally defined as a basis for describing drug susceptibility of tumor explanted cells [62], renewed interest in clonogenic assays has appeared in relation to emerging biological descriptors of tumor “stem cells” [63]. While more laborious than “bulk” proliferation assays, clonogenic assays do allow a modeling of pharmacologically relevant durations of exposure to the agent and estimation of anticipated tumor killing potential at times remote from the presence of the compound. Indeed, a body of evidence suggests that such tumor-derived clonogenic assays cannot only correctly pick agents of value to particular patients, but also inform decision making about progressing to detailed xenograft testing [64]. Clonogenic assays might therefore usefully complement as a “gold standard” the information from the in  vitro area under the concentration × time curve methodology described above and provide a counterpoint to ATP based and colorimetric or fluorescence survival laboratory “kits”, which while adaptable to high throughput as a survival assay, ultimately measure some aspect of cell metabolism (ATP itself, mitochondrial activity on tetrazolium-type compounds, protein mass, etc.) and therefore are at best derivative indicators of effects of a drug on tumor cell viability. An additional question is how important should evidence of antiangiogenic mechanisms be in prioritizing a compound’s development. While beyond the scope of this chapter, it is well established that many conventional anticancer agents have antiangiogenic properties [65], especially when long exposure times of endothelial cells are used. While various in  vitro assays of antiangiogenic potential can be defined, including human umbilical vein endothelial cell (HUVEC) proliferation, tube formation in matrigel, growth from explanted at aortic rings, etc. bona fide antiangiogenic activity requires a living host in which neovasculature can be defined [66–68]. A caveat is that assessment in a local in vivo model, such as the cornea micropocket assay, can overstate the antiangiogenic potential, while the use of matrigel plugs impregnated with endothelial growth factors requires a molecule suitable for in vivo administration. So unless the screening campaign has as its focus a primary endothelial cell target, early study of in vitro defined screening leads would be somewhat premature. Finally, evidence for “synergy” of novel agents with established chemotherapeutic or molecularly targeted agents can be sought. It is beyond the scope of this

102

E.A. Sausville

monograph to review the issues in testing for synergy, other than to note that in vitro methodologies do allow for the application of a variety of synergy algorithms [69–71], and it is of importance to consider the concentrations studied in relation to the anticipated concentrations of use of the agent. One should examine all portions of the concentration effect curves of the test agents, with the conventional agent studied only at pharmacologically relevant concentrations, and with a sequence of exposure that would adequately assay for potential antagonism (test agent administered before, after, or simultaneously with the conventional agent), as this information would clearly influence how to pursue combination strategies in the clinic.

4.3 Methods for In Vivo Evaluation 4.3.1 Overview of In Vivo Testing Goals The term “in vivo evaluation” as used here will refer to the assay for evidence of antitumor activity, or some attribute of antitumor activity, e.g., antiangiogenic or anti-invasive properties, in an intact animal, most commonly the mouse. The information received from in vivo studies depends on the question asked: not all models are appropriate for all questions. Drugs need different types of models at different times in their discovery and development life cycle: “Pharmacology” models to qualify a compound for further study do not require tumored animals, but should be of the same strain as anticipated for ultimate efficacy studies. “Efficacy” models to define biologic effect can be of two types: target enriched, where some aspect of the drug’s intended target is preselected to be present or absent, in the model, therefore leading to the inference that some aspect of the drug’s action is related to the function of the target in the model. Alternatively, the selected models may not have reference to a particular target (target unselected), which may be more relevant to the drug’s eventual use. “Biological” models can be derived, which utilize genetic approaches to mirror aspects of the human disease, and thus allow definition of the drug’s capacity to modulate target(s) relevant to disease incidence and progression. As will be described below, their value may be greatest after a clinical candidate has been selected and not in early screening stages. There are a number of issues to consider in choosing the animal model to be used in a compound’s development. From a tumor perspective, the “take rate” of the model influences the number of animals that must be implanted with tumor [72] and that factor plus the variability of tumor size as a function of time influences the size of the treatment and control groups. It is beyond the scope of this monograph to discuss the statistical features in developing treatment groups [73], but these are ideally based on the actual experience of the laboratory performing the studies with the model in hand. Variation in model performance in a particular laboratory in comparison to published experience with the model is common. The doubling time

4  Preclinical Models for Anticancer Drug Development

103

of tumor models, reflecting the proliferative state of the tumor cells, varies also [74], and this influences not only the duration of the experiment, but also the relation to the human clinical circumstance. Rapidly proliferating tumor models may overstate the potential relevance of outcomes to the human clinical circumstance, where except in a few tumor types, doubling times in the 30–90 day range are common. Early in compound selection and qualification, models that allow assay of a number of candidate molecules in a reliable and predictable fashion call for models amenable to comparison of perhaps up to a dozen analogs. Later in the qualification process, extension to a range of schedules in a variety of models may be of value, as will be discussed below. From a “drug” perspective, one cannot emphasize enough the need to have a pharmaceutically tractable form of the agent, not necessarily the final dose formulation for human use, but one where drug is compatible with administration, e.g., does not precipitate out in plasma, or in a body compartment (e.g., peritoneum of treated animals). Taken to this position’s logical conclusion, some discovery teams insist on known preliminary murine pharmacology prior to embarking on in vivo efficacy studies, assuring that the drug candidate will have an plasma area under the concentration × time curve in the experimental test species at least similar to what is necessary from in  vitro studies described above. This is particularly a tenable position if an analogous chemotype has not been studied previously. An additional part of experiment planning at this stage is the relation of the intended doses for study to the maximal tolerated dose (MTD) of the agent, conventionally one dose level below that which causes a defined proportion of the animals to develop ­unacceptable toxicity, minimally described as loss of 10–20% of body weight ­during the course of the experiment. Cytotoxic agents are conventionally administered even in the human close to their maximum tolerated doses, while “targeted” agents, as exemplified by, for example, hormone receptor antagonists are used at orders of magnitude of dose below their MTD, but where their respective receptors are ­saturated. The intended “style” of use in humans should shape the dose selection­ in animals. Algorithms for the definition of MTD have been described by others [74] and will not repeat here. From an overall strategic perspective, a primary value of demonstrating activity of a drug candidate in an intact animal model remains the ability to display the potential of efficacy across physiological barriers, e.g., IV or IP administration vs. subcutaneously located tumor. There is really little value of intratumoral administration, and IP administered drug activity against IP growing tumor is likewise at best a crude indicator of potential of a drug candidate’s potential.

4.3.2 Types of Mouse Models for Cancer Drug Evaluation A more detailed discussion of methods of evaluating tumor models in animals has been described by Corbet et al. [36], Alley et al. [74], and others [75–77], but a brief outline of methodological consideration would include the following points.

104

E.A. Sausville

Nonsubcutaneous, but compartmental models have a tumor confined initially to a particular body location, e.g., peritoneum or liver, but do not have a basis for assessing continuous change of tumor size. Systemic models may utilize tail vein or cardiac injection of tumor cells. In either of these cases, evidence of drug effect is assayed by ILS increased life spem:

é (T - C )ù % ILS = ê ú ´ 100, ë C û where T and C are the median survival times of animals in treated (T) or control (C) groups times until death, or some other relevant endpoint, e.g., hind limb paralysis or activity level of the animals (see Fig. 4.4 for examples). ILS is used in any intraperitoneal model or other nonmeasurable models such as intracranial glioblastoma or intravenous lymphoma or leukemia models. This parameter may also be relevant

Non - Measureable Tumor

Fraction Alive

Good activity

Marginally active control

Time

Advanced Stage Model

Early Stage Model

Control Vehicle control

Control

Active compound T/C <40%

Time

T/.C~40%

Tumor Volume

Tumor Volume

Vehicle control

Cytostatic Regression with Later regrowth Time

Fig.  4.4  Styles of in  vivo evaluation. Times of treatment can vary from once at the start of treatment to successive treatments throughout the period of observation. The upper panel shows a tumor model that is nonmeasurable and therefore evaluated in terms of time to an endpoint, such as survival, preset degree of weight loss, hindlimb paralysis, etc. The lower left panel describes an “early stage” tumor model, usable only in tumors with a high “Take rate” [74], which in comparison to a vehicle alone treatment group shows significant tumor growth suppression. The lower right panel shows an advanced stage model, where tumor of a definable weight either shows modest effect (T/C ~40%), evidence of tumor stasis, or an actual regression. Advanced stage models allow more clear delineation of the dose–response of antitumor effect in comparison to early stage models

4  Preclinical Models for Anticancer Drug Development

105

to tumors arising in transgenic animals, where the latency of tumor development may be uncertain in a particular animal. Modern technologies such as luminescence techniques to image tumor cells in such inaccessible compartments have provided an additional strategy to follow tumor growth in the case of compartmental growth in a not easily accessed compartment [78–80]. In contrast, subcutaneous placement of tumors either as syngeneic (tumor cells derived from the same species and strain in which the cells are implanted) tumors or xenografts (tumor cells arise from a different strain or species than the test strain, usually immunocompromised to facilitate tumor growth such as athymic “nude” or severe combined immunodeficient “SCID” mice) allows the potential for uniform commencement of therapy when tumors are at defined initial size, and effects of longitudinal growth followed throughout the course of the experiment. In that event, evidence of drug efficacy can be apparent through a variety of measures:

or

æ T ö æ DTumor weight (median )of treatedmice ö %ç ÷ = ç ´ 100 è C ø è DTumor weight (median )of controlmice ÷ø æ T ö æ DTumor weight (median )of treatedmice ö %ç ÷ = ç ÷ø ´ 100 for T < 0. è Cø è Initial tumor weight Additional parameters can be defined such as:

é (T - C )ù Growth delay = ê ú ´ 100, ë C û where T and C are median times to reach predetermined tumor weight in treated and control groups, respectively, or

ìï é (T - C ) - duration of Rx ´ 0.31ù üï Net log cell kill = í ê úý Doubling time û þï îï ë Tumor weight may be estimated by various formulas derived from tumor dimensions measured by calipers during the course of the experiment. Subcutaneous models also allow the scoring of tumor regressions and apparent tumor-free animals. The major criticism of such models is that the subcutaneous compartment, while convenient for the investigator, is very artificial and does not contain the nourishing stroma that would reflect the tumor’s site of origin. Thus, various “orthotopic” approaches after direct injection of tumor cells, e.g., into the axillary fat pad to mimic a breast site of origin, cecal to represent intestinal primary sites, etc. [81]. Key issues in the operation of any subcutaneous tumor xenograft model systems include assuring whether any special growth requirements exist for the tumors, e.g., estrogen or androgen supplementation for hormone-dependent tumors; the presence of a truly untreated control group as well as a vehicle-treated control

106

E.A. Sausville

group (certain particularly lipid containing vehicles can influence the growth of tumors); and adequate statistics as described above. The standard National Cancer Institute animal testing protocols in use since 1990 conventionally defined preliminary evidence of activity as a dose and schedule of agent that imparted a T/C of £ 40, using tumor models that would require ten animals per control group and six per treatment group [72, 74]. In all cases, animals were randomized so that each treatment or control group had a comparable range of tumor burden at the start of an experiment. An additional variable is when to stage a tumor model or initiate treatment [74]. An “early treatment” approach initiates treatment before the tumor is measurable. Where tumor weight is estimated from volume of prolate ellipsoid:

(

Length (mm) ´ Width (mm) 2 2 (assuming unit - specific gravity).

)

Tumor weight = Volume mm3 =

An “early stage tumor model” commences treatment at 63–200 mg. The advantage of early stage treatment models is that the effect of chronic dosing can be understood and that by design all entumored mice are used. Disadvantages are that such models do not represent the majority of clinically encountered human tumors, which are treated with novel agents when a patient already has advanced stage, bulky disease. Also “early stage treatment” models are really suitable in a statistical sense where the model has > 90% “Take” Rate with < 10% spontaneous remissions. In contrast, “advanced stage” tumor models utilize starting tumors at >200  mg. Advantages of studying advanced stage tumors include the possibility of scoring parameters relevant to the clinic (partial and complete responses) as well as the capacity for scoring effects of the test agent on some angiogenesis or stromal endpoints. Disadvantages of such models include the long running times (6–12 weeks/ experiment) and the need for larger numbers of mice to be entumored to allow randomization of different tumor sizes, creating the possibility of nonusable animals in an experiment. The “hollow fiber” model is a departure from xenograft testing that has the goal of minimizing the number of animals exposed to compounds with a low likelihood of in vivo activity; of decreasing the amount of compound utilized per tumor model tested; and the completion in a relatively rapid fashion of initial screening for the potential to prioritize among a series of leads. The hollow fiber model uses tumor cells grown in porous (to Mr < 500,000) fibers that can be placed in a variety of animal compartments (IP and SC), exposed to systemically administered compartments, and then removed after, e.g., 1 week or less and assayed for viable cell numbers by colorimetric approaches. The National Cancer Institute USA converted to this initial prescreen for in vivo activity in the late 1990s, and its operation and standard procedures have been described elsewhere [82]. It is not intended as a definitive model to support further development. Activity of a compound in a large number of cell lines in the hollow fiber system is in fact correlated with subsequent activity in more conventional xenograft systems[83], and hollow fiber models have

4  Preclinical Models for Anticancer Drug Development

107

been utilized by various commercial parties in analogous efforts to increase efficiency in the selection of development leads [84] prior to xenograft testing experiences.

4.3.3 Clinical Correlation with In Vivo Screening and Model Results In a retrospective analysis of activity in NCI USA screening systems for 39 predominantly classical cytotoxic agents entered into clinical trials under NCI sponsorship in the 1980 to approximately 1998 time period, there was essentially no correlation between activity in a particular histologic type of animal model and activity in the corresponding clinical neoplasm [60]. However, of those agents that were active in at least 33% of models tested across different histologic model types, there was a 40–50% likelihood of phase II clinical activity in at least two disease types. In contrast, those agents with activity in <33% of models tested had no evidence clinical activity in more than one Phase II trial. The NCI of Canada took a somewhat different approach, considering the predictive value of cell line activity, mouse allograft and xenograft approaches. Similar to the NCI USA study, no particular xenograft system or indeed panels of breast or colon cancer models were clearly predictive of human clinical activity; however, activity of an agent in panels of nonsmall cell lung and ovarian models was somewhat predictive of clinical activity [85]. Why might the xenograft models widely used have such a poor predictive value? If their value is so limited, why conduct animal testing at all? In response to the first question, a number of features intrinsic to mouse models potentially limit their predictive value. First and foremost, classical cytotoxic agents are conventionally studied and used at close to their MTD in both mouse and humans. Thus, small differences between the species in pharmacological features of the agent can connote major differences in ultimate clinical applicability to humans. For example, radically different plasma protein binding of the active lactone form of camptothecins between mice and humans, combined with intrinsic refractoriness of mouse target organs vs. human target organs conspire to allow mice to over predict activity of camptothecins in humans [86, 87]. Differential species-of-origin cell susceptibilities also occur with brefeldins [88], modulators of protein secretion from endoplasmic reticulum sites, and the bizelesin class minor groove DNA binders [89]. A different half life in mice vs. humans (longer) overstates the potential for the antitumor effect of MS-275, a recently introduced histone deacetylase inhibitor [90], which is very active in mice on a daily schedule which causes excessive toxicity in humans [91]. Intrinsic differences between the susceptibility of murine and human ATPases [92] likewise overstate the developmental potential of cardiotoxic glycoside-like agents, such as neriifolins. One implication of these data is that an important use of animal model studies of efficacy of novel anticancer agents may actually be after the conclusion of initial phase I testing in humans. Assuming that the phase I studies offer an understanding of tolerated human pharmacology, armed with this information, a return to the animal

108

E.A. Sausville

models may occur with conscientious effort to model the human pharmacology in mice. This may allow a more realistic selection of dose, schedule, and tumor types for more involved phase II testing. A more full description of this strategy in pediatric neoplasms has been described by Houghton and colleagues [93]. An additional refinement in the use of animal models is to define the drug exposure parameters (peak concentration, Cmax; area under the concentration × time of elimination curve, etc.) in the animals and correlate with evidence of antitumor effect, and evidence of drug effect on its intended target (pharmacodynamics) in both the tumor and a surrogate compartment. This will allow a Phase II experience in humans to be developed in an informed and efficient fashion, closely guided by this information. This approach utilizing primarily pharmacological information is well illustrated in the clinical development of dasatinib [94]. The use of a pharmacodynamic approach in a surrogate compartment, viz., peripheral blood mononuclear cells, to define schedule and dose escalation limits, is exemplified by the development of bortezomib [95, 96], where the preclinically developed assay of proteosome activity guided the conduct of dose escalation in subsequent clinical trials. A somewhat different point of view than described above has been offered by those who criticize the use of cells propagated from cell lines, and rather would suggest that mouse models derived from tumor fragments propagated directly from human tumors are superior to the predictive value of models derived from cell lines. Certainly, Fiebig et al. have consistently championed the value of such models in defining agent activity in such systems [64], and it is true that such models do offer a closer correspondence of histology of the mouse borne tumor models to the human. Additional criticisms of subcutaneous xenografts come from the perspective that they do not in a particular sense represent tumors arising in the organ of origin, with defined genetic lesions that would tie to particular subsets of human disease. Genetically modified mouse strains [97, 98] to express oncoproteins of defined prevalence in human tumors have been developed and do represent attractive models in which to explore particular agents targeted to pathways of importance in human disease. However, such transgenic models are often by their long latency and variable penetrance difficult to apply to a series of candidate drug molecules arising from a discovery and development campaign and are perhaps best reserved in their use for candidates at a more advanced stage of development, where in particular tolerated pharmacology is known. In that event, the model could usefully confirm the capacity of the test agent to modify the relevant oncoprotein target or pathway driving the tumor [99]. Thus, the perspective that emerges is that animal model activity by itself is actually just the first step in designing an early clinical development campaign. Such information must be ideally correlated with pharmacological information in both the test species and with initial behavior of the drug in humans. Additional systems to predict intestinal cell permeability (the CACO intestinal cell model) [100], cardiotoxicity after exposure to cardiomyocytes ex vivo [101] or effect on cloned cardiac ion channels[102], or neuronal toxicity models [103] can also be applied in an effort to further refine the likelihood of patterns of likely drug toxicity, but the

4  Preclinical Models for Anticancer Drug Development

109

best application of these additional test systems is in the framework of a certain relationship of antitumor efficacy in murine hosts in order to refine the basis for entering agents into Phase II testing in humans.

4.4 Summary and Conclusions Ultimately, in vitro and in vivo drug development models all function in an effort to “stack the deck” in favor of selecting a drug ultimately active in humans. It is important to emphasize that current FDA regulations do not require any degree of efficacy of an agent in any animal or other model prior to entry of the agent in initial human studies. Rather, safety prior to allowing an investigational new drug application (IND) is the paramount consideration. Yet, no drug thus far approved as the result of a New Drug Application process for use in humans in the treatment of cancer has been devoid of easily definable empirical activity in at least some series of models such as those described in this chapter. Thus the drug developer who would propose proceeding to the clinic, particularly beyond Phase I, despite poorly understood preclinical bases for activity, does not have a clear precedent for that course. This chapter has sought to give an overview of the preclinical aspects of a compound that would increase enthusiasm for further development, ranging from very early screening results to advanced in  vivo model development. The future will undoubtedly see modifications in these approaches both from the standpoint of model systems, for example, the creation of mice that have conscientiously altered drug metabolizing systems [104] to more closely resemble human metabolic potential, to the use of cell models that have been either chosen or constructed to represent human tumors with genetic abnormalities that are representative of a compound’s intended tumor type for registrational strategies. For example, it is increasingly recognized that BRCA1-deficient tumors are selectively sensitive to poly ADP ribose polymerase inhibitors [105] and that platinating agents are selectively active in cells deficient in certain members of the Fanconi anemia associated gene cluster [106]. Thus screens for novel agents of value in these contexts can be designed with genetically defined target cell classes. Additional refinements in animal testing may emerge from the routine use of imaging strategies (e.g., luminescent or fluorescent cells or drug probes labeled for PET or other analogous detection strategies) that will allow minimal residual disease to be legitimately modeled [107] in an effort to define compounds that would usefully induce a state of tumor latency or dormancy. The prospect of “personalized” medicine anticipated to emerge from the characterization of a patient’s tumor by expression of genes or proteins will only increase the need for corresponding animal models that mirror these efforts to refine and select treatments offered to patients. The coming century of cancer drug discovery and development will hopefully introduce models with increased predictive value for applicability to human patients, the ultimate species of interest for our preclinical model efforts.

110

E.A. Sausville

References 1. Kuh HJ, Jang SH, Wientjes MG, et al: Determinants of paclitaxel penetration and accumulation in human solid tumor. J Pharmacol Exp Ther 290:871–80, 1999. 2. Shepherd FA, Rodrigues Pereira J, Ciuleanu T, et al: Erlotinib in previously treated non-small-cell lung cancer. N Engl J Med 353:123–32, 2005. 3. Tan AR, Headlee D, Messmann R, et al: Phase I clinical and pharmacokinetic study of flavopiridol administered as a daily 1-hour infusion in patients with advanced neoplasms. J Clin Oncol 20:4074–82, 2002. 4. Demo SD, Kirk CJ, Aujay MA, et  al: Antitumor activity of PR-171, a novel irreversible inhibitor of the proteasome. Cancer Res 67:6383–91, 2007. 5. Dowsett M, Nicholson RI, Pietras RJ: Biological characteristics of the pure antiestrogen fulvestrant: overcoming endocrine resistance. Breast Cancer Res Treat 93 (Suppl 1):S11–8, 2005. 6. Janus A, Robak T, Smolewski P: The mammalian target of the rapamycin (mTOR) kinase pathway: its role in tumourigenesis and targeted antitumour therapy. Cell Mol Biol Lett 10:479–98, 2005. 7. Lipinski CA, Lombardo F, Dominy BW, et al: Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv Drug Deliv Rev 46:3–26, 2001. 8. Newman DJ, Cragg GM: Natural products as sources of new drugs over the last 25 years. J Nat Prod 70:461–77, 2007. 9. Cragg GM, Boyd MR, Cardellina JH 2nd, et al: Ethnobotany and drug discovery: the experience of the US National Cancer Institute. Ciba Found Symp 185:178–90, 1994 (discussion 190–6). 10. Chen YH, Lipes BD, Kenan DJ, et al: Identification of recombinant antibodies against multiple distinct toll-like receptors by homolog mining a single immune scFv phage library. J Immunol Methods 340(2):144–53, 2009. 11. Wang B, Chen YB, Ayalon O, et al: Human single-chain Fv immunoconjugates targeted to a melanoma-associated chondroitin sulfate proteoglycan mediate specific lysis of human melanoma cells by natural killer cells and complement. Proc Natl Acad Sci USA 96:1627– 32, 1999. 12. Strome SE, Sausville EA, Mann D: A mechanistic perspective of monoclonal antibodies in cancer therapy beyond target-related effects. Oncologist 12:1084–95, 2007. 13. Foss FM, Bacha P, Osann KE, et al: Biological correlates of acute hypersensitivity events with DAB(389)IL-2 (denileukin diftitox, ONTAK) in cutaneous T-cell lymphoma: decreased frequency and severity with steroid premedication. Clin Lymphoma 1:298–302, 2001. 14. Sahdev S, Khattar SK, Saini KS: Production of active eukaryotic proteins through bacterial expression systems: a review of the existing biotechnology strategies. Mol Cell Biochem 307:249–64, 2008. 15. Waehler R, Russell SJ, Curiel DT: Engineering targeted viral vectors for gene therapy. Nat Rev Genet 8:573–87, 2007. 16. Aurigemma R, Tomaszewski JE, Ruppel S et al: Regulatory aspects in the development of gene therapy, in Curiel DT, Douglas J: Cancer Gene Therapy, Humana Press, Totowa NJ, 2005, pp 441–472. 17. Chabner BA, Roberts TG Jr. Timeline: Chemotherapy and the war on cancer. Nat Rev Cancer 5:65–72, 2005. 18. Waud WR: Murine L1210 and P388 leukemias, in Teicher BA & Andrews PA: Anticancer Drug Development Guide Preclinical Screening, Clinical Trials, and Approval, 2nd ed. Humana Press, Totowa, NJ, 2004, pp 79–97. 19. Boyd MR: The NCI human tumor cell line (60 Cell)screen, in Teicher BA & Andrews PA: Anticancer Drug Development Guide Preclinical Screening, Clinical Trials, and Approval, 2nd ed. Humana Press, Totowa, NJ, 2004, pp 41–61.

4  Preclinical Models for Anticancer Drug Development

111

20. Druker BJ, Tamura S, Buchdunger E, et al: Effects of a selective inhibitor of the Abl tyrosine kinase on the growth of Bcr-Abl positive cells. Nat Med 2:561–6, 1996. 21. Pegram MD, Konecny G, Slamon DJ: The molecular and cellular biology of HER2/neu gene amplification/overexpression and the clinical development of herceptin (trastuzumab) therapy for breast cancer. Cancer Treat Res 103:57–75, 2000. 22. Baumann CK, Castiglione-Gertsch M: Estrogen receptor modulators and down regulators: optimal use in postmenopausal women with breast cancer. Drugs 16:2335–53, 2007. 23. Taplin ME: Drug insight: role of the androgen receptor in the development and progression of prostate cancer. Nat Clin Pract Oncol 4:236–44, 2007. 24. Kim DH, Sabatini DM: Raptor and mTOR: subunits of a nutrient-sensitive complex. Curr Top Microbiol Immunol 279:259–70, 2004. 25. Schellens JH: Capecitabine. Oncologist 12:152–5, 2007. 26. Weinstein IB, Joe A: Oncogene addiction. Cancer Res 68:3077–80, 2008. 27. Reddy A, Kaelin WG Jr.: Using cancer genetics to guide the selection of anticancer drug targets. Curr Opin Pharmacol 2:366–73, 2002. 28. Davies H, Bignell GR, Cox C, et al: Mutations of the BRAF gene in human cancer. Nature 417:949–54, 2002. 29. Willmore-Payne C, Holden JA, Tripp S, et  al: Human malignant melanoma: detection of BRAF- and c-kit-activating mutations by high-resolution amplicon melting analysis. Hum Pathol 36:486–93, 2005. 30. Aina OH, Liu R, Sutcliffe JL, et al: From combinatorial chemistry to cancer–targeting peptides. Mol Pharm 4:631–51, 2007. 31. Messeguer A, Cortés N: Combinatorial chemistry in cancer research. Clin Transl Oncol 9:83–92, 2007. 32. Tan DS: Current progress in natural product-like libraries for discovery screening. Comb Chem High Throughput Screen 7:631–43, 2004. 33. Schepartz SA: Introduction and Historical Background, in Foye WO: Cancer Chemotherapeutic Agents, American Chemical Society, Washington DC 1995, pp 1–7. 34. Elion GB, Hitchings GH: Metabolic basis for the actions of analogs of purines and pyrimidines. Adv Chemother 2:91–177, 1965. 35. Grever MR, Schepartz SA, Chabner BA: The National Cancer Institute: cancer drug discovery and development program. Semin Oncol 19:622–38, 1992. 36. Corbett T, Polin L, LoRusso P et al: In vivo methods for screening and preclinical testing, in Teicher BA & Andrews PA: Anticancer Drug Development Guide Preclinical Screening, Clinical Trials, and Approval, 2nd ed. Humana Press, Totowa, NJ 2004, pp 99–123. 37. Shoemaker RH: The NCI60 human tumour cell line anticancer drug screen. Nat Rev Cancer 6:813–23, 2006. 38. Paull KD, Hamel E, Malspeis, L: Prediction of biochemical mechanism of action from the in vitro antitumor screen of the national cancer institute, in Foye WO: Cancer Chemotherapeutic Agents, American Chemical Society, Washington, DC, 1995 pp 9–45. 39. Weinstein JN, Kohn KW, Grever MR, et al: Neural computing in cancer drug development: predicting mechanism of action. Science 258:447–51, 1992. 40. Scherf U, Ross DT, Waltham M et al: A gene expression database for the molecular pharmacology of cancer. Nat Genet 24:236–44, 2000. 41. Rabow AA, Shoemaker RH, Sausville EA et  al: Mining the National Cancer Institute’s tumor-screening database: identification of compounds with similar cellular activities. J Med Chem 45:818–40, 2002. 42. Zaharevitz DW, Holbeck SL, Bowerman C, et  al: COMPARE: a web accessible tool for investigating mechanisms of cell growth inhibition. J Mol Graph Model 20:297–303, 2002. 43. Keinan S, Paquette WD, Skoko JJ, et  al: Computational design, synthesis and biological evaluation of para-quinone-based inhibitors for redox regulation of the dual-specificity phosphatase Cdc25B. Org Biomol Chem 6:3256–63, 2008. 44. Bender A, Bojanic D, Davies JW et al: Which aspects of HTS are empirically correlated with downstream success? Curr Opin Drug Discov Devel 11:327–37, 2008.

112

E.A. Sausville

45. Hancock CN, Macias A, Lee EK et al: Identification of novel extracellular signal-regulated kinase docking domain inhibitor. J Med Chem 48:4586–95, 2005. 46. Fischer PM: Computational chemistry approaches to drug discovery in signal transduction. Biotechnol J 3:452–70, 2008. 47. Alonso H, Bliznyuk AA, Gready JE: Combining docking and molecular dynamic simulations in drug design. Med Res Rev 26:531–68, 2006. 48. Kirchmair J, Distinto S, Schuster D, et al: Enhancing drug discovery through in silico screening: strategies to increase true positives retrieval rates. Curr Med Chem 15:2040–53, 2008. 49. Rowlands MG, Newbatt YM, Prodromou C, et  al: High-throughput screening assay for inhibitors of heat-shock protein 90 ATPase activity. Anal Biochem 327:176–83, 2004. 50. Jobson AG, Cardellina JH 2nd, Scudiero D, et al: Identification of a Bis-guanylhydrazone [4,4¢-Diacetyldiphenylurea-bis(guanylhydrazone); NSC 109555] as a novel chemotype for inhibition of Chk2 kinase. Mol Pharmacol 72:876–84, 2007. 51. Wendt MD, Shen W, Kunzer A, et al: Discovery and structure-activity relationship of antagonists of B-cell lymphoma 2 family proteins with chemopotentiation activity in  vitro and in  vivo. J Med Chem 49:1165–81, 2006. 52. Rapisarda A, Uranchimeg B, Scudiero DA et al: Identification of small molecule inhibitors of hypoxia-inducible factor 1 transcriptional activation pathway. Cancer Res 62:4316–24, 2002. 53. Bommi-Reddy A, Almeciga I, Sawyer J, et  al: Kinase requirements in human cells: III. Altered kinase requirements in VHL−/− cancer cells detected in a pilot synthetic lethal screen. Proc Natl Acad Sci USA 105:16484–9, 2008. 54. Kaelin WG Jr.: The concept of synthetic lethality in the context of anticancer therapy. Nat Rev Cancer 5:689–98, 2005. 55. Webb CP, Hose CD, Koochekpour S, et al: The geldanamycins are potent inhibitors of the hepatocyte growth factor/scatter factor-met-urokinase plasminogen activator-plasmin proteolytic network. Cancer Res 60:342–9, 2000. 56. Simon JA, Bedalov A: Yeast as a model system for anticancer drug discovery. Nat Rev Cancer 4:481–92, 2004. 57. Liu X, Kramer JA, Swaffield JC, et al: Development of a highthroughput yeast-based assay for detection of metabolically activated genotoxins. Mutat Res 653:63–9, 2008. 58. Tran TC, Sneed B, Haider J, et al: Automated, quantitative screening assay for antiangiogenic compounds using transgenic zebrafish. Cancer Res 67:11386–92, 2007. 59. van Haaften G, Romeijn R, Pothof J, et al: Identification of conserved pathways of DNA-damage response and radiation protection by genome-wide RNAi. Curr Biol 16:1344–50, 2006. 60. Johnson JI, Decker S, Zaharevitz D, et al: Relationships between drug activity in NCI preclinical in vitro and in vivo models and early clinical trials. Br J Cancer 84:1424–31, 2001. 61. Hollingshead MG, Alley MC, Kaur G, et al: NCI specialized procedures in preclinical drug evaluations, in Teicher BA & Andrews PA: Anticancer Drug Development Guide Preclinical Screening, Clinical Trials, and Approval, 2nd ed. Humana Press, Totowa, NJ, 2004 ,pp 153–182. 62. Hamburger AW, Salmon SE: Primary bioassay of human tumor stem cells. Science 197:461–3, 1977. 63. Chumsri S, Phatak P, Edelman MJ et  al: Cancer stem cells and individualized therapy. Cancer Genomics Proteomics 4:165–74, 2007. 64. Fiebig HH, Maier A, Burger AM: Clonogenic assay with established human tumour xenografts: correlation of in vitro to in vivo activity as a basis for anticancer drug discovery. Eur J Cancer 40:802–20, 2004. 65. Schirner M, Hoffmann J, Menrad A, Schneider MR: Antiangiogenic chemotherapeutic agents: characterization in comparison to their tumor growth inhibition in human renal cell carcinoma models. Clin Cancer Res 4:1331–6, 1998. 66. Taraboletti G, Giavazzi R: Modelling approaches for angiogenesis. Eur J Cancer 40:881–9, 2004. 67. Auerbach R, Lewis R, Shinners B, et al: Angiogenesis assays: a critical overview. Clin Chem 49:32–40, 2003. 68. Zogakis TG, Costouros NG, Kruger EA, et  al: Microarray gene expression profiling of angiogenesis inhibitors using the rat aortic ring assay. Biotechniques 33: 664–6, 668, 2002.

4  Preclinical Models for Anticancer Drug Development

113

69. Chou TC, Talalay P: Quantitative analysis of dose–effect relationships: the combined effects of multiple drugs or enzyme inhibitors. Adv Enzyme Regul 22:27–55, 1984. 70. Teicher BA: Preclinical models for combination therapy, in Teicher BA & Andrews PA: Anticancer Drug Develoopment Guide Preclinical Screening, Clinical Trials, and Approval, 2nd ed. Humana Press, Totowa, NJ, 2004, pp 213–242. 71. Decker S, Sausville EA: Preclinical modeling of combination treatments: fantasy or requirement? Ann N Y Acad Sci 1059:61–9, 2005. 72. Hollingshead MG: Antitumor efficacy testing in rodents. J Natl Cancer Inst 100:1500–10, 2008. 73. Tan M, Fang HB, Tian GL, et al: Small-sample inference for incomplete longitudinal data with truncation and censoring in tumor xenograft models. Biometrics 58:612–20, 2002. 74. Alley MC, Hollingshead M, Dykes DJ, et al: Human tumor xenograft models in nci drug development, in Teicher BA & Andrews PA: Anticancer Drug Development Guide Preclinical Screening, Clinical Trials, and Approval, 2nd ed. Humana Press, Totowa, NJ 2004 pp 125–152. 75. Fiebig HH, Dengler WA, Roth T: Human tumor xenografts: predictivity, characterization and discovery, in Fiebig HH and Burger AM: Relevance of Tumor Models for Antiocancer Development. vol 54, Karger, Contributions to Oncology Basel, 1999, pp 29–50. 76. Kelland LR: Of mice and men: values and liabilities of the athymic nude mouse model in anticancer drug development. Eur J Cancer 40:827–36, 2004. 77. Fichtner I, Becker M, Zeisig R, et al: In vivo models for endocrine-dependent breast carcinomas: special considerations of clinical relevance. Eur J Cancer 40:845–51, 2004. 78. Hollingshead MG, Bonomi CA, Borgel SD, et al: A potential role for imaging technology in anticancer efficacy evaluations. Eur J Cancer 40:890–8, 2004. 79. Hoffman RM: The multiple uses of fluorescent proteins to visualize cancer in vivo. Nat Rev Cancer 5:796–806, 2005. 80. Sahai E: Illuminating the metastatic process. Nat Rev Cancer 7:737–49, 2007. 81. Bibby MC: Orthotopic models of cancer for preclinical drug evaluation: advantages and disadvantages. Eur J Cancer 40:852–7, 2004. 82. Hollingshead MG, Alley MC, Camalier RF et al: In vivo cultivation of tumor cells in hollow fibers. Life Sci 57:131–141, 1995. 83. Decker S, Hollingshead M, Bonomi CA, et al: The hollow fibre model in cancer drug screening: the NCI experience. Eur J Cancer 40:821–6, 2004. 84. Zabludoff SD, Deng C, Grondine MR, et al: AZD7762, a novel checkpoint kinase inhibitor, drives checkpoint abrogation and potentiates DNA-targeted therapies. Mol Cancer Ther 7:2955–66, 2008. 85. Voskoglou-Nomikos T, Pater JL, Seymour L: Clinical predictive value of the in  vitro cell line, human xenograft, and mouse allograft preclinical cancer models. Clin Cancer Res 9:4227–39, 2003. 86. Mi Z, Burke TG: Marked interspecies variations concerning the interactions of camptothecin with serum albumins: a frequency-domain fluorescence spectroscopic study. Biochemistry 33:12540-5, 1994. 87. Thompson J, Stewart CF, Houghton PJ: Animal models for studying the action of topoisomerase I targeted drugs. Biochim Biophys Acta 1400:301–19, 1998. 88. Ishii S, Nagasawa M, Kariya Y, et  al: Selective cytotoxic activity of brefeldin A against human tumor cell lines. J Antibiot (Tokyo) 42:1877–8, 1989. 89. Volpe DA, Tomaszewski JE, Parchment RE, et  al: Myelotoxic effects of the bifunctional alkylating agent bizelesin on human, canine and murine myeloid progenitor cells. Cancer Chemother Pharmacol 39:143–9, 1996. 90. Saito A, Yamashita T, Mariko Y, et al: A synthetic inhibitor of histone deacetylase, MS-27275, with marked in vivo antitumor activity against human tumors. Proc Natl Acad Sci USA 96:4592–7, 1999. 91. Ryan QC, Headlee D, Acharya M, et al: Phase I and pharmacokinetic study of MS-275, a histone deacetylase inhibitor, in patients with advanced and refractory solid tumors or lymphoma. J Clin Oncol 23:3912–22, 2005.

114

E.A. Sausville

92. Gupta RS, Chopra A, Stetsko DK: Cellular basis for the species differences in sensitivity to cardiac glycosides (digitalis). J Cell Physiol 127:197–206, 1986. 93. Peterson JK, Houghton PJ: Integrating pharmacology and in vivo cancer models in preclinical and clinical drug development. Eur J Cancer 40:837–44, 2004. 94. Luo FR, Yang Z, Camuso A, et al: Dasatinib (BMS-354825) pharmacokinetics and pharmacodynamic biomarkers in animal models predict optimal clinical exposure. Clin Cancer Res 12:7180–6, 2006. 95. Adams J, Palombella VJ, Sausville EA, et al: Proteasome inhibitors: a novel class of potent and effective antitumor agents. Cancer Res 59:2615–22, 1999. 96. Orlowski RZ, Stinchcombe TE, Mitchell BS, et al: Phase I trial of the proteasome inhibitor PS-341 in patients with refractory hematologic malignancies. J Clin Oncol 20: 4420–7, 2002. 97. Omer CA, Chen Z, Diehl RE, et al: Mouse mammary tumor virus-Ki-rasB transgenic mice develop mammary carcinomas that can be growth-inhibited by a farnesyl:protein transferase inhibitor. Cancer Res 60:2680–8, 2000. 98. Nørgaard P, Law B, Joseph H, et  al: Treatment with farnesyl-protein transferase inhibitor induces regression of mammary tumors in transforming growth factor (TGF) alpha and TGF alpha/neu transgenic mice by inhibition of mitogenic activity and induction of apoptosis. Clin Cancer Res 5:35–42, 1999. 99. Hansen K, Khanna C: Spontaneous and genetically engineered animal models; use in preclinical cancer drug development. Eur J Cancer 40:858–80, 2004. 100. Cheng KC, Li C, Uss AS: Prediction of oral drug absorption in humans – from cultured cell lines and experimental animals. Expert Opin Drug Metab Toxicol 4:581–90, 2008. 101. Dorr RT: Cytoprotective agents for anthracyclines. Semin Oncol 23(4 Suppl 8):23–34, 1996. 102. Dorn A, Hermann F, Ebneth A, et  al: Evaluation of a high-throughput fluorescence assay method for HERG potassium channel inhibition. J Biomol Screen 10:339–47, 2005. 103. Joseph EK, Chen X, Bogen O, et al: Oxaliplatin acts on IB4-positive nociceptors to induce an oxidative stress-dependent acute painful peripheral neuropathy. J Pain 9:463-72, 2008. 104. Cheung C, Gonzalez FJ: Humanized mouse lines and their application for prediction of human drug metabolism and toxicological risk assessment. J Pharmacol Exp Ther 327:288– 99, 2008. 105. Farmer H, McCabe N, Lord CJ, et al: Targeting the DNA repair defect in BRCA mutant cells as a therapeutic strategy. Nature 434:917–21, 2005. 106. Taniguchi T, Tischkowitz M, Ameziane N, et al: Disruption of the Fanconi anemia-BRCA pathway in cisplatin-sensitive ovarian tumors. Nat Med 9:568–74, 2003. 107. Boss DS, Olmos RV, Sinaasappel M, et al: Application of PET/CT in the development of novel anticancer drugs. Oncologist 13:25–38, 2008.

Part III

Chapter 5

Phase I Clinical Trials with Anticancer Agents Stephen Leong, Justin Call, Alex A. Adjei, and Wells Messersmith

5.1 Introduction Although the term “phase I” is used to describe numerous trial designs, the overarching goal of a phase I study is to determine the optimal dose and/or schedule of a therapy for evaluation in the phase II setting. Phase I trials typically test different doses of an anticancer agent(s) in various neoplastic diseases, with safety evaluation as a main objective. These studies range from first-in-human trials of novel single agents to new combinations of FDA-approved therapies. Frequently, pharmacokinetic studies are incorporated in phase I clinical trials, in order to determine drug exposure and clearance. In addition, phase I studies may include biomarkers of drug effects such as functional imaging or direct measurement of drug effects on either tumor biopsies and/or normal tissue samples. Since both toxicity and efficacy generally exhibit dose-dependency with ­oncology drugs, phase I studies usually escalate doses from a low, safe dose (e.g., one-tenth the lethal dose in the most sensitive species tested preclinically) to a “maximum tolerated dose” (MTD) in preplanned steps. The MTD is generally defined as the dose level at which less than one-third of the subjects experience dose-limiting toxicity (DLT). The dose escalation steps and number of patients per dose cohort vary depending on the type of phase I design. With the advent of ­“targeted” oncologic therapies, which tend to have less severe acute toxicities, a number of novel phase I designs have been proposed, which constitute a more advantageous safety profile compared to classic “cytotoxic agents,” where many of the phase I designs were initially conceived. The risks and benefits of phase I studies for cancer patients have been discussed extensively. Although response rates in single-agent studies has remained low in the past three decades, phase I trials conducted between 1999 and 2002 were found to

A.A. Adjei (*) Department of Medicine, Roswell Park Cancer Institute, Elm & Carlton Streets, Buffalo, NY 14263, USA e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_5, © Springer Science+Business Media, LLC 2011

117

118

S. Leong et al.

have approximately one-tenth the death rate from study drug(s) than between 1991 and 1994 [1]. In addition, response rates for multiagent trials tend to be much higher, frequently exceeding that seen for FDA-approved second- and third-line therapies. In this chapter, the design, analysis, and reporting of phase I oncology studies will be discussed.

5.2 Design Options and Dose Escalation There are numerous phase I designs and dose escalation options, some of which have evolved in the era of “targeted therapies.” The classical “3 + 3” modified Fibonacci design involves preplanned dose escalation steps based on a ­mathematical series (Fig. 5.1). After the first dose level, which is typically determined by animal toxicology studies, the next levels increase in a stepwise fashion, for instance, the next dose level may increase by the increment of 100% (level 2), 67% (3), 50% (4), 40% (5), and 33% (6) with all subsequent levels increasing by 33% [2]. Each dose level is explored initially with three patients, and the patients are observed for ­predefined toxicities defined as “dose-limiting toxicities” (DLTs). While DLT ­definitions vary based on the study, typically these are serious toxicities that require medical management and may be life-threatening. If none of the first three patients at a dose level experience DLT, the dose is increased to the next level. Since the goal is to limit the occurrence of DLT to <33% (at the MTD), if one of the three patients experiences DLT the same dose level is expanded to six patients by adding three more patients at the same dose. If none of the next three patients develop DLT, the dose level is increased since the proportion of patients’ experiencing DLT is  <20%. If one or more of the patients experience DLT, the 33% threshold is crossed and the dose level is decreased to the prior dose level. Another three patients are treated at the lower dose level to confirm that the proportion of DLTs is <33%. Often, phase I studies will include an expansion cohort called the “MTD confirmation phase,” where up to 12 patients are treated to gather additional safety and pharmacokinetic data. There are other versions of the 3 + 3 designs, which follow different dose escalation rules. The “fixed dose” escalation is a predetermined dose escalation (e.g., 20–40% dose escalation) based on various factors including preclinical pharmacokinetic (PK) modeling, manufacturing practices, and convenience. The “doubling method” uses an escalation scheme whereby the dose is doubled until any grade 2 drug-related toxicity, where upon escalation is reversed to modified Fibonacci or fixed dose escalation. Another variation is the enrollment of four patients per cohort, but only three evaluable patients are needed to proceed to meet dose escalation criteria. The advantage of this method is to prevent lengthy delays because of patients withdrawing consent, becoming “not evaluable” or failing eligibility criteria. The fourth patient (“extra patient”) may provide additional safety information that would allow the study to proceed to the next dose level. If the dose level needs to be expanded, only two more patients are needed to complete the cohort.

5  Phase I Clinical Trials with Anticancer Agents

119

Treat 3 patients at K dose

0 DLT

Escalate dose k+1 level: return to top of flow chart with dose k+1

1 DL T

Treat 3 more patients at dose k

1 out of 6 patients with DLT

> 1 DL T

>1 of 6 with DLT

If 6 patients already treated at dose k-1 level

Stop Study or Dose expand patients at this dose level per protocol

De-escalate to dose k-1

If 3 patients already treated at dose k-1

Treat 3 more patients at dose k-1

Fig. 5.1  A standard “3 + 3” dose escalation design starting a dose k. MTD is usually defined as the highest dose at which 0 or 1 DLT is observed in six patients (although some “3 + 3” rules call the highest dose with 2 or fewer DLTs in six patients the MTD). If de-escalation occurs at the first dose level, then the study is discontinued

Although the “3 + 3” method has generally been considered safe, it is often criticized for several reasons. First, many patients receive subtherapeutic levels of study drugs. Second, the trials can take a long time to complete due to the large number of patients and dose levels. Third, although DLTs are recorded throughout the time a patient is on study, only those occurring within the first course are used for the purposes of dose escalation, thus cumulative toxicities may not be recognized in deriving the MTD. Fourth, intrapatient dose escalation is generally not allowed, thus preventing patients entered early in the study from receiving doses closer to the MTD, where a greater chance of clinical benefit may be present. Finally, using statistical modeling, more efficient designs have been developed, which theoretically can result in a more rapid study completion with fewer patients and with a greater proportion of patients treated near the MTD.

120

S. Leong et al.

In an effort to address these weaknesses, alternative phase I study designs have been proposed. • Accelerated titration design (rule-based): a constellation of designs to utilize single-patient toxicity-guided escalations • Continual reassessment method (CRM): statistically guided dose escalation that uses real-time toxicity data to predict the dose closest to the MTD • Pharmacologically guided: use of real-time PK data to guide escalation with intent to achieve drug exposure • Rolling six design. In the accelerated titration method [2], subjects are enrolled in single-patient cohorts, and the dose is increased by 100% until mild or moderate toxicities are seen. This allows rapid escalation, with few patients, at low doses of study drugs. Once toxicity is seen, the escalation is lowered to 40% for each dose level, and the design reverts back to a 3 + 3 design. Intrapatient dose escalation may be allowed. Variations of this design have been statistically modeled against the traditional 3 + 3 designs, and the number of patients required to determine the MTD may be cut in half, with fewer patients treated at subtherapeutic levels. Another design that allows aggressive dose escalation and varying dose escalations is the CRM [3]. In this method, data observed in patients early in the trial are used to help select doses to be given later in the trial. This is considered an adaptive trial because the design is adapted based on accumulating data in the trial. Supporters of CRM design believe that patients are less likely to be treated at toxic doses and more likely to treat at efficacious doses with the assumptions that dose increases also increase the probability of efficacy and toxicities. Critics have concerns about this study design’s safety, worrying that escalations may occur too quickly and that patients may be treated at unsafe doses based on reliance on mathematical model. These concerns are largely unfounded. There are various modifications to the CRM that have been proposed [4, 5]. Although there is little doubt that designs with more rapid dose increases such as the accelerated titration or CRMs outperform traditional “conservative” 3 + 3 designs on computer simulations, the clinical application of these designs is more complicated. One retrospective study of 150 phase I trials from 2002 to 2004 found that one-third used aggressive dose escalation schemes, which were associated with higher rates of toxicity but no increase in response rates nor greater proportions of patients receiving the recommended phase II dose [6]. In addition, studies with aggressive titration (defined as two successive dose increases of 100%) did not demonstrate a reduction in the number of patients needed to complete the studies and had a higher proportion of patients experienced severe or life-threatening toxicities. A novel study design, “the rolling six,” has been recently proposed [7]. This study was initially proposed as a pediatric phase I study design because of the overall safety profile of pediatric phase I trials, the extended periods of time that studies are suspended to accrual, and the observation that the large majority of dose levels are ultimately expanded to accrue six patients. This design allows for the accrual of 2–6

5  Phase I Clinical Trials with Anticancer Agents

121

patients concurrently onto a dose level. The decision as to which dose level to enroll a patient is based on the number of patients currently enrolled and evaluable, the number of patients experiencing DLTs, and the number of patients still at risk of developing a DLT at the time of new patient entry. One of the pitfalls for this design is that it assumes that MTD and adverse event profiles will be similar in adults and children. However, there have been incidents where children experience a different toxic adverse event profile in adults. For example, children were especially vulnerable to the CNS toxicities of all-­trans-retinoic acid, which was not predictable from the adult phase I data [8]. Thus, in this “rolling six design,” more children could potentially experience adverse effects than in a traditional 3 + 3 study format. Despite all these new trial designs, none have replaced the traditional 3 + 3 design in most phase I studies. However, with the development of “targeted drugs” and their relatively benign toxicity profiles (monoclonal antibodies are an example), novel study designs will need to be considered since these drugs may not have a traditional MTD.

5.3 Selection of Starting Dose and Schedule Before new drugs are tested in humans, animal testing is performed to determine an agent’s antitumor activity, clinical pharmacology, and toxicity profile. Each of these issues must be carefully addressed to provide safe and reasonable starting doses and schedules for phase I trials.

5.3.1 Preclinical Pharmacology Studies Preclinical pharmacology studies in animals provide information about the pharmacokinetic properties of the agent, including bioavailability, absorption, distribution, metabolism, and clearance. These studies, usually performed in nontumor-bearing animals, also provide information about the achievable plasma concentrations. Understanding these pharmacokinetic properties is necessary to rationally select the drug administration schedule to be employed in the initial phase I trial in humans. Historically, most new anticancer agents were tested using two relatively fixed schedules of drug administration: single-bolus intravenous dose every 3–4 weeks and 5 consecutive days of treatment repeated at 3–4 week intervals. The most ­frequently used preclinical pharmacology protocols reflect each of these schedules, which are common for traditional cytotoxic drugs that require a period of recovery time (from myelosuppression, mucositis, etc.) between doses. However, other schedules of administration, including weekly intravenous infusion, continuous intravenous infusion, and continuous oral dosing are now being used, particularly for biologically targeted therapies. Thus, selecting the schedule of administration

122

S. Leong et al.

depends on the pharmacokinetic properties of the agent as well as the proposed mechanism of action, and whether continuous drug exposure is deemed necessary for an anticancer effect. The selected schedule also depends on animal toxicity observed with the proposed schedule of administration.

5.3.2 Preclinical Toxicology Studies Preclinical toxicology studies in animals are essential to characterize the potential adverse effects of an investigational agent. The major objectives of preclinical toxicology studies include defining the organ toxicities including dose and schedule dependencies, identifying the reversibility of these effects, and determining the initial safe starting dose for humans. In general, the ideal approach is to ensure that the preclinical toxicology studies accurately reflect the intended clinical investigations in humans, including the use of identical drug formulation, schedules and routes of drug administration, and dose levels anticipated to reflect the likely experience in patients. Traditionally, toxicology investigations have involved a simplified two-step approach. The initial step focused on acute toxicity in small animals (usually mice or rats), with the primary endpoint being determination of the dose level lethal in 10% of animals (LD10). The second step consisted of more extensive toxicologic assessment, including careful determination of any organ-specific toxicities in animals associated with the schedule and route of administration that was to be used in the initial clinical trial. Because substantial variation may exist between species in tolerance to a given drug, the safety of a projected starting dose in humans is confirmed by examining the preclinical toxicities in at least two animal species (rodent and nonrodent species). Both the qualitative and quantitative toxicities are usually well defined after studies in a small-animal model (such as mice) and a larger animal (such as dogs or rabbits). On occasion, additional testing is needed in a large-animal species such as monkeys. Certain organ-specific toxicities such as myelosuppression and gastrointestinal toxicity are reliably detected with current toxicology models. In contrast, hepatic and renal toxicities are often difficult to assess in animal testing. Toxicities involving the heart, lung, nervous system, pancreas, and skin are even less reliably predicted. Although preclinical studies can help to establish a safe starting dosage for humans and may predict acute organ toxicity, the true safety profile of a new agent emerges only after it has been investigated in the initial phase I study.

5.3.3 Selection of the Phase I Starting Dose In general, selection of the starting dose for the initial phase I clinical trial is based on animal toxicology studies. Even though toxicology studies are performed in

5  Phase I Clinical Trials with Anticancer Agents

123

multiple species prior to initiating the first-in-human phase I clinical trial, uncertainty remains about the potential toxicities for humans. In selecting the starting dose, the guiding principle is one of safety – begin with a safe starting dose, and avoid placing initial study participants at risk of treatment-related adverse events. The traditional method used to select the starting dose is based on the animal LD10, while the more modern method is based on the no observed adverse effect level (NOAEL) in animals. Both methods are described below, with emphasis given to the more modern method. 5.3.3.1 Traditional Method Toxicology studies are typically performed in rodents (usually mice) and another nonrodent large species (usually dogs). In the mice studies, the dose at which approximately 10% of the mice die (the murine LD10) is defined. One-tenth of this murine LD10 (0.1 MLD10), expressed in mg/m2, has historically been a safe starting dose in humans when toxicology studies in a second species (such as dog) did not show substantial differences in the dose–toxicity relationship. Therefore, conventional phase I trial design has been conducted by selecting a starting dose of 0.1 MLD10 or lower [9]. However, for phase I trials, ethical concerns exist about treating excessive numbers of patients at subtherapeutic doses of a new agent. Although the overall response rate in phase I trials is generally low, the majority of responses occur within 80–120% of the recommended phase II dose, at least for cytotoxic agents [10]. These considerations raise ethical pressures to treat fewer patients at the initial dose levels in the absence of toxicity. Increasing the starting dose could potentially reduce the number of patients treated at subtherapeutic doses. However, an important principle in phase I design is the protection of patients from exposure to unacceptable levels of risk (toxicity), so evaluation of a new method of starting dose selection must include not only a measure of its relative efficiency, but also a determination of its relative safety. Furthermore, the primary objective of all phase I trials is to permit determination of the phase II dose. To complete a trial quickly with few patients receiving nontoxic doses is not helpful, if the recommended phase II dose is subsequently shown to be inaccurate. To explore the question of increasing the starting dose based on MLD10, a review of compounds evaluated in phase I trials in the late 1990s was undertaken [9]. Agents selected for review were cytotoxic drugs studied as single agents in initial phase I trials performed to determine the MTD. All published trials of such agents were included, provided their starting dose was based on MLD10 information. With the knowledge of the “true” MTD determined in each trial, the number of doseescalation steps to achieve MTD was calculated based on the actual starting dose of 0.1 MLD10 and theoretical starting doses of 0.2 and 0.3 MLD10. To assure comparability, dose escalation was performed in all cases according to the modified Fibonacci scheme. The objectives of the analysis were to determine whether

124

S. Leong et al.

increasing the starting dose shortened dose escalation and trial length and to assess the safety of the use of higher starting doses. A trial was arbitrarily considered unsafe, if three or fewer dose levels (including the starting dose) were required to reach MTD. This was based on the notion that escalation schemes that reached the MTD in three or fewer steps would occasionally be expected to result in serious toxicity at the starting dose level. Fourteen agents studied in 21 trials met the criteria for inclusion. For this group of agents and trials, a starting dose of 0.1 MLD10 led to a median of seven dose levels to attain the MTD (range, 4–14 dose levels). A starting dose of 0.2 MLD10 yielded a median of five dose levels (range, 3–11 dose levels) to attain the MTD, and when the starting dose level was increased to 0.3 MLD10, a median of three dose levels was required to reach the MTD (range, 2–9 dose levels). If an unsafe trial was defined as three or fewer levels to attain the MTD, then 0 of 21 trials (0 of 14 agents) were considered unsafe with 0.1 MLD10 starting dose, 5 of 21 (2 of 14 agents) were considered unsafe at the 0.2 MLD10, and 11 of 21 (6 of 14 agents) were considered unsafe at the 0.3 MLD10 starting dose. No agent in any of the three starting doses would have entered a phase I trial at a dose level above the MTD [9]. The authors concluded that a starting dose of 0.2 MLD10 may be a reasonable approach to shorten duration of phase I trials and limit the number of patients who are treated at very low doses, when the dose escalation scheme is modified Fibonacci. However, when combined with a more aggressive dose escalation scheme (for example, one patient per cohort and/or 100% increase in initial dose levels), the authors felt that a starting dose of 0.2 MLD10 does not offer a large advantage over the usual starting dose of 0.1 MLD10 in this setting [9]. 5.3.3.2 Modern Method To increase the margin of safety, the current recommended method to determine the starting dose for phase I trials is based on the NOAEL. The US Department of Health and Human Services, Food and Drug Administration (FDA), and Center for Drug Evaluation and Research (CDER) publish Guidance for Industry document for estimating the maximum safe starting dose for first-in-human clinical trials for therapeutics in adult healthy volunteers. This document was most recently published in 2005 and is updated periodically. The guidance outlines a standardized process for deriving the maximum recommended starting dose (MRSD). Although phase I oncology trials typically enroll patient volunteers rather than normal volunteers, many principles and the approaches recommended are applicable in designing such trials. The major elements of this process are the determination of the NOAELs in the tested animal species, conversion of NOAELs to the human equivalent dose (HED) using allometric scaling factors (see Table 5.1), selection of the most appropriate animal species, and application of a safety factor. This method of selecting the starting dose is outlined in Fig. 5.2 and described in detail below.

5  Phase I Clinical Trials with Anticancer Agents

125

Table 5.1  Conversion of animal doses to human equivalent doses To convert animal dose in mg/kg to HED a in To convert animal mg/kg, either dose in mg/kg to Species Humana Mouse Hamster Rat Ferret Guinea pig Rabbit Dog Primates Monkeys Marmoset Squirrel monkey Baboon

HED mg/m2, multiply by km

Divide animal dose by

Multiply animal dose by

37  3  5  6  7  8 12 20

12.3   7.4   6.2   5.3   4.6   3.1   1.8

0.08 0.13 0.16 0.19 0.22 0.32 0.54

12  6  7 20

  3.1   6.2   5.3   1.8

0.32 0.16 0.19 0.54

a Assumes 60 kg human. For species not listed or for weights outside the standard ranges, HED can be calculated from the following formula: HED = Animal Dose In Mg/Kg × (Animal Weight In Kg/Human Weight In Kg)0.33

Step 1: NOAEL Determination The first step in determining the phase I starting dose is to evaluate the ­available animal data so that the NOAEL can be determined for each toxicology study. Several definitions of NOAEL exist, but for selecting a starting dose, the ­following is used: the highest dose level that does not produce a significant increase in adverse effects in comparison to the control group. In addition, adverse effects that are biologically significant (even if they are not statistically significant) should be considered in the determination of the NOAEL. The NOAEL is not the same as the no observed effect level (NOEL), which refers to any effect, not just an adverse one, although in some cases the two might be identical. The definition of the NOAEL, in contrast to that of the NOEL, reflects the view that some effects observed in the animal may be acceptable pharmacodynamic effects of the agent and may not raise a safety concern. The NOAEL should also not be confused with the lowest observed adverse effect level (LOAEL). Findings in preclinical toxicology studies that can be used to determine the NOAEL include overt toxicity (for example clinical signs, macroscopic and microscopic lesions), surrogate markers of toxicity (for example serum liver enzyme levels), and exaggerated pharmacodynamic effects. As a general rule, an adverse effect used to define the NOAEL should be based on an effect that would be unacceptable if produced by the initial dose in a phase I clinical trial. In general, the NOAEL for each toxicology study is reported in mg/kg.

126 Fig. 5.2  General process for the selection of starting dose

S. Leong et al. Step 1 Determine No Observed Adverse Effect Level (NOAEL)s (mg/kg) in animal toxicity studies

Step 2 Convert each animal NOAEL to Human Equivalent Dose (HED)

Step 3 Select HED from most appropriate species

Step 4 Choose safety factor and divide HED by that factor

Maximum Recommended Starting Dose (MRSD)

Step 2: HED Calculation The NOAEL for each species tested should be identified, and then converted to the HED using appropriate scaling factors. For most systemically administered agents, this conversion should be based on the normalization of doses to body surface area (which assumes that doses scale 1:1 between species when normalized to body surface area). The body surface area normalization and the conversion of the animal dose to human dose are done by using the allometric scaling factors listed in Table  5.1 (also see Example Calculations below). The species that generates the lowest HED is called the most sensitive species. Scaling between species based on body surface area is not recommended for therapeutics administered by alternative routes (topical, intranasal, subcutaneous, and intramuscular) for which the dose is limited by local toxicities. Similarly, therapeutics administered into anatomical compartments that have little subsequent distribution outside the compartment (intrathecal, intraocular, and intrapleural) should not be scaled by body surface area. For recombinant proteins with molecular weight >100,000  Da administered intravascularly, the available data often show

5  Phase I Clinical Trials with Anticancer Agents

127

that the NOAEL occurs at a similar mg/kg dose across species. Conversion of the animal NOAEL to the HED should be normalized to mg/kg (see Table 5.1). Step 3: Most Appropriate Species Selection Without any additional information to guide the choice of the most appropriate species for assessing human risk, the most sensitive species is designated the most appropriate, because using the lowest HED would generate the most conservative starting dose. When information indicates that a particular species is more relevant for assessing human risk (and deemed the most appropriate species), the HED for that species may be used in subsequent calculations, regardless of whether this species is the most sensitive. This situation is more applicable to biologic therapies, many of which have high selectivity for binding to human target proteins and limited reactivity in species commonly used for toxicity testing. Factors that could influence the choice of the most appropriate species rather than the default to the most sensitive species include differences in the absorption, distribution, metabolism and excretion of the agent between the species, and class experience that may indicate a particular animal model is more predictive of human toxicity. Class experience implies that previous studies have demonstrated that a particular animal model is more appropriate for the assessment of safety for a particular class of therapeutics. Step 4: Application of Safety Factor A safety factor should then be applied to the HED to increase assurance that the first dose in humans will not cause adverse effects. The use of the safety factor should be based on the possibility that humans may be more sensitive to the toxic effects of a therapeutic agent than predicted by the animal models, that bioavailability may vary across species, and that the models tested do not evaluate all possible human toxicities. These factors can be accommodated by lowering the human starting dose from the HED. The starting dose should be obtained by dividing the HED by the safety factor. In practice, the default safety factor that should normally be used is 10. This is a historically accepted value, but should be evaluated based on available information. For example, the safety factor should be increased for steep dose– response curve, severe toxicities (organ damage and neurological toxicity), nonmonitorable toxicity (histopathologic changes in animals not detected by clinical markers), and toxicities without premonitory signs. If the onset of significant toxicities is not reliably associated with premonitory signs in animals, it may be difficult to know when toxic doses are approached in human trials. Other factors include variable bioavailability, irreversible toxicities, unexplained mortality, variability in doses or drug levels eliciting effects, nonlinear pharmacokinetics, and novel therapeutic targets.

128

S. Leong et al.

Example Calculations for Converting Animal Doses to HEDs Example 1: To convert an animal dose from mg/kg to the HED in mg/m2, the animal dose in mg/kg is multiplied by the conversion factor (km) as listed in the second column in Table 5.1. The km has units of kg/m2 and is equal to the body weight in kg divided by the surface area in m2. To convert a dose of 30 mg/kg in a dog: 30 mg/kg × 20 kg/m2 = 600 mg/m2 in humans To convert a dose of 2.5 mg/kg in a human: 2.5 mg/kg × 37 kg/m2 = 92.5 mg/m2 in humans Example 2: To convert an animal dose from mg/kg to the HED in mg/kg, either divide the animal dose by the ratio of the human/animal km factor (third column in Table 5.1) or multiply the animal dose by the ratio of the animal/human km factor (fourth column in Table 5.1). To convert a dose of 50 mg/kg in rat: 50 mg/kg ÷ 6.2 = 8 mg/kg in humans or 50 mg/kg × 0.16 = 8 mg/kg in humans 5.3.3.3 Other Methods to Select the Starting Dose Other methods used to select the initial starting dose for phase I trials have been published, including pharmacokinetic-based methods [11] and quantitative structure–activity relationship (QSAR) modeling [12]. However, neither of these methods obviates the need for dedicated toxicology animal studies.

5.4 Phase I Evaluation and Endpoints The primary endpoint of phase I studies is to determine a recommended dose and/ or schedule for evaluation in the phase II setting. In order to achieve this goal, the primary objectives of phase I studies generally are to: • • • •

Evaluate the safety of the drug or combination of drugs Describe the toxicity profile Identify the maximum tolerated dose (MTD) Describe the pharmacokinetics of the drug(s)

However, phase I studies are now incorporating additional objectives to: • Investigate laboratory and ancillary correlative studies including pharmacodynamics markers • Describe preliminary evidence of antitumor activity

5  Phase I Clinical Trials with Anticancer Agents

129

This section will focus on the evaluation of the safety and toxicity, identification of the MTD and evaluating antitumor activity in phase I clinical trials. The pharmacokinetic and pharmacodynamic in early drug development will be discussed in later chapters.

5.4.1 Reporting of Toxicities When describing toxicities, a standard terminology and grading system, known as the National Cancer Institute (NCI) Common Terminology Criteria (CTC), is utilized. The original NCI CTC was developed in 1982 because of the need of formal toxicity reporting and to document severity in the evaluation of new treatments and modalities, which were lacking in the studies at that time. In 1997, the Cancer Therapy and Evaluation Program (CTEP) with representatives from the pharmaceutical industry, FDA, the Committee for Proprietary Medicinal Products (CPMP), and the major clinical trial groups in the USA, Canada, Europe, and Japan convened to improve accuracy, precision, and completeness of the CTC and to standardize reporting. A year later, CTC v2.0 was published with 28 Categories and over 250 AEs associated with a grading scale for reporting severity. The CTC v2.0 became the worldwide standard “library” for reporting acute AEs in cancer therapy clinical trials. In March 2003, CTEP NCI released the third version, known as Common Terminology Criteria for Adverse Events (CTCAE). The CTCAE includes more than 1,000 terms with improved anatomic site specificity and expanded criteria for surgical effects. The CTCAE represents the first comprehensive grading system for reporting both acute and late effects in oncology and was the first attempt to cover AEs associated with all therapeutic interventions, including radiation and surgery.

5.4.1.1 Adverse Events The CTCAE describes an adverse event (AE) as any unfavorable and unintended sign (including an abnormal laboratory finding), symptom, or disease temporally associated with the use of a medical treatment or procedure that may or may not be considered related to the medical treatment or procedure. An AE is a term that is a unique representation of a specific event used for medical documentation and scientific analyses. Each AE term is mapped to a MedDRA term and code. The Medical dictionary for regulatory activities (MedDRA) is a comprehensive medical terminology system used for regulatory reporting and drug labeling. MedDRA does not provide severity ranking of AEs. Examples of adverse events include but are not limited to: • Abnormal test findings • Clinically significant symptoms and signs • Changes in physical examination findings

130

S. Leong et al.

• Hypersensitivity • Progression/worsening of underlying disease Additionally, they may include the signs or symptoms resulting from: • • • • • • • •

Drug overdose Drug withdrawal Drug abuse Drug misuse Drug interactions Drug dependency Extravasation Exposure in utero

Adverse events outlined in the CTCAE v3.0 are graded based on a five-point scale, which generally corresponds to mild, moderate, severe, life threatening, and death. Table  5.2 is an example of how an adverse event is graded, in this case nausea.

5.4.1.2 Serious Adverse Events A serious adverse event (SAE) or serious adverse drug reaction is any untoward medical occurrence at any dose that: • • • • •

Results in death Is life-threatening (immediate risk of death) Requires inpatient hospitalization or prolongation of existing hospitalization Results in persistent or significant disability/incapacity Results in congenital anomaly/birth defect

Medical and scientific judgment should be exercised in determining whether an event is an important medical event. An important medical event may not be immediately life threatening and/or result in death or hospitalization. However, if it is determined that the event may jeopardize the subject and may require intervention to prevent one of the other outcomes listed in the definition above, the important medical event should be reported as serious. Examples of such events are intensive treatment in an emergency room or at home for allergic bronchospasm; blood dyscrasias or convulsions that do not result in hospitalization; or development of drug dependency or drug abuse. Adverse events reported from clinical trials associated with hospitalization or prolongation of hospitalization are all considered serious. Any initial admission (even if less than 24 h) to a healthcare facility meets these criteria. Admission also includes transfer within the hospital to an acute/intensive care unit (e.g., from the psychiatric wing to a medical floor, medical floor to a coronary care unit, neurological floor to a tuberculosis unit).

5  Phase I Clinical Trials with Anticancer Agents Table 5.2  Adverse event Nausea

Common terminology criteria for adverse events v3.0 grading of nausea MedDRA term Grade 1 Grade 2 Grade 3 Grade 4 Inadequate Life-threatening Oral intake Nausea Loss of oral caloric consequences decreased appetite or fluid without without intake, IV significant alteration in fluids, tube eating habits weight loss, dehydration, feeding, or TPN or malnutrition, indicated ³24 h IV fluids indicated <24 h

131

Grade 5 Death

MedDRA Medical dictionary for regulatory activities

Hospitalization does not include the following: • • • • • • •

Rehabilitation facilities Hospice facilities Respite care (e.g., caregiver relief) Skilled nursing facilities Nursing homes Routine emergency room admissions Same day surgeries (as outpatient/same day/ambulatory procedures)

Hospitalization or prolongation of hospitalization in the absence of a precipitating, clinical adverse event is not in itself an SAE. Examples include: • Admission for treatment of a preexisting condition not associated with the development of a new adverse event or with a worsening of the preexisting condition (e.g., for work-up of persistent pretreatment lab abnormality) • Social admission (e.g., subject has no place to sleep) • Administrative admission (e.g., for yearly physical exam) • Protocol-specified admission during a clinical trial (e.g., for a procedure required by the trial protocol) • Optional admission not associated with a precipitating clinical adverse event (e.g., for elective cosmetic surgery) • Preplanned treatments or surgical procedures should be noted in the baseline documentation for the entire protocol and/or for the individual subject • Diagnostic and therapeutic noninvasive and invasive procedures, such as surgery, should not be reported as adverse events. However, the medical condition for which the procedure was performed should be reported, if it meets the definition of an adverse event. For example, an acute appendicitis that begins during the adverse event reporting period should be reported as the adverse event, and the resulting appendectomy should be recorded as treatment of the adverse event.

132

S. Leong et al.

5.4.2 Radiographic Evaluation Antitumor activity is not a primary endpoint of a phase I trial; however, most phase I clinical trials require patients to have evaluable or measurable disease by a radiological exam or a tumor marker. Response Evaluation Criteria in Solid Tumors (RECIST) was created in 2000 by an international committee, which provided standardized guidelines for measuring tumor response using X-ray, CT, and/or MRI. RECIST was based on simplifying former evaluation methods such as WHO and ECOG. The RECIST criteria are recommended but not mandatory for NCI sponsored trials and involves formal rules for the measurement of tumor lesions. RECIST classifies lesions as measurable and nonmeasurable. Measurable lesions: lesions that can be accurately measured in at least one dimension with longest diameter ³20  mm using conventional techniques or ³10 mm with spiral CT scan. These lesions are also known as “target lesions.” Nonmeasurable lesions: all other lesions, including small lesions (longest diameter <20 mm with conventional techniques or <10 mm with spiral CT scan), i.e., bone lesions, leptomeningeal disease, ascites, pleural/pericardial effusion, inflammatory breast disease, lymphangitis cutis/pulmonis, cystic lesions, and also abdominal masses that are not confirmed and followed by imaging techniques. These lesions are also known as “nontarget lesions.” RECIST criteria combines the assessment of all lesions both target and nontarget lesions into one of four categories. • CR (complete response) = disappearance of all target lesions • PR (partial response) = 30% decrease in the sum of the longest diameter of target lesions • PD (progressive disease) = 20% increase in the sum of the longest diameter of target lesions • SD (stable disease) = small changes that do not meet above criteria With the discovery of new imaging techniques and its potential use to evaluate anticancer therapeutics, the NCI has developed guidelines for the use of DCE-MRI, magnetic resonance spectroscopy and fluorodeoxyglucose PET (FDG PET) in clinical trials of anticancer therapeutics [13, 14].

5.4.3 Correlative Studies Because most current anticancer compounds have been designed to selectively inhibit molecular targets, there is considerable interest in assessing their cellular and subcellular effects in the clinical setting. Conceptually, assays of drug action in vivo can serve two purposes. In early clinical trials, these assays can determine whether the drug target has been inhibited at drug concentrations that are achievable in the clinical setting. In later clinical trials, these assays could potentially provide an early marker of drug efficacy if a strong correlation between assay

5  Phase I Clinical Trials with Anticancer Agents

133

results and clinical outcome can be established. More importantly, assays could improve the efficacy of a given compound by selecting the patient population that is likely to respond to therapy. In the phase I context, these are the critical issues to address as one designs biomarker or other correlative studies. Identify the specific question to be addressed • Pharmacokinetic (What is the right dose and schedule?) • Pharmacodynamic (Does the drug inhibit a known target?) • Pharmacodiagnostic (Is there a way to identify a specific patient subpopulations that will benefit from this treatment?) Determine what is needed to measure biomarker • • • •

Tissues (Can needed tissue be obtained safely?) Timing (Can tissues be obtained at the right time point?) Assay methodology (Is a suitable assay available?) Patients (In what patients and in how many?)

These topics are discussed in depth in subsequent chapters.

5.4.4 The Concept of Optimal Biologic Dose for Novel, Nontoxic Agents Advances in molecular biology have led to a new generation of anticancer agents that inhibit aberrant and cancer-specific proliferative and antiapoptotic pathways. These agents may be cytostatic and may produce relatively minimal organ toxicity compared with standard cytotoxics. This has fueled interest in alternatives to toxicity as a surrogate endpoint in phase I trials. The concept of an “optimal biologic dose” defined as a dose that reliably inhibits a drug target or achieves a target plasma concentration, is seen as desirable and appropriate for the phase I study of mechanism-based, relatively nontoxic novel agents. This idea is appropriate, if certain inherent problems can be resolved. In the case of a pharmacokinetic endpoint, it has to be shown that the target concentration chosen can inhibit the drug target in patient tumors. This requires accounting for plasma protein binding, which determines the amount of free drug available to interact with the target, as well as interindividual variations in drug absorption and metabolism. When target modulation is chosen as the endpoint, the drug target, as well as the magnitude of inhibition necessary for clinical benefit, has to be known. Finally, while target inhibition in normal tissue may provide important supplementary information, critical drug development decisions will need to be made with information gleaned from target suppression in tumor samples. An optimal biologic dose should inhibit the target in patient tumors. Most importantly, there should be absolute certainty of the drug target, and there should be evidence that modulating the target in tumors consistently leads to growth inhibition. The selected dose should incorporate the fact that there will be wide variations in steady-state drug levels in patients.

134

S. Leong et al.

Having outlined these issues, how does one select a phase II dose of a drug with minimal dose-dependent organ toxicities? Apart from immunotherapeutic agents, it is debatable if many such agents currently exist. Most small molecule inhibitors of cellular proteins will demonstrate chronic low-grade toxicities at high doses that preclude continuous dosing. Thus, the concept of MTD may need to be redefined as a dose that can be safely administered chronically. With this definition, an MTD can be established for most drugs. In the rare case of truly minimally toxic agents, the optimal dose may be defined by saturation in absorption, quantity of tablets to be ingested or volume of drug to be infused, and other practical issues that would preclude dose escalation. In a review of 60 phase I studies of targeted agents, Parulekar and Eisenhauer found that the optimal biologic dose rarely formed the basis of dose selection [15]. Ultimately, until studies are able to incorporate modulation of validated drug targets in tumors and utilize information on allelic variants of polymorphic genes responsible for drug transport and metabolism to select doses for individual patients, efforts have to be made to define the MTD of phase I agents based on classical definitions of first cycle toxicity, as well as the feasibility and tolerability of chronic administration.

5.5 Ethical Considerations of Phase I Oncology Trial Ethical controversy surrounding phase I oncology trials has existed and been discussed in the medical literature for the past several decades, as well as in more recent publications [16–19]. The specific ethical issues frequently discussed include the nature of the scientific objectives of phase I oncology trials (whether or not there is a therapeutic intent), the risk–benefit ratio and the likelihood that participants will experience direct personal benefit, and the validity of the informed consent process for terminally ill cancer patients. These issues are discussed further in the respective sections below.

5.5.1 Therapeutic Intent The primary objectives of phase I trials are to determine the appropriate dose and schedule for phase II testing, as well as to define an agent’s preliminary toxicity profile. Although phase I trials also seek evidence of tumor response, they are not designed to draw conclusions about clinical benefit. Critics charge that since phase I trials are dose-finding and safety studies, they are by definition nontherapeutic trials and offer no intended direct patient benefit [18]. However, a distinction can be made between therapeutic intent and the prospect of direct participant benefit – that is, distinguishing the objectives or intent of a phase I study from the possibility (or probability) of direct patient benefit from the study treatment [16, 18]. For example, even though therapeutic benefit may not be explicitly stated in the study objectives, phase I trials offer each participant a prospect of direct benefit,

5  Phase I Clinical Trials with Anticancer Agents

135

as all patients enrolled in a phase I study receive treatment with the investigational agent (albeit at different doses). In contrast, some efficacy studies with the explicit objective of determining therapeutic benefit, such as randomized placebo-controlled trials, may not offer even a prospect of direct benefit, at least for a subset of participants (those receiving placebo). Therefore, it has been argued that phase I trials are neither more nor less therapeutic than other phases of trials [18]. Determining preliminary evidence of efficacy is now an explicit secondary objective for many phase I oncology trials.

5.5.2 Risk–Benefit Ratio Since phase I oncology trials involve the initial administration of new anticancer agents to humans based on apparent safety and potential efficacy from preclinical animal studies, the precise absolute risks and benefits for an individual participant enrolling in a particular phase I study may be largely unknown. However, recent analyses have been performed in an attempt to quantify the risk–benefit ratio for cancer patients participating in phase I trials [1, 20]. Roberts et  al. searched abstracts and journal articles reporting the results of phase I cancer treatment trials originally submitted to annual meetings of the American Society of Clinical Oncology (ASCO) from 1991 to 2002 [1]. The analysis was limited to published studies of single agents not yet approved by the US FDA for any indication at the time of ASCO submission. Trials that included radiotherapy or that enrolled patients with hematological malignancies (e.g., leukemia and lymphoma) were excluded. The analysis included 213 studies involving 6,474 cancer patients. The overall toxic death rate was 0.54%, while the overall objective response rate (both complete and partial responses) was 3.8%. Serious treatmentrelated toxicity (defined as grade 3 or 4 toxic events) occurred in 10.3% of patients. The authors noted that the toxic death rate decreased over the study period, from 1.1% from 1991 to 1994 to 0.06% from 1999 to 2002. Although the overall objective response rate also decreased during the study period, the magnitude was less compared with the decrease in the toxic death rate. Therefore, the authors concluded that the ratio of benefit to risk may have improved from 1991 to 2002 for patients enrolled to phase I oncology trials during that time [1]. Another analysis was reported by Horstmann et al., who analyzed 460 non-pediatric phase I oncology trials sponsored by the Cancer Therapy Evaluation Program at the NCI between 1991 and 2002. The trials analyzed included 11,935 participants, all of whom were assessed for toxicity and 10,402 of whom were assessed for a response to therapy. The overall rate of death due to toxic events was 0.49%, while approximately 14% of patients experienced at least one grade 4 toxic event. The overall response rate was 10.6% with considerable variation among trials. The overall response rate was 4.4% for trials with single investigational agents, 17.8% for combination studies that included at least one anticancer agent approved by the FDA, and 27.4% for studies with only FDA-approved chemotherapeutic agents. An additional 34.1% of the participants had stable disease or a less than partial response.

136

S. Leong et al.

The overall likelihood of a tumor response was about 20-fold greater than the likelihood of a drug-related toxic death. Given the variability of overall response rate among the different studies, determining the risk–benefit ratio for an individual patient may depend, at least in part, on the specific phase I trial (for example, whether the trial includes an anticancer agent already approved by the FDA). Under standard phase I trial design, investigators enroll successive cohorts of patients at increasing doses of an experimental therapy with the primary goal of determining DLT and then backing down to a dose appropriate for phase II testing. An analysis of response rates and dose–response effects demonstrated that, at least for traditional cytotoxic agents, most responses seen in phase I oncology trials occur between 80 and 120% of the maximum-tolerated dose [10]. This analysis suggests that the probability of benefit may partly depend on where in the doseescalation scheme the study participant enrolls (low-dose levels versus high-dose levels). This dose–response effect has been recognized and alternative phase I trial designs have been proposed, specifically to decrease the proportion of participants exposed to subtherapeutic doses. However, as dose escalation proceeds, while the chance of benefit may increase so may the risk of toxicity. Therefore, for the individual patient considering participation in a phase I study, the risk–benefit ratio may change over the course of the trial, as higher doses are reached. Another factor in determining the risk–benefit ratio for an individual patient is where in the patient’s anticancer treatment journey phase I trials are introduced. Ninety percent of agents studied in phase I trials fail to gain FDA approval, most commonly due to a lack of efficacy [18]. Therefore, if a cancer patient was to consider a phase I trial of a single experimental agent as initial therapy, the risk–benefit ratio would likely be unfavorable compared to the FDA-approved standard treatment or to participation in a tumor type-specific phase II or III clinical trial. However, in many cases, patients are referred for phase I studies when their only other option is hospice care, in which case the risk–benefit ratio often becomes more favorable in the minds of patients and physicians, even though the absolute chance of benefit may be small. Furthermore, some individuals may choose to participate in phase I oncology trials on altruistic grounds, even if the risk–benefit ratio they achieve is unfavorable. 5.5.2.1 Informed Consent Whereas phase I trials in other areas of medicine enroll healthy participants, phase I trials in oncology typically enroll patients who have terminal cancer and who have exhausted standard treatment options. Some have argued that these may be vulnerable patients at the end of life, and question the informed consent process in which patients choose to participate in experimental research with a very low chance of clinical benefit [16, 21]. Indeed, there is some evidence that phase I study participants commonly misconstrue the purpose of and overestimate the benefits from phase I trials, thus raising doubts about the validity of consent. Daugherty et al. administered a survey to 144 phase I oncology trial participants

5  Phase I Clinical Trials with Anticancer Agents

137

in order to obtain empiric information about patient motivation and the informed consent process [21]. Regarding patient motivation, 73% of respondents stated the main reason for phase I trial participation was to seek an anticancer response, while 17% stated the main reason for participation was to improve quality of life (decrease pain or other cancer-related symptoms). Thus, 90% of participants reported that direct personal benefit was the main reason for trial participation. In addition, while the vast majority of participants (93%) reported understanding most or all of the information given to them about the phase I study in which they had agreed to participate, only 31% were able to state accurately the purpose of phase I studies as dose finding [21]. One explanation of this apparent discrepancy requires separating participant understanding of the purpose of a phase I study from their personal motivations to enroll and their understanding of the probability of benefit. A participant’s inability to state the purpose of a phase I study as dose finding may reflect that patients care more about the probability of receiving benefits and the requirements of the study than about scientific methodology or the researcher’s intent in conducting the study [16]. This interpretation is supported by data from a study by Joffe et al., in which a standard questionnaire was sent to 287 adult patients with cancer who had recently enrolled in a clinical trial [22]. This study showed that while 84% of participants reported reading the consent form carefully and 73% considered it an important source of information, only 37% considered the consent form itself important in their decision to participate in the study [22]. In addition, when asked about the potential benefits to themselves, 75% reported that the main reason why cancer clinical trials are carried out is to improve the treatment of future cancer patients, and 71% acknowledged that they may not receive direct medical benefit from participation in the clinical trial [22]. Another issue relates to the adequacy of the informed consent document itself. One study evaluated the substantive content of the informed consent document for 272 phase I oncology studies, and found that 99% explicitly stated that the study was research and that in 86% this statement was prominent [23]. Furthermore, 92% of consent forms indicated that safety testing was the research goal. Overall, an average of 35 lines were dedicated to potential risks in contrast with an average of four lines dedicated to potential benefits. In addition, 67% of the forms mentioned death as a potential consequence of participation in the study, while only 5% mentioned cure as a possible benefit [23]. Only one consent form indicated that any benefits were expected.

5.6 Conclusions The overall purpose of all anticancer research is to develop improved methods to detect, diagnose, and treat all types of cancer. In the development of new anticancer drugs, phase I trials constitute the critical first step of human testing. The primary endpoint of phase I studies is to determine the optimal dose and/or schedule for evaluation in the phase II setting. A number of different conservative and ­aggressive

138

S. Leong et al.

dose escalation schemes may be employed depending on the expected clinical properties of the agent(s) tested. With the development of cancer selective ­compounds, i.e., “targeted” agents, novel phase I designs incorporating pharmacodynamic endpoints and other correlative studies are become more common. Although they are complicated ethical issues inherent to phase I studies, careful attention to consent discussions and forms help ensure that study subjects have an adequate understanding of the risks and goals of early clinical testing.

References 1. Roberts TG, Jr., Goulart BH, Squitieri L, et al: Trends in the risks and benefits to patients with cancer participating in phase 1 clinical trials. JAMA 292:2130–40, 2004 2. Simon R, Freidlin B, Rubinstein L, et al: Accelerated titration designs for phase I clinical trials in oncology. J Natl Cancer Inst 89:1138–47, 1997 3. O’Quigley J, Pepe M, Fisher L: Continual reassessment method: a practical design for phase 1 clinical trials in cancer. Biometrics 46:33–48, 1990 4. Chevert S: The continual reassessment method in cancer phase I clinical trials: a simulation study. Stat Med 12:1093–108, 1993 5. Goodman SN, Zahurak ML, Piantadosi S: Some practical improvements in the continual reassessment method for phase I studies. Stat Med 14:1149–61, 1995 6. Koyfman SA, Agrawal M, Garrett-Mayer E, et al: Risks and benefits associated with novel phase 1 oncology trial designs. Cancer 110:1115–24, 2007 7. Skolnik JM, Barrett JS, Jayaraman B, et al: Shortening the timeline of pediatric phase I trials: the rolling six design. J Clin Oncol 26:190–5, 2008 8. Mahmoud HH, Hurwitz CA, Roberts WM, et  al: Tretinoin toxicity in children with acute promyelocytic leukaemia. Lancet 342:1394–5, 1993 9. Eisenhauer EA, O’Dwyer PJ, Christian M, et al: Phase I clinical trial design in cancer drug development. J Clin Oncol 18:684–92, 2000 10. Von Hoff DD, Turner J: Response rates, duration of response, and dose response effects in phase I studies of antineoplastics. Invest New Drugs 9:115–22, 1991 11. Mahmood I: A Bayesian approach for the estimation of pharmacokinetic parameters in children. Am J Ther 10:88–92, 2003 12. Contrera JF, Matthews EJ, Kruhlak NL, et al: Estimating the safe starting dose in phase I clinical trials and no observed effect level based on QSAR modeling of the human maximum recommended daily dose. Regul Toxicol Pharmacol 40:185–206, 2004 13. http://imaging.cancer.gov/clinicaltrials/guidelines 14. Shankar LK, Hoffman JM, Bacharach S, et  al: Consensus recommendations for the use of 18F-FDG PET as an indicator of therapeutic response in patients in National Cancer Institute Trials. J Nucl Med 47:1059–66, 2006 15. Parulekar WR, Eisenhauer EA: Phase I trial design for solid tumor studies of targeted, noncytotoxic agents: theory and practice. J Natl Cancer Inst 96:990–7, 2004 16. Agrawal M, Emanuel EJ: Ethics of phase 1 oncology studies: reexamining the arguments and data. JAMA 290:1075–82, 2003 17. Freireich EJ: Ethical considerations in cancer chemotherapy. Annu Rev Pharmacol Toxicol 19:547–57, 1979 18. Joffe S, Miller FG: Rethinking risk-benefit assessment for phase I cancer trials. J Clin Oncol 24:2987–90, 2006 19. Lipsett MB: On the nature and ethics of phase I clinical trials of cancer chemotherapies. JAMA 248:941–2, 1982

5  Phase I Clinical Trials with Anticancer Agents

139

20. Horstmann E, McCabe MS, Grochow L, et al: Risks and benefits of phase 1 oncology trials, 1991 through 2002. N Engl J Med 352:895–904, 2005 21. Daugherty CK, Banik DM, Janish L, et al: Quantitative analysis of ethical issues in phase I trials: a survey interview of 144 advanced cancer patients. IRB 22:6–14, 2000 22. Joffe S, Cook EF, Cleary PD, et al: Quality of informed consent: a new measure of understanding among research subjects. J Natl Cancer Inst 93:139–47, 2001 23. Horng S, Emanuel EJ, Wilfond B, et al: Descriptions of benefits and risks in consent forms for phase 1 oncology trials. N Engl J Med 347:2134–40, 2002

Chapter 6

Phase II Trials with Anticancer Agents Hui K. Gan, J. Jack Lee, and Lillian L. Siu

6.1 Introduction Following the determination of drug pharmacology, tolerability, and maximum tolerated dose (MTD) in Phase I trials, investigational drugs usually proceed to Phase II trials. As succinctly summarized by the FDA (21 CFR 312.21) [1], these trials are “conducted to (a) evaluate the effectiveness of the drug for a particular indication or indications in patients with the disease or condition under study and (b) determine the common short-term side effects and risks associated with the drug. Phase II studies are typically well controlled, closely monitored, and conducted in a relatively small number of patients, usually not more than several hundred subjects.” Phase II trials act as a filter, helping to determine which of the investigational drugs emerging from Phase I testing are sufficiently effective to warrant further assessment in definitive Phase III trial. While seemingly straightforward, it is important that Phase II trials be prospectively designed and executed, with appropriate statistical and clinical methodology, rather than being a report of an ad hoc treatment that was subsequently found to be of interest. Furthermore, the appropriate reporting of the statistical design of a Phase II trial is as important as the reporting of its results, as it is difficult to judge the validity of the latter without knowing the former [2]. Two recent reviews have shown that only 23–35% of Phase II trials reported their statistical design adequately [3, 4]. In one of these reviews [3], those studies that were well designed and well reported were significantly more successful than their counterparts, as reflected by faster completion (on average, by 1 year) and greater likelihood of publication in higher impact journals [3]. Data such as these, emphasizing that good Phase II trial design results in improved clinical results and easier publication, may be starting to influence clinical trial conduct as both reviews also found that the levels of appropriate statistical design and reporting progressively increased over the course of the 1990s [3, 4]. L.L. Siu (*) Division of Medical Oncology and Hematology, Princess Margaret Hospital, University of Toronto, 610 University Avenue, Suite 5-718, Toronto, ON, Canada, M5G 2M9 e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_6, © Springer Science+Business Media, LLC 2011

141

142

H.K. Gan et al. Table 6.1  Factors influencing the design of Phase II trials Major variables in the design of Phase II trials Endpoint   Objective clinical response or disease control   Time-to-event based endpoints, e.g., time to progression   Toxicity   Biomarker response   Pharmacological response Number of study arms   Single-arm studies   Multiple-arm studies (randomized and nonrandomized) Statistical framework   Hypothesis testing   Estimation   Ranking and selection   Frequentist or Bayesian paradigm   Decision theoretic approach   Univariate or bivariate testing Number of stages   One stage (early termination not possible)   Multiple stages (early termination for lack of efficacy or promising efficacy possible)   Continuous monitoring Characteristics of investigational agents   Single agent vs. combination of agentsa   Chemotherapy vs. targeted therapies vs. others Adapted from [3, 5] a  These can include any combination of chemotherapy, targeted agents, radiotherapy, surgery, or other anticancer treatment

6.2 Factors Influencing the Design of Phase II Trials Several types of Phase II trials are commonly described. Which particular type is ultimately chosen depends on a number of factors (see Table 6.1), and a decision regarding the study design should be made before the commencement of the study. Several recent reviews of published Phase II studies found that 92–96% of studies were single-armed studies (with the remainder being randomized studies) [3, 5], 79–89% had objective response as the primary criterion [3, 6], 50–71% were multicenter studies [3, 5], and 85% employed chemotherapy (of which 57% employed drug combinations) [3]. The average study enrollment was 39–52 patients [3, 5].

6.3 Endpoints in Phase II Trials 6.3.1 Objective Response Rate Objective response is the most commonly used endpoint in Phase II trials, whether it be with chemotherapy or newer targeted agents [3, 6, 7]. There are several guidelines for assessing clinical response. The Response Evaluation Criteria in Solid

6  Phase II Trials with Anticancer Agents

143

Tumors (RECIST) guidelines are probably the most commonly used guidelines in clinical trials [8, 9]. In the original RECIST 1.0 guidelines, responses are based upon the sum of the maximal diameter of measurable lesions (³20  mm with ­conventional CT or ³10 mm with spiral CT) and the behavior of nonmeasurable lesions (including bone lesions, leptomeningeal disease, ascites, pleural/pericardial effusions, inflammatory breast cancer, lymphangitis cutis/pulmonis, cystic lesions, and tumor markers). All measurable lesions up to a maximum of five lesions per organ and ten lesions in total, representative of all involved organs, should be identified as target lesions, recorded, and measured at baseline. For target lesions, a complete response requires the disappearance of all lesions; partial response is indicated by a 30% decrease in the sum of the longest diameters of target lesions; progressive disease is indicated by a 20% increase in the sum of the longest diameters of target lesions or the appearance of new lesions; and stable disease is composed of all other cases. Confirmation of complete or partial response requires repeat assessment at least 4 weeks after the initial determination of response. For nontarget lesions, a complete response requires the disappearance of all nontarget lesions and normalization of tumor marker levels; an partial response or stable disease requires the persistence of one or more nontarget lesion(s) and/or the maintenance of tumor marker level above the normal limits; and progressive disease requires the appearance of one or more new lesions and/or unequivocal progression of existing nontarget lesions. An update to RECIST has now been published (RECIST 1.1) [10]. Key changes include a reduction in the maximum number of assessable lesions to five in total (and two per organ); lymph nodes with a short axis of ³15 mm are now considered measurable and assessable as target lesions; lymph nodes which are included as assessable lesion at baseline will be considered to have had a complete remission to treatment if their short axis shrinks to <10 mm; confirmation of complete or partial response is not required in randomized trials since the control arm serves as an appropriate means of interpreting the data; finally, the definition of disease progression now requires a 20% increase in the sum of the tumor measurements and an absolute increase of 5 mm. RECIST 1.1 also provides further guidance regarding what constitutes “unequivocal” progression of nonmeasurable/nontarget lesions and regarding the interpretation of tumor response with newer imaging techniques such as FDGPET. While RECIST is based on one-dimensional measurement, an alternate method for categorizing clinical responses to treatment is the World Health Organization (WHO) criterion which is based on two-dimensional tumor measurements [11, 12].

6.3.2 Toxicity Toxicity is often another endpoint of interest in Phase II trials because it offers the opportunity to assess adverse events in a more homogeneous patient population over a longer treatment duration, compared with Phase I trials. One common system of assessing toxicity is the National Cancer Institute Common Terminology Criteria for Adverse Events (http://evs.nci.nih.gov/ftp1/CTCAE/CTCAE_4.03_2010-06-14_ QuickReference_8.5x11.pdf), version 4.0). Although it is beyond the scope of this

144

H.K. Gan et al.

chapter to discuss such a comprehensive assessment tool, suffice to say that toxicity is often graded on a scale of 0–5 where toxicities graded as 1–2 are mild to moderate, 3–4 are severe to life-threatening or disabling toxicities, and 5 is death. Many clinicians focus on the frequency of grade 3–5 toxicities to gauge the tolerability of an investigational new agent.

6.3.3 Disease Progression Progression-based endpoints are becoming increasingly common in Phase II trials. While a review of trials in breast cancer suggested that 4.1% of trials would have used some sort of progression endpoint [3], more recent reviews of randomized Phase II trials and trials of targeted agents show that the use of such endpoints (either alone or in combination with other endpoints) had increased to 9% [4] and 29%, respectively [7]. Progression-based endpoints may be particularly relevant where rapid or substantial changes in tumor size are not expected. Examples of such situations are where tumors are relatively slow growing (e.g., adenoid cystic carcinoma) or where treatment is with the newer targeted agents, many of which are cytostatic rather than cytocidal [13]. There also remains debate about which measure of progression is most appropriate, with some being more in favor of time-to-event assessment [e.g., time-to-progression, progression-free survival (PFS)] [14] while others prefer this is to be a percentage of a binary measurement at a prespecified time (e.g., the percent of patients who are progression free at 6 months) [15]. Specifically, PFS is the length of time from randomization or treatment to objective disease progression or death from any cause. Disease-free survival (DFS) is similar but is defined as the time from being disease free to disease recurrence or death from any cause. Time to failure is defined as the time from randomization or treatment to progression, recurrence, toxicity, or any other reason for discontinuation of treatment. Certain issues may arise during the use of time-to-event endpoints in Phase II trials. The appropriate management of patients who are lost to follow-up or who suffer substantial deviations from the planned therapy needs to be considered and reported. To avoid selection bias, analysis based on the entire intent-to-treat population is preferable to analysis based on certain subsets. However, for Phase II trials, it is not uncommon to define evaluability based on patients receiving at least a minimum number of treatment cycles. Furthermore, the time until event may be influenced by the frequency of assessments, e.g., PFS will seem longer in a trial where assessments are done every 4 months as compared to a trial where assessments are done every 2 months, as detection of progression will be delayed in the former. It is therefore important that, wherever possible, the frequency of assessments be identical across the different arms of a trial or when comparing to historical control data. Detection of progression between assessments is also a problem, especially if treatment is not blinded. Finally, a certain degree of interpretation is required to determine whether the increase in tumor size is sufficient to be ­considered progression. To address this latter issue, the use of independent reviewers blinded to treatment assignment is important.

6  Phase II Trials with Anticancer Agents

145

6.3.4 Other Endpoints Other endpoints which have been occasionally reported in Phase II trials include the change in tumor marker levels (7%) and overall survival (4%) [6].

6.4 Phase II Trials Based on the Hypothesis-Testing Framework The vast majority of Phase II trials are based upon the hypothesis-testing statistical framework so these trials will be considered first. The central tenet of this framework is to establish a null hypothesis (H0) and an alternate hypothesis (H1) and then gather sufficient data to decide between these two hypotheses.

6.4.1 Single-Stage Phase II Trials Arguably the simplest Phase II trial design using the hypothesis-testing framework is the single-stage Phase II trial. A trial is prospectively designed to recruit a group of patients for treatment with the investigational agent. Once the proper sample size has been calculated, the trial recruits patients until such time as the requisite number of patients has been enrolled and evaluated. After study closure, the single-stage Phase II trial then examines the data from all patients in the trial in order to test the hypothesis:

H 0 : p < p0 vs. H1 : p > p1 where the null hypothesis (H0) is that the drug is ineffective because the response rate observed in the study (p) is less than a prespecified value (p0) and therefore not worthy of further investigation. The alternate hypothesis (H1) is that the drug shows interesting efficacy (i.e., p > p1) and should therefore proceed to further investigation, with p1 being some prespecified level of activity that is clinically desirable. The definition of H0 depends on the clinical context. Where a very low response rate can be reliably expected (e.g., second- or third-line treatment in a tumor with few therapeutic options such as pancreatic cancer) then p0 is conventionally set at 0.05 or 0.10 [6]. Otherwise, the definition of H0 should be based on published clinical data, which should be explicitly stated and referenced [6]. Unfortunately, this was only the case in 49% of published studies in one recent review [6]. A major consideration aside from the values assigned to p0 and p1 is the acceptable risk of a false-positive trial (a or type I error) and of a false-negative trial (b or type II error). In the context of a clinical trial, a is the risk that an ineffective agent is considered to be effective and proceeds to a Phase III trial whereas b is the risk that an effective

146

H.K. Gan et al.

agent is rejected as being ineffective and development is ­prematurely terminated. If stringent a (e.g., 0.05) and b values (e.g., 0.1) are desired, then a larger sample size may be required. Some have argued that minimization of b (i.e., the risk of failing to develop an effective agent) is more important than minimizing a (i.e., the risk of developing an agent with minimal activity) as the latter problem will be detected at the Phase III stage [16]. Hence, some trials may allow an a rate as high as 0.2 while continuing to restrict the b rate to 0.1 in an attempt to reduce the sample sizes required. Once the values of p0, p1, a, and b have been specified, it is straightforward to calculate the minimum sample size for a single-stage Phase II trial under the constraints required. The power of the trial can also be derived from these values and it is defined as 1 − b. Assuming that b = 0.1, the power of the trial is 0.9. This means that the trial has a 90% chance of detecting an active agent (H1: p > p1) if the agent is in fact active. The reader is referred to Chap. 1 for further discussion of these concepts. Note that rejection of H0 does not automatically imply that H1 is true, i.e., there are situations where p0 < p < p1 [17]. When the sample size is determined based on these constraints, it only guarantees that type I and type II errors will not exceed a and b under H0 and H1, respectively. The hypothesis-testing framework described above corresponds to a one-sided test, which is most commonly applied in singlearm Phase II trials. Occasionally, two-sided test (H0: p = p0 vs. H1: p ¹ p0) could be considered. Two-sided tests are regularly applied in randomized settings and are more conservative than one-sided tests.

6.4.2 Two-Stage Phase II Trials More commonly, Phase II trials are conducted using a two-stage design wherein lack of efficacy in the first stage of the trial results in early study termination, thereby sparing patients from being futilely treated. However, should some early promise be seen in the first stage, then recruitment continues until the full sample size has been achieved to provide a better estimate of the treatment efficacy. Hence, although a conclusion that an agent is ineffective can be reached after the first stage, a conclusion about the degree of efficacy can only be determined after the second and final stage of the trial. Probably the earliest example of the two-stage design was described by Gehan in 1961 [18] wherein a trial could be terminated early if a predefined series of consecutive failures were seen in the first stage of the trial. Otherwise, the trial proceeds to the second stage of enrolling more patients to provide a better estimate of the response rate. This is predominantly of historical interest because the design is based on the assumption that very few investigational agents will show any efficacy, which was the case at the time this trial design was conceived. Gehan’s design can still be applied nowadays in settings when it is reasonable to assume that the null hypothesis response rate (p0) is 0. The more common two-stage design used today is based on the designs by Simon [19]. As with a single-stage Phase II trial, Simon’s designs [19] allow for the testing of the null and the alternate hypotheses with

6  Phase II Trials with Anticancer Agents

147

prespecified type I and type II error rates. Furthermore, Simon’s two-stage design can be constructed so as to minimize the average sample size (the optimal design) or to minimize the maximum sample size of the study (the minimax design) assuming the null hypothesis is true. For example, when p0 = 0.1, p1 = 0.3, and a = b = 0.1, the optimal two-stage design enrolls 12 patients in the first stage. If no responses are seen or only one response is found, then the trial is stopped and the agent is declared ineffective. Otherwise, 23 more patients are enrolled for a total study size of 35 patients. At the end of the trial, if five or fewer responses are observed, the agent is deemed ineffective. Otherwise (six or more responses in 35 patients), the agent is considered effective and worthy of further evaluation. Under the null hypothesis (H0), there is a 66% chance that the trial will be stopped early and the expected sample size under H0 is 19.8 patients, i.e., when H0 is true, the averaged sample size is 12 × 0.66 + 35 × 0.34. Note that the maximum study sample size is 35 patients. In comparison, the minimax design would enroll 16 patients into the first stage. If no response or only one response is seen, the trial is stopped early and then agent is considered ineffective. Otherwise, nine more patients are enrolled in the second stage to reach a total of 25 patients. At the end of the trial, the agent is considered ineffective if four or fewer responses are seen. The probability of stopping early is 51% under the null hypothesis (H0). The expected sample size is 20.4 patients (slightly higher than 19.8 as shown in the optimal design) but the maximum study sample size (25 patients) is smaller. The minimax design is preferable where the total number of eligible patients is limited and/or the increase of the expected sample size under H0 from the optimal design is small. Otherwise, optimal design is preferred. In both examples above, early stopping was possible for futility but not for activity of the experimental agent because it was not necessary to stop a trial early in this situation as patients were being treated with an active agent, and continued enrolment allowed for collection of more data to provide a better estimation of treatment response and toxicity. Not everyone agrees with this position and there are designs which allow for early stopping of a trial for both demonstrated inefficacy as well as demonstrated efficacy, e.g., Fleming’s two-stage design [20, 21]. One problem with fixed sample size two-stage designs is the difficulty in study conduct caused by the rigid requirement of examining the outcome at the specified sample size in each fixed stage. This may be particularly true in multicenter trials due to the complexity of coordinating patient accrual and follow-up across multiple study sites. Temporarily halting the accrual for evaluating outcomes impedes the study momentum and can lower the enthusiasm for investigators to participate in the trial. In addition, when the actual conduct deviates from the original design, it results in undefined stopping boundaries and the planned statistical properties no longer hold. To address this problem, Green and Dahlberg [22] examined the performance of planned vs. attained designs and gave an empirical solution by adapting the stopping rules to achieve desirable statistical properties. Herndon proposed a hybrid design by blending the one-stage and two-stage designs, thus allowing for uninterrupted accrual between the stages [23]. Chen and Ng [24] gave a collection of two-stage designs by varying sample sizes in both the first and second stages to construct optimal flexible designs.

148

H.K. Gan et al.

Several authors have described other refinements of the two-stage design. Jung et al. [25, 26] proposed a graphical method for searching all design parameters to attain admissible two-stage designs with good design characteristics. Others have proposed designs where trials may simultaneously evaluate objective response and another endpoint such as toxicity (see Sect. 6.6.3 below).

6.4.3 Multistage Phase II Design Phase II trial designs with more than two stages have also been described, e.g., the Fleming multistep procedure [20] and Chen’s optimal and minimax three-stage design [27]. In essence, the study is assessed at the end of several stages to see if the observed response rate to date has reached a minimum threshold or a maximum threshold. Although such multistage designs are theoretically superior to two-stage designs as there are more opportunities for the trial to be closed early for either proven efficacy or inefficacy, this is usually offset by the increased difficulty in execution, the longer time to completion, and increased cost [28, 29]. At this time, such designs are not widely used [3].

6.5 Randomized Phase II Trials Although still relatively uncommon [3, 5], the use of randomized Phase II trials has increased by 17-fold between 1986 and 2002 [4]. In essence, patients in such trials are randomly assigned at study entry into the different arms of the Phase II study. Common reasons why patients might be randomized in a Phase II trial are to conduct two or more single-arm trials simultaneously; to provide a concurrent “comparator” arm against which to informally compare the experimental arm(s); to provide a formal statistical comparison between a concurrent “comparator” arm and the experimental arm(s); or finally, to select the best among a number of experimental arm(s) for further development. Lee and Feng [4] reviewed 266 randomized Phase II studies conducted from 1986 to 2002 and found that most studies employed a randomized Phase II design to provide a concurrent “comparator” arm.

6.5.1 Randomized Phase II Trials to Provide a Concurrent “Comparator” Arm In such trials, patients are randomized between one or more experimental arms and a “comparator” arm wherein patients are treated with a standard treatment. An example is the calibrated design by Herson and Carter [30]. An appropriate response

6  Phase II Trials with Anticancer Agents

149

rate in the comparator arm based on historical data lends credence to any efficacy seen in the experimental arm(s) as it suggests that selection bias or other confounding factors have not artificially inflated the response rate(s) in the study [31]. Otherwise, a repeat Phase II trial should ideally be performed to ascertain the true efficacy of the experimental arm. This design allows for a greater than 90% recovery of the power of a fixed sample, noncontrolled design but at the expense of a three- to fivefold increase in sample size. The importance of such attempts to control for selection bias is underscored by a paired comparison of Phase II and Phase III trials which used identical chemotherapeutic regimens [5]. On average, the response rates of a regimen in the Phase II setting was 12.9% higher than that of the same regimen in the Phase III setting, presumably reflecting selection bias in the smaller Phase II studies [5]. In the example given above, the comparison between the comparator arm and the experimental arm did not undergo a formal statistical test [30]. Many groups feel that randomized Phase II trials should not undergo a formal comparison [31, 32]. This is because randomized Phase II trials are not small Phase III randomized controlled trials and their limited sample size, and statistical power increases the chance of type II error.

6.5.2 Randomized Phase II Trials to Select from a Number of Experimental Arms Another application of randomized Phase II trials is to select one of a variety of experimental treatments for further development. For example, an investigational agent may be planned for Phase III testing but the optimal schedule is unknown. A randomized Phase II trial is undertaken to select between several schedules of the experimental agent. The statistical framework used in such randomized Phase II trials is based on ranking and selection theory. One such “pick-the-winner” design is the Simon, Wittes, and Ellenberg (SWE) method [33]. Unlike the hypothesis-testing framework that controls both type I and type II errors, the ranking and selection procedure of the SWE method controls only type II errors. Basically, the response rate of each treatment arm is estimated and the arm with the highest response rate is picked as the winner. The design is appealing because the required sample size is much smaller than that for a randomized Phase III trial under the hypothesis-testing framework. For example, a Phase III trial may require 146 patients per arm to test between response rates of 10% vs. 25%, with 90% power, and a two-sided 5% type I error. On the other hand, the SWE method requires only 21 patients per arm with the same power. The tradeoff is that the false-positive rate can range from 20 to over 40% as reported in simulation studies [34]. The SWE method works best when it is likely that there will be one true “winner” with all other contenders falling substantially below par. However, this design is less useful where there are several regimens of similar activity. At the end of the trial, this method always picks the ­treatment arm with the best-observed outcome as the winner, regardless of whether none of the regimens worked, some

150

H.K. Gan et al.

of them worked, or all of them worked. In addition, there is no early stopping rule for futility. Possibly because of these issues, only about 11% of randomized Phase II designs have used this method [4].

6.5.3 Randomized Phase II Trials as “Screening Trials” Another approach to Phase II randomized trials is to design them as “screening trials.” In essence, these trials are less concerned with definitively defining the activity of an agent but more concerned with “screening” for a group of drugs which are more likely to succeed in the Phase III setting, at which time efficacy (or lack of it) will be definitively defined [14, 15]. While this might seem a subtle difference, adoption of a screening philosophy has been used to justify relaxing the a value to 0.2, or the b to 0.2 or increasing the expected difference (D) between the treatment arms [15]. If objective response is the primary endpoint, one might stipulate an increase of 20% as being likely. In turn, this reduces the sample size required for the study. Anecdotally, one problem with such trials is that they may be widely viewed as a practice-changing “mini-Phase III” trials should positive results be reported, resulting in a premature change in clinical practice and potentially impairing patient recruitment into any subsequent confirmatory Phase III trials.

6.5.4 Randomized Discontinuation Trials Yet another form of randomized Phase II trial is the randomized discontinuation trial [35]. In this two-stage trial design, all patients are treated with the investigational agent in the first stage. At the conclusion of the first stage, all patients displaying disease progression, excessive toxicity, or treatment noncompliance are taken off-study. Those who have shown an objective response to the investigational agent continue on open-label treatment with the investigational agent. The remaining patients (i.e., those who experience disease stabilization only) are randomized to treatment with the same investigational agent or a placebo. One key advantage of this design is its application to reduce the variability of cases where there is substantial heterogeneity within the target population as to who is sensitive to the experimental agent [36, 37]. By focusing on a more homogeneous group of patients who are more likely to respond to treatment, such randomized discontinuation trials may require a smaller sample size than a standard Phase II trial [35–37] although this is disputed by others [38]. The randomized discontinuation design is not as efficient as upfront randomization if treatment has a fixed effect on tumor growth or if treatment benefit is confined to slower-growing tumors [37]. There are also issues regarding the ethics of ceasing active treatment

6  Phase II Trials with Anticancer Agents

151

in patients deriving some benefit (i.e., disease stabilization) and about the generalizability of such trials.

6.5.5 Randomized Phase II/III Trials Randomized Phase II/III trial designs allow for a positive randomized Phase II trial to proceed into a definitive Phase III design, which might be an advantage when one is examining the activity of an investigational agent in a small patient population [39–41]. However, it should be stressed that such designs should be decided upon prior to the initiation of the trial rather than a Phase III component being arbitrarily added on to the end of an ambivalent or inadequately powered Phase II trial. Seamless Phase II/III trials have been proposed in the literature [42, 43] with the idea of using a Phase II endpoint to guide the “Go/No Go” decision for pursuing Phase III. A positive correlation between a short-term Phase II endpoint and a longer-term Phase III endpoint is assumed.

6.5.6 Discussion Regarding Randomized Phase II Trials Overall, the use of randomized Phase II trials remains a very contentious issue. Some groups have strongly advocated the use of randomized Phase II trials in general [31, 32] while others have made a case for their being strongly indicated only in certain situations such as in the assessment of targeted agents which are cytostatic. Others advice caution in their use [34] or continue to support the use of single-arm Phase II trials in certain situations, e.g., where the sample size is necessarily small [44]. At this time, there is no actual data comparing the efficacy (e.g., percentage of times a correct “Go/No Go” decision is made) of single-arm trials vs. randomized Phase II trials. As such, the choice of which design to use remains a matter of clinical choice based on the theoretical benefits of a particular design in that particular type of experimental agent in that particular patient population. Until such time as such comparative data is available, it is possible to outline situations where certain Phase II trials designs may be advantageous. Single-arm Phase II trials may be most appropriate where there are limited patients available, especially if there are very robust and reproducible data about standard treatments which can inform the subsequent decision about whether to proceed to a Phase III trial. Randomized Phase II trials are increasingly being used where there are concerns regarding selection bias, inadequate historic data, or where cytostatic targeted agents are being used. They may also be relevant where an experimental agent is being added to an established agent. Specialized examples of randomized Phase II trials address specific clinical needs such as selection trials to choose which of several experimental treatments should be pursued further or randomized discontinuation trials that facilitate the assessment of treatments in populations where there are differential responses in different subgroups.

152

H.K. Gan et al.

6.6 Other Theoretical Frameworks for Phase II Trials 6.6.1 Bayesian Designs Another statistical approach to the design of Phase II trials (applicable to both single-arm studies and randomized studies) is the Bayesian framework. Unlike the hypothesis-testing/frequentist framework in which the parameter of interest (e.g., treatment effect) is considered fixed but the data random, the Bayesian framework assumes that the parameter is random but data fixed. Bayesian designs begin by defining a plausible range for the efficacy of the investigational (and any standard) treatment based on the previous experience and all available data to date, thereby acknowledging that there is often some degree of uncertainty about the true efficacy of any treatment. This putative efficacy range is formally described in an a priori probability distribution p(q). As patients are entered into the trial and new data are generated, the cumulative information is used to update p(q) by computing the posterior distribution. At the investigators’ discretion, interim analyses can be performed. Bayes factors and credible intervals can be calculated to perform hypothesis testing and confidence interval estimation as in the frequentist approach. The success of Bayesian approach hinges upon the proper choice of the prior distribution and the correct specification of the likelihood function. Given data, the posterior distribution can be computed for all inference purposes. Various authors have developed specific Bayesian designs for Phase II trials. Herson [45] developed a Bayesian design to allow for early termination of a Phase II trial secondary to inefficacy of the experimental agent. This design also has the advantage of fixing the sample size and always reaches a conclusion regarding the efficacy of the investigation agent at the end of the study. Mehta and Cain [46] described a design that would allow for early stopping for either demonstrated inefficacy or demonstrated efficacy. However, their design has a higher chance of being unable to reach a definitive conclusion regarding the efficacy of the investigational agent. Lee and Liu [47] proposed a predictive probability design which allows the flexibility of continuously monitoring the trial outcome. The design is efficient and remains robust in controlling type I and type II error rates. Johnson and Cook proposed a Bayesian hypothesis testing approach to address the problem. Thall and Simon [48] also described a randomized Phase II design where data are monitored continuously (i.e., after each patient’s results are received) for a trial in which an investigation agent is compared with a standard arm. Sample size for a Bayesian design can be determined by yielding certain precision in estimating a parameter of interest or attaining certain accuracy in hypothesis testing. Although controlling type I and type II errors are frequentist’s properties and are not required by Bayesian methods, a good design would have low type I and type II errors. Therefore, simulation studies can be performed to calibrate the design parameters such that the design will yield desirable frequentist operating characteristics including controlling type I and type II errors. The main advantage of the Bayesian design is its flexibility. It is amendable to performing interim analyses on an ongoing, continuous basis rather than at prespecified time points, although a proper a priori distribution must be specified to counter

6  Phase II Trials with Anticancer Agents

153

the ­statistical price to be paid for excessive interim analyses, as is also the case in the frequentist approach. Bayesian design is also more flexible in encompassing ­deviations from study design during the conduct of the trial. It allows for incorporation of information from external sources to reach a decision. Some disadvantageous of this design are that it is considered inherently more subjective because the need for specifying the a priori distribution. Bayesian approach is also inherently more computationally intensive and there are relatively fewer software available for its implementation. However, the computational hurdle is less of an obstacle these days, thanks to better algorithms, faster computers, and more interest and activity in developing Bayesian tools [49]. In conclusion, the Bayesian framework is a viable alternative to the frequentist framework. There are Bayesian alternatives for most frequentist Phase II trial designs, including those that allow for the analysis of time-to-event variables [50]. As such, use of Bayesian Phase II trial designs may vary more with local expertise and familiarity with these designs rather than with any inherent superiority of the frequentist framework. An excellent tutorial for Bayesian clinical trials was provided by Berry [51].

6.6.2 Decision Theoretic Designs There are designs that attempt to incorporate additional parameters into the results of a Phase II trial, specifically the costs which reflect the gains and losses that would be involved in accepting or rejecting the experimental agent. Broadly speaking, these additional parameters include the number of patients that would be needed in any subsequent Phase III trial, the number of patients whom might be treated with the experimental agent if it was accepted, and the difference in cost (utility) of patients who are treated with an effective agent rather than an ineffective agent. While the basic design first suggested by Sylvester and Staquet [52] has been modified by other authors, this framework has not been widely used for a number of reasons, not least being that some of the additional parameters required are subjective and difficult to quantify.

6.6.3 Bivariate Analysis Although antitumor efficacy as evidenced by tumor shrinkage has been the endpoint of greatest interest in Phase II trials to date, it is rare that a decision is made solely on the objective response rate of the investigational agent. Factors such as toxicity and cost often enter into the final decision about whether an investigational agent progresses to further testing, although this is often based on the investigator’s assessment rather than on a statistical analysis. However, there are statistical designs that allow for bivariate analysis (using two endpoints) of a Phase II trial to decide whether it should proceed to Phase III testing. The utility of incorporating the cost of an investigational drug in a formal analysis was touched upon above in trials using a decision theory framework. Some authors have described designs to allow for

154

H.K. Gan et al.

bivariate analysis based on both response and toxicity [53–57] while others have developed designs to monitor for both response and early progression [16, 58].

6.7 Evolving Challenges in the Age of Targeted Therapies More recently, the search for new anticancer treatments has explored the use of agents that selectively target specific antigens or pathways in tumor cells. In many cases, these agents are expected to be cytostatic agents which results in disease stabilization rather than agents which cause tumor shrinkage [59]. As such, concerns have been raised about the ability of traditional Phase II trials to assess such agents, given that these have generally used objective response rate as their primary endpoint [59]. Some groups have suggested that other endpoints, such as early disease progression or DFS, should be used as endpoints for trials of targeted agents. Others have suggested that a multinomial approach be adopted where the stopping rule for futility or efficacy is based on bivariate analysis of objective response and lack of disease progression [16, 58]. A recent review of Phase II trials employing 31 targeted agents in six solid tumor types was recently published [7]. Of the 89 trials identified, 65 studies (69%) still employed objective response rate as the primary or coprimary endpoint. Of these, 62% detected an objective response, albeit modest. The mean response rate was approximately 10% with a range of 3–28%. Although relatively modest, evidence of objective response was significantly correlated with increased likelihood of FDA approval. The FDA did not approve any agent that had a 0% objective response rate in Phase II testing. Although all the agents approved by the FDA showed some objective response rate, almost all were relatively modest – more than half of the approved agents had a response rate less than 10%. It is interesting to contrast this with a recent review of Phase II trials employing chemotherapy where the mean response rate was 43% (range 16–87%) but where these large response rates were not predictive of subsequent success in Phase III trials [5]. In addition to supporting the continued use of objective response rate in Phase II trials of targeted agents, the study by El-Maraghi and Eisenhauer [7] also identified nonprogression rate as a useful trial endpoint as all agents which were subsequently approved by the FDA also has a nonprogression rate of >30%. Some weaknesses were identified in the trials reported to date, as only a minority of studies enriched for the presence of the target of interest (20% of studies) and only 50% of studies reported the statistical underpinnings of their sample size calculations. The taskforce on the methodology for the development of innovative cancer therapies (MDICT) recently published some guidelines for the future conduct of Phase II trials using targeted agents [59]. They concluded that Phase II trials remain relevant in the development of innovative cancer therapies. Objective response rates as defined by the RECIST criteria also remain appropriate endpoints in Phase II trials, but they suggest that multinomial designs (assessing both response and lack of progression) should be considered for all future studies. The use of multinomial designs may be particularly useful where the objective response rate was likely to

6  Phase II Trials with Anticancer Agents

155

be low. They encouraged the incorporation and validation of innovations into traditional Phase II trial design. These include the analysis of response as a continuous variable rather than a dichotomous response variables (e.g., using assessment tools such as spider plots and waterfall plots), the appropriate use of randomized Phase II trial designs, and enrichment strategies for patients who express the target of the experimental agent(s). They identified that there should be improved reporting of Phase II trials in this area, including the publication of negative trials and the thorough description of all relevant information such as patient characteristics, design, endpoints, and predefined “Go/No Go” criteria.

6.8 Conclusions Phase II trials remain a pivotal phase in the development of chemotherapeutic and targeted anticancer treatments. Their main objective is to assess whether new treatments, whether used alone or in combination with established treatments, show sufficient efficacy to warrant continuing on to costly and lengthy Phase III trials. Objective response rates remain the commonest endpoint of these trials, although the incorporation of other endpoints is becoming increasingly relevant in the era of molecularly targeted therapy. Innovations in Phase II trial designs, such as the increasing use and diversity of randomized Phase II trial designs, reflect an appropriate and on-going need for Phase II trial design to evolve in response to the changing clinical landscape. Table 6.2 lists key elements to consider in choosing an appropriate Phase II trial design. Table 6.3 summarizes Table 6.2  Checklist of elements to be incorporated in the (a) design and (b) reporting of a Phase II trial Mandatory   The rationale of the experimental treatment is defined, within the context of relevant clinical information and trials   Inclusion and exclusion criteria are defined   Primary endpoint   Secondary endpoints(s)   Type of Phase II trial   Number of treatment arms and description of each arm   Method of assessing endpoints – both the frequency and type of assessments   Definition of the endpoints used   Statistical design of trial especially sample size calculations and preplanned subset analyses   Explicit statement of the results of all specified primary endpoints, secondary endpoints, and preplanned analyses   All patients entered into the study should be accounted for in the analysis and results, even those who are subsequently withdrawn for any reason or who experience significant protocol violations   A discussion of the results within the context of all relevant clinical information and trials   A conclusion about whether the experimental treatment should proceed to further testing or not (Go/No Go decision) Optional   The rationale and types of any correlative studies in the study

Table 6.3  Characteristics of commonly used Phase II trials designs using the hypothesis-testing framework Trial design Advantages Disadvantages Single-arm trials One-stage Small sample size; simplicity of execution Inability to stop trial early if drug is clearly very effective or clearly ineffective Inability to stop the trial early if drug is Two-stage Small sample size; relatively simple to clearly very effective (optional efficacy execute; ability to stop trial early if stopping is allowed) lack of efficacy is evident Multiple stages Ability to stop trial early if drug is clearly Increased complexity of execution very effective or clearly ineffective Randomized trials Approximately four-fold increase in sample Comparator trials The presence of a comparator arm may size compared with equivalent single-arm improve the interpretation about drug trial. These trials are not recommended efficacy in some circumstances, e.g., where the number of patients available for lack of historical data; concerns exist trial recruitment is limited regarding selection or other biases; where an experimental agent is being combined with an accepted therapeutic agent; where a cytostatic agent is being assessed using time-to-event endpoints (controversial) Unless one arm is likely to be clearly Selection trials Useful in late Phase II testing if a drug is superior to the rest (at least 15% intended for Phase III testing but uncertainly difference), there is a high remains about the optimal schedule false-positive rate

Data are awaited to confirm that the increased complexity of randomized Phase II trials result in more accurate estimates about drug efficacy

Not commonly used

Probably the commonest Phase II design in use

Comments

156 H.K. Gan et al.

Phase II/III trials

Randomized discontinuation trials

Screening trials

Where a drug is highly likely to be active, this design allows for an integrated approach to Phase II and III testing

Similar to comparator trials but often smaller sample sizes are required (through relaxation of a and b) and a formal statistical comparison is performed Useful if substantial patient heterogeneity in response to drug treatment is expected

Possibly misinterpreted as being “mini-Phase III trials” with subsequent premature acceptance of drug and/or hindrance of a confirmatory Phase III trial Increased sample size; difficulty in accruing patients to the second (randomized) part of trial may occur if more patients than expected respond/progress in the open-label part of the trial Must be specified a priori rather than at the end of a promising Phase II trial Their use is controversial

Their use is controversial

6  Phase II Trials with Anticancer Agents 157

158

H.K. Gan et al.

Table 6.4  Illustrative sample sizes for trials with specified designs and parameters Single-arm Phase II Two-arm randomized trial Phase II trial Predicted efficacy of investigational Sample sizea (first drug (p0 vs. p1) stage/total) a and b Sample sizeb per arm 0.05 vs. 0.20 0.05 and 0.10 21/41 114 0.10 and 0.10 12/37 95 0.05 and 0.20 10/29 88 0.20 and 0.20 9/16 56 0.20 vs. 0.40

0.05 and 0.10 0.10 and 0.10 0.05 and 0.20 0.20 and 0.20

19/54 17/37 13/43 11/16

119 98 91 56

0.40 vs. 0.60

0.05 and 0.10 0.10 and 0.10 0.05 and 0.20 0.20 and 0.20

25/66 18/46 16/46 10/25

140 115 107 66

 Based on Simon optimal two-stage design with one-sided a  Based on chi-square test with continuity correction and two-sided a

a

b

the commonly used Phase II designs while Table  6.4 shows typical sample sizes required. These tables are given for illustrative purposes only and it is strongly recommended that a biostatistician be consulted early during the design of a Phase II trial to ensure that appropriate trial designs and parameters are used.

References 1. U.S. Food and Drug Administration, Title 21 – Food and Drugs; Chapter 1 – Food and Drug Administration; Subchapter D – Drugs for Human Use, Department of Health and Human Services. 2007, U.S. Food and Drug Administration, Silver Spring, MD. 2. Baar, J. and I. Tannock, Analyzing the same data in two ways: a demonstration model to illustrate the reporting and misreporting of clinical trials. Journal of Clinical Oncology, 1989. 7(7): pp. 969–978. 3. Perrone, F., et  al., Statistical design in phase II clinical trials and its application in breast cancer. Lancet Oncology, 2003. 4(5): pp. 305–311. 4. Lee, J.J. and L. Feng, Randomized phase II designs in cancer clinical trials: current status and future directions. Journal of Clinical Oncology, 2005. 23(19): pp. 4450–4457. 5. Zia, M.I., et  al., Comparison of outcomes of phase II studies and subsequent randomized control studies using identical chemotherapeutic regimens. Journal of Clinical Oncology, 2005. 23(28): pp. 6982–6991. 6. Vickers, A.J., V. Ballen, and H.I. Scher, Setting the bar in phase II trials: the use of historical data for determining “go/no go” decision for definitive phase III testing [see comment]. Clinical Cancer Research, 2007. 13(3): pp. 972–976. 7. El-Maraghi, R.H. and E.A. Eisenhauer, Review of phase II trial designs used in studies of molecular targeted agents: outcomes and predictors of success in phase III. Journal of Clinical Oncology, 2008. 26(8): pp. 1346–1354.

6  Phase II Trials with Anticancer Agents

159

8. Therasse, P., et  al., New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. Journal of the National Cancer Institute, 2000. 92(3): pp. 205–106. 9. Therasse, P., E.A. Eisenhauer, and J. Verweij, RECIST revisited: a review of validation studies on tumour assessment. European Journal of Cancer, 2006. 42(8): pp. 1031–1039. 10. Eisenhauer, E.A., et al., New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). European Journal of Cancer, 2009. 45(2): pp. 228–247. 11. World Health Organization, WHO Handbook for Reporting Results of Cancer Treatment. 1979, World Health Organization, Geneva. 12. Miller, A.B., et al., Reporting results of cancer treatment. Cancer, 1981. 47(1): pp. 207–214. 13. Stone, A., et  al., Optimizing randomized phase II trials assessing tumor progression. Contemporary Clinical Trials, 2007. 28(2): pp. 146–152. 14. Stone, A., C. Wheeler, and A. Barge, Improving the design of phase II trials of cytostatic anticancer agents. Contemporary Clinical Trials, 2007. 28(2): pp. 138–145. 15. Rubinstein, L.V., et al., Design issues of randomized phase II trials and a proposal for phase II screening trials. Journal of Clinical Oncology, 2005. 23(28): pp. 7199–7206. 16. Dent, S., et al., Application of a new multinomial phase II stopping rule using response and early progression. Journal of Clinical Oncology, 2001. 19(3): pp. 785–791. 17. Ratain, M.J. and T.G. Karrison, Testing the wrong hypothesis in phase II oncology trials: there is a better alternative. Clinical Cancer Research, 2007. 13(3): pp. 781–782. 18. Gehan, E.A., The determination of the number of patients required in a preliminary and a followup trial of a new chemotherapeutic agent. Journal of Chronic Diseases, 1961. 13: pp. 346–353. 19. Simon, R., Optimal two-stage designs for phase II clinical trials. Controlled Clinical Trials, 1989. 10(1): pp. 1–10. 20. Fleming, T.R., One-sample multiple testing procedure for phase II clinical trials. Biometrics, 1982. 38(1): pp. 143–151. 21. Chang, M.N., et al., Designs for group sequential phase II clinical trials. Biometrics, 1987. 43(4): pp. 865–874. 22. Green, S.J. and S. Dahlberg, Planned versus attained design in phase II clinical trials. Statistics in Medicine, 1992. 11(7): pp. 853–862. 23. Herndon, J.E. II, A design alternative for two-stage, phase II, multicenter cancer clinical trials. Controlled Clinical Trials, 1998. 19(5): pp. 440–450. 24. Chen, T.T. and T.H. Ng, Optimal flexible designs in phase II clinical trials. Statistics in Medicine, 1998. 17(20): pp. 2301–2312. 25. Jung, S.H., M. Carey, and K.M. Kim, Graphical search for two-stage designs for phase II clinical trials. Controlled Clinical Trials, 2001. 22(4): pp. 367–372. 26. Jung, S.-H., et al., Admissible two-stage designs for phase II cancer clinical trials. Statistics in Medicine, 2004. 23(4): pp. 561–569. 27. Chen, K. and M. Shan, Optimal and minimax three-stage designs for phase II oncology clinical trials. Contemporary Clinical Trials, 2008. 29(1): pp. 32–41. 28. Lee, J.J., Clinical trial design for anticancer therapies, in The Cancer Handbook, M.R. Alison, Editor. 2007, John Wiley and Sons, Ltd., Chichester, pp. 1330–1344. 29. Tomblyn, M.R. and J.D. Rizzo, Are there circumstances in which phase 2 study results should be practice-changing? Hematology, 2007. 2007(1): pp. 489–492. 30. Herson, J. and S.K. Carter, Calibrated phase II clinical trials in oncology. Statistics in Medicine, 1986. 5(5): pp. 441–447. 31. Van Glabbeke, M., W. Steward, and J.P. Armand, Non-randomised phase II trials of drug combinations: often meaningless, sometimes misleading. Are there alternative strategies? European Journal of Cancer, 2002. 38(5): pp. 635–638. 32. Anonymous, Phase II trials in the EORTC. The Protocol Review Committee, the Data Center, the Research and Treatment Division, and the New Drug Development Office. European Organization for Research and Treatment of Cancer. European Journal of Cancer, 1997. 33(9): pp. 1361–1363.

160

H.K. Gan et al.

33. Simon, R., R.E. Wittes, and S.S. Ellenberg, Randomized phase II clinical trials. Cancer Treatment Reports, 1985. 69(12): pp. 1375–1381. 34. Liu, P.Y., M. LeBlanc, and M. Desai, False positive rates of randomized phase II designs. Controlled Clinical Trials, 1999. 20(4): pp. 343–352. 35. Kopec, J.A., M. Abrahamowicz, and J.M. Esdaile, Randomized discontinuation trials: utility and efficiency. Journal of Clinical Epidemiology, 1993. 46(9): pp. 959–971. 36. Temple, R.J., Enrichment designs: efficiency in development of cancer treatments. Journal of Clinical Oncology, 2005. 23(22): pp. 4838–4839. 37. Freidlin, B. and R. Simon, Evaluation of randomized discontinuation design. Journal of Clinical Oncology, 2005. 23(22): pp. 5094–5098. 38. Capra, W.B. and W.B. Capra, Comparing the power of the discontinuation design to that of the classic randomized design on time-to-event endpoints. Controlled Clinical Trials, 2004. 25(2): pp. 168–177. 39. Schaid, D.J., et al., A design for phase II testing of anticancer agents within a phase III clinical trial. Controlled Clinical Trials, 1988. 9(2): pp. 107–118. 40. Ellenberg, S.S. and M.A. Eisenberger, An efficient design for phase III studies of combination chemotherapies. Cancer Treatment Reports, 1985. 69(10): pp. 1147–1154. 41. Storer, B.E., A sequential phase II/III trial for binary outcomes. Statistics in Medicine, 1990. 9(3): pp. 229–235. 42. Inoue, L.Y.T., P.F. Thall, and D.A. Berry, Seamlessly expanding a randomized phase II trial to phase III. Biometrics, 2002. 58(4): pp. 823–831. 43. Lachin, J.M. and N. Younes, A composite design for transition from a preliminary to a fullscale study. Statistics in Medicine, 2007. 26(27): pp. 5014–5032. 44. Taylor, J.M.G., T.M. Braun, and Z. Li, Comparing an experimental agent to a standard agent: relative merits of a one-arm or randomized two-arm phase II design. Clinical Trials, 2006. 3(4): pp. 335–348. 45. Herson, J., Predictive probability early termination plans for phase II clinical trials. Biometrics, 1979. 35(4): pp. 775–783. 46. Mehta, C.R. and K.C. Cain, Charts for the early stopping of pilot studies. Journal of Clinical Oncology, 1984. 2(6): pp. 676–682. 47. Lee, J.J. and D.D. Liu, A predictive probability design for phase II cancer clinical trials. Clinical Trials, 2008. 5(2): pp. 93–106. 48. Thall, P.F. and R. Simon, Practical Bayesian guidelines for phase IIB clinical trials. Biometrics, 1994. 50(2): pp. 337–349. 49. Berry, D.A., Introduction to Bayesian methods III: use and interpretation of Bayesian tools in design and analysis. Clinical Trials, 2005. 2(4): pp. 295–300; discussion 301–304. 50. Thall, P.F., L.H. Wooten, and N.M. Tannir, Monitoring event times in early phase clinical trials: some practical issues. Clinical Trials, 2005. 2(6): pp. 467–478. 51. Berry, D.A., Bayesian clinical trials. Nature Reviews Drug Discovery, 2006. 5(1): pp. 27–36. 52. Sylvester, R.J. and M.J. Staquet, Design of phase II clinical trials in cancer using decision theory. Cancer Treatment Reports, 1980. 64(2–3): pp. 519–524. 53. Bryant, J. and R. Day, Incorporating toxicity considerations into the design of two-stage phase II clinical trials. Biometrics, 1995. 51(4): pp. 1372–1383. 54. Conaway, M.R. and G.R. Petroni, Designs for phase II trials allowing for a trade-off between response and toxicity. Biometrics, 1996. 52(4): pp. 1375–1386. 55. Jin, H., Alternative designs of phase II trials considering response and toxicity. Contemporary Clinical Trials, 2007. 28(4): pp. 525–531. 56. Conaway, M.R. and G.R. Petroni, Bivariate sequential designs for phase II trials. Biometrics, 1995. 51(2): pp. 656–664. 57. Thall, P.F., R.M. Simon, and E.H. Estey, New statistical strategy for monitoring safety and efficacy in single-arm clinical trials. Journal of Clinical Oncology, 1996. 14(1): pp. 296–303. 58. Zee, B., et al., Multinomial phase II cancer trials incorporating response and early progression. Journal of Biopharmaceutical Statistics, 1999. 9(2): pp. 351–363.

6  Phase II Trials with Anticancer Agents

161

59. Booth, C.M., et  al., Design and conduct of phase II studies of targeted anticancer therapy: recommendations from the task force on methodology for the development of innovative cancer therapies (MDICT). European Journal of Cancer, 2008. 44(1): pp. 25–29. 60. Johnson, V.E. and J.D. Cook, Bayesian design of single-arm phase II clinical trials with continuous monitoring. Clinical Trials, 2009. 6(3): pp.217–226.

Chapter 7

Phase III Clinical Trials with Anticancer Agents Wendy R. Parulekar and Daniel J. Sargent

7.1 Introduction The primary objective of a phase III clinical trial is to demonstrate or confirm the therapeutic benefit of a new treatment or treatment strategy when compared with an alternate treatment, usually the current standard of care. Due to the modest incremental benefits seen with most new anticancer agents, phase III studies generally require large sample sizes and considerable resource allocation. Thus, the decision to conduct a phase III study must be carefully considered. Comprehensive pharmacokinetic and pharmacodynamic supporting data should be present, including preclinical and clinical information regarding absorption, distribution, metabolism, and excretion of different dosing schedules as well as drug interactions including any ethnic/gender specific differences in these parameters. Pharmacodynamic data of interest includes information on toxicity and antitumor activity. The traditional measure of antitumor activity has been tumor response as measured by standardized criteria such as RECIST [1]. While tumor shrinkage was frequently considered an adequate activity measure for cytotoxic chemotherapy, it may not be entirely appropriate for solid tumors with difficult to measure radiological endpoints such as prostate or ovarian cancer or molecularly targeted anticancer agents whose main mechanism of action may be described as cytostatic rather than cytocidal. Additional signals of activity have been proposed including changes in serum tumor marker levels [2, 3] or measures of progression such as progression rates [4, 5] or risk of progression [6] Research to validate these and other endpoints is ongoing. For many clinicians, the ultimate goal of a phase III study is to inform and influence clinical practice by comparing the efficacy of a new treatment or treatment regimen against the current standard of care. The calibration of treatment effects from clinical trials to target populations requires consideration of multiple issues

W.R. Parulekar (*) Department of Oncology, Queen’s University, Kingston, ON, Canada e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_7, © Springer Science+Business Media, LLC 2011

163

164

W.R. Parulekar and D.J. Sargent Null Hypothesis is True

Alternative Hypothesis is True

Decision: Conclude No Difference

True Negative

False Negative; Type II Error

Decision: Conclude A Difference

False Positive;

True Positive;

Type I Error

Fig. 7.1  Hypothesis testing in a therapeutic trial and decision making

such as the relative homogeneity of the trial population based on inclusion and exclusion criteria compared with the “real world,” feasibility of administration and compliance with the new therapy in actual practice setting, and the long-term adverse events associated with a new therapy which may not be known prior to uptake by the medical community. The frequentist approach to clinical trial design and interpretation will be the focus of this chapter. This approach is familiar to many clinicians and consists of hypothesis testing and generation of tests of statistical significance and precision (confidence intervals) as a guide to the conduct and interpretation of data from a single trial (Fig. 7.1). In contrast, the Bayesian approach to inference incorporates prior knowledge of the probability of a specific outcome into the interpretation of the results of a single trial, leading to the generation of a revised probability.

7.2 Population of a Phase III Clinical Trial In general, the population of subjects entered on a phase III trial is one that is expected to benefit from the intervention(s) being tested and to comply with study procedures. Selection of study participants is mostly based on patient factors (e.g., age, general well being or functional status, and organ function) and disease ­characteristics such as type and extent of cancer. In the current era of molecularly targeted drug therapy, there is great interest in further defining the study population using patient or tumor molecular characteristics such as genes or gene products. The development of trastuzumab highlights this strategy. The identification of tumoral HER2/neu gene amplification as an adverse prognostic factor and characterization of the gene product as a transmembrane receptor provided the rationale for the development of an antibody directed against the receptor [7–9]. In the early clinical

7  Phase III Clinical Trials with Anticancer Agents

165

and pivotal phase III studies, the patient populations were enriched for HER2/neu positive disease [10–14]. While this strategy appears to have been appropriate given the demonstrated efficacy of trastuzumab in this group of patients, by excluding the (much larger) group of patients with HER2/neu negative disease, it presupposes that the mechanism of action of this agent is fully known and is dependent solely on the presence of gene amplification and/or overexpression. Due to the complexity of cellular signaling pathways and the likelihood of redundancy of pathways needed to maintain the malignant phenotype, early enrichment of patient populations for clinical studies based on the knowledge of the putative target must be done with caution and with sufficient evidence supporting this strategy.

7.3 Randomization Randomization is the single most important tool available to reduce bias in study conduct and interpretation of results by ensuring comparability between patient groups in different treatment arms. Simple randomization refers to random allocation that occurs with a known probability, i.e. “flipping a coin.” Limitations to this method include the potential for imbalances in patient numbers and prognostic factors between the treatment groups at the end of the randomization period. The likelihood of such imbalances is reduced as the trial’s sample size increases. Constrained randomization methods are available that minimize the likelihood of imbalance such as block randomization. Each “block” contains a prespecified number and proportion of treatment allocations. The order of the allocations is varied between blocks as is block length to provide an additional layer of blindness to the procedure. Upon completion of each block, there will be an equal number of patients in each treatment arm [15]. Minimization is another method that is commonly used. It is an example of adaptive randomization in which each randomization is determined by the composition of the patient population entered onto the study until that point in time. The driving principle is that each randomization will minimize the difference in the numbers and composition of patients enrolled between study arms [16].

7.3.1 Multiple Randomization Multiple randomizations of a single patient may occur on a study. This may arise with a factorial design which seeks to address more than one treatment question within the same clinical study or when patients already participating on a ­randomized study are randomized to a second treatment strategy after completion of the first one. An increasingly common scenario is the participation of a single patient on another study which addresses a separate clinical question. In general, participation on multiple concurrent clinical trials should only proceed if there is no

166

W.R. Parulekar and D.J. Sargent

conflict with respect to study procedures and data collection methods and after careful consideration of the impact of such participation on the study endpoints.

7.3.2 Stratification To reduce the potential for imbalances of prognostic factors between treatment groups which may affect the outcome independent of treatment effect, stratification may be used in the randomization methods described above. In permuted block randomization, blocks are completed within each stratum or prognostic subgroup. As this may be logistically challenging and may result in imbalances in treatment allocation, efforts should be made to keep the stratification factors as few as possible. While the same logistic challenge does not apply to the minimization method, the analysis and interpretation of data can be challenging in the case of multiple stratification factors. Analysis of and adjustment for potential prognostic variable (covariate analysis) can be performed at the time of final analysis if stratification is not done at randomization. Although this method may be less challenging to implement, it cannot correct the extreme imbalances in prognostic factors, and the covariates to adjust for must be prespecified.

7.4 General Trial Design 7.4.1 Endpoints in Phase III Clinical Trials 7.4.1.1 Primary Versus Secondary The phase III clinical trial is designed to be the definitive test of efficacy of a new therapy or treatment combination. Since practice standards are defined by phase III data, the primary endpoint must be clinically meaningful such as prolongation of survival, increased cure rate, quality of life, or palliation of disease-related ­symptoms. Secondary endpoints serve multiple purposes such as the provision of supporting data for treatment efficacy, demonstration of safety and tolerability, as well as the exploration of biological hypotheses regarding the cancer or treatment strategy under evaluation as a means of guiding future research questions. The primary endpoint guides the statistical design of the study (e.g., the sample size) and constitutes the basis for the hypothesis testing and estimates of efficacy.

7.4.1.2 Criteria for Measurement The definition of all endpoints must be concise, and the means of determination must be objective. Standardized criteria should be used to measure the endpoints

7  Phase III Clinical Trials with Anticancer Agents

167

whenever possible such as the RECIST criteria for tumor response and the CTCv3.0 [17] for toxicity assessment. The use of disease-specific composite endpoints which include multiple events in one term is common in oncology; however, this practice poses challenges to the interpretation of clinical trial results as measures of clinical benefit and the comparison of results between studies. Using breast cancer as an example, multiple definitions exist in the literature for the term “disease-free survival” which encompasses various combinations and definitions of the following events: contralateral breast cancer, in situ carcinoma, second primary cancers, and death from other causes. Efforts to adopt standardized endpoints during all stages of study conduct from protocol development to results reporting are underway [18]. 7.4.1.3 Surrogate Endpoint The use of surrogate or substitute endpoints on clinical trials is based on the realization that although overall survival is considered the least ambiguous and most relevant outcome measure of treatment efficacy in oncology, in many cases it may be impractical or impossible to evaluate new agents in a timely manner using overall survival as the primary endpoint. Advantages to inclusion of a surrogate endpoint in a study include a reduced sample size (and thus financial and administrative burden) due to the use of a more frequent and/or proximal endpoint, and better assessment of treatment efficacy due to a reduction in the length of followup and resulting decrease in the risk of competing causes of death. The use of a surrogate endpoint may also be more acceptable to study participants with diseases for which multiple treatment options or salvage therapies exist because it can be evaluated earlier in the disease course, thus allowing patients to be treated with other “standard” therapies should the experimental therapy fail to demonstrate an effect on the surrogate endpoint. A formal definition proposed by Prentice [19] requires that the surrogate endpoint must yield a valid test of the null hypothesis of no association between treatment and the true endpoint. Operationally, the surrogate endpoint must be disease-specific and accurately measured, have biological relevance and face validity (reflects the true disease burden), have a short latency compared with the natural history of the disease, have a strong statistical association with the true endpoint, and be responsive to treatment. Limitations associated with the use of surrogate endpoints include the need for compelling data regarding the relationship of the surrogate endpoint with the true endpoint and with the disease process itself, the potential for increased missing data compared with an overall survival endpoint, and more restrictive eligibility criteria if the surrogate endpoint such as a tumor marker is expressed differentially in a given patient population. In addition, there may be a loss of power to study important and less common events due to a reduced sample size [20, 21]. The validation of a surrogate endpoint requires a meta-analysis of trial results, preferably using individual patient data to enable a complete analysis including

168

W.R. Parulekar and D.J. Sargent

adjustment for covariates. In addition, studies with and without efficacy differences between study arms are needed to assess the positive and negative predictive value of the surrogate endpoint. Buyse et al. investigated the use of progression-free survival as a surrogate marker for overall survival in patients with advanced colorectal cancer [22]. Individual data from ten historical trials and three validation trials comprising 3,089 and 1,263 patients, respectively, were analyzed. The analysis demonstrated a high correlation between progression-free survival and overall survival [rank correlation coefficient 0.82 (95% CI 0.82–0.83)] as well as treatment effects on both endpoints (correlation coefficient 0.74 when one influential study was excluded and 0.99 for all studies). In the validation trials, progression-free survival predicted overall survival. A similar technique was used to demonstrate the validity of 3-year disease-free survival (DFS) as a surrogate marker for 5-year overall survival in the adjuvant colon cancer setting [23]. Although this analysis led to the definition of a new regulatory endpoint for full drug approval in the adjuvant setting for colon cancer, the role of DFS as a surrogate marker for overall survival continues to be explored in other disease settings.

7.4.2 Masking Masking or blinding refers to the process of concealment of treatment allocation from study participants (single) or study participants and study personnel (double) as a means of reducing bias in study conduct and analysis. Potential sources of bias that may be reduced using masking include patient selection, endpoint determination, treatment withdrawal, and assessment of eligibility status. Instances exist in which masking may not be feasible such as studies in which the treatment comparisons involve therapies with different modes of administration (intravenous versus oral) or unique toxicity profiles. Masking entails certain requirements including the use of a placebo, an appropriate drug coding and distribution system, as well as a standard operating procedure for unblinding the treatment allocation should an urgent situation arise in which identification of the therapy is needed. A placebo is a chemically inert substance identical in appearance to the active drug under evaluation in a clinical study. A placebo may be used as the sole intervention on a treatment arm (placebo controlled) or in addition to an active control. Specific circumstances where the use of a placebocontrolled design may be appropriate include the lack of an effective anticancer therapy, the presence of an anticancer therapy with limited efficacy and significant toxicity, or disease states characterized by waxing and waning of disease severity or spontaneous remissions or regression [24]. An unbalanced randomization between placebo and the agent under investigation (favoring the latter) may be preferable to a balanced 1:1 randomization from a study participant point of view and as a means of maximizing the safety data on the investigational agent accumulated in the study. The addition of a placebo to an active control is useful when the experimental arm consists of a combination of a new

7  Phase III Clinical Trials with Anticancer Agents

169

agent with standard therapy which has limited efficacy and pharmacological data exists supporting the combination.

7.4.3 Multiple Arm Studies The parallel group design for phase III studies may be expanded to include more than two concurrent treatment arms. Such a design may allow the simultaneous comparison of multiple promising treatment regimens to the current standard of care or alternatively, comparison of a new regimen against regimens considered to be the current standard of care. The statistical considerations are based on the comparisons between the different treatment arms, and thus adjustment of the Type I error rate for each comparison is needed to maintain the overall alpha level of 0.05. The advantages of this design include efficiency of time due to the ability to evaluate multiple regimens concurrently and increased confidence in interpretation of outcome data since each treatment group will contain comparable patients and standardized methodology will be used for endpoint assessments. As an example, a multiple arm study design was used to compare the activity and toxicity of three different two drug combinations in patients with previously untreated, advanced colorectal carcinoma [25]. The primary outcome was time to progression, where the control regimen was compared with two experimental regimens. The final sample size (795) allowed for 80% power to detect a hazard ratio of 0.75 between each experimental regimen and the control using a two-sided log rank test at a level of 0.025 for each comparison, maintaining the overall type I error rate at 0.05. The final analysis demonstrated a significantly better time to progression of one of the experimental regimens compared with the control but not the other.

7.4.4 Factorial Designs Factorial designs are used to increase the efficiency of phase III studies by testing the efficacy of multiple treatment strategies and interactions between treatments within one trial. Using a 2 × 2 factorial design of treatments A and B, patients will be randomized with equal probability to one of the four treatment groups: neither treatment, A alone, B alone, or A and B. The analysis will involve testing for the effects of treatment A by comparing those patients who received treatment A with those who did not, stratifying for the presence of B. The same process is repeated for the analysis of the efficacy of treatment B. In the absence of interaction, two treatments may be studied using a factorial design with half the sample size as two single factor studies. The factorial design is the most appropriate method to study the interaction effects between treatments although some efficiency is lost in the process [26, 27]. The major limitation of a factorial design is if there is indeed an interaction between the two treatments, the study sample size will likely be too

170

W.R. Parulekar and D.J. Sargent

small to provide sufficient power to detect treatment effects for A and B in the presence and absence of each other. Factorial designs can be generalized. For example, one of the factors may involve different doses or schedules of a new agent rather than treatment versus no treatment with a particular therapy. The Cancer and Leukemia Group B (CALGB) used a factorial design in an adjuvant, phase III breast cancer study which evaluated two treatment strategies: sequence of doxorubicin, cyclophosphamide, and paclitaxel chemotherapy (concurrent versus sequential) and interval between the doses of chemotherapy (every 2 versus every 3 weeks). Patients were allocated with equal probability to the four treatment groups as shown in Fig.  7.2. DFS (the primary endpoint) was significantly prolonged with the 2 weekly dosing compared with the 3 weekly dosing [risk ratio (RR) = 0.74, P = 0.010]. DFS was not associated with treatment sequence (P = 0.58). No interaction between the density of therapy and the sequence of therapy was detected (P = 0.40) [28].

Fig. 7.2  Factorial design utilized in CALGB Trial 9741. From Citron et al. [28]. Reprinted with permission. © 2008 American Society of Clinical Oncology. All rights reserved

7  Phase III Clinical Trials with Anticancer Agents

171

7.4.5 Equivalence and Noninferiority Design Equivalence studies confirm the absence of a meaningful difference between treatments – either positive or negative. A margin of equivalence, expressed as ∆, must be chosen as an indicator of the largest difference in efficacy between the treatments that would be considered clinically acceptable to practicing physicians and patients and thus allow adoption of this treatment into current practice. The results of the trial including 95% confidence intervals must lie entirely within the range of −∆ to +∆ to be declared equivalent. Conversely, any difference that exceeds this predefined margin would lead to rejection of the new therapy as an equivalent treatment option. Noninferiority studies confirm the absence of a meaningful loss of efficacy when a new treatment or treatment strategy is compared with the current standard of care. In this scenario, the noninferiority margin ∆ is the degree of inferiority of the new treatment to the control that the trial will attempt to exclude statistically. The new treatment will be declared noninferior if the confidence interval of the treatment differences excludes a margin of delta or greater. Determination of the margin is clinically and statistically based and should be conservative given the uncertainties inherent in the process. An important principle that guides the selection of the delta is that it should be smaller than the effect size of the active control compared with placebo to prove that there is a preservation of some degree of the efficacy of the active control; preservation of at least 50% of the effect size is generally utilized [29]. In general, the equivalence or noninferiority designs are useful when evaluating a new treatment that may be less toxic, less costly, or more convenient to administer than the standard treatment which would be recommended on these grounds as long as efficacy would not be compromised too much. As the sample size and statistical assumptions for these types of studies differ from superiority studies, declaration of a trial as an equivalence or noninferiority study is required prior to study commencement [30, 31]. However, these studies generally require significantly greater sample sizes and determining the appropriate ∆ can be challenging.

7.5 Biomarkers in Phase III Trials A biomarker is a characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention [32]. Thus, the term biomarker is a broad one which must be interpreted according to the context of its use. Biomarkers are increasingly incorporated into clinical trials as a means of characterizing tumor biology and behavior, drug pharmacodynamics using laboratory or radiology based investigations, or as a means of predicting response to drug therapy. This interest in biomarkers has been fueled by the current emphasis in oncology on the development of noncytotoxic, targeted agents which interact with specific molecules or

172

W.R. Parulekar and D.J. Sargent

molecular pathways involved with the malignant state including cellular proliferation, differentiation, and apoptosis as well as extracellular process such as angiogenesis, invasion, and migration. Biomarkers can be classified as prognostic or predictive. A prognostic marker is one that provides information about the outcome of an individual based on marker status, in the absence of therapy or when treated with empiric therapy. Prognostic biomarkers can be used to select patients for whom treatment is necessary (poor prognosis) or conversely, for whom treatment is not warranted (good prognosis). In contrast, a predictive biomarker predicts a differential efficacy of a particular therapy in the presence versus the absence of the biomarker. The ability to select patients for a specific targeted therapy based on a molecular profile is critical to increasing the efficiency of the drug development process. Based on the experience with HER2/neu in breast cancer, it is clear that there are targets of new agents which may not be present or relevant within histologically similar tumors. Furthermore, modest but important treatment effects in a subgroup might be masked by a lack of effect when the agent is tested in an unselected population. Three basic strategies can be used to assess the role of biomarkers in the context of a phase III study. The first, labeled retrospective validation involves the use of data from a study that enrolled a population with no application of selection criteria based on marker status and on whom biospecimens were collected in a rigorous fashion as part of study conduct. Candidate biomarkers of interest that are pertinent to the study population and treatment intervention are identified, and the biospecimens are analyzed with respect to biomarker status. The study population is classified into distinct groups based on marker levels using previously defined cutpoints, e.g., high/intermediate/low or positive/negative. All aspects of the biomarker analysis must be specified prospectively, and the biomarker assays performed blinded to the clinical outcomes of interest. A common and important limitation of this type of strategy is the lack of power associated with the reduced number of study subjects with available biospecimens that undergo successful assay testing compared with the number of patients randomized onto the study, issues of bias with respect to tissue collection, and the potential for poor biomarker assay performance if the biospecimens of interest were not collected, handled, or stored under the proper conditions. Methodology issues aside, this approach may be the most practical one to use when there is uncertainty regarding target validation or assay standardization that prevents the incorporation of biomarker studies in a clinical trial prospectively. Furthermore, the statistical limitations associated with power may be largely overcome with a high rate of biospecimen retrieval and testing. This strategy was utilized by Paik et al. in a prospectively defined multigene assay study that utilized biospecimens and a clinical database previously compiled on a phase III trial [33]. A second type of design using a biomarker involves enrollment of an enriched patient population selected on the basis of a molecular characteristic of the tumor. The potential for a greater therapeutic effect with a particular treatment strategy in an enriched compared with an unselected population may allow a reduction in sample size due to a lower targeted hazard ratio (see Sect. 6) and increased power to address the question of treatment efficacy. The usefulness of this design depends

7  Phase III Clinical Trials with Anticancer Agents

173

on several factors, including the proportion of responsive patients, the accuracy of the assay for predicting responsiveness, and the degree to which the mechanism of action of the drug is understood [34, 35]. The last point is especially important. By excluding patients without the biomarker of interest, one may miss a therapeutic opportunity if the treatment under investigation has multiple mechanisms of action that are not linked to the expression of the putative target. A third type of biomarker design facilitates the evaluation of predictive markers in a prospective fashion without patient selection [36]. The indirect method of biomarker assessment involves division of the patient population by biomarker status (positive versus negative) followed by randomization to the experimental and standard therapies in both sets of patients. The statistical analysis may involve tests of interaction between marker status and treatment or a formal comparison of outcome between the treatment effects in each marker group (Fig. 7.3). The direct method of biomarker assessment involves randomization of patients, and subsequent comparison of outcomes between a marker-based strategy to select therapy and a selection process which is independent of marker status. At least two variations of this type of design are possible. In one variation, the marker-based strategy involves randomization between two treatments, and the patients in the nonmarker-based arm all receive the same treatment. In a second possible implementation, randomization to two treatments occurs in the nonmarker-based arm,

Fig. 7.3  Marker by treatment interaction design. From Sargent et al. [36]. Reprinted with permission. © 2008 American Society of Clinical Oncology. All rights reserved

174

W.R. Parulekar and D.J. Sargent

Fig.  7.4  Marker-based strategy design. From Sargent et  al. [36]. Reprinted with permission. © 2008 American Society of Clinical Oncology. All rights reserved

thus allowing an assessment of the experimental therapy in both groups (Fig. 7.4). This allows clarification of whether any finding regarding the efficacy of the marker-directed approach to therapy is due to a true effect of marker status or to an improved regimen regardless of marker status. This design may also allow a retrospective assessment of an alternative classification for the marker. An ongoing study provides an example of a clinical trial designed to validate the prognostic and/or predictive role of a biomarker in patients with nonsmall-cell lung cancer, randomized to either pemetrexed or erlotinib [37]. Prior to randomization, the epidermal growth factor receptor (EGFR) status of the tumor is determined on all patients using fluorescent in situ hybridization (FISH). Patients are stratified by EGFR status and randomized to erlotinib, a small molecule tyrosine kinase inhibitor of the EGFR, or premetrexed, an antifolate chemotherapeutic agent. The primary goal of the study is to compare the progression-free survival of EGFR positive and negative patients treated with either agent. Secondary outcome measures will clarify the predictive and prognostic roles of EGFR expression (by FISH and immunohistochemistry) and of EGFR mutations.

7.6 Statistical Considerations 7.6.1 Hypothesis Testing and Confidence Intervals Two methods of statistical inference form the basis of phase III trial design and analysis – hypothesis testing and estimation. Using an example of a new anticancer agent A to be compared with the standard of care B, the outcomes resulting from each treatment intervention are captured (e.g., disease relapse or death). Hypothesis testing involves the identification of a null and an alternative hypothesis. In phase III trials, the null hypothesis (H0) is generally a statement claiming that there is no difference between the treatments with respect to a parameter of interest, such as

7  Phase III Clinical Trials with Anticancer Agents

175

median survival. The alternative hypothesis HA indicates the opposite – a genuine difference exists. Further specification of the alternative hypothesis as one- or twosided is required. This refers to the direction of the treatment difference (better or worse) that would be categorized as extreme when the appropriate test statistic is applied to the data. Use of a two-sided alternative hypothesis is generally recommended in phase III trials since a new therapy might be better or worse when compared with the standard or care. However, there are some cases where the one-sided approach may be justified, such as when a combination therapy is compared with the treatment having one of the agents in the combination. In that case, it may be argued that the direction of the treatment difference is likely to be better with the combination of agents. As discussed in Chap. 1 (Basic Biostatistics for the Clinical Trialist), the type I and type II errors also need to be specified to determine the appropriate sample size. The power or sample size calculations are a critical part of the study design and involve the determination of how large a sample size is required to detect a difference of a specific magnitude. This should be determined prior to study conduct. The statistical significance of the difference in outcomes between treatments is expressed by the P-value, which is the probability that a difference as large or larger than that observed would have arisen by chance alone. A limitation of use of a P-value for the interpretation of data is that it does not provide information regarding the magnitude of treatment effect attributed to a therapy given the results of a particular study. This type of information is provided by another method of statistical inference that involves the estimation of a treatment’s effectiveness using a parameter such as the hazard ratio and a measure of precision of the hazard ratio estimate using a confidence interval. The confidence interval provides a range of reasonable values in which the true value is expected to lie, as discussed in Chap. 1.

7.6.2 Sample Size The endpoint of a phase III trial may be defined as a continuous measurement (laboratory values), dichotomous outcome (cure versus no cure), or as the time to a clinically relevant outcome (event) of interest. The majority of phase III clinical trials are designed around a time-to-event endpoint, and this will be discussed in detail below. When planning a phase III study, the investigator must specify certain parameters that will determine the sample size. The importance of this cannot be overstated. The failure to reject the null hypothesis may be due to the true lack of efficacy of a new therapy or the lack of power of the study to detect the difference. Determinants of sample size include the difference in efficacy between treatments that the trial is designed to detect delta (∆) as well as the type I and type II errors. The most challenging to define is ∆ because it reflects a clinical opinion regarding what the difference is likely to be and what is clinically meaningful. This can be expressed as a ratio of survival at a particular time point or a relative risk reduction of an event between treatments which can then be converted into a hazard ratio. The hazard ratio

176

W.R. Parulekar and D.J. Sargent

(see Sect. 10.1) will determine the specific number of events that need to be observed prior to analysis of the data. Other factors influencing sample size include how rapidly the events will occur in the population of interest, accrual rate, and the length of follow-up. For example, a phase III study comparing adjuvant chemotherapy to observation in patients with nonsmall-cell lung cancer targeted an absolute improvement of 10% in 3-year survival from 60% favoring the chemotherapy arm [38]. To detect the equivalent hazard ratio of 1.43, 198 events were required. Using a onesided 5% significance level test and 80% power, the total sample size was estimated to be 450 with 6.75 years of accrual and less than 1 year of follow-up. Table 7.1 provides examples of sample size calculations using different statistical parameters, and a more detailed discussion of sample size calculation is included in Chap. 1.

7.6.3 Interim Analyses The relatively modest therapeutic gains anticipated with new treatment strategies coupled with increasing survival rates for many cancers have resulted in the need for large sample sizes and long durations of follow-up for most phase III studies. This is especially true for adjuvant studies. The concept of an interim analysis (an analysis conducted prior to the protocol-specified final analysis) is thus intuitively appealing as it may lead to early disclosure of important results or closure of the study. Interim analyses may demonstrate treatment futility (i.e., the study has a low probability of rejecting the null hypothesis based on the interim analysis), evidence of treatment harm that could not have been predicted, or a positive efficacy result of sufficient magnitude that disclosure of results to study participants and the scientific community

Table 7.1  Calculation of sample sizes for a two-arm trial Control median survival 1 year Control median survival 2 years Number of patients/ Number of total Number of Number of total arm events patients/arm events Hazard ratio Power = 0.80 1.3 276 456 385 456 1.4 170 277 239 277 1.5 191 256 168 191 Power = 0.90 1.3 369 611 515 611 1.4 228 371 320 371 1.5 159 256 225 256 Number of patients per arm and total number of events for a trial with an assumed median timeto-event of 1 or 2 years in the control group, power = 0.80 or 0.90, Type I error 5%, and various hazard ratios. Calculations assume an accrual time of 2  years and 2-year minimum follow-up. Note that the control group median survival impacts the number of patients required but not the total number of events requested, as a longer median survival relative to the follow-up period implies a greater number of patients need to be followed to attain the same number of events

7  Phase III Clinical Trials with Anticancer Agents

177

is medically and ethically justified. Since multiple looks at the data inflate the type I error and thus the probability of incorrectly rejecting the null hypothesis, the statistical design must be modified to support the strategy of interim analyses. Group sequential stopping rules have been developed which result in a modest increase in sample size compared with that required for a single analysis at the end of the study. These rules involve the specification of boundary values on the type I error at each interim analysis (also known as spending functions). Commonly used approaches include the O’Brien–Fleming and Pocock boundaries [39, 40]. The O’Brien–Fleming bound results in an uneven spread of the type I error such that the data must be more convincing at earlier than later stages of a study to result in rejection of the null hypothesis and declaration of efficacy at the final analysis with only a small penalty or change in the type I error rate due to the earlier analyses. The Pocock approach spends more of the type I error prior to the final analysis compared with the O’Brien– Fleming method and consequently, the nominal type I error at the final analysis is substantially smaller than those with only one analysis. Stopping rules for futility also exist when there is a low probability that the ­difference in efficacy targeted by the study design will be demonstrated given the data observed at the time of the interim analysis [41]. The statistical parameters tend to be less conservative when stopping for futility compared with efficacy as reflected in the less stringent type 1 error rates. A frequently used approach described by Wieand et al. [42] involves the conduct of a futility analysis when one half of the target events has been observed in a time-toevent analysis and consideration of early study termination if the hazard ratio comparing the experimental to the control arm is >1. Using this approach will result in a <2% loss of power compared with the initial design specifications. This method is most appropriate for advanced disease settings where the length of accrual may be more than twice the median survival for the control regimen. Stopping boundaries may also be applied to point estimates and confidence intervals generated at the interim analyses [43]. A confidence interval which includes data that supports both the null and alternative hypotheses would lead to the recommendation that the study should be continued. Conversely, a point estimate and confidence interval that supports one hypothesis over the other may lead to early termination of the study.

7.7 Phase II/III Design A phase II/III design is one that incorporates the principle of examination of preliminary data prior to embarkation on a full phase III study as part of the conduct of a single study. In this scenario, the “early look” involves analysis of the data from a randomized phase II study rather than an interim analysis of an ongoing phase III study that is adequately powered to target a difference of interest. The major advantage to this type of design is that it provides a degree of objectivity to the interpretation of the phase II results, and allows use of an established trial ­infrastructure for the conduct of the phase III study. In addition, there is the added efficiency that all

178

W.R. Parulekar and D.J. Sargent

randomized patients contribute data for the final analysis. However, if all patients are to be included in the final analysis, the ability to alter the protocol to use what is learned in the phase II portion of the trial is severely constrained. A study that examined the efficacy of a matrix metalloproteinase inhibitor (MMPI) in the treatment of nonsmall-cell lung cancer (NSCLC) nicely illustrates the concepts discussed above [44]. The phase II portion of this randomized phase II/III placebo-controlled study demonstrated similar response rates and acceptable toxicity profiles of the treatment arms consisting of chemotherapy plus the MMPI and chemotherapy plus placebo [45]. Based on these results, the study was expanded to a phase III design. The results of a planned interim analysis showed no survival advantage and increased toxicity in the experimental arm. The independent Data Safety Monitoring Committee (DSMC) reviewed the results and recommended that the study treatment be stopped and the results disclosed to the trial committee and investigators. The recommendations were accepted and a final analysis was performed when the protocol specified required number of events was reached. Toxicity was higher in the experimental arm and there were no statistically significant differences in efficacy between the treatment groups.

7.8 Independent Data Safety Monitoring Committee Independent Data Safety Monitoring Committee [also known as the Data Safety Monitoring Board (DSMB)] oversight is considered an essential for the conduct of a phase III study with mortality or major morbidity as a primary or secondary endpoint, or in a setting where trial participants are at increased risk for adverse outcomes. This body, created prior to the study and whose members have no conflict of interest with the trial, typically reviews the accrual, cumulative toxicity, and efficacy data from ongoing phase III studies as well as relevant information from external sources that may impact the design or the decision to continue the study. It acts in an advisory role to the sponsor of the study. Transparency of process and independence of the committee or board members from trial conduct are essential operational criteria for the DSMC as is the existence of a charter defining the scope of responsibilities. The composition of the Committee may vary but members must have the requisite expertise (e.g., scientific, legal, ethical, statistical, patient advocacy). Reference documents are available from the Food and Drug Administration [46] and the United States National Cancer Institute [47].

7.9 Termination of a Clinical Trial Prior to the Final Analysis It is important to make the distinction between the term “stopping rules” and the actual decision to stop a study. Any decision that results in termination of a study prior to the planned final analysis is a complex one which should be guided by,

7  Phase III Clinical Trials with Anticancer Agents

179

but not solely reliant on statistical tests regarding efficacy or futility (see Sect.  6.3). Disadvantages to early termination of a clinical study include the inability to study important secondary objectives, a loss of credibility with investigators resulting in a failure to influence medical practice, an inability to study toxicity and efficacy associated with longer duration therapy, and biased estimates of treatment effects. This is especially true in those instances where early termination is due to the perception of improved efficacy of the new treatment over standard therapy. A systematic review of randomized controlled trials that stopped early for apparent benefit demonstrated an inadequacy of reporting why the decision was made as well as a propensity to overestimate the treatment effect in many of the studies [48].

7.10 Data Analysis and Reporting Fundamental to the analysis of phase III studies is the concept that outcome data for all patients should be included and that any exclusions should be detailed in the methods section and in subsequent publications. This concept underlies the use of the term “intent to treat” (ITT) analysis which refers to the process of inclusion of data from all randomized patients, regardless of the eligibility status, the treatment actually received on study, and compliance with the treatment or protocol procedures. In a broad sense, the ITT principle applies to the design, conduct, and analysis of a study since it requires a robust method of patient follow-up, outcome measurement, data capture, and analysis. Instances where the application of the ITT principle to analysis may not be preferred arise with the equivalence and noninferiority designs. In these scenarios, analysis of data according to the randomized treatment allocation instead of treatment actually received may bias the results of the study toward a positive result, i.e., failing to reject the null hypothesis of no treatment difference. In this instance, analysis based on actual treatment received (also known as per protocol) is recommended in addition to the ITT analysis [49–51].

7.10.1 Measures of Effectiveness in Results Reporting One approach to measuring the relative efficacy of one treatment to another involves comparisons of the proportions of patients who have experienced the event of interest (e.g., relapse of disease or death) in each treatment arm at a specific time point such as 5 or 10 years after randomization. Another approach involves measuring the time to an event and the generation of survival curves. Measures of treatment effectiveness commonly used in the phase III oncology clinical trials will be briefly described below.

180

W.R. Parulekar and D.J. Sargent

7.10.1.1 Measurements for Proportions Risk and Relative Risk The risk of an event in a treatment group is calculated by dividing the number of people to whom a predefined event occurs by the total number of people in that group. The relative risk is defined as the ratio between the risks of the event occurring in the two different groups under comparison. For example, if death is defined as the event of interest, a relative risk of <1 when comparing an experimental therapy with a control therapy indicates better survival in the experimental group. Relative Risk Reduction This term represents the difference in event rates between the control and experimental groups divided by the event rate in the control group. It can also be calculated by subtracting the relative risk from 1. Absolute Risk Reduction The absolute risk reduction is the difference in risks between the control and experimental groups. It may be expressed as a decimal or percentage. Number Needed to be Treated The number needed to treat represents the number of patients needed to treat to prevent one event (adverse outcome). It can be readily calculated by dividing 1 by the absolute risk reduction expressed as a decimal value. Odds Ratio The odds ratio of an event is calculated by dividing the number of people to whom an event occurs by the number of people to whom it does not occur. The odds ratio measures the relative likelihood of an outcome between the two treatment groups and is calculated by dividing the odds ratio of an event in one group by the corresponding value in the other group. It is clear that fluency with the terminology used to report the outcomes in phase III clinical trial is critical to the practice of medicine and to the process of informed decision making from a patient and health policy point of view. As an example, the impact of a relative risk reduction of a given magnitude can only be interpreted if one knows the risk in the control group (baseline risk). A relative risk reduction of 50% in a population with a risk of an event of 5% translates into an absolute risk

7  Phase III Clinical Trials with Anticancer Agents

181

reduction of 2.5% and the need to treat 40 patients to avoid one adverse outcome. The same relative risk reduction in a group of patients with a baseline risk of 40% translates into an absolute risk reduction of 20% and a corresponding number needed to treat of 5. Although relative measures of the effectiveness of a specific intervention are generally constant across different levels of baseline risk, absolute measures are not. 7.10.1.2 Measurement for Time-to-Event Outcomes Many studies in oncology are designed to study the long-term outcomes associated with treatment interventions. Due to the pattern of enrollment, individual patients will be followed for different periods of time and not every patient will experience the event of interest such as death (or relapse of disease). Survival duration in clinical trials is measured as the time interval from randomization to death. The most common method of estimating the survival distribution based on the data of those who have experienced the event as well as those who have not (censored data) is the Kaplan–Meier method [52]. The relative survival experience between the two groups is often represented by the hazard ratio, which is defined as the ratio of the instantaneous risk of failure at any given time in one treatment arm (numerator) and the other treatment arm (denominator). To maximize the power of a study and avoid the inefficiency of comparing the survival curves at one specific time point, the log rank test is used to compare the entire survival curves of both treatment arms. Use of this test will have the maximum statistical power when there is a relatively constant proportionality of the hazard rates between the treatment groups over time. The utility of this approach to survival analysis is demonstrated with a double blind, placebo-controlled study of gemcitabine versus gemcitabine plus a small molecule tyrosine kinase inhibitor in pancreatic cancer (Fig.  7.5) [53]. The hazard ratio for overall survival demonstrated a significant improvement favoring the experimental arm (HR 0.82 95% CI 0.67–0.99 P = 0.038), in contrast to the marginal increase in median survivals observed (6.24 versus 5.91 months). More detailed description of analytic methods and interpretation for time-to-event outcomes is included in Chap. 1.

7.10.2 Univariable and Multivariable Testing Univariable and multivariable analyses are performed in phase III clinical trials to characterize the strength of association between candidate prognostic factors (including treatment) and outcome. Univariable testing analyzes this relationship for each factor in isolation. Although relatively simple to perform, univariable analyses do not adjust for simultaneous effects of more than one variable on outcome. Multivariable testing can be used to assess the association between a factor and an outcome, adjusting for additional factors thought to be related to the outcome as well. Ideally, prognostic factors of importance in a therapeutic study should be used

182

W.R. Parulekar and D.J. Sargent

Fig. 7.5  Kaplan–Meier curve for overall survival. From Moore et al. [53]. Reprinted with permission. © 2008 American Society of Clinical Oncology. All rights reserved

to stratify treatment allocation at the time of randomization. However, stratification may not be implemented due to the issues of feasibility. Furthermore, the statistical plan should state what factors are to be included in planned multivariable analyses and, if appropriate, the algorithm that will be used to test them. The most common approach to multivariable analyses is regression. Depending on the outcome of interest, there are different types of regression approaches that can be used (e.g., linear regression for continuous outcomes and logistic regression for binary outcomes). For time-to-event outcomes, Cox proportional hazards regression is a method of multivariable modeling of the impact of potential prognostic factors on the outcome, e.g., progression-free survival. There are limitations and assumptions to this method, which are discussed in Chap. 1. However, there are alternative regression modeling approaches for time-to-event outcomes which can relax or use different assumptions (e.g., additive hazards regression) [54].

7.10.3 Subgroup Analyses Subgroup analysis frequently occurs in randomized phase III studies as a means of testing biological hypotheses and estimating treatment effects in groups of patients selected by baseline characteristics. The limitations associated with subgroup analysis are well documented, including small patient numbers and the resulting loss of power to detect meaningful differences in outcome, failure to acknowledge the uncertainty of estimates of treatment effect sizes in subgroups, inflation of the type I error rate due to multiplicity of testing, and the failure to perform tests of interaction between the treatment and subgroup characteristic of interest [55–57]. The lack of transparency regarding the type of subgroup analysis, i.e., prespecified versus post

7  Phase III Clinical Trials with Anticancer Agents

183

hoc is a concern. While issues such as multiplicity apply to both types of subgroup analysis, post hoc analyses are of particular concern due to the inability of the reader to quantify how many analyses were actually undertaken and if the decision to perform them was motivated by prior knowledge of the data. When properly conducted and interpreted, subgroup analyses may provide useful information for future research, and guidelines have been developed for reporting subgroup analysis [58]. These guidelines pertain to criteria for inclusion of subgroup analysis data in the abstract, requirements for addressing methodology issues, inclusion of tests for interaction, effect estimates, and confidence intervals in the results section, and a cautious interpretation of the results of the subgroup analysis in the discussion section of a report. Consistency of subgroup findings across multiple independent clinical trials strengthens the conclusions that are appropriate. An example of this is the identification of KRAS status as a predictive marker for response to EGFR antibody therapy in the advanced colorectal cancer setting. Multiple analyses have indentified mutant KRAS status as a negative predictive factor for response to this class of agents [59–62].

7.11 Transparency and Consistency in Clinical Trial Conduct and Reporting 7.11.1 Trial Registries The goal of a clinical trials’ registry is to provide transparency to the conduct of clinical studies and the disclosure of results as a means of advancing the public good by increasing the body of knowledge available to clinical research stakeholders and countering the phenomenon of selective publication of trials which show a positive result or demonstrate noninferiority. The importance of this goal is underscored by the International Committee of Medical Journal Editors (ICMJE) initiative which requires prior entry of phase III clinical trials in a public registry as a condition for publication [63]. Criteria defining a suitable registry have been articulated: it must be electronically searchable, accessible to the public free of cost, open to all registrants, and not for profit. In addition, a mechanism to ensure the validity of the registration data must be in place. A minimal registration data set involving 20 elements has been defined as follows: unique trial number, trial registration date, secondary IDs, funding source(s), primary sponsor, secondary sponsors, responsible contact person, research contact person, title of the study, official scientific title of the study, research ethics review, condition, intervention(s), key inclusion and exclusion criteria, study type, anticipated trial start date, target sample size, recruitment status, primary outcome, and key secondary outcomes [64]. An example of such registry is Clinicaltrials.gov [65], which contains information on more than 97,253 clinical trials with locations in 174 countries. Expansion of the registry was mandated by US Public Law 110-85 Food and Drug

184

W.R. Parulekar and D.J. Sargent

Administration Amendments Act of 2007 Title VIII, Section 801. Implications for the registry included the change in status of several data elements from optional to required (in keeping with the recommendations of the ICMJE) and the addition of a new section for the reporting of basic results.

Fig.  7.6  Flow diagram of the progress through the phases of a randomized trial (enrollment, intervention, follow-up, and data analysis). From Moher et al. [66]. Reprinted with permission. All rights reserved

7  Phase III Clinical Trials with Anticancer Agents

185

7.11.2 CONSORT Statement The CONSORT (Consolidated Standards of Reporting Trials) statement is an evidence-based, minimum set of recommendations for reporting clinical trials [66]. The statement standardizes the preparation of reports of trial findings as a means of ensuring completeness and transparency of reporting, thereby aiding their critical appraisal and interpretation. It is composed of a 22-item checklist and flow ­diagram. The checklist items pertain to the title, abstract, introduction, methods, results and discussion sections. As an example, the recommendation for the title of the report is to include how patients were allocated to the interventions using words such as random allocation, randomization, or randomly assigned. The results section should include information regarding the flow of participants (preferably using a diagram), a description of protocol deviations, the dates of recruitment and follow-up, baseline demographics, the number of participants, summary of results, a report of other analyses and adverse events. The flow diagram shows the flow of participants through four stages of a trial – enrollment, intervention allocation, follow-up, and analysis (Fig.  7.6), thus demonstrating the extent of adherence to the ITT principle.

7.12 Summary Phase III clinical trials are fundamental to the practice of medicine, the advancement of research, and the empowerment and engagement of patients and the scientific community in the research agenda. A successful clinical trial is not defined by the nature of the result, i.e., positive or negative, but by the relevance of the clinical question, the use of an appropriate design and statistical methodology to address the question, and proper oversight during the study conduct. Finally, advancement of the health of a society can only be achieved through a critical appraisal of the full body of available evidence. Transparency of conduct and analysis is necessary to achieve this goal.

References 1. Therasse P, Arbuck SG, Eisenhauer EA, et  al: New guidelines to evaluate the response to treatment in solid tumors (RECIST guidelines). J Natl Cancer Inst 92:205–216, 2000 2. Scher HI, Halabi S, Tannock I, et al: Design and end points of clinical trials for patients with progressive prostate cancer and castrate levels of testosterone: recommendations of the Prostate Cancer Clinical Trials Working Group. J Clin Oncol 26:1148–1159, 2008 3. Rustin GJ, Quinn M, Thigpen T, et al: Re: new guidelines to evaluate the response to treatment in solid tumors (ovarian cancer). J Natl Cancer Inst 96:487–488, 2004 4. Zee B, Melnychuk D, Dancey J, et  al: Multinomial phase II cancer trials incorporating response and early progression. J Biopharm Stat 9:351–363, 1999

186

W.R. Parulekar and D.J. Sargent

5. Dent S, Zee B, Dancey J, et al: Application of a new multinomial phase II stopping rule using response and early progression. J Clin Oncol 19:785–791, 2001 6. Rosner GI, Stadler W, Ratain MJ: Randomized discontinuation design: application to ­cytostatic antineoplastic agents. J Clin Oncol 20:4478–4484, 2002 7. Slamon DJ, Clark GM, Wong SG, et  al: Human breast cancer: correlation of relapse and survival with amplification of the HER-2/neu oncogene. Science 235:177–182, 1987 8. Venter DJ, Tuzi NL, Kumar S, et al: Overexpression of the c-ERB-2 oncoprotein in human breast carcinomas: immunohistological assessment correlates with gene amplification. Lancet 2:269–273, 1987 9. van de Vijver MJ, Mooi WJ, Wisman P, et  al: Immunohistochemical detection of the neu protein in tissue sections of human breast tumors with amplified neu DNA. Oncogene 2: 175–178, 1988 10. Cobleigh ME, Vogel CL, Tripathy D, et al: Multinational study of the efficacy and safety of humanized Anti-HER2 monoclonal antibody in women who have HER2-overexpressing metastatic breast cancer that has progressed after chemotherapy for metastatic disease. J Clin Oncol 17:2639–2648, 1999 11. Vogel CL, Cobleigh ME, Tripathy D, et  al: Efficacy and safety of trastuzumab as a single agent in first-line treatment of HER2-overexpressing metastatic breast cancer. J Clin Oncol 3:719–726, 2002 12. Slamon DJ, Leyland-Jones B, Shak S, et al: Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2. N Engl J Med 344: 783–792, 2001 13. Romond EH, Perez EA, Bryant J, et al: Trastuzumab plus adjuvant chemotherapy for operable HER2-positive breast cancer. N Engl J Med 353:1673–1684, 2005 14. Piccart-Gebhart MJ, Procter M, Leyland-Jones B, et al: Trastuzumab after adjuvant chemotherapy in HER2-positive breast cancer. N Engl J Med 353:1659–1672, 2005 15. Matts JP, Lachin JM: Properties of permuted-block randomization in clinical trials. Control Clin Trials 9:327–334, 1988 16. Begg CD, Iglewicz BA: A treatment allocation procedure for sequential clinical trials. Biometrics 36:81–90, 1980 17. http://ctep.cancer.gov (last accessed October 2010) 18. Hudis CA, Barlow WE, Constantino JP, et al: Proposal for standardized definitions for efficacy end points in adjuvant breast cancer trials. The STEEP system. J Clin Oncol 25: 2127–2132, 2007 19. Prentice RL: Surrogate endpoint in clinical trials: definition and operational criteria. Stat Med 8:431–440, 1989 20. Herson J: The use of surrogate endpoints in clinical trials (An introduction to a series of 4 papers). Stat Med 8:403–404, 1989 21. Ellenberg SS, Hamilton JM: Surrogate endpoints in clinical trials: cancer. Stat Med 8:405– 413, 1989 22. Buyse M, Burzykowski T, Carroll K, et al: Progression-free survival is a surrogate for survival in advanced colorectal cancer. J Clin Oncol 25:5218–5224, 2007 23. Sargent DJ, Wieand HS, Haller DG, et al: Disease-free survival versus overall survival as a primary end point for adjuvant colon cancer studies: individual patient data from 20,898 patients on 18 randomized trials. J Clin Oncol 23:8664–8670, 2005 24. Daugherty CK, Ratain MJ, Emanuel EJ, et al: Ethical, scientific, and regulatory perspectives regarding the use of placebos in cancer clinical trials. J Clin Oncol 26:1371–1378, 2008 25. Goldberg RM, Sargent DJ, Morton RF, et  al: A randomized controlled trial of fluorouracil plus leucovorin, irinotecan, and oxaliplatin combinations in patients with previously untreated metastatic colorectal cancer. J Clin Oncol 22:23–30, 2004 26. Piantadosi S: Clinical trials: a methodologic perspective. John Wiley and Sons: New York, 1997 27. Peterson B, George SL: Sample size requirements and length of study for testing interaction in a 2 × k factorial design when time-to-failure is the outcome. Control Clin Trials 14: 511–522, 1993

7  Phase III Clinical Trials with Anticancer Agents

187

28. Citron ML, Berry DA, Cirrincione C, et al: Randomized trial of dose-dense versus conventionally scheduled and sequential versus concurrent combination chemotherapy as postoperative adjuvant treatment of node-positive primary breast cancer: first report of intergroup trial C9741/Cancer and Leukemia Group B Trial 9741. J Clin Oncol 21:1431–1439, 2003 29. Rothmann M, Li N, Chen G, et al: Design and analysis of non-inferiority mortality trials in oncology. Stat Med 22:239–264, 2003 30. Blackwelder WC: “Proving the null hypothesis” in clinical trials. Control Clin Trials 3: 345–353, 1982 31. International Conference on Harmonization. http://www.ich.org (last accessed October 2010) 32. Biomarkers Definitions Working Group. Biomarkers and surrogate endpoints: preferred definitions and conceptual framework. Clin Pharmacol Ther 69:89–95, 2001 33. Paik S, Shak S, Tang G, et al: A multigene essay to predict recurrence of tamoxifen-treated, node-negative breast cancer. N Engl J Med 351:2817–2826, 2004 34. Simon R, Maitournam A: Evaluating the efficiency of targeted designs for randomized clinical trials. Clin Cancer Res 10:6759–6763, 2004 35. Maitournam A, Simon R: On the efficiency of targeted clinical trials. Stat Med 24:329–339, 2005 36. Sargent DJ, Conley BA, Allegra C, et al: Clinical trial designs for predictive marker validation in cancer treatment trials. J Clin Oncol 23:2020–2027, 2005 37. http://www.cancer.gov/clinicaltrials/NCCTG-N0723 (last accessed October 2010) 38. Winton T, Livingston R, Johnson D, et  al: Vinorelbine plus cisplatin vs. observation in resected non-small-cell lung cancer. N Engl J Med 352:2589–2597, 2005 39. O’Brien PC, Fleming TR: A multiple testing procedure for clinical trials. Biometrics 35:549– 556, 1979 40. Pocock SJ: Group sequential methods in the design and analysis of clinical trials. Biometrika 64:191–199, 1977 41. Pampallona S, Tsiatis AA: Group sequential designs for one-sided and two-sided hypothesis testing with provision for early stopping in favor of the null hypothesis. J Stat Plan Inference 42:19–35, 1994 42. Wieand S, Schroeder G, O’Fallon JR: Stopping when the experimental regimen does not appear to help. Stat Med 13:1453–1458, 1994 43. Jennison C, Turnbull BW: Repeated confidence intervals for group sequential clinical trials. Control Clin Trials 5:33–45, 1984 44. Leighl NB, Paz-Ares L, Douillard JY, et al: Randomized phase III study of matrix metalloproteinase inhibitor, BMS 275291 in combination with paclitaxel and carboplatin in advanced non-small-cell lung cancer: National Cancer Institute of Canada-Clinical Trials Group Study BR.18. J Clin Oncol 23:2831–2839, 2005 45. Douillard JY, Peschel C, Shepherd F, et  al: Randomized phase II study of combining the matrix metalloproteinase inhibitor BMS-275291 with paclitaxel plus carboplatin in advanced non-small cell lung cancer. Lung Cancer 46:361–368, 2004 46. FDA Draft Guidance for Clinical Trial Sponsors on the Establishment and Operation of Clinical Trial Data Monitoring Committees. U.S. Department of Health and Human Services, Food and Drug Administration, November, 2001 www.fda.gov/cber/gdlns/clindatmon.pdf (last accessed October 2010) 47. Smith MA, Ungerleider RS, Korn EL, et al: Role of independent data-monitoring committees in randomized clinical trials sponsored by the National Cancer Institute. J Clin Oncol 15:2736–2743, 1997 48. Montori VM, Devereaux PJ, Adhikari NK, et al: Randomized trials stopped early for benefit: a systematic review. JAMA 294:2203–2209, 2005 49. Lewis JA, Machin D: Intention-to treat – who should use ITT? Br J Cancer 68:647–650, 1993 50. Matilde SM, Chen X: Choosing the analysis population in non-inferiority studies: per protocol or intent-to-treat. Stat Med 25(7):1169–1181, 2006 51. Lee YJ, Ellenberg JH, Hirtz DG, et  al: Analysis of clinical trials by treatment actually received: is it really an option? Stat Med 10:1595–1605, 1991

188

W.R. Parulekar and D.J. Sargent

52. Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53:457–481, 1958 53. Moore MJ, Goldstein D, Hamm J, et al: Erlotinib plus gemcitabine compared with gemcitabine alone in patients with advanced pancreatic cancer: a phase III trial of the National Cancer Institute of Canada Clinical Trials Group. J Clin Oncol 25:1960–1966, 2007 54. Grambsch PM, Therneau TM, Fleming TR: Diagnostic plots to reveal functional form for covariates in multiplicative intensity models. Biometrics 51(4):1469–1482, 1995 55. Assmann SF, Pocock SJ, Enos LE, et al: Subgroup analysis and other (mis)uses of baseline data in clinical trials. Lancet 355:1064–1069, 2000 56. Pocock SJ, Assmann SF, Enos LE, et al: Subgroup analysis, covariate adjustment and baseline comparisons in clinical trial reporting: current practice and problems. Stat Med 21: 2917–2930, 2002 57. Hernández A, Boersma E, Murray G, et al: Subgroup analyses in therapeutic cardiovascular clinical trials: are most of them misleading? Am Heart J 151:257–264, 2006 58. Wang R, Lagakos SW, Ware JH, et al: Statistics in medicine – reporting of subgroup analyses in clinical trials. N Engl J Med 357:2189–2194, 2007 59. Van Cutsem E, Lang I, D’haens G, et al: KRAS status and efficacy in the first line-treatment of patients with metastatic colorectal cancer treated with FOLFIRI with or without cetuximab: the CRYSTAL experience. J Clin Oncol 26(May 20 suppl; abstr 2): 2008 60. Bokemeyer C, Bondarenko I, Hartmann JT, et al: KRAS status and efficacy of first-line treatment of patients with metastatic colorectal cancer with FOLFOX with or without cetuximab: the OPUS experience. J Clin Oncol 26(May 20 suppl; abstr 4000): 2008 61. Teipar S, Peeters M, Humblet Y, et al: Relationship of efficacy with KRAS status (wild type versus mutant) in patients with irinotecan-refractory metastatic colorectal cancer, treated with irinotecan (q2w) and escalating doses of cetuximab (q1w): the EVEREST experience ­(preliminary data). J Clin Oncol 26(May 20 suppl; abstr 4001):2008 62. Karapetis CS, Khambata-Ford S, Jonker DJ, et al: K-ras mutations and benefit from ­cetuximab in advanced colorectal cancer. N Engl J Med 359:1757–1765, 2008 63. De Angelis C, Drazen JM, Frizelle F, et al: Clinical trial registration: a statement from the international committee of medical journal editors. N Engl J Med 351:1250–1251, 2004 64. De Angelis C, Drazen JM, Frizelle F, et al: Is this clinical trial fully registered? A statement from the international committee of medical journal editors. N Engl J Med 352:2436–2438, 2005 65. Clinicaltrials.gov available at http://www.clinicaltrials.gov (last accessed October 2010) 66. Moher D, Schulz KF, Altman DG: The CONSORT statement: revised recommendations for improving the quality of reports of parallel-group randomized trials. Ann Intern Med 134:657–662, 2001

Chapter 8

Pharmacokinetic Studies in Early Anticancer Drug Development Alex Sparreboom and Sharyn D. Baker

8.1 Introduction The last two decades has seen a substantial change in the approaches to the clinical development of anticancer drugs. This paradigm shift was instigated by the discovery of a significant number of highly innovative and distinct classes of molecules in terms of both mechanism of action and chemical structure, along with the deployment of an array of new approaches to clinical trial design. Moreover, a rising understanding of the relationships between pharmacokinetics and pharmacodynamics has encouraged a more systematic and rigorous analysis of the potential role of clinical pharmacology in the day-to-day management of cancer patients. Historically, pharmacokinetic studies have been an important integrated component of the various stages of oncology drug development, have demonstrated utility as a guide to dose-escalation strategies, trial design issues associated with dosing frequency, and have provided pertinent information on interindividual pharmacokinetic variability of novel agents. It has been recently suggested that lack of efficacy and/or inadequate safety of investigational agents are currently the primary reasons for attrition in oncology drug development, with poor pharmacokinetic properties representing only less than 10% of total attrition [38]. However, it has become widely appreciated at the same time that the small therapeutic window of most anticancer drugs, including many molecularly targeted agents, demands that a rigorous effort be made to characterize their pharmacokinetic properties, optimize their regimens, and that the failure to recognize this early on in drug development may lead to suboptimal dosing strategies, eventually resulting in lack of efficacy. In this chapter, we review the importance of pharmacokinetic studies in oncology drug development, highlight causes of variability in the response to chemotherapeutic treatment, and discuss methods for dose selection in order to decrease the

S.D. Baker (*) Department of Pharmaceutical Sciences, St. Jude Children’s Research Hospital, 262 Danny Thomas Place, CCC Room I5308, Mail Stop 313, Memphis, TN 38105-3678, USA e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_8, © Springer Science+Business Media, LLC 2011

189

190

A. Sparreboom and S.D. Baker

occurrence of significant interindividual pharmacokinetic variability as a source of treatment failure and/or unacceptable toxicity.

8.2 Importance of Pharmacokinetic Studies in Oncology Drug Development The current approach where a chemotherapeutic agent is administered at the maximum dose a patient can tolerate before the onset of unacceptable toxicity is still in wide clinical use, even for many of the molecularly targeted agents. This approach is supported by a series of retrospective analyses involving cytotoxic chemotherapy which indicates that the greater the dose intensity of an anticancer drug, the better the outcome [68]. However, the therapeutic range for most anticancer agents is rather narrow, and in many cases no information is available on the intrinsic sensitivity of a patient’s tumor to a particular agent and the patient’s tolerability of a given dose prior to therapy. Hence, the dosage of chemotherapeutic agents remains largely empirical, and has generally been derived from the kind of information shown in Fig. 8.1. Since the effect of a therapeutic agent in the body is generally a function of its concentration at the (molecular) site of action, it is obvious that a description of the Management of therapy Pharmacokinetics

- Multiple drug therapy

- Absorption

- Convenience of regimen

- Distribution

- Complience of patient

- Metabolism - Excretion

Toxicity - efficacy

Dosage regimen

- Therapeutic window

- Existence of other disease states

- Concentration-response relationship

- Costs - Route of administration - Dosage form - Tolerance – dependence

- Age - Weight

- Side effects

Other factors

State of patient

Pharmacogenetics - Target genes - Drug-metabolizing enzymes - Drug transporters

- Drug interactions

Fig. 8.1  Determinants of a dosage regimen for an anticancer drug

8  Pharmacokinetic Studies in Early Anticancer Drug Development

191

spatial-temporal behavior of the drug in the body will be helpful, if not essential, in understanding and predicting normal tissue toxicity and optimizing tumor response. In view of the multiple factors that can cause drug concentrations to vary after administration of a fixed dose, it is clearly much more meaningful to have knowledge of drug exposure measures, usually expressed as the area under the concentration-time curve (AUC), rather than of absolute dose only. Ideally, this would be at the molecular locus of action or at least at the tissue or tumor level but, with the exception of certain hematological malignancies, drug concentrations are commonly measured only in the plasma or blood cells as the only readily accessible surrogate. As mentioned previously, there is often a marked variation in drug handling between individual patients resulting in variability in AUC (or clearance). Figure 8.2 illustrates interindividual variability in apparent drug clearance, expressed as the coefficient of variation, for selected anticancer agents given either intravenously or orally. The coefficient of variation (CV) is defined as the standard deviation divided by the mean, providing a scaled version of standard deviation. Variation in drug clearance often leads to variability in the pharmacodynamic effects of a given dose

Intersubject variability in clearance (%)

100

80

60

40

20

B B us Ca orte ulfa Cy pe zo n clo c m ph C itab ib os is ine ph pla Cy ami tin ta de D ra Do oc bine xo eta r x Et ubi el Fl op cin uo os ro id Iri ura e no c il M Mel teca et ph n ho a t la Pa rexa n TePem clita te m e xe Te ozo trex l m lom ed sir i o de Tr Top limu ox ot s ac ec Vi itab an nc in ris e tin e Da sa t Ev Erlo inib er tin oli ib Ge mu f s Im itini La atin b pa ib N tin So ilotinib ra ib S fen Vo unit ib rin inib os ta t

0

Intravenous drugs

Oral drugs

Fig. 8.2  Interindividual pharmacokinetic variability of selective anticancer agents administered intravenously (white bars) or orally (black bars). Data are expressed as percent coefficient of variation (CV × 100%) in apparent (oral) clearance. Modified from [65]

192

A. Sparreboom and S.D. Baker

of a drug. That is, an identical dose of drug may result in acceptable toxicity in one patient, and unacceptable and possibly life-threatening toxicity in another; or clinical response in one individual and cancer progression in another. Although the coefficient of variation in clearance is a useful descriptive statistic that provides a relative measure of the dispersion in a given population, a more clinically informative measure is the fold difference (or range) in clearance. For example, Fig.  8.3 illustrates steady-state exposure for the tyrosine kinase inhibitor, imatinib, in a population of 82 patients with gastrointestinal stromal tumors [26]. This analysis indicates that up to 60-fold differences between patients can be observed in steadystate concentration of imatinib, suggesting that a subset of individuals with very low steady-state concentrations is unlikely to obtain clinical benefit from continued administration of imatinib at the approved standard dose, and a subset at the higher end of concentrations is more likely to experience severe toxicities [6, 52, 39]. A combination of physiological variables, genetic characteristics, and environmental factors are known to alter the relationship between the absolute dose and the concentration-time profile in plasma. Indeed, the correlation between, for example, the AUC or steady-state concentration of a drug in plasma and the intensity of pharmacodynamic effects is commonly better than that between absolute dose and the intensity of such effects. The definition of the relationships between the pharmacokinetic variables of a drug and the drug’s pharmacodynamic endpoints may allow the administration of the optimum dosage of that drug in any given patient. The optimum dosage is in fact that dose which maximizes the likelihood of response and simultaneously minimizes the likelihood of toxicity in a particular patient. In incidental cases, it is possible to define the optimum dosage to achieve a required drug exposure measure a priori for an individual patient from measurable physiological variables such as renal or hepatic function. In most cases, however, dose

Cumulative frequency (%)

60.8-fold 100 75 50 25 0 100

1000 10000 Imatinib Css (ng/mL)

100000

Fig. 8.3  Cumulative frequency of imatinib steady-state concentration in 82 patients with gastrointestinal stromal tumors treated with oral imatinib. Data were normalized to a total dose of 400 mg. The vertical lines represent the median (solid line) and the 25th and 75th percentile (dotted lines). Data obtained from [26]. Note x-axis is plotted on a log-scale

8  Pharmacokinetic Studies in Early Anticancer Drug Development

193

adjustments will be required in the light of pharmacokinetic or pharmacodynamic data obtained after an initial dose and also possibly subsequent doses of the drug in the individual patient.

8.3 Establishing Pharmacokinetic–Pharmacodynamic Relationships 8.3.1 Preclinical Development In general, the preclinical data available before a candidate anticancer drug is entered into clinical trials include the in vitro cytotoxic activity in various animal and human tumor cell lines, and the in  vivo antitumor activity in mouse tumor models or in human tumors xenografted in nude mice. In addition to these antitumor evaluations, toxicological studies are conducted in a variety of species to determine the toxic doses such as the lethal dose for 10% of the animals (LD10). These in vitro evaluations clearly have value in defining target plasma drug concentrations, although direct extrapolation is often difficult because of oversimplification of these models, as usually no correction is applied for the lack of metabolic transformation (or any other form of drug elimination), the presence of physiological barriers, and/or differences in the extent of serum protein binding [2]. Nonetheless, these studies may guide the choice of administration schedules and starting dosage in phase I clinical trials. For example, if drug activity is S-phase specific, prolonged exposure may be required that would be better achieved using a prolonged intravenous infusion if that agent has a relatively short terminal half-life. The preclinical data have also shown to be valuable in defining a therapeutic window to be reached in patients. Theoretically, it is possible to define two-dimensional plasma-concentration time windows, the lower limit (or threshold concentration) of which is associated with antitumor activity, and the upper (or toxic concentration) with an unacceptable degree of toxic side effects [12]. This approach is the basis of therapeutic drug monitoring which is widely used and accepted also for numerous noncancer drugs, including calcineurin inhibitors and antiretroviral agents [35].

8.3.2 Clinical Development 8.3.2.1 Choice of a Starting Dose Chapter 5 deals with the design, conduction, and analysis of phase I clinical trials with anticancer agents. Historically, the empiric starting dose for anticancer agents is based on toxicological studies in rodents and dogs, and 1/10 of the murine LD10

194

A. Sparreboom and S.D. Baker

is often chosen for the first dose level to be used in patients because at this dose intolerable toxicity is rarely encountered [43]. Based on theoretical considerations in addition to an extensive review of preclinical toxicological studies from literature data, it has been proposed that limited toxicological models using mouse and rat data only can appropriately and safely be used for anticancer drug development [50, 19]. It is important to realize, however, that compared with patients, much higher plasma concentrations of (experimental) anticancer agents can generally be achieved in animals, which cannot always be explained by a proportionally higher drug clearance. When trying to extrapolate results from efficacy studies performed in tumor-bearing animals to the clinical situation, it should thus be taken into consideration that tumors are exposed to drug levels which can, in most cases, never be achieved in patients. Clearly, this is particularly relevant when the relationships between plasma levels and antitumor effects are poorly understood. 8.3.2.2 Dose-Escalation Schemes If the starting dose is not severely toxic in patients, further dose escalation is usually based on the modified Fibonacci sequence, in which escalating steps have decreasing relative increments (e.g., 100, 67, 50%, etc.). Each succeeding step is typically continued in cohorts of three patients until dose-limiting toxicity is reached. The major limitations of this standard design related to ethical considerations and efficiency have been reviewed [19], and numerous alternative methodologies have been proposed. One of these approaches was designed to rationalize and accelerate dose escalation by making use of preclinical pharmacologic data. This process of pharmacokinetically guided dose escalation assumes that for agents showing no major differences in target cell sensitivity, schedule dependence, or toxicity between animals and patients, the AUC at the murine LD10 and the AUC at the human maximum-tolerated dose should be similar. This approach has proven useful in choosing a more appropriate dose levels to be studied in trials of several agents over the past decade [19], thereby reducing the number of patients exposed to theoretically subtherapeutic dosages of the drug involved. Other groups have advocated similar approaches to escalate systemic exposure measures rather than dose to determine the maximum-tolerated systemic exposure in pediatric populations, and have used it in trials with a variety of agents, including topotecan, teniposide, carboplatin, and paclitaxel [22, 72]. Both concepts have evident value in reducing the number of patients required for trials, but have proven difficult to adopt in the clinical setting because of a number of pitfalls, including the assumption of linear pharmacokinetics between species, the substantial interpatient variability in results, as well as logistic issues of obtaining real-time data enabling subsequent escalation steps. Currently, there is clearly a need for more in vitro and in vivo preclinical pharmacologic studies in an effort to rationalize the transition between animals and patients, which still remains a difficult and hazardous task, and a rather empirical exercise. Because of this, even the most recently proposed dose-escalation methods are still largely based on empiric experience.

8  Pharmacokinetic Studies in Early Anticancer Drug Development

195

8.3.2.3 Obtaining Parameter Estimates Pharmacokinetic studies performed during early clinical development should be applied to define pharmacokinetic parameters at different dose levels, including peak concentration, drug clearance, half-lives, volume of distribution, and metabolic profiles. This information can be used subsequently to design more rational schedules of administration and define the therapeutic window of the agent, and may provide clues as to potentially cumulative toxicity or specific toxicity associated with organ dysfunction. The detection of metabolites and the description of their pharmacokinetic behavior are also extremely important, particularly in case of agents that require metabolic activation, such as irinotecan and capecitabine. The generated parameters are tentatively correlated with the observed clinical outcome, particularly in terms of toxicity, in an effort to define pharmacokinetic– pharmacodynamic relationships that may be of use in further clinical testing of the agent involved. In addition, dose–effect relationships, at least in terms of toxicity, can best be assessed at this early step in clinical development, where a wide range of doses is being administered to the patients, contrary to the other clinical phases of drug development. Once a dose is selected, the same dose is usually maintained throughout the treatment unless serious toxicity occurs, in which case the dose is decreased for subsequent treatment courses. In contrast, the dose of chemotherapy is rarely increased in the absence of toxicity, even though this might be a reason for treatment failure. Although wide interpatient variability has been demonstrated in all aspects of anticancer drug pharmacokinetics, several studies have demonstrated reasonably predictive relationships between some measure of drug exposure and either toxicity or antitumor efficacy, which would add support to the arguments to increase the dose in case of limited toxicity.

8.4 Sources of Pharmacokinetic Variability 8.4.1 Drug Scheduling and Administration Sequencing The antitumor activity of certain chemotherapeutic agents is highly schedule dependent. For these drugs, the dose fractionated over several days can produce a different antitumor response or toxicity profile compared with the same dose given over a shorter period. In the first definitive demonstration of schedule dependency in oncology, investigators documented markedly increased efficacy in patients with small-cell lung cancer when an identical total dose of etoposide was administered by a 5-day divided-dose schedule rather than a 24-h infusion [57]. Pharmacokinetic analysis in that study showed that both schedules produced very similar overall drug exposure (as measured by AUC), but that the divided-dose schedule produced twice the duration of exposure to an etoposide plasma concentration of >1 mg/mL.

196

A. Sparreboom and S.D. Baker

This observation was consistent with preclinical data, and the authors speculated that exposure to this threshold concentration was important in achieving clinical efficacy, while exposure to higher plasma concentrations augmented drug-induced toxicity. Subsequently, this finding has led to the use of prolonged oral administration of etoposide to treat patients with cancer [31]. Similar schedule dependency has also been demonstrated for a number of other anticancer agents, notably paclitaxel and topotecan. For both these agents, the variability in clinically tested treatment schedules is enormous, ranging from short i.v. infusions of less than 30 min to 21-day or even 7-week continuous infusion administrations, with large differences in experienced toxicity profiles and with fortuitous implications for the drug’s pharmacokinetic profiles. In most cases, the mechanisms underlying the schedule dependencies are not completely understood, and further investigation will be essential to determine the nature of this and to define optimal durations of drug administration. An opposite effect has been observed with the administration of purine and pyrimidine analogs where toxicity is accentuated or different with prolonged infusions relative to shorter infusions.

8.4.2 Body Size and Body Composition The traditional method of individualizing anticancer drug dosage is by using body-surface area (BSA) [30]. A number of considerations have contributed to BSA having conventionally become the single variable to dose anticancer agents. First, it is assumed that a correlation exists between BSA and some particular characteristics of each patient, such as glomerular filtration rate (GFR), blood volume, and basal metabolic rate, and certainly this provides a condition to individualize doses [24]. A similar relationship with liver function has not been made, which is particularly noteworthy as the metabolic pathways of many drugs are strictly related to the activity of hepatic enzymes. More recently, however, these basic principles have been, in part, questioned by a study where a poor correlation between BSA and GFR was reported [16]. Second, as mentioned previously, the starting dose of agents calculated in phase I studies is based on data derived from animal models where drug dose is calculated relative to weight (mg/ kg) or BSA (mg/m2). In animals, doses are usually tested until the LD10 (10% of lethal dose), and in human phase I studies the first dose employed is 1/10 of LD10. Third, some studies published in the 1950s suggested a role of BSA in drug dose calculation, when attempts were made to define a more accurate method for the administration of cytotoxic drugs in children. Among these, a retrospective analysis was reported in 1958: the results of applying a BSA-based formula in adults and in children, and Meeh’s formula in animals, to determine the conventional pediatric and adult dose of ­several cytotoxic drugs [53]. Pinkel calculated the doses per unit surface area, and found similar figures for most agents tested between children and adults, and recommended normalizing the doses of cytotoxic agents using BSA.

8  Pharmacokinetic Studies in Early Anticancer Drug Development

197

Estimation of BSA is most commonly achieved using a formula that was derived primarily for its use of basal metabolism by the use of weight and height alone, and more recently, it has been confirmed that the original formula was surprisingly accurate considering the small sample size used in its derivation [30]. The usefulness of normalizing anticancer drug dose to BSA in adults has been questioned recently, since it clearly has been shown that for most drugs, there is no relationship between BSA and anticancer drug clearance in adults [3]. Likewise, attempts to replace BSA as a size metric in dose calculation with alternate descriptors such as lean body weight, either in an average population or in individuals at the outer extremes of weight (i.e., frail or severely obese patients), have failed for many anticancer agents [66, 44]. It should be pointed out that BSA is probably a much more important consideration in drug dose calculation for pediatric patients as compared to adults, because of the larger size range in the former population [4]. In part based on the failure to reduce interindividual pharmacokinetic variability with the use of BSA-normalization to obtain a starting dose, many of the more recently developed molecularly targeted agents such as the tyrosine kinase inhibitors are currently administered using a flat-fixed dose irrespective of an individual’s BSA.

8.4.3 Age Changes in body composition and organ function at the extremes of age can affect both drug disposition and drug effect [47]. For example, maturational processes in infancy may alter absorption and distribution of drugs as well as change the capacity for drug metabolism and excretion. The importance of understanding the influence of age on the pharmacokinetics and pharmacodynamics of individual anticancer agents has increased steadily as treatment for the malignancies of infants and the elderly has advanced. Although the influence of age has been evaluated formally for a limited number of drugs, reviewing our current understanding of how maturation from birth to young adulthood, and subsequent senescence into advanced age influence the various approaches for individualizing drug treatment to enhance the chance of therapeutic success. Although pediatric cancer remains a rare disease compared with cancer in adults and particularly the elderly population, optimizing treatment in a patient group with a high cure rate and a long expected survival becomes critical to minimize the incidence of preventable late complications while maintaining efficacy. 8.4.3.1 Age-Related Absorption Changes Gastric emptying time varies with gestational age and may be prolonged in premature infants and neonates compared with older children [67]. Gastric pH is neutral in the first few weeks after birth, and gradually declines to adult values by the age

198

A. Sparreboom and S.D. Baker

of 2, which may affect the bioavailability of compounds [69]. The implication of higher pH in delayed absorption of weak acids and an increased absorption of weak bases has been demonstrated more recently for tyrosine kinase inhibitors and other oral agents, as highlighted in Table 8.1. In addition, drugs may be administered in altered forms to small children. For example, crushed tablets in food or slurries administered by naso-gastric tube may alter their rate and extent of absorption. 8.4.3.2 Age-Related Volume of Distribution Changes Changes in body composition among birth, adolescence, and advanced age may alter a number of pharmacokinetic parameters. For example, skeletal muscle mass and subcutaneous fat is reduced in the newborn compared with the older infant. Changes in the proportion of body water compartments, particularly extracellular fluid volume, are dramatic between birth and adulthood; extracellular fluid volume represents 50% of body weight for premature infants, 35% for infants 4–6 months old, and 20% of adolescent and adult volume. Polar drugs, which distribute primarily in body water, will therefore have a larger volume of distribution in infants compared with older children and adults. The net result of an isolated increase in volume of distribution is a lower peak concentration and prolonged terminal halflife, which occurs if the drug clearance remains unchanged [63].

Table 8.1  Interactions between oral anticancer agents and drugs that alter the pH of the upper gastrointestinal tract (selected list) Drug pH Solubility pH elevating agent Effect on drug exposure Dasatinib 2.6 18.4 mg/mL Antacid (concurrent) ↓ AUC 55% ↓ Cmax 58% 6.0 0.008 mg/mL Antacid (2 h prior) No change Famotidine ↓ AUC 61% ↓ Cmax 63% Erlotinib >5.0 Decreased Omeprazole ↓ AUC 46% ↓ Cmax 61% Gefitinib >5.0 Drops sharply Ranitidine ↓ AUC 44% Imatinib 5.8 Very good Antacid (15 min prior) No change Omeprazole No change NR Lapatinib 0.001 mg/mL in 0.1 N HCl 0.007 mg/mL in water Nilotinib Decreases with NR increasing pH Sorafenib Decreases with NR increasing pH Sunitinib £6.8 >25 mg/mL NR NR Everolimus <0.1 mg/mL in water, 0.1 N HCl, and citrate buffer (pH 2.0–10.0) Vorinostat 7.4 0.1–0.2 mg/mL NR AUC area under the plasma concentration-time curve, Cmax maximum plasma concentration, NR not reported

8  Pharmacokinetic Studies in Early Anticancer Drug Development

199

Drug distribution may also be affected by alternations in plasma proteins. Protein binding may be reduced because of persistence of fetal albumin and reduced plasma protein content, particularly low levels of gamma-globulin. Although adult values of protein for binding of acidic drugs may be achieved between 1 and 2 years of age, adult values for gamma-globulin are not reached until age 7–12. One unique aspect of the immature infant is the gradual maturation of specific organs. For example, the myelin content of the brain is lower in newborns. Because of incomplete maturation of the blood–brain barrier, membrane permeability is greater in the infant, and the brain to plasma ratio of some drugs has been shown to vary with age [1]. Similarly, renal tubular function develops later than glomerular filtration, so that drugs that undergo substantial tubular resorption in older children and adults may have a larger clearance than is expected in the infant. 8.4.3.3 Age-Related Changes in Renal Function The immature kidney has a diminished GFR. For full-term infants, the GFR is 40  mL/min/1.73  m2, with substantial interindividual variability [48]. The GFR decreases to maximal values between the age of 6 and 12 months, and may further deteriorate with advanced age. Tubular function and passive resorption may also be significantly lower in the neonate. Because tubular resorption matures at a slower rate than glomerular filtration, toddlers may have remarkably high clearance rates for compounds that undergo tubular resorption in the older children. The GFR assessed by 99mTc-DTPA in children between 2 months and 17 years was correlated with BSA, but when GFR was normalized to BSA, there was no correlation between GFR and age [56]. Premature and term infants require dose adjustments based on measurements or estimation of GFR for drugs that are eliminated primarily by glomerular filtration, such as carboplatin. Formulas are available for the estimation of creatinine clearance for infants and older children based on creatinine and height, or by using some other endogenous surrogate marker of GFR, like cystatin C, but height can be difficult to measure accurately in infants [29]. 8.4.3.4 Age-Related Changes in Hepatic Metabolism Immature organ systems in preterm and term infants and young children can cause altered disposition for many classes of drugs, including antibiotics, anticonvulsants, and antineoplastic agents [63]. Although the activity of cytochrome P450 (CYP) microsomal enzymes responsible for the oxidation may be low in neonates, a dramatic increase in the metabolic capacity occurs between 8 weeks and 3 years, with rates increasing from 20% of adult clearance to two- to sixfold higher rates than adults. Values then decline gradually to reach adult clearance rates in puberty. Glucuronidation enzymes increase from birth to 30% of adult activity by the age of 3  months. Esterase activity, metabolically important for

200

A. Sparreboom and S.D. Baker

agents such as irinotecan and capecitabine, gradually increases during the first year of life; reduced hydrolysis rates have been reported for drugs like procaine in premature infants and neonates, with normal activity measurable by 12  months. Studies of hepatic metabolism and elimination in adolescents are very limited. In general, however, adolescents have clearances that are intermediate between the high clearance rates observed in toddlers and the lower clearance rates measured in adults.

8.4.4 Pathophysiological Changes 8.4.4.1 Effects of Disease Pathophysiologic changes associated with particular malignancies may cause dramatic alterations in drug disposition. For example, increases in the clearance of both antipyrine and lorazepam were noted after remission induction compared with the time of diagnosis in children with acute lymphoblastic leukemia (ALL) [55]. The clearance of unbound teniposide is lower in children with ALL in relapse than during first remission [23]. Because leukemic infiltration of the liver at the time of diagnosis is common, drugs metabolized by the liver may have a reduced clearance, as has been documented in preclinical models [54]. Furthermore, in mouse models, certain tumors elicited an acute phase response that coincided with downregulation of human CYP3A4 in the liver as well as the mouse ortholog Cyp3a11 [10]. The reduction of murine hepatic Cyp3a gene expression in tumor-bearing mice resulted in decreased Cyp3a protein expression and consequently a significant reduction in Cyp3a-mediated metabolism of midazolam. These findings support the possibility that tumor-derived inflammation may alter the pharmacokinetic and pharmacodynamic properties of CYP3A4 substrates, leading to reduced metabolism of drugs in humans [49], and suggest the possible need for disease-specific design of early clinical trials. 8.4.4.2 Effects of Renal Impairment The potential impact of pathophysiological status on interindividual pharmacokinetic variability can be due to either the disease itself or to dysfunction of specific organs involved in drug elimination. For example, if urinary excretion is an important elimination route for a given drug, any decrement in renal function could lead to decreased drug clearance and may result in drug accumulation and toxicity. It would therefore be logical to decrease the drug dose relative to the degree of impaired renal function, in order to maintain plasma concentrations within a target therapeutic window. The best known example of this a priori dose adjustment of an anticancer agent remains carboplatin, which is excreted renally almost entirely by glomerular filtration. Various strategies have been developed

8  Pharmacokinetic Studies in Early Anticancer Drug Development

201

to estimate carboplatin doses based on renal function among patients, either using creatinine clearance [18] or GFRs as measured by a radioisotope method [8]. Application of these procedures has led to a substantial reduction in pharmacokinetic variability, such that carboplatin is currently one of the few drugs routinely administered to achieve a target exposure rather than on a mg/m2 or mg/kg basis. The US Food and Drug Administration is currently developing a guidance for industry to replace a previous guidance on the impact of renal impairment on the pharmacokinetics, dosing, and labeling of drugs issued in May 1998 [34]. The impact of this initial guidance was recently assessed following a survey of 94 new drug applications for small-molecule new molecular entities approved over the past 5  years (2003–2007). The survey results indicated 41% of the applications that included renal impairment data resulted in a recommendation of dose adjustment in renal impairment [73]. Interestingly, the survey results provided evidence that renal impairment can affect the pharmacokinetics of drugs that are predominantly eliminated by nonrenal processes such as metabolism and/or active transport. The latter finding supports the FDA recommendation to evaluate pharmacokinetic–pharmacodynamic alterations in renal impairment for those drugs that are mainly eliminated by nonrenal processes, in addition to those that are mainly excreted unchanged by the kidney. A striking example of a drug in the former category is imatinib, an agent that is predominantly eliminated by hepatobiliary pathways but where predialysis renal impairment is associated with dramatically reduced drug clearance [27]. 8.4.4.3 Effects of Hepatic Impairment In contrast to the predictable decline in renal clearance of drugs when glomerular filtration is impaired, it is not as easy to make a general prediction on the effect of impaired liver function on drug clearance. The major problem is that commonly applied criteria to establish hepatic impairment are typically not good indicators of drug-metabolizing activity and that several alternative hepatic function tests, such as indocyanine green and antipyrine, have limited value in predicting anticancer drug pharmacokinetics (and metabolism). An alternative dynamic measure of liver function has been proposed which is based on totaled values (scored to the WHO grading system) of serum bilirubin, alkaline phosphatase, and either alanine aminotransferase or aspartate aminotransferase to give a hepatic dysfunction score [70]. Based on pharmacokinetic studies in patients with normal and impaired hepatic function, guidelines have been proposed for dose adjustments of several agents when administered to patients with severe liver dysfunction. Examples for more recently approved agents are shown in Table 8.2. It should be emphasized that no uniform criteria have been used in the conduct of these studies, and that ultimately substantial advances could be made through a priori determination of the hepatic activity of enzymes of relevance to the chemotherapeutic drug(s) of interest, as was done for docetaxel [32].

↓ 5 mg (50% reduction)

No dose adjustment

400 mg twice daily 200 mg twice daily <200 mg every third day 200 mg once daily

No dose adjustment

↓ 25% No dose adjustment ↓ 750 mg (40% reduction)

No dose adjustment

No dose adjustment

↓ 75 mg (50% reduction)

Recommended dose adjustment No dose adjustment

NR not reported, AUC area under the plasma concentration-time curve (dose-normalized), CL clearance, HF hepatic function, ULN upper limit of institutional normal, AlkPhos alkaline phosphatase, bili bilirubin, SGOT serum glutamic oxaloacetic transaminase (also named AST, aspartate aminotransferase), Tbili total bilirubin

Table 8.2  Recommended dosing of anticancer agents based on varying degrees of hepatic impairment (selected list) Hepatic function group (defined by the degree of Drug impairment) Effect of hepatic impairment on drug exposure Dasatinib Moderate (Child-Pugh class B) ↓ AUC 8%, ↓ Cmax 47% Severe (Child-Pugh class C) ↓ AUC 28%, ↓ Cmax 43% Erlotinib SGOT ³ 3 × ULN ± albumin < 25 g/dL ↓ CL 41–63% Direct bili 1–7 mg/dL Gefitinib Hepatic metastasis and↑ SGOT, Tbili, AlkPhos: Moderate Similar to normal HF Severe Imatinib Any or no elevations in SGOT and: Mild (Tbili >1.0–1.5 × ULN) Similar to normal HF Moderate (Tbili >1.5–3 × ULN) Severe (Tbili >3–10 × ULN) ↑ Steady-state AUC 50% Lapatinib Moderate (Child-Pugh class B) ↑ AUC 14% Severe (Child-Pugh class C) ↑ AUC 63% Sorafenib Patients with HCC and: Mild (Child-Pugh class A) Exposure similar between class A and class B; ↓ 23–65% compared with non-HCC patients Moderate (Child-Pugh class B) Mild (bili >1 to £1.5 × ULN; and/or AST > ULN) No significant associations between HF groups and AUC Moderate (bili >1.5 to £3 × ULN; any AST) Patients with moderate-to-severe hepatic Severe (bili >3 to 10 × ULN; and any AST) impairment did not tolerate doses Very severe (albumin <2.5 mg/dL; any bili and AST) recommended for patients with normal HF Sunitinib Moderate (Child-Pugh class B) Similar to normal HF Severe (Child-Pugh class C) Everolimus Moderate (Child-Pugh score B) ↓ CL 53%

202 A. Sparreboom and S.D. Baker

8  Pharmacokinetic Studies in Early Anticancer Drug Development

203

8.4.4.4 Effects of Serum Proteins The binding of drugs to serum proteins, particularly those that are highly bound, may also have significant clinical implications for therapeutic outcome [28]. Although protein binding is a major determinant of drug action, it is clearly only one of a myriad of factors that influence the disposition of most anticancer drugs, with very few exceptions (e.g., the staurosporine analog UCN-01) [59]. The extent of protein binding is a function of drug and protein concentrations, the affinity constants for the drug–protein interaction and the number of proteinbinding sites per class of binding site. Since only the unbound (or free) drug in plasma water is available for diffusion from the vascular compartment to the tumor interstitium, the therapeutic response will correlate with free drug concentration rather than total drug concentration. Several clinical situations, including liver and renal disease, can significantly decrease the extent of serum binding and may lead to higher free drug concentrations and possible risk of unexpected toxicity, although the total (free plus bound forms) plasma drug concentrations are unaltered [62]. It is important to realize, however, that after therapeutic doses of most anticancer drugs, plasma binding is drug concentration-independent, suggesting that the total plasma concentration is reflective of the unbound concentration. Thus, other physiologic changes, for instance, decreased renal and hepatic function, generally produce more clinically significant alterations in drug disposition than that seen with alterations in plasma protein binding. For some anticancer agents, including etoposide [51] and paclitaxel [64], it has been shown that protein binding is highly dependent on dose- and schedule-varying plasma concentrations.

8.4.5 Sex Dependence Additional changes in body composition, particularly changes that differ in males and females, occur with the onset of puberty. In adolescence, boys actually lose body fat (to a mean of 12% of body weight), whereas girls increase the proportion of body weight composed of fat by up to 25%. These changes suggest that during adolescence, sex-related differences in both volume of distribution and clearance could be more prominent than they are in either children or older adults. A number of population pharmacokinetic analyses have suggested that male gender is positively correlated with the higher elimination capacity of various anticancer drugs (e.g., paclitaxel) or with increased clearance (e.g., imatinib) compared with female gender. These observations have added to a growing body of evidence that the pharmacokinetic profile of various anticancer drugs exhibits significant sexual dimorphism, which is rarely considered in the design of clinical trials during oncology drug development.

204

A. Sparreboom and S.D. Baker

8.4.6 Drug Interactions 8.4.6.1 Coadministration of Other Chemotherapeutic Drugs The vast majority of pharmacologic studies of anticancer agents have assessed the effects of a single drug. However, as a consequence of somatic mutations, tumor cell kill tends to decrease with subsequent courses of treatment, and since genetically resistant cell types are selected out, single-agent treatment is rarely curative. Therefore, and for a variety of other reasons, modern cancer chemotherapy is most frequently given as a combination of different drugs. Favorable and unfavorable interactions between drugs must be considered in developing such combination regimens. These interactions may influence the effectiveness of each of the components of the combination, and typically occur when the pharmacokinetic behavior of one drug is altered by the other. These interactions are important in the design of trials evaluating drug combinations because, occasionally, the outcome of concurrent drug administration is diminished therapeutic efficacy or increased toxicity of one or more of the administered agents. Pharmacokinetic interactions between chemotherapeutic agents are usually evaluated and identified in phase I and II trials. In addition to pharmacokinetic interactions, combinations of drugs might also show pharmacodynamic interactions that cannot be explained by altered pharmacokinetic profiles. Some of these interactions are at the cellular level or are cell cycle related and can be classified as synergistic, additive, or antagonistic. If pharmacokinetic and pharmacodynamic interactions exist, the drug doses and sequences that allow safe administration of combination chemotherapy are typically defined during early clinical evaluation.

8.4.6.2 Coadministration of Nonchemotherapeutic Drugs Many prescription and over-the-counter medications have the potential to pharmacokinetically interact with anticancer agents, altering their pharmacokinetic characteristics and leading to clinically significant interactions. Over 100,000 deaths per year in the USA alone can be attributed to such drug interactions, placing drug interactions between the fourth and sixth leading cause of death [40]. It is obvious that all aspects of pharmacokinetics might be affected when a drug is given in combination with another drug, including absorption (resulting in altered absorption rate or oral bioavailability), distribution (mostly caused by protein-binding displacement), metabolism, and excretion. However, most known drug interactions are due to changes in metabolic routes related to altered expression or functionality of cytochrome P450 (CYP) isozymes. This class of enzymes, particularly the CYP3A4 isoform, is responsible for the oxidation of the majority of currently prescribed anticancer drugs, resulting in more polar and usually inactive metabolites. Elevated CYP activity (induction), translated into a more rapid metabolic rate, may result in

8  Pharmacokinetic Studies in Early Anticancer Drug Development

205

a decrease in plasma concentrations and to total loss of therapeutic effect. For example, anticonvulsant drugs (e.g., phenytoin, phenobarbital, and carbamazepine) induce drug-metabolizing enzymes and thereby increase the clearance in children of various anticancer agents. Conversely, suppression (inhibition) of CYP activity may trigger a rise in plasma concentrations and lead to exaggerated toxicity commensurate with overdose. It should be born in mind that several pharmacokinetic parameters could be altered simultaneously. Especially in the development of anticancer agents given by the oral route, oral bioavailability plays a crucial role [15]. This parameter is contingent on adequate intestinal absorption and the circumvention of intestinal and, subsequently, hepatic metabolism of the drug. One of the principal mechanisms that can explain interactions with anticancer agents given orally is the affinity for solute carriers and/or ATP-binding-cassette transporters expressed in the intestinal epithelium and directed toward the gut lumen (Fig. 8.4). Extraction of anticancer drugs by extensive metabolism in the gut wall and/or the liver during first-pass (i.e., prior to reaching the systemic circulation) is another potential mechanism involved in suspected interactions for various agents. An ideal chemotherapeutic drug would have an adequate absolute bioavailability and little interpatient and intrapatient variability in absorption. For the most commonly used oral agents, including Oral dose IV dose

Liver

SLC

SLC

ABC

drug

drug

SLC

drug

metabolite

SLC

CYP

portal vein ABC

ABC Systemic Circulation

CYP GI Tract

metabolite

uct ile d

ABC

b

Tumor

SLC drug

CYP

metabolite

ABC

Fig.  8.4  Role of ATP-binding-cassette transporters, solute carriers, and drug-metabolizing enzymes expressed in the intestinal epithelium, liver, and tumor, and in drug absorption, distribution, and elimination. ABC ATP-binding-cassette transporters, SLC solute carriers, CYP cytochrome P450 enzymes

206

A. Sparreboom and S.D. Baker

i­matinib, EGFR tyrosine kinase inhibitors, etoposide, cyclophosphamide, and methotrexate, however, these criteria are not met. An additional consideration adding to the complexity is related to a possible influence of food intake on the extent of drug absorption after oral administration, which can increase, decrease, or remain unchanged depending on specific physicochemical properties of the drug in question (Table 8.3). The relatively narrow therapeutic index of most of these agents means that significant inter- and intrapatient variability would predispose some individuals to excessive toxicity or, conversely, inadequate efficacy. 8.4.6.3 Coadministration of Complementary and Alternative Medicine In recent years, interest in complementary and alternative medicine (CAM) has grown rapidly in the industrialized world. Some of the reasons for this increase relate to dissatisfaction with conventional allopathic therapies, a desire of patients and parents to be involved more actively in their own and their children’s health care, and because patients/parents find these alternatives to be more congruent with their own philosophical orientations [60]. Surveys within the past decade estimate the prevalence of CAM use in pediatric oncology to be between 31 and 87%, and in many cases the treating physician is unaware of the patients’ CAM use. With a larger number of children [46] and participants to phase I clinical trials [17] using herbal treatments combined with allopathic therapies, the risk for herb–drug interactions is a growing concern, and there is an increasing need to understand possible adverse drug interactions in oncology at early stage of drug development. During the last decade, a wealth of evidence has been generated showing that many herbal preparations interact extensively with drug-metabolizing enzymes and drug transporters. A number of clinically important pharmacokinetic interactions have now been recognized, although causal relationships have not always been established and confirmatory studies in children are currently lacking. Most of the observed interactions point to the herbs affecting several isoforms of the CYP family, either through inhibition or induction. These enzymes have a crucial role in the elimination of the majority of investigational and approved anticancer drugs, and concurrent use of some herbs with chemotherapy is destined to have serious clinical and toxicological implications. Therefore, rigorous testing for possible pharmacokinetic interactions of anticancer drugs with widely used herbs is urgently required. In the context of chemotherapeutic drugs, only St. John’s wort [45], garlic [13], and milk thistle [71] have been formally evaluated for their drug-interaction potential in vivo. However, various other herbs have the potential to significantly modulate the expression and/or activity of drug-metabolizing enzymes and drug transporters, including ginkgo, echinacea, ginseng, and kava [60]. Because of the high prevalence of herbal medicine use in the USA, physicians should include herb usage in their routine drug histories in order to appropriately advise individual patients in which potential hazards should be taken into consideration prior to participation in a clinical trial.

High-fat breakfast High-fat meal High-fat meal

Low-fat meal (5% fat, 500 calories) High-fat meal (50% fat, 1,000 calories) High-fat meal Moderate-fat meal (30% fat, 700 calories) High-fat meal (50% fat, 900 calories) High-fat, high-calorie meal High-fat meal High-fat meal

Gefitinib

Lapatinib

Effect on drug exposure ↑ AUC 14% Single dose: ↑AUC 200% Multiple dose: ↑AUC 37–66% ↓ AUC 14%, ↓ Cmax 34% ↑ AUC 32%, ↑ Cmax 37% No change Variability (% CV) ↓ 37% ↑ AUC 167%, ↑ Cmax 142% ↑ AUC 325%, ↑ Cmax 203% ↑ AUC 82% No change in bioavailability ↓ Bioavailability 29% ↑ AUC 18% ↓ AUC 16%, ↓ Cmax 60% ↑ AUC 37% With or without food With or without food With foodd

Without food Without food

Without foodc

With food and a large glass of waterb

With or without food

Manufacturer’s recommendations for administration With or without food Without fooda

AUC area under the plasma concentration-time curve, Cmax maximum plasma concentration a  Recommended without food as the approved dose is the maximum-tolerated dose b  Recommended with food to reduce nausea c  Recommended without food to achieve consistent drug exposure; was taken without food in clinical trials d  Was taken with food in clinical trials

Sunitinib Everolimus Vorinostat

Nilotinib Sorafenib

Imatinib

Food High-fat meal High-fat, high-calorie breakfast

Drug Dasatinib Erlotinib

Table 8.3  Effect of food on exposure to oral anticancer agents (selected list)

8  Pharmacokinetic Studies in Early Anticancer Drug Development 207

208

A. Sparreboom and S.D. Baker

8.4.7 Inherited Genetic Factors The discipline of pharmacogenetics describes differences in the pharmacokinetics and pharmacodynamics of drugs as a result of inherited variation in drug-metabolizing enzymes, drug transporters, and drug targets between patients. These inherited differences in enzymes and receptors are occasionally responsible for extensive interpatient variability in drug disposition (systemic exposure) or effects (normal tissue and tumor exposure). Severe toxicity might occur in the absence of typical metabolism of active compounds, while the therapeutic effect of a drug could be diminished in the case of absence of activation of a prodrug. The importance and detectability of polymorphisms for a given enzyme depends on the contribution of the variant gene product to pharmacological response, the availability of alternative pathways of metabolism, and the frequency of occurrence of the least common variant allele. Although many substrates have been identified for the known polymorphic drugmetabolizing enzymes and transporters, the contribution of a genetically determined source of interindividual pharmacokinetic variability has been established for only very few cancer chemotherapeutic agents. Most of these cases involve agents for which elimination is critically dependent on a rate-limiting breakdown by a polymorphic enzyme (e.g., 6-mercaptopurine by thiopurine-S-methyltransferase, flurouracil by dihydropyrimidine dehydrogenase, and SN-38 by UDP-glucuronosyltransferase 1A) or when a polymorphic enzyme is involved in the formation of a toxic metabolite (e.g., tamoxifen by CYP2D6) [33]. In addition to drug metabolism, pharmacokinetic processes are highly dependent on the interplay with drug transport in organs such as the intestine, kidney, and liver. Genetically determined variation in drug transporter function or expression is now increasingly recognized to have a significant role as a determinant of intersubject variability in response to various commonly prescribed drugs [20]. The most extensively studied class of drug transporters are those encoded by the family of ATP-binding cassette (ABC) genes, which also play a role in the resistance of malignant cells to anticancer agents. Among the 48 known ABC gene products, ABCB1 (P-glycoprotein), ABCC1 [multidrug-resistance associated protein-1 (MRP1)] and its homologue ABCC2 (MRP2; cMOAT), and ABCG2 [breast cancer resistance protein (BCRP)] are known to influence the oral absorption and disposition of a wide variety of drugs [61]. As a result, the expression levels of these proteins in humans have important consequences for an individual’s susceptibility to certain anticancer drug-induced side effects, interactions, and treatment efficacy, for example, in the case of genetic variation in ABCG2 in relation to gefitinib-­ induced diarrhea [14]. In recent years, various naturally occurring variants in these ABC transporter genes have been identified that might affect the function and/or expression of the corresponding proteins, although this type of information is rarely incorporated into prospective clinical trials [61]. Similar to the discoveries of functional genetic variations in drug efflux transporters of the ABC family, there have been considerable advances in the identification of inherited variants in transporters that facilitate cellular drug uptake in tissues

8  Pharmacokinetic Studies in Early Anticancer Drug Development

209

that play an important role in drug elimination, such as the liver and kidney. Among these, members of the organic anion-transporting polypeptides (OATP), organic anion transporters (OAT), and organic cation transporters (OCT) can mediate the cellular uptake of a large number of structurally divergent compounds [37, 58]. Accordingly, functionally relevant polymorphisms in these influx transporters may contribute to interindividual and interethnic variability in drug disposition and response, for example, in the case of the impact of polymorphic variants in the OCT1 gene SLC22A1 on survival of patients with chronic myeloid leukemia receiving treatment with imatinib [36].

8.5 Dose Adaptation Using Pharmacokinetic– Pharmacodynamic Principles 8.5.1 Therapeutic Drug Monitoring Prolonged infusion schedules of anticancer drugs offer a very convenient setting for dose adaptation in individual patients. At the time required to achieve steady-state concentration, it is then possible to modify the infusion rate for the remainder of the treatment course if a relationship is known between this steady-state concentration and a desired pharmacodynamic endpoint. This method has been successfully used to adapt the dose during continuous infusions of 5-FU and etoposide, and for repeated oral administration of etoposide or repeated i.v. administration of ­cisplatin [9]. Methotrexate plasma concentrations are monitored routinely to identify patients at high risk of toxicity and to adjust leucovorin rescue in patients with delayed drug excretion. This monitoring has significantly reduced the incidence of serious toxicity (including toxic death) and in fact, improved outcome by eliminating unacceptably low systemic exposure levels [21]. Therapeutic drug monitoring has also been applied to or is currently under investigation for several more recently developed anticancer drugs, including imatinib [6, 52, 39] and sorafenib [5].

8.5.2 Feedback-Controlled Dosing It remains to be determined how information on interindividual pharmacokinetic variability can be used to devise an optimal dosage regimen of a drug for the treatment of a given disease in an individual patient. Obviously, the desired objective would be most efficiently achieved if the individual’s dosage requirements could be calculated prior to administering the drug. While this ideal cannot be met completely in clinical practice, with the notable exception of carboplatin, as described above, some success may be achieved by adopting feedback-controlled dosing. In the adaptive dosage with feedback control, population-based predictive models are used initially, but allow the possibility of dosage alteration based on

210

A. Sparreboom and S.D. Baker

feedback revision. In this approach, patients are first treated with standard dose and during treatment pharmacokinetic information is estimated by a limited-sampling strategy and compared with that predicted from the population model with which dosage was initiated. On the basis of the comparison, more patient-specific pharmacokinetic parameters are calculated, and dosage is adjusted accordingly to maintain the target exposure measure producing the desired pharmacodynamic effect. It has been proposed that, despite its mathematical complexity, this approach may be the only way to deliver the desired precise exposure of an ­anticancer agent. The study of population pharmacokinetics seeks to identify the measurable factors that cause changes in the dose–concentration relationship and the extent of these alterations so that, if these are associated with clinically significant shifts in the therapeutic index, dosage can be appropriately modified in the individual patient. It is obvious that a careful collection of data during the development of drugs and subsequent analyses could be helpful to collect some essential information on the drug. Unfortunately, important information is often lost by failing to analyze this data or due to the fact that the relevant samples or data were never collected. Historically, this has resulted in the notion that tools for the identification of patient population subgroups are inadequate for most of the currently approved anticancer drugs. However, the use of population pharmacokinetic models is increasingly studied in an attempt to accommodate as much of the pharmacokinetic variability as possible in terms of measurable characteristics. This type of analysis has been conducted for a number of clinically important anticancer drugs, including carboplatin [11], docetaxel [7], topotecan [25], gefitinib [41], and erlotinib [42], and provided mathematical equations based on morphometric, demographic, phenotypic enzyme activity, and/ or physiologic characteristics of patients, in order to predict drug clearance with an acceptable degree of precision and bias.

8.6 Conclusions Substantial progress has been made in recent years toward optimization of cancer chemotherapy during early phases of drug development with the use of pharmacokinetics, pharmacodynamics, and pharmacogenetics, although various aspects of anticancer drug pharmacology deserve more work before they become more useful clinically. Indeed, incorporation of pharmacologic principles in drug development and clinical trials is essential to maximize the clinical potentials of new anticancer agents, as improvements in outcome can be anticipated using these principles to individualize anticancer drug administration. In addition to its importance with classical anticancer drugs, incorporation of these principles will also be essential for the rational development of new agents designed to exploit advances in molecular oncology and those acting on oncogenes, tumor suppressor genes and related signal transduction pathways, invasion, angiogenesis and metastasis, as well as agents used for chemoprevention.

8  Pharmacokinetic Studies in Early Anticancer Drug Development

211

References   1. Assael BM (1982) Pharmacokinetics and drug distribution during postnatal development. Pharmacol Ther 18:159-197.   2. Baker SD and Hu S (2009) Pharmacokinetic considerations for new targeted therapies. Clin Pharmacol Ther 85:208-211.   3. Baker SD, Verweij J, Rowinsky EK, Donehower RC, Schellens JH, Grochow LB and Sparreboom A (2002) Role of body surface area in dosing of investigational anticancer agents in adults, 1991-2001. J Natl Cancer Inst 94:1883-1888.   4. Bartelink IH, Rademaker CM, Schobben AF and van den Anker JN (2006) Guidelines on paediatric dosing on the basis of developmental physiology and pharmacokinetic considerations. Clin Pharmacokinet 45:1077-1097.   5. Blanchet B, Billemont B, Cramard J, Benichou AS, Chhun S, Harcouet L, Ropert S, Dauphin A, Goldwasser F and Tod M (2009) Validation of an HPLC-UV method for sorafenib determination in human plasma and application to cancer patients in routine clinical practice. J Pharm Biomed Anal 49:1109-1114.   6. Blasdel C, Egorin MJ, Lagattuta TF, Druker BJ and Deininger MW (2007) Therapeutic drug monitoring in CML patients on imatinib. Blood 110:1699-1701; author reply 1701.   7. Bruno R, Hille D, Riva A, Vivier N, ten Bokkel Huinnink WW, van Oosterom AT, Kaye SB, Verweij J, Fossella FV, Valero V, Rigas JR, Seidman AD, Chevallier B, Fumoleau P, Burris HA, Ravdin PM and Sheiner LB (1998) Population pharmacokinetics/pharmacodynamics of docetaxel in phase II studies in patients with cancer. J Clin Oncol 16:187-196.   8. Calvert AH, Newell DR, Gumbrell LA, O’Reilly S, Burnell M, Boxall FE, Siddik ZH, Judson IR, Gore ME and Wiltshaw E (1989) Carboplatin dosage: prospective evaluation of a simple formula based on renal function. J Clin Oncol 7:1748-1756.   9. Canal P, Chatelut E and Guichard S (1998) Practical treatment guide for dose individualisation in cancer chemotherapy. Drugs 56:1019-1038. 10. Charles KA, Rivory LP, Brown SL, Liddle C, Clarke SJ and Robertson GR (2006) Transcriptional repression of hepatic cytochrome P450 3A4 gene in the presence of cancer. Clin Cancer Res 12:7492-7497. 11. Chatelut E, Canal P, Brunner V, Chevreau C, Pujol A, Boneu A, Roche H, Houin G and Bugat R (1995) Prediction of carboplatin clearance from standard morphological and biological patient characteristics. J Natl Cancer Inst 87:573-580. 12. Collins JM, Zaharko DS, Dedrick RL and Chabner BA (1986) Potential roles for preclinical pharmacology in phase I clinical trials. Cancer Treat Rep 70:73-80. 13. Cox MC, Low J, Lee J, Walshe J, Denduluri N, Berman A, Permenter MG, Petros WP, Price DK, Figg WD, Sparreboom A and Swain SM (2006) Influence of garlic (Allium sativum) on the pharmacokinetics of docetaxel. Clin Cancer Res 12:4636-4640. 14. Cusatis G, Gregorc V, Li J, Spreafico A, Ingersoll RG, Verweij J, Ludovini V, Villa E, Hidalgo M, Sparreboom A and Baker SD (2006) Pharmacogenetics of ABCG2 and adverse reactions to gefitinib. J Natl Cancer Inst 98:1739-1742. 15. DeMario MD and Ratain MJ (1998) Oral chemotherapy: rationale and future directions. J Clin Oncol 16:2557-2567. 16. Dooley MJ and Poole SG (2000) Poor correlation between body surface area and glomerular filtration rate. Cancer Chemother Pharmacol 46:523-526. 17. Dy GK, Bekele L, Hanson LJ, Furth A, Mandrekar S, Sloan JA and Adjei AA (2004) Complementary and alternative medicine use by patients enrolled onto phase I clinical trials. J Clin Oncol 22:4810-4815. 18. Egorin MJ, Van Echo DA, Olman EA, Whitacre MY, Forrest A and Aisner J (1985) Prospective validation of a pharmacologically based dosing scheme for the cisdiamminedichloroplatinum(II) analogue diamminecyclobutanedicarboxylatoplatinum. Cancer Res 45:6502-6506.

212

A. Sparreboom and S.D. Baker

19. Eisenhauer EA, O’Dwyer PJ, Christian M and Humphrey JS (2000) Phase I clinical trial design in cancer drug development. J Clin Oncol 18:684-692. 20. Evans WE and McLeod HL (2003) Pharmacogenomics--drug disposition, drug targets, and side effects. N Engl J Med 348:538-549. 21. Evans WE, Relling MV, Rodman JH, Crom WR, Boyett JM and Pui CH (1998) Conventional compared with individualized chemotherapy for childhood acute lymphoblastic leukemia. N Engl J Med 338:499-505. 22. Evans WE, Rodman JH, Relling MV, Crom WR, Rivera GK, Pratt CB and Crist WM (1991) Concept of maximum tolerated systemic exposure and its application to phase I-II studies of anticancer drugs. Med Pediatr Oncol 19:153-159. 23. Evans WE, Rodman JH, Relling MV, Petros WP, Stewart CF, Pui CH and Rivera GK (1992) Differences in teniposide disposition and pharmacodynamics in patients with newly diagnosed and relapsed acute lymphocytic leukemia. J Pharmacol Exp Ther 260:71-77. 24. Felici A, Verweij J and Sparreboom A (2002) Dosing strategies for anticancer drugs: the good, the bad and body-surface area. Eur J Cancer 38:1677-1684. 25. Gallo JM, Laub PB, Rowinsky EK, Grochow LB and Baker SD (2000) Population pharmacokinetic model for topotecan derived from phase I clinical trials. J Clin Oncol 18:24592467. 26. Gardner ER, Burger H, van Schaik RH, van Oosterom AT, de Bruijn EA, Guetens G, Prenen H, De Jong FA, Baker SD, Bates SE, Figg WD, Verweij J, Sparreboom A and Nooter K (2006) Association of enzyme and transporter genotypes with the pharmacokinetics of imatinib. Clin Pharmacol Ther 80:192-201. 27. Gibbons J, Egorin MJ, Ramanathan RK, Fu P, Mulkerin DL, Shibata S, Takimoto CH, Mani S, LoRusso PA, Grem JL, Pavlick A, Lenz HJ, Flick SM, Reynolds S, Lagattuta TF, Parise RA, Wang Y, Murgo AJ, Ivy SP and Remick SC (2008) Phase I and pharmacokinetic study of imatinib mesylate in patients with advanced malignancies and varying degrees of renal dysfunction: a study by the National Cancer Institute Organ Dysfunction Working Group. J Clin Oncol 26:570-576. 28. Grandison MK and Boudinot FD (2000) Age-related changes in protein binding of drugs: implications for therapy. Clin Pharmacokinet 38:271-290. 29. Gretz N, Schock D, Sadick M and Pill J (2007) Bias and precision of estimated glomerular filtration rate in children. Pediatr Nephrol 22:167-169. 30. Gurney H (1996) Dose calculation of anticancer drugs: a review of the current practice and introduction of an alternative. J Clin Oncol 14:2590-2611. 31. Hainsworth JD (1999) Extended-schedule oral etoposide in selected neoplasms and overview of administration and scheduling issues. Drugs 58 Suppl 3:51-56. 32. Hooker AC, Ten Tije AJ, Carducci MA, Weber J, Garrett-Mayer E, Gelderblom H, McGuire WP, Verweij J, Karlsson MO and Baker SD (2008) Population pharmacokinetic model for docetaxel in patients with varying degrees of liver function: incorporating cytochrome P4503A activity measurements. Clin Pharmacol Ther 84:111-118. 33. Huang RS and Ratain MJ (2009) Pharmacogenetics and pharmacogenomics of anticancer agents. CA Cancer J Clin 59:42-55. 34. Huang SM, Temple R, Xiao S, Zhang L and Lesko LJ (2009) When to conduct a renal impairment study during drug development: US Food and Drug Administration perspective. Clin Pharmacol Ther 86:475-479. 35. Jaquenoud SE, van dV, Rentsch K, Eap CB and Baumann P (2006) Therapeutic drug monitoring and pharmacogenetic tests as tools in pharmacovigilance. Drug Saf 29:735-768. 36. Kim DH, Sriharsha L, Xu W, Kamel-Reid S, Liu X, Siminovitch K, Messner HA and Lipton JH (2009) Clinical relevance of a pharmacogenetic approach using multiple candidate genes to predict response and resistance to imatinib therapy in chronic myeloid leukemia. Clin Cancer Res 15:4750-4758. 37. Kim RB (2003) Organic anion-transporting polypeptide (OATP) transporter family and drug disposition. Eur J Clin Invest 33 Suppl 2:1-5.

8  Pharmacokinetic Studies in Early Anticancer Drug Development

213

38. Kola I and Landis J (2004) Can the pharmaceutical industry reduce attrition rates? Nat Rev Drug Discov 3:711-715. 39. Larson RA, Druker BJ, Guilhot F, O’Brien SG, Riviere GJ, Krahnke T, Gathmann I and Wang Y (2008) Imatinib pharmacokinetics and its correlation with response and safety in chronicphase chronic myeloid leukemia: a subanalysis of the IRIS study. Blood 111:4022-4028. 40. Lazarou J, Pomeranz BH and Corey PN (1998) Incidence of adverse drug reactions in hospitalized patients: a meta-analysis of prospective studies. J Am Med Assoc 279:1200-1205. 41. Li J, Karlsson MO, Brahmer J, Spitz A, Zhao M, Hidalgo M and Baker SD (2006) CYP3A phenotyping approach to predict systemic exposure to EGFR tyrosine kinase inhibitors. J Natl Cancer Inst 98:1714-1723. 42. Lu JF, Eppler SM, Wolf J, Hamilton M, Rakhit A, Bruno R and Lum BL (2006) Clinical pharmacokinetics of erlotinib in patients with solid tumors and exposure-safety relationship in patients with non-small cell lung cancer. Clin Pharmacol Ther 80:136-145. 43. Mahmood I and Balian JD (1999) The pharmacokinetic principles behind scaling from preclinical results to phase I protocols. Clin Pharmacokinet 36:1-11. 44. Mathijssen RH and Sparreboom A (2009) Influence of lean body weight on anticancer drug clearance. Clin Pharmacol Ther 85:23; author reply 24. 45. Mathijssen RH, Verweij J, De Bruijn P, Loos WJ and Sparreboom A (2002) Effects of St. John’s wort on irinotecan metabolism. J Natl Cancer Inst 94:1247-1249. 46. McLean TW and Kemper KJ (2006) Complementary and alternative medicine therapies in pediatric oncology patients. J Soc Integr Oncol 4:40-45. 47. McLeod HL, Relling MV, Crom WR, Silverstein K, Groom S, Rodman JH, Rivera GK, Crist WM and Evans WE (1992) Disposition of antineoplastic agents in the very young child. Br J Cancer Suppl 18:S23-S29. 48. Milsap RL and Jusko WJ (1994) Pharmacokinetics in the infant. Environ Health Perspect 102 Suppl 11:107-110. 49. Moore MM, Chua W, Charles KA and Clarke SJ (2010) Inflammation and cancer: causes and consequences. Clin Pharmacol Ther 87:504-508. 50. Newell DR, Burtles SS, Fox BW, Jodrell DI and Connors TA (1999) Evaluation of rodent-only toxicology for early clinical trials with novel cancer therapeutics. Br J Cancer 81:760-768. 51. Perdaems N, Bachaud JM, Rouzaud P, Murris-Espin M, Hermant C, Mihura J, Lochon I, Houin G, Canal P and Chatelut E (1998) Relation between unbound plasma concentrations and toxicity in a prolonged oral etoposide schedule. Eur J Clin Pharmacol 54:677-683. 52. Picard S, Titier K, Etienne G, Teilhet E, Ducint D, Bernard MA, Lassalle R, Marit G, Reiffers J, Begaud B, Moore N, Molimard M and Mahon FX (2007) Trough imatinib plasma levels are associated with both cytogenetic and molecular responses to standard-dose imatinib in chronic myeloid leukemia. Blood 109:3496-3499. 53. Pinkel D (1958) The use of body surface area as a criterion of drug dosage in cancer chemotherapy. Cancer Res 18:853-856. 54. Powis G, Harris RN, Basseches PJ and Santone KS (1986) Effects of advanced leukemia on hepatic drug-metabolizing activity in the mouse. Cancer Chemother Pharmacol 16:43-49. 55. Relling MV, Crom WR, Pieper JA, Cupit GC, Rivera GK and Evans WE (1987) Hepatic drug clearance in children with leukemia: changes in clearance of model substrates during remission-induction therapy. Clin Pharmacol Ther 41:651-660. 56. Rodman JH, Maneval DC, Magill HL and Sunderland M (1993) Measurement of Tc-99m DTPA serum clearance for estimating glomerular filtration rate in children with cancer. Pharmacother 13:10-16. 57. Slevin ML, Clark PI, Joel SP, Malik S, Osborne RJ, Gregory WM, Lowe DG, Reznek RH and Wrigley PF (1989) A randomized trial to evaluate the effect of schedule on the activity of etoposide in small-cell lung cancer. J Clin Oncol 7:1333-1340. 58. Smith NF, Figg WD and Sparreboom A (2005) Role of the liver-specific transporters OATP1B1 and OATP1B3 in governing drug elimination. Expert Opin Drug Metab Toxicol 1:429-445.

214

A. Sparreboom and S.D. Baker

59. Sparreboom A, Chen H, Acharya MR, Senderowicz AM, Messmann RA, Kuwabara T, Venzon DJ, Murgo AJ, Headlee D, Sausville EA and Figg WD (2004) Effects of alpha1-acid ­glycoprotein on the clinical pharmacokinetics of 7-hydroxystaurosporine. Clin Cancer Res 10:6840-6846. 60. Sparreboom A, Cox MC, Acharya MR and Figg WD (2004) Herbal remedies in the United States: potential adverse interactions with anticancer agents. J Clin Oncol 22:2489-2503. 61. Sparreboom A, Danesi R, Ando Y, Chan J and Figg WD (2003) Pharmacogenomics of ABC transporters and its role in cancer chemotherapy. Drug Resist Updat 6:71-84. 62. Sparreboom A, Nooter K, Loos WJ and Verweij J (2001) The (ir)relevance of plasma protein binding of anticancer drugs. Neth J Med 59:196-207. 63. Sparreboom A, van AJ, Mayer U, Schinkel AH, Smit JW, Meijer DK, Borst P, Nooijen WJ, Beijnen JH and van TO (1997) Limited oral bioavailability and active epithelial excretion of paclitaxel (Taxol) caused by P-glycoprotein in the intestine. Proc Natl Acad Sci USA 94:20312035. 64. Sparreboom A, van ZL, Brouwer E, Loos WJ, de BP, Gelderblom H, Pillay M, Nooter K, Stoter G and Verweij J (1999) Cremophor EL-mediated alteration of paclitaxel distribution in human blood: clinical pharmacokinetic implications. Cancer Res 59:1454-1457. 65. Sparreboom A and Verweij J (2009) Advances in cancer therapeutics. Clin Pharmacol Ther 85:113-117. 66. Sparreboom A, Wolff AC, Mathijssen RH, Chatelut E, Rowinsky EK, Verweij J and Baker SD (2007) Evaluation of alternate size descriptors for dose calculation of anticancer drugs in the obese. J Clin Oncol 25:4707-4713. 67. Strolin BM, Whomsley R and Baltes EL (2005) Differences in absorption, distribution, metabolism and excretion of xenobiotics between the paediatric and adult populations. Expert Opin Drug Metab Toxicol 1:447-471. 68. Tannock IF, Boyd NF, DeBoer G, Erlichman C, Fine S, Larocque G, Mayers C, Perrault D and Sutherland H (1988) A randomized trial of two dose levels of cyclophosphamide, methotrexate, and fluorouracil chemotherapy for patients with metastatic breast cancer. J Clin Oncol 6:1377-1387. 69. Tetelbaum M, Finkelstein Y, Nava-Ocampo AA and Koren G (2005) Back to basics: understanding drugs in children: pharmacokinetic maturation. Pediatr Rev 26:321-328. 70. Twelves C, Glynne-Jones R, Cassidy J, Schuller J, Goggin T, Roos B, Banken L, Utoh M, Weidekamm E and Reigner B (1999) Effect of hepatic dysfunction due to liver metastases on the pharmacokinetics of capecitabine and its metabolites. ClinCancer Res 5:1696-1702. 71. van Erp NP, Baker SD, Zhao M, Rudek MA, Guchelaar HJ, Nortier JW, Sparreboom A and Gelderblom H (2005) Effect of milk thistle (Silybum marianum) on the pharmacokinetics of irinotecan. Clin Cancer Res 11:7800-7806. 72. Woo MH, Relling MV, Sonnichsen DS, Rivera GK, Pratt CB, Pui CH, Evans WE and Pappo AS (1999) Phase I targeted systemic exposure study of paclitaxel in children with refractory acute leukemias. Clin Cancer Res 5:543-549. 73. Zhang Y, Zhang L, Abraham S, Apparaju S, Wu TC, Strong JM, Xiao S, Atkinson AJ, Jr., Thummel KE, Leeder JS, Lee C, Burckart GJ, Lesko LJ and Huang SM (2009) Assessment of the impact of renal impairment on systemic exposure of new molecular entities: evaluation of recent new drug applications. Clin Pharmacol Ther 85:305-311.

Chapter 9

Pharmacodynamic Studies in Early Phase Drug Development D. Ross Camidge, Robert C. Doebele, and Antonio Jimeno

9.1 Introduction: The Role of Pharmacodynamic Biomarkers in Oncology and in Oncology Drug Development Broadly speaking, a biomarker is something derived from biological material that can be objectively measured. Measurements may be categorical (e.g., present/ absent or 1+/2+/3+), or they may be continuous. While a biomarker can be something as mundane as gender, body surface area, or blood pressure, it can also represent the quantitation of an imaging technique or of a specific molecule or group of molecules from various solid tissues or body fluids. Within recent years, a proliferation of biomarker use and the qualifying terminologies associated with them – including such terms as predictive, prognostic and pharmacodynamic – has occurred within oncology. The most commonly described biomarkers represent measurements taken at a single time point prior to an intervention or, if the measurements are retrospective, on a sample taken prior to the intervention, then they may be correlated with general outcomes (prognostic) or outcomes specific to that intervention (predictive) (cf. Chap. 10). In contrast to these static biomarkers, pharmacodynamic (PD) biomarkers involve at least two temporally separated measurements to determine changes that occur over time in response to a pharmacological intervention. PD biomarkers have, in fact, been used since the beginnings of clinical oncology, although they have not always been referred to as such. Early and persistent use of PD biomarkers in clinical decision making has included assessments of both efficacy [e.g., measuring radiological, clinical, or tumor marker (e.g., PSA, CEA, and CA125) responses to determine the duration of treatment] and toxicity (e.g., adjusting the dose or timing of chemotherapy or radiotherapy in reaction to alterations in blood cell counts, liver function tests, skin erythema, etc.). Imaging-related PD

D.R. Camidge (*) Developmental Therapeutics and Thoracic Oncology Programs, Clinical Thoracic Oncology Program, University of Colorado Cancer Center, Aurora, CO, USA e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_9, © Springer Science+Business Media, LLC 2011

215

216

D.R. Camidge et al.

biomarkers, including modern functional imaging, are discussed in detail in Chap. 11. This chapter addresses advances in the use of PD biomarkers in the past 10 years that predominantly reflect the entry of the so-called “molecularly targeted” or “rationally designed” anticancer drugs into clinical trials and practice. Rationally designed drugs are the result of preclinical drug development based on the identification of a potentially relevant molecular target, followed by evolution of a specific therapeutic intervention. This is in contrast to the traditional reverse ordering of events that occurs through the phenotypic screens used to identify most traditional cytotoxics, whereby the exact mechanism of action may only be discovered some time later, if at all. As the relationship between the interaction of the drug with its target (drug–target interaction) and downstream efficacy is often much better understood for the rationally designed drugs than it is for traditional cytotoxic chemotherapy, looking at specific molecular changes related to the site of action of the drug offers several possible roles for PD biomarkers in modern oncology drug development (Table  9.1). Whether these roles have been or can be successfully achieved depends on a combination of understanding the limitations of specific analysis techniques on specific biomarkers, the tissues to which they are applied, and the design of the clinical studies in which they are incorporated (Table 9.2). Addressing these issues in advance of the inclusion of a proposed PD biomarker Table 9.1  Potential roles for PD biomarkers in oncology drug development 1. Proof of mechanism in early phase clinical trials (does the drug hit its target at clinically achievable doses) 2. Proof of concept in early phase clinical trials (does hitting the target have potentially relevant cellular consequences) 3. Dose/regimen estimation (extent and duration of action on drug target by drug at given dose/ administration schedule) 4. Lead vs. backup compound prioritization within overall drug target development program 5. Prediction of clinical outcome in relation to specific drug intervention (potential to develop as relevant endpoint for regulatory purposes; potential to use for individual patient decision making) Table  9.2  Theoretical factors to consider in the use of any PD biomarker in oncology drug development 1. Background intra- and intersubject variability of biomarker in tissue of interest (reflecting tissue heterogeneity, temporal, and/or circadian change) 2. Background intra- and interobserver variability of biomarker assay (reflecting tissue preparation techniques and/or biomarker analysis and quantitation techniques) 3. Dynamic range of assay readout/determination of cut-points for categorical data presentation 4. Proximity of biomarker to site of action of drug and to site of phenotypic response commitment 5. Relationship of tissue assessed to the clinically relevant site of action of drug [reflecting potential differences among tissues in drug penetration, drug–target number, drug–target status (e.g., mutated/wild-type), and drug–target context (e.g., upstream and downstream modifiers of drug effect)] 6. Optimal time points for PD readouts [reflecting time course of drug exposure (blood vs. tissue), drug action, and phenotypic change in tissues of interest]

9  Pharmacodynamic Studies in Early Phase Drug Development

217

within a definitive drug-intervention clinical study to ensure that the results achieved are as informative as possible cannot be overemphasized.

9.2 Choosing the Right Biomarker for PD Studies of a Specific Drug The preclinical development of rationally designed drugs offers the promise of understanding the relationship between drug–target interaction and anticancer efficacy, such that initially it would appear that if you could prove that your new drug had activity against its proposed target in humans at clinically achievable doses, and that the requisite duration of activity was achievable through a clinically viable dosing regimen, the new drug ought to be a sure success. Unfortunately, the reality of the situation is, predictably, not as simple as this. First, the specificity of any drug is always a relative term. Kinase inhibitors have dominated the first wave of rationally designed drugs, the vast majority of which contain structural ATP-mimetic motifs that facilitate entry into and blockage of the ATP-binding kinase domain of the molecule. Relative specificity, expressed in terms of differences in binding affinity or inhibitory concentrations (ICs), is usually achieved by the presence of additional motifs that either facilitate or block interaction with surrounding sequences in the target or other similar molecules. However, as exposure to drug in a particular tissue increases, the potential for targets other than the primary molecule to be affected also increases. While this may bring about additional toxicities, it may also bring about significant increases in efficacy. Although it may be possible to use PD biomarkers to demonstrate activity against the specific target molecule that a drug has been engineered to hit in humans, it is important to ask the question as to whether that target is still the most or the only relevant target to consider in the clinic. The potential of the drug designer’s target to not necessarily be the most relevant clinical target for that drug has recently increased given new data emerging about the molecular aberrations driving some cancers. Some malignant tissues have sequence aberrations in specific molecular drivers, and the IC50s or IC90s of these mutant drivers to an ATP-mimetic type drug can be very different from those of the wild-type forms that are more commonly used as the drug target in preclinical work [1]. Since a range of other wild-type kinases are also usually used to demonstrate drug specificity preclinically and several different mutant drivers may coexist in a patient’s cancer, the possibility for occult but highly relevant off-target effects in the clinic also has to be considered. These data emphasize the importance of directly demonstrating PD activity on any given target in the tissue of interest, rather than assuming that the pharmacokinetic exposures with activity (or lack of activity) in a preclinical model will automatically translate to equivalent target effects in the clinic. Even for monoclonal antibodies, the presence of the epitope on cross-reacting molecules or of molecules associated with the target molecule [e.g., through dimer formation, such as

218

D.R. Camidge et al.

that between the IGF-1R and epidermal growth factor receptor (EGFR), or the IGF-1R and Her 2] that may also be affected by the monoclonal has to be considered [2–4]. The recent advent of deliberately “multitargeted” kinase inhibitor drugs (cf. Chaps. 14 and 15) that have activity against a range of potentially useful targets (and, potentially, some innocent bystanders) further emphasizes the importance of the correct drug target(s) selection for use in any PD studies in clinical drug development. Secondly, even if the appropriate drug target(s) can be identified in advance, it is important to recognize that the context of any such target will determine its true role as a driver of the malignant process, and/or as a site of selective tumor vulnerability in a patient, and/or its optimal method (extent and duration) of interference/ activation by a drug. This context may reflect upstream factors such as ligand levels, interacting receptors through heteromultimerization, the presence or absence of specific activating gene mutations, and downstream factors such as the presence or absence of damage recognition pathways or prosurvival pathways which may vary significantly between patients, between tissues in the same patient, and between preclinical models. Beyond the obvious plea that preclinical drug development models should recapitulate the clinical scenario as much as possible, and that the techniques used for PD biomarker assessment should be comparable between preclinical and clinical work (cf. next subsection), the above points also help us to think about how the “right” biomarkers for PD studies in early phase drug development should be chosen. Just because a biomarker changes does not mean it is either necessary or sufficient to induce clinically relevant phenotypic changes in a patient’s tumor (Fig. 9.1). The closer a PD biomarker is to the site of action of the drug (“early” PD biomarker) the more likely it is to inform proof of mechanism (i.e., that the drug hits a specific target at clinically achievable exposures) but the less likely it is to determine the clinical relevance of that target. In contrast, “later” PD biomarkers – which may be molecular, for example, markers of committed phenotypic change such as cell-cycle progression or cell death, or metabolic change (cf. functional imaging studies, Chap. 11); or radiological changes in tumor cell mass – are more likely to reflect proof of concept, that is, an association of drug activity with ultimate clinical benefit, but are less likely to pinpoint the specific molecular targets of the drug responsible. However, it is important to remember that in early phase drug development all our endpoints are approximations of benefit, and in many common tumors even objective tumor shrinkages assessed by formal response criteria only loosely correlate with overall or progression-free survival benefit [5, 6]. Beyond the issues associated with methodology and presence/levels/ variability/relevance of detectable markers in the tissues of interest, the “right” PD biomarkers for early phase oncology studies should clearly contain both “early” and “late” elements of PD change. After adjusting for the number of variables assessed, ideally correlations between different early and late PD biomarkers could then be sought. This would potentially allow a fuller exploration of the links between a drug’s actions on specific molecular targets with changes in relevant markers of potential downstream benefit.

9  Pharmacodynamic Studies in Early Phase Drug Development

219

Drug exposure Pharmacodynamic change Baseline biomarker

a

Neither necessary nor sufficient

b

Necessary but not sufficient

c

Necessary and sufficient

(additional factors required)

Relevant clinical outcome

Fig. 9.1  Schematic diagram of PD biomarkers that are (a) neither necessary nor sufficient (epiphenomena), (b) necessary but not sufficient, or (c) both necessary and sufficient for clinically relevant change. While early PD biomarkers close to the site of action of the drug provide good information on proof of drug mechanism in the clinical setting, they are less likely to distinguish between these three scenarios. In contrast, later PD biomarkers associated with either committed cellular phenotypic change or gross radiological or clinical changes in tumor mass are more likely to reflect the changes both necessary and sufficient for clinical benefit (such as improved symptoms or prolonged life), but provide less information about the specific molecular targets hit by the drug that are driving this change. Information on both early and late PD biomarkers within early phase trials, and their direct correlation, is recommended

9.3 Choosing the Right Tissue for PD Studies: Tumor-Derived Tissue and Methodologies Clearly, directly studying drug effects on tumor tissue represents the ultimate in PD readouts. As a result, tumor biopsies have become an important part of many early phase clinical trials. However, performing nondiagnostic tumor biopsies raises technical and ethical concerns mostly related to the use of a potentially harmful procedure with little or no clinical benefit to the patient. In this section, we describe examples of methodologies oriented toward the acquisition, processing and analysis of tumor samples for the purpose of exploring PD effects on tumors. Occasionally, archival [usually formalin-fixed paraffin embedded (FFPE)] samples from original diagnostic/resection specimens may be used as the baseline for PD readouts. However, because of concerns about both nonstandardized sample preparation and storage techniques, and the accurate representation of the tumor’s current biology at the time of study entry, fresh baseline samples are much

220

D.R. Camidge et al.

p­ referred. In general, ideal PD studies therefore comprise two or more study-specific biopsies, one acquired prior to starting therapy (baseline sample) and at least one acquired after some predefined treatment period [post-treatment sample(s)].

9.3.1 Methods for Tissue Acquisition A biopsy is a test involving the removal of cells or tissues for examination. Surgical sampling of tissue suitable for PD exploration may result in either an incisional or excisional biopsy, depending on whether only a part or all of a lesion/anatomical unit is removed. When a sample of tissue or fluid is removed with a needle, the procedure is called either a core biopsy or a needle aspiration biopsy/fine-needle aspiration (FNA), depending on the gauge and type of the needle. Preexisting cell suspensions contained within malignant pleural effusions or ascites can also be captured by needle aspirations with the yield being a direct proportion of the volume of fluid captured and then centrifuged to generate a cell pellet, which may then be handled as other solid tissues. To maximize sample preparation standardization, clear instructions relating to the size of lesions to cut for processing and the utilization of core biopsies should be considered. 9.3.1.1 Core Biopsies Core biopsies are usually performed using 18-G core biopsy needles, and one to two samples are routinely obtained from lesions during diagnostic procedures (Fig.  9.2). In research procedures, this number will vary, but it is frequently impractical to acquire more than three cores per procedure. They are generally obtained with radiological support for lesion identification (see below). The use of larger bore (14-G) needles has been advocated to obtain a sufficient amount of tissue and to maintain histological architecture. The Case Western University group performed a study of 190 sequential liver biopsies in a pig preclinical model using 14-, 18-, and 20-G needles [7]. This study showed that 14-G needles recovered an average DNA content per sample of 40 mg vs. only 12 mg for 18-G needles. Furthermore, 14-G Tru-Cut needles resulted in acceptable preservation of nodal tissue architecture in cases of lymphoma [8]. The maintenance of tissue architecture may be important when the PD endpoint uses immunohistochemistry (IHC) techniques, but in general practice 18-G needles are used more widely, and IHC quality seems adequate [9]. 9.3.1.2 Fine-Needle Aspirate Biopsy Fine-needle aspirate (FNA) biopsy is a cytopathologic diagnostic tool that has gained popularity due to its cost-effectiveness, efficiency, and safety [10], and is

9  Pharmacodynamic Studies in Early Phase Drug Development

221

Fig  9.2  FNA and core biopsies. Imaging to guide biopsies usually involves either computed tomography (above) or ultrasound (below) modalities. A fine-needle aspirate (FNA) biopsy can be used to identify and localize the site for later core biopsy or act as a primary tissue acquisition method itself. FNA produces a suspension of cells and/or a collection of small tissue fragments that can be used as such or converted to a pellet and embedded in paraffin. Core biopsies generate a larger cylinder of tissue, complete with preserved histological architecture

being increasingly incorporated in oncology research studies. It can precede a regular core biopsy for site identification or can itself be the procedure of choice for biomarker studies [11] (Fig. 9.2). FNAs are usually performed with radiological support, but for superficial lesions direct sampling is acceptable. FNAs are performed using a 22- to 27-G needle, usually with on-site real-time microscopic evaluation by a cytopathologist or specially trained cytotechnologist. The number of FNA “passes” performed ranges from 2 to 6, depending on the endpoints and materials needed. The first pass is used for on-site examination of the material, to determine the presence and quality of lesional tissue (adequacy assessment). Most studies show that the on-site cytology evaluation increases the total diagnostic yield by 10–15% [12, 13]. The aspirate is smeared onto 2–3 slides, stained, for example, with DiffQuik, according to standard protocols, and examined microscopically on-site. Slides are usually not cover-slipped and are stored at room temperature. Subsequent passes are then performed either to acquire sufficient tissue procurement for the endpoint assays or to further guide the operator to the appropriate area of the lesion to perform a core biopsy (if applicable).

222

D.R. Camidge et al.

The advantage of FNA vs. standard core biopsies in a clinical research setting is that although its total tissue yield is less, its morbidity is felt to be lower than regular core biopsies, multiple passes can be easily performed with real-time supervision of lesion sample quality, and material easily suitable for forming a cell suspension is obtained (which can be advantageous in some tissue preparation scenarios).

9.3.1.3 Third-Space Collections (Ascites, Pleural Fluid) Large amounts of tumor material can sometimes be obtained by processing thirdspace collections such as malignant pleural fluid or ascites. Relatively simple processing can yield either cells in suspension or a pellet that can then be frozen and/or embedded in paraffin. The major limitations are that the yield in terms of tissue per unit of volume can vary largely between individuals, and large-volume centrifuge equipment is required. Cell-free exudates can also be assessed, for example, for proteomics, although these have not yet been used for PD evaluations [14, 15].

9.3.2 Anatomical Sites to Be Biopsied Tumor biopsies for PD studies are generally obtained only from sites that constitute minimal risk to the subject. The limited data available suggest that biopsies can even be safely performed from irradiated tissues [16]. Acceptable sites include: • Cutaneous lesions; for example, from skin cancers, skin metastases, or lesions directly involving the skin, such as underlying breast cancers, without evidence of active local infection. • Peripherally accessible lymph nodes; for example, from cervical, axillary, or inguinal locations. • Liver lesions that are not immediately adjacent to major vessels. • Peripheral pleural lesions. • Accessible oral or nasopharyngeal lesions. • Easily accessible rectal or upper gastrointestinal lesions. Sites that are commonly not deemed acceptable for sequential biopsies for PD studies include (but are not limited to): • • • • •

Lung. Brain. Mediastinum. Intra-abdominal or intrapelvic sites. Any site deemed by the patient’s primary oncologist, study investigator, or radiologist to represent greater than minimal risk to the subject.

9  Pharmacodynamic Studies in Early Phase Drug Development

223

It is critical that the team performing the biopsies (e.g., interventional radiology, surgery, and cytopathology) is involved in devising the biopsy-related plan, both for the overall study and for individual patients. The final decision regarding the conduct of any specific procedure should also be made by the biopsy team to maximize patient safety. For study subjects with cancer, it is expected that the liver will be the most common site for repeat tumor biopsies. For liver biopsies, in general, published reviews quote a complication rate of 0.06–0.32% [17]. The complication rate does not appear to be affected by perihepatic ascites [18]. Available data indicate a similar complication rate in a phase I setting [19]. Biopsies are routinely conducted under ultrasound (US) or computed tomography (CT) guidance. The choice of imaging technique is selected depending on the imaging characteristics of the area to be biopsied. Imaging studies for the patient should be evaluated prior to biopsy by the research team with a radiologist. The assessment should include the identification of optimal target lesion(s) for biopsy and a safety evaluation of the proposed procedure. Identifying an alternative or secondary lesion beforehand allows for rapid modification of the plan within the same procedure should the primary target lesion be unidentifiable or negative by cytology. For relatively superficial lesions in solid organs or in soft tissues, ultrasoundguided fine-needle aspiration (US FNA) is usually the preferred method. This ensures innocuous real-time acquisition of anatomical data (relevant to surrounding vessels) and allows repeated/extended procedures. For lesions located in a cavity (usually primary cancers such as rectal, esophageal, and gastric) or adjacent to a cavity (e.g., pancreatic), endoscopic ultrasoundguided fine-needle aspiration (EUS FNA) is usually the preferred method for diagnostic biopsies and may also be feasible for repeated biopsies for PD assessments [20]. However, because of the need for an operator with specialized training and the additional invasiveness of the procedure compared with simpler imageguided biopsies of more superficial body sites, endoscopic biopsies are uncommon in the PD setting, with the possible exception of rectal lesions accessible via sigmoidoscopy or proctoscopy. Most diagnostic biopsy procedures will have preestablished specific patient care and safety recommendations depending on the anticipated use of sedation, the discomfort of the procedure, and the bleeding risk, which should be strictly followed.

9.3.3 Tissue Heterogeneity A recurring question when discussing PD biomarkers in cancer is that of heterogeneity both within the individual lesion and between the primary (which is the tissue most frequently available in archival form) and the metastasis (which is the cancer manifestation most frequently targeted by treatment in clinical trials). Intralesional heterogeneity may reflect both variability between tumor cells and

224

D.R. Camidge et al.

variability in the amount of contaminating nontumor (stromal) components in the sample. The ­significance of stromal contaminants will vary depending on the PD assays used. For most immunohistochemical assays, it is usually not a problem because the scorer can delineate tumor vs. nontumor at the time of observation. However, for automated IHC quantitation, without some aspect of observer-driven or dual-label-mediated area delineation, quantitation of stromal components may still be an issue. For assays based on molecular extractions from tissue (e.g., ELISAs, transcript level assessments, or specific DNA sequence detection through PCR), stromal contamination will have the largest theoretical effect. For this reason, macrodissection of biopsies preprocessing, and in some instances, laser microdissection of tumor from nontumor material from the slide sections, should be considered. Intralesional heterogeneity cannot be avoided but it can be addressed through preintervention assessments of the background variability of any given marker with any given detection technique in archival tumor samples to ensure that the study is adequately sized to determine a true PD effect. Unfortunately, such prestudy diligence is rare, increasing the risk of spurious false-positive and -negative results in many tumor-based PD assessments. With regard to interlesional variability, there is very limited data available, with reports arising from autopsy studies in breast cancer showing extensive heterogeneity between the primary and metastases, as well as among multiple metastatic lesions from the same patient [21]. When planning PD studies in early clinical trials, sequential biopsies from the same area (usually metastasis in nodes or liver) should be the first choice to minimize nondrug-related variability in the endpoints assessed. Samples taken from any tissue from the same area in the same individual over a time course of drug exposure offer the most robust method to reliably ascribe changes in a biomarker to being a true PD effect.

9.3.4 Technical Aspects of Processing and Preservation It is the research team’s responsibility to ensure that a detailed tissue management and preservation standard operating procedure (SOP) exists for each study. This is key to ensuring accurate and consistent endpoints for analysis. The plan should be guided by the kind of analyses intended for the samples. Preclinical validation of sample processing should ideally be conducted well in advance to the critical clinical study. A general rule is if storage stability information derived from preclinical studies is not available or is questionable, samples should be preserved with the easiest platform available and the lowest degree of processing to minimize commitment to one particular assay or another. For IHC, this means FFPE tissue; for most other molecular biology assays this usually means intact frozen preservation at either −70°C or in liquid nitrogen. Large and intermediate-volume solid tissue samples (surgical and core biopsies) will be discussed separately from cell suspension specimens (FNA and third-space fluids).

9  Pharmacodynamic Studies in Early Phase Drug Development

225

9.3.4.1 Processing of Surgical and Core Biopsies Frozen Tissues Flash freezing in liquid nitrogen and rapid transfer to a deep freeze (−70°C or lower) are the preferred method for sample preservation, especially if genome-wide or microarray transcript analysis is contemplated [22]. Although recent techniques are now allowing the use of archival tissue for RNA analysis [23], fresh frozen preservation is still preferred. The results from one animal study clearly illustrate the detrimental impact of formalin fixation on RNA quantity and transcript size [24]. While the yield of total RNA from frozen liver was 625 mg/100 mg tissue, the yield in formalin-fixed tissue was only 30  mg/100  mg. It is recommended that approximately 0.1 cm3 of the tissue specimen, which should yield sufficient mRNA for most studies, be snap frozen in liquid nitrogen within at most half an hour, but preferably within minutes, of surgical resection and stored at −80°C or below [25]. Samples should be kept in small, airtight containers and kept from drying out during frozen storage by placing fragments of ice within the sample [26]. Since brief transport of tissue on ice before fixation and processing appears to have minimal impact on RNA quality and expression [27], if liquid nitrogen is not freely available within the area of the biopsy procedure, a practical approach is to send the samples by routine iced transport for centralized processing. Tumor samples can be frozen and thawed at least three times without compromising the RNA integrity and genetic expression profile [28]. Another potential compromise for settings without access to adequate freezers is saving a portion of the biopsy in an RNA preservative, as discussed below. RNA-preserving methods such as RNAlater involve the use of solutions that precipitate out RNases thereby helping to preserve intact RNA. In a study that correlated the impact of alternative tissue handling procedures on the quality of RNA [29], freshly excised normal skin was taken from patients during Mohs surgery. One portion of the tissue was placed immediately into RNAlater (Ambion, Austin, TX) for 24 h and then stored at 4°C for 2–6 weeks. Another portion of the tissue was held for 30 min at room temperature during dissection in surgical pathology before being flash frozen at −20°C. Purification and analysis of the total RNA in these specimens showed that the tissues preserved immediately after biopsy in RNAlater produced distinct ribosomal bands of high-molecular-weight RNA. In contrast, the tissues held at room temperature resulted in smearing of lower-molecular weight RNA species typical of extensive RNA degradation. Paraffin-Embedded Tissues for IHC Commonly, archival specimens available from regular clinical specimens are FFPE tissue blocks. These can offer some information on RNA, DNA, and even proteomics for prognostic/predictive biomarker purposes. For PD readouts, however, FFPE is most commonly used for IHC endpoints. Although IHC is routinely performed

226

D.R. Camidge et al.

by many pathology laboratories, it is rarely standardized. A major cause of ­variation in the reproducibility of IHC staining is induced by tissue fixation and, to a lesser degree, tissue processing. Efforts have been undertaken to promote standardization of fixation and processing [30]. Most laboratories use neutral-buffered formalin (10%) for tissue fixation which introduces cross-links, whereas coagulative fixatives are less popular. Problems with formalin fixation comprise delay of fixation and variations in the duration of the fixation. Solutions to these problems could be to start fixation soon (<30  min) after surgical removal of the tissue and to avoid overfixation (>24–48 h). For tissue processing, the most important problem is inadequate tissue dehydration prior to paraffin embedding. This can be prevented by preparing all solutions freshly every week, and to accurately titrate the volume of fixative to the volume of tissue being processed. For example, if formalin is the selected fixative, an entire core from one of the 18- to 20-G passes should be immediately submerged in a 50  ml container ­containing not less than 15 ml of 10% formalin. Larger samples require larger containers and more fixative, but in the setting of clinical research excess tissue should probably be frozen to maximally preserve its potential for different assays in the future. 9.3.4.2 Cell Suspensions and FNAs: Special Handling Characteristics DNA/mRNA Collection For adequate DNA/mRNA collection, the FNA material is gently expelled from the FNA needle with a 10cc syringe (by air pressure) directly into a 1.5 ml cryovial containing, for example, 500  ml of RLT buffer (with 1% b-mercaptoethanol). Subsequent passes are handled in the same manner, usually combining the material in the same cryovial. After gentle shaking, the specimen is frozen in dry ice or liquid nitrogen, and stored at −70°C. Concentrated cellular material from larger volumes of cell suspensions is processed similarly. Protein Analysis For protein endpoints, a similar processing as described above is used, only using an adequate protein lysis buffer. For IHC studies derived from cell suspensions, the collected sample is centrifuged to form a cell pellet that is then processed into a FFPE block. Because of the comparatively minute amount and fragmented nature of the specimen from FNAs, after dispersal either an agarose gel or fibrin-clot is used to hold the specimen together on centrifugation prior to dehydration and embedding in paraffin. The cell blocks are used to generate numerous slides and can be used to perform immunostaining without any differences from the established protocols for tumor blocks.

9  Pharmacodynamic Studies in Early Phase Drug Development

227

Viable Cell Collection For collection of viable cells with the potential for later ex vivo culture, the FNA material (ideally from at least two passes) or concentrated (centrifuged) malignant effusion is collected into a screw-cap conical containing an appropriate medium such as 10  ml of warm sterile RPMI [supplemented with 10–20% fetal bovine serum (FBS) and 1% penicillin/streptomycin] [31]. The materials from more than one pass are usually combined in the same conical. The vial is tightly capped and transported at room temperature for immediate processing in a tissue culture facility. An aliquot of these materials is also usually preserved in a vial containing cell freezing medium (e.g., RPMI + 10% FBS + 10% DMSO solution) for potential later use.

9.3.5 Operational and Planning Aspects The expected success rate of paired tumor biopsies is a critical component that should be factored in when designing PD-based studies, especially to prevent under-powered pilot studies. In one of the largest reports, a single center ­experience comprising 192 biopsies in 107 patients in seven clinical trials was reported [19]. All but eight patients had sequential pre- and posttreatment ­biopsies. Seventyeight (73%) of the 107 patients had liver lesion biopsies. In eight patients, either one or both biopsies contained insufficient viable tumor tissue or no tumor tissue at all for analysis. From a total of 99 patients in whom paired biopsies were attempted, a total of 87 (88%) were successful. Reasons for failure included patient refusal for a second biopsy (n = 2), vasovagal ­reaction with first biopsy precluding a second biopsy (n = 1), subcapsular hepatic bleeding (n = 1), and, most commonly, obtaining necrotic tumor, fibrous, or normal tissue in one of the two sequential biopsies (n = 8). This was the first reported series demonstrating that with adequate precautions and experience, sequential tumor biopsies are feasible and safe during early phase clinical trials. However, subsequent reports have failed to confirm this high success level, with completion rates across studies showing considerable variability [32]. In a broader clinical setting, the rate of nonevaluable results due to other causes is high. The timing between tumor biopsies may vary but is usually before and after 1–6 weeks of treatment. Longer lapse times are complicated by increased risk of noncompliance due to loss of follow-up or disease progression. The incidence of withdrawal of informed consent for biopsies is likely to be higher in subjects taken off-study due to progression than in those patients still participating actively in the study. For easily accessible tissues – usually normal tissues (cf. subsection 9.4), but occasionally malignant tissues – more frequent sampling within the first few hours or days of treatment to generate a more detailed time course for analyses may also be possible.

228

D.R. Camidge et al.

9.3.6 Examples of Analytical Techniques for PD Endpoints 9.3.6.1 DNA Analysis DNA sequencing looking for stable mutations or amplification strategies ­looking at stable gene copy number alterations may be used as potentially predictive or prognostic static biomarkers. During prolonged exposure to drug, selection ­pressures may also alter the underlying dominant mutant or gene copy number status of the tumor such that investigations of these at progression may inform mechanisms of acquired resistance, but DNA analysis per se is not usually ­considered suitable for use as a PD biomarker. However, with the advent of therapies ­targeting epigenetic changes, DNA methylation patterns are increasingly being studied. For example, the DNA methyltransferase inhibitor 5-aza-2¢deoxycytidine (decitabine) induces DNA demethylation and re-expression of epigenetically silenced genes and increases carboplatin sensitivity of tumor xenograft models. A recently reported clinical study determined the feasibility of delivering a dose of decitabine, combined with carboplatin, that would be capable of producing equivalent biologic effects in patients with solid tumors [33]. Decitabine induced dose-dependent, reversible demethylation in ­peripheral-blood mononuclear cells (PBMCs). Furthermore, decitabine induced ­demethylation of the MAGE1A CpG island in PBMCs, buccal cells, and tumor biopsies, as well as elevation of fetal hemoglobin expression. This provided proof of principle and gave the basis for exploring these biomarkers further. Similar studies have been conducted ­exploring other methylation-dependent biomarkers in hematological malignancies [34].

9.3.6.2 Messenger-RNA Analysis As microarrays have become more powerful and reliable, researchers and clinicians have begun applying full-genome expression analyses in a variety of challenging clinical situations, including to identify previously unrecognized subsets of cutaneous melanomas [35], breast carcinomas [36, 37], and colorectal cancer [38]. Microarrays also have now been used to make multiclass distinctions among highly related tumor types, such as the adenocarcinomas [39], to stratify patients with cancer into subgroups with distinct clinical manifestations and different responses to therapy [40–43] and to predict chemotherapy response [44, 45]. Their use as PD biomarkers in clinical studies has not been explored as extensively, limited by the requirement of acquiring fresh tissue after treatment. A clinical trial in esophageal cancer patients treated with the EGFR inhibitor gefitinib showed that, in addition to documenting a decrease in proliferation markers such as Ki67, microarray experiments on tumor biopsies showed that gefitinib also downregulated oncogenes associated with tumor progression [46].

9  Pharmacodynamic Studies in Early Phase Drug Development

229

9.3.6.3 Protein Analysis: Immunohistochemistry IHC refers to the process of localizing proteins in cells of a tissue section exploiting the principle of antibodies binding specifically to antigens in biological tissues. Visualizing an antibody–antigen interaction can be accomplished in a number of ways. In the most common instance, an antibody is conjugated to an enzyme, such as peroxidase, that can catalyze a color-producing reaction. Alternatively, the antibody can also be tagged to a fluorophore, such as FITC, or rhodamine (immunofluorescence). Readouts may be qualitative – location of antibody binding – and/or quantitative. In terms of quantitation, the assumption is that the amount of fluorescence or of colorimetric substrate deposition reflects the amount of primary antibody that is bound, and by extension the amount of antigen present. The primary antibodies used can be polyclonal (a heterogeneous mix of antibodies recognizing various epitopes created by injecting animals with antigens, which may or may not be enriched further through the use of specific epitope binding procedures and elutions) or pure monoclonals (generally created by genetic engineering). Polyclonals are easier to produce, but of limited supply and with a greater potential for reacting with a range of different epitopes (some of which may be present on more than just the intended target molecule) than monoclonal antibodies. Any staining should be put into context by pertinent positive and negative controls run in parallel. The preservation of proteins under various fixation conditions has to be considered. A prospective study on a panel of commonly used fixatives was undertaken to determine optimal tissue preservation protocols for EGFR [47]. The stability of the epitope on cut tissue sections stored for a period of up to 24 months was also tested using material originating from patients with head and neck cancer, nonsmall-cell lung carcinomas, and colorectal adenocarcinomas. Depending on the fixative used and the time of storage of cut tissue sections, significant variation in the determined level of EGFR expression was demonstrated. Quantification of the signal may be by eye. For example, measurement of a continuous variable such as percentage cell positivity (which may have a qualitative element to count a cell as positive or not, such as full membrane staining, nuclear staining, etc. depending on the epitope); categorical scoring of the average intensity of the whole section, for example, 0, 1+, 2+, or 3+; or H-scoring [a composite of intensity of staining and the percentage of cells displaying each intensity creating a continuous variable usually ranging from 0 (no reactive cells) to 300 (100% of cells being 3+ positive)]. Advantages of categorical scoring include a simpler implementation in a broad clinical setting and are the method used for detection of EGFR and HER2/neu for treatment selection [48, 49]. Advantages of the H-scoring include the ability to factor in both the relevance of the number of cells and the intensity of the biomarker staining, which is perceived as providing better dynamic range and thus seems the more appropriate system to consider for PD studies (Fig. 9.3) [9, 50]. Alternatively, automated or semiautomated quantitation may be used. These novel computer-aided IHC techniques are typically used in conjunction with tissue

230

D.R. Camidge et al.

9  Pharmacodynamic Studies in Early Phase Drug Development

231

microarrays (TMA) for better field standardization­. Studies using conventional staining by regular IHC followed by automated image interpretation using, for example, the Automated Cellular Imaging System (ACIS II, Chromavision, Inc.) have shown highly reproducible results comparable to Western blotting, ELISA, and visual scoring (0–3+) by a single pathologist [51]. Absolute levels of target proteins, assessable by IHC, are rarely altered sufficiently by drug intervention for use as PD biomarkers in drug development. Notable exceptions to this include proteins with short half-lives related to the cell cycle that reflect cell turnover, or when the intervention is anticipated to produce direct target downregulation, for example, the internalization and destruction of IGF-1R with anti-IGF-1R antibodies [52]. In contrast, phosphospecific antibodies, with the potential to reflect changes in the activation status of proteins, a far more dynamic variable and one potentially very closely related to the action of many new signal transduction inhibitors being explored in clinical trials, represent a real advance in the study of cancer. They bind to a certain protein, but only when this protein is phosphorylated in a specific site (usually known to be indicative of overall protein activation/inactivation state). Phosphorylation-dependent binding may be confirmed through the use of, for example, alkaline phosphatase treatment of the samples to produce a negative control. Unfortunately, the changeability of phospho-endpoints that make them so attractive as PD biomarkers also makes them highly labile and prone to producing both false-positive and -negative reads. A recent report assessed the impact of delays in fixation on phospho-biomarkers of Src kinase activity [53]. A total of 20 patients with locally advanced breast cancer and 5 with early bladder cancer had multiple tissue samples taken which were fixed at documented time points up to 60 min after biopsy. These were examined to determine if the amount of paxillin, phosphopaxillin, phospho-focal adhesion kinase, and total phospho-tyrosine changed over time, using a quantitative lysate immunoassay. In breast cancer, there was an increase in the amount of phospho-paxillin (60% per hour; P = 0.019) up to 60 min after biopsy. The amount of total paxillin decreased (28% per hour; P = 0.034) over the same time course. In early bladder cancer, no changes were noted in any endpoints up to 45 min.

Fig.  9.3  Pharmacodynamics of rapamycin in tumor biopsies. (a) Images from pre- and posttherapy tumor biopsies (9  mg/kg/day) evaluating phospho-p70, phospho-S6 ribosomal protein, and phospho-4EBP1. All images were taken at ×40. (b) Graphs of the numeric assessment of the three endpoints in the four tumor biopsy pairs. The intensity (0–3) and percentage (0–100%) of cells positive were considered, and an index multiplying both was calculated. Two patients had benefit from the drug, defined as prolonged treatment without progression (#1 and #2), whereas two patients did not (#3 and #4). Only phospho-S6 showed significant decrease in the patients with clinical benefit (#1 and #2) compared with the two subjects showing rapidly progressing disease (#3 and #4) (Jimeno et al. [9])

232

D.R. Camidge et al.

9.4 Choosing the Right Tissue for PD Studies: Use of Surrogate Tissues (Nontumor/Normal Tissues) in PD Studies Although some malignancies, notably leukemias, are very easily accessible for repeated molecular analyses, for most solid tumors, taking repeated biopsies for the determination of PD biomarkers, especially over prolonged time courses, is problematic. Given that the same molecule to which a designer drug is directed may be present in normal, nonmalignant tissues, as well as in the tumor itself, the potential to use more easily accessible normal tissues as tumor-surrogates for PD studies has been explored. Exploration of mechanism-based toxicities as evidence of drug action and of appropriate drug exposure in easily accessible and assessable normal tissues is one example of this logic in action. For example, EGFR tyrosine kinase inhibitors, such as erlotinib, commonly induce an acneiform skin rash due to the effects of EGFR inhibition on skin cell maturation and/or skin secretion production increasing the chances of skin pore blockage. Smokers have higher rates of erlotinib metabolism and manifest both lower drug exposures and lower rates of skin rash than nonsmokers [54, 55]. Doubling the standard dose of erlotinib in smokers increases drug exposure and the incidence of rash to approximately those of ­nonsmokers [55]. In addition to mechanism-based toxicities, specific molecular analyses comparable to those performed within tumor biopsies may also be performed within normal tissues. Easily accessible normal tissues that are commonly utilized in this context include cellular material of both mesenchymal (e.g., PBMCs or platelets) and epithelial [e.g., skin, hair sheath cells, buccal or other intestinal mucosa, and circulating endothelial cells (CEC)] origin, and cell-free material such as circulating ligands. Since there is a potential choice of tissues, generally speaking, if the anticipated effect of the interventional drug is to reduce the levels of a biomarker, the tissue with the highest basal levels should be the most desirable, assuming equivalent background variability, to minimize the chances of producing false-positive results. Conversely, if the anticipated effect is an elevation of the biomarker, then low basal levels would be more desirable, all other things being equal. Normal tissue PD biomarkers offer a series of theoretical advantages over solid tumor-derived PD biomarkers, as well as some theoretical disadvantages (Table 9.3). For drugs with preclinical data suggesting that they are nongenotoxic, with predictable and reversible side-effects, surrogate tissue PD biomarkers may also be explored in healthy volunteers, dramatically increasing the speed and reducing the cost of this aspect of clinical development. We explore some examples of the ­different normal tissues that have been used in oncology PD studies in turn:

9.4.1 Skin Punch biopsies of human skin usually sample both the epidermis (stratified keratinized squamous epithelium) and skin adnexae, such as sweat glands and hair follicles, contained within the underlying dermis. From a 4 mm punch biopsy, some

9  Pharmacodynamic Studies in Early Phase Drug Development

233

Table 9.3  Normal tissue surrogates compared with solid tumor tissues for PD biomarker assessments in oncology drug development (theoretical) Normal tissue surrogates (pros) Solid tumor tissue (cons) 1. More easily accessible for multiple biopsies, full prospective collection, completely standardized sample preparation, and fuller time course assessments are more feasible

2. More uniform tissue architecture, so less heterogeneity contributing to background variability 3. Present in healthy volunteers, offering greater potential for early phase clinical development for minimally toxic drugs Normal tissue surrogates (cons) 1. Penetration of drug into normal tissues may differ from that in tumor, misleading any dose/exposure–response assessments 2. Amount of drug target present in normal tissue may be less than in tumors, making bioassays harder, and if drug not in significant excess potentially misleading any dose/exposure– response assessments 3. Form of drug target present in normal tissue may be different from that in tumors (e.g., mutant vs. wild-type), potentially misleading any dose/exposure–response assessments 4. Upstream and downstream context of drug target present in normal tissue may be different from that in tumors (e.g., pathway addiction vs. bystander pathway), such that phenotypic consequences will not correlate between tissues

1. Repeated biopsies of malignant tissue are problematic; baseline samples are often archival (nonstandardized preparation and possible biological changes between time of archival specimen and onset of study) 2. Greater tissue heterogeneity contributing to higher background variability 3. Studies can only be conducted in cancer patients Solid tumor tissue (pros) 1. Drug effects are assessed in tissue where the outcomes are potentially directly clinically relevant 2. Drug effects are assessed in tissue where the outcomes are potentially directly clinically relevant

3. Drug effects are assessed in tissue where the outcomes are potentially directly clinically relevant 4. Drug effects are assessed in tissue where the outcomes are potentially directly clinically relevant

9–28 mg of tissue can be obtained, suitable for IHC or potentially other molecular biology techniques; punches of this size yielding a mean amount of protein of 160 mg (range: 80–270 mg) [56]. The major achievements to date for using skin as a surrogate tissue have utilized IHC to demonstrate proof of drug mechanism for the EGFR inhibitors. The EGFR is expressed at relatively high levels in skin, and both small molecule inhibitors and monoclonal antibodies against the receptor have been shown through IHC to decrease signaling at the level of the drug target itself (reduced EGFR phosphorylation = proof of mechanism), as well as altering downstream signaling in terms of reduced proliferation markers and increased cell-cycle arrest markers (proof of concept, at least in skin cells) [57–61]. The location on the body surface from which skin biopsies are taken may be important for two reasons. Firstly, for cosmetic reasons. Although the upper chest

234

D.R. Camidge et al.

or back has often been used, these areas also have the highest risk of keloid scar formation [58, 59, 62]. Secondly, the frequency of relevant skin adnexae and sun exposure will vary depending on the site chosen. Higher levels of EGFR signaling and proliferation (e.g., Ki67) have been reported in sweat glands, and particularly in hair follicles, compared with basal epidermal rates [56, 58], although the statistical method for analysis can alter the ordering of these results within small series [56]. Ki67 rates and sweat gland frequency do not appear to differ significantly between biopsies taken from Caucasians from the inner upper arm, inner thigh, lower back, or upper outer buttock. However, hair follicle frequency and melanocyte count do. The best compromise of highest hair follicle count and lowest melanocyte counts appears to be represented by biopsies taken from the upper outer buttock [56]. As with IHC in tumor biopsies, the exact method of quantitation of any signal remains controversial. In several studies, single observers blinded to the timing of the specimens recorded the percentage of keratinocytes that were positive or negative across ten high-power fields (400×). Pre- and post-treatment samples (usually after approximately 1 month of dosing) were then compared using Wilcoxon rank testing [58, 60, 61]. While these methods provided some of the earliest proof of drug mechanism for EGFR inhibitors, the exact quantitation used could be improved upon. Ideally, IHC measurements would be quantitated by more than one observer, and an assessment of both inter- and intra-observer variability should be included. While a positive/negative cell on each individual cell may facilitate rapidity of counting, and be an accurate reflection for specific on/off phenomena such as cell-cycle markers, for surface receptor signaling it seems more of a biological abstraction. Preclinically, in cell lines and xenografts, phospho-EGFR change in response to EGFR inhibitors covers the full dynamic range of inhibition with an onset within minutes and an offset over many hours [63]. Suggestions of clinical dose–response relationships using these percentage cell positive counts have been reported, but rarely [57]. Attempts to measure staining across a broader, more biologically representative dynamic range with automated IHC densitometry, have been reported, with minor suggestions of a dose–response relationship [59]. However, what general densitometry per se gains in objectivity, it potentially loses in terms of the subtlety of observer-driven IHC reads for differentiating specific from nonspecific-binding patterns. It may, for example, be particularly susceptible to the additional variable effect of melanin content, secondary to sun exposure differences.

9.4.2 Hair Hair follicles may have an elevated epithelial proliferation rate compared with the basal epidermis of skin [56, 58]. When hairs are plucked, some cells of the hair follicle remain attached to the hair shaft within the so-called hair sheath [64, 65]. Ki67 positive proliferating cells can be seen within the sheath of plucked human hairs by immunofluorescence [64], and epithelial outgrowth can be demonstrated in explant cultures of such plucked hairs in vitro [65]. Plucked human hair (hair sheath cells)

9  Pharmacodynamic Studies in Early Phase Drug Development

235

have therefore also been explored as potentially easily accessible sources of PD biomarkers for oncology drug development. All human hair follicles undergo cycles of growth, degeneration, and regeneration at regular intervals. The normal human hair cycle can be divided into three separate phases: anagen, catagen, and telogen [66]. Active growth (anagen) in the scalp lasts for 3–7 years [67]. During a brief transitional period (catagen), the hair sheath then becomes fibrous and the lower parts of the follicle involute, after which the hair enters a resting phase (telogen), characterized both by the absence of hair growth and by a fully keratinized basal bulb surrounded by an epithelial sac (a so-called clubbed hair). Rigorous combing preplucking is recommended to remove the clubbed hairs, to try to limit the yield on plucking to actively growing anagen hairs. In healthy male volunteers, cell-cyclerelated markers – which conveniently support a simple on/off count for each cell – have been assessed in both scalp and eyebrow hair [68]. Consistent with the clinically observed slower rate of eyebrow growth compared with scalp hair and the tendency for certain cytotoxic chemotherapies to produce hair loss from the scalp significantly more than that from the eyebrows, there is a clear statistically significant difference favoring scalp over eyebrow for Ki67, total p27, and phospho-p27 expression; with total Rb, phospho-Rb, and phospho-HH3 also having higher expression in scalp hairs than in eyebrows [68]. Given the increased number of hairs on the scalp compared with the eyebrows in most patients, the scalp is usually the more obvious body site of choice for assessing antiproliferative PD endpoints in plucked hairs. However, using ANOVA models to analyze the data, the variability of scalp hair proliferative marker signals both between and within subjects appears to be relatively high, with the specific value dependent on the marker assessed (Fig. 9.4) [68]. This variability inevitably increases the chances of spurious positive or negative results and makes the determination of any dose–response aspects of PD change harder to determine. It also emphasizes the importance of sizing a study to the basal variability of a specific marker in a specific tissue to maximize the utility of the data derived from a subsequent intervention study. Although scalp hair may seem the more attractive hair source to use, some patients entering oncology clinical trials may have elements of natural or iatrogenic alopecia, limiting the availability of scalp hair. Therefore, it is interesting to note that proof of mechanism for PARP inhibitors was first demonstrated in plucked eyebrow hairs from ovarian cancer patients (who may have been relatively recently treated with taxanes with associated induction of scalp alopecia). After only a single dose of KU-0059436, induction of gamma-H2AX bodies (a result of PARP inhibition) was clearly noted in the sheath cell nuclei of patients’ plucked eyebrow hairs [69].

9.4.3 Buccal Mucosa The inside of the buccal cavity is covered with a moist epithelium structurally similar to the skin. Its basal proliferation rate is thought to be higher than that of the skin, making it a potentially more attractive accessible tissue for assessing ­antiproliferative

236

D.R. Camidge et al.

Fig.  9.4  Normal tissue PD biomarkers – hair. Demonstration of cell-cycle marker expression KI67 staining of whole (left panels) and sectioned (right panels) eyebrow (a, b) and scalp (c, d) hairs. (Camidge et al. [68])

PD biomarkers [70, 71]. Superficial buccal squamous cells may be relatively easily exfoliated from the inside of the cheeks using a wooden or plastic scraper. Using a simple positive/negative discrimination for proof of mechanism, buccal mucosal scrapes have been used to show evidence of farnesyl-transferase inhibition after a week of treatment with R11577 through the accumulation of prelamin A staining in the nuclei of buccal squames [72]. However, vital dyes showing lack of membrane integrity and SDS-PAGE showing the presence of ­significant proteolysis suggest that almost all buccal scrape cells are dead or dying (DR Camidge, unpublished data). While it is clear that hard-wired information from DNA remains intact enough to be utilized, for example, for so-called “DNA-fingerprinting” in forensic science, the applicability of buccal scrapes for most other PD reads remains uncertain. Punch biopsies of the buccal mucosa, which sample the basal proliferating layers, are harder to perform, but offer more reliable material to work with for most PD endpoints. Utilizing a full preintervention biomarker exploration, including the optimal number of stepped histological sections to count for each marker; of different antibodies directed against the same marker; of the impact of anatomical location within the mouth; and of circadian rhythms; the true background intraand inter-individual variability of a range of specific proliferation biomarkers by IHC in 3  mm buccal punch biopsies was assessed in healthy human volunteers,

9  Pharmacodynamic Studies in Early Phase Drug Development

237

demonstrating much lower coefficients of variation than, for example, in plucked hair cells [68, 73]. Multiple biopsies over a 24-h period were clearly feasible, and the information generated from this background work was then used to help size a subsequent intervention study utilizing a cell-cycle inhibitor. In conjunction with the established single dose pharmacokinetic profile of the drug, biopsies were timed to occur at baseline, at Tmax and at a later time point when drug levels were anticipated to be falling [74]. Perhaps as a consequence of this extensive prospective work-up to maximize the utility of the data derived from the intervention study, not only was proof of mechanism established for AZD5438, a CDKinhibitor, but also dose–response relationships and time–response relationships were established helping to determine both a minimally effective dose and a proposed frequency of dosing that could inform the regimen chosen for subsequent clinical studies (Fig. 9.5) [75].

9.4.4 White Blood Cells and Platelets Circulating mesenchymal cell counts, particularly the white blood cells and platelets, derived from the bone marrow have often been used as rough PD biomarkers (of toxicity) to reflect cytotoxic drug exposure [76]. Their quantification is well established and covers a broad dynamic range. For some of the newer, more specific cell-cycle inhibitors, documentation of myelosuppression in early phase studies may also constitute proof of mechanism as well as offer an easy assessment of a dose/exposure–response relationship through this surrogate tissue. [77] In comparison to the earlier example of mechanism-based toxicity in the skin (rash with EGFR inhibitors), as the cells being assessed for this mechanism-based toxicity are contained within the blood itself, correlations with plasma pharmacokinetic exposures may, in theory, be far closer than for those seen with cells from solid tissues. In addition to plasma drug exposure correlations, correlations between myelosuppression and anticancer efficacy have also been reported, for several tumor types with several different traditional cytotoxic chemotherapies [78]. Blood cell count is a true late PD biomarker, reflecting multiple factors both upstream (e.g., stem cell numbers/stem cell function) and downstream (e.g., DNA damage identification and processing) from the point of drug action. Therefore, as these may differ dramatically between blood cells and tumor, beyond confirming that underdosing with drug is a bad thing, the overall relationship between myelosuppression and anticancer efficacy is likely to be relatively loose. However, blood cells also offer the potential for more molecularly specific interrogations of PD action. PD change in drug target molecules present in PBMCs may be assessed at large numbers of different time points using this easily accessible surrogate tissue, as well as offering a tissue more easily amenable to standard wet lab molecular biological manipulations including FACS analyses, western blots, ELISAs, and RT-PCR. Proof of drug mechanism using

238

D.R. Camidge et al.

Fig. 9.5  Normal tissue PD biomarkers - buccal mucosa. (a) Buccal punch biopsy removed for PD studies. Histograms demonstrating time and dose-response effect of CDK-inhibitor on unit length labellling index (ULLI) of phospho-pRb. Different phospho-epitopes on phospho-pRb are shown at 1.5 hours (b) and 6 hours (c) post-dosing with a CDK inhibitor. Statistical significance is marked with asterixes. (Camidge et al. [75])

9  Pharmacodynamic Studies in Early Phase Drug Development

239

PBMCs has been reported for several different novel agents, including anti-Bcl2 antisense (reduction in Bcl2 protein levels), farnesyl-transferase inhibitors (reduced ras farnesylation), anti-XIAP antisense (reduction in XIAP message levels), MEK inhibitors (reduction of phospho-ERK), mTOR inhibitors (inhibition of p70 S6 kinase activity), and CDK-inhibitors (reduction of phospho-Rb), with suggestions of dose–responses being noted for some of these [75, 77, 79– 82]. As PBMCs are usually quiescent, it is important to realize that in assays exploring proliferation endpoints, cells often require ex vivo stimulation, for example, with phorbol esters, to establish a high enough proliferative baseline, potentially stretching the surrogacy of their relationship to the in  vivo tumor further [75, 81]. Recently, it has been discovered that platelets resist senescence through upregulation of BCl-Xl and that continuous dosing of dogs with a Bcl-2 family inhibitor with activity against BCl-Xl cause a rapid decrease in platelet levels with a slow return to normal levels as younger platelets repopulate the blood [83, 84]. Demonstration of the same phenomenon within first-in-human phase I studies has now established both proof of drug mechanism for ABT-263 and the beginnings of a dose–response relationship in humans [85].

9.4.5 Circulating Ligands, Shed Receptors, and Endothelial Cells Proof of concept for a range of different drugs has been demonstrated from PD effects on circulating cytokines, on molecules shed from the surface of normal cells, and even from changes in the circulating numbers of some normal nonmesenchymal cells. Most examples to date from this area relate to antiangiogenic drugs. The number of cancer therapeutic agents that target angiogenesis is increasing after the initial success of the antivascular endothelial growth factor (VEGF) monoclonal antibody, bevacizumab, in metastatic colorectal cancer [86]. One of the earliest methods used to examine angiogenesis was the measurement of microvascular density in tumor samples stained using IHC techniques (typically CD31 or CD34) to identify and quantify the number of blood vessels in a given area [87, 88]. Although this might represent the most direct measurement of the intended target, the requirement for rebiopsy of patients and the heterogeneity of tumors with respect to stromal and blood vessel involvement render this technique cumbersome and imprecise. In contrast, direct measurement of blood-derived VEGF levels after treatment offers a far more user friendly PD method of assessing a drug’s antiangiogenic effects. Studies have demonstrated that total serum levels of VEGF-A increase after treatment with bevacizumab [89]; however, most or all of the total VEGF is bound directly to the therapeutic agent [90]. Indeed, when free VEGF is measured following immunodepletion, a decrease is seen following treatment; however, this has not been correlated to a therapeutic response [89, 91]. Other cytokines, such as placental growth factor

240

D.R. Camidge et al.

(PlGF), hepatocyte growth factor (HGF), IL-8, and basic fibroblast growth factor (bFGF (FGF-2)) have also been measured following antiangiogenic therapy, and although they change, changes in either direction have not been successfully or consistently correlated with response [90, 92–94]. Increases in serum or plasma VEGF are not limited to agents that directly bind VEGF, but have also been observed with tyrosine kinase inhibitors that target the vascular endothelial growth factor receptors (VEGFR) [95], suggesting that some rises in VEGF or other cytokines such as PlGF [94] may reflect feedback loops, for example, as a response to increased hypoxia. Soluble forms of VEGF receptors have been detected in the blood of patients with cancer and act as endogenous negative regulators of angiogenesis by sequestering proangiogenic cytokines. Measurement of these shed receptors, which arise from alternative mRNA splicing [96], has demonstrated that sVEGFR-2 [95] or sVEGFR-3 [94] levels decrease significantly following treatment with antiangiogenic agents, but this has not been correlated with clinical benefit. The observed decrease in soluble receptors also fits with a feedback mechanism, potentially aimed at increasing free ligand levels in response to the drug’s antiangiogenic effects. Much attention has recently been focused on the role of CEC or endothelial progenitor cells (EPC), which can be detected in the blood of patients using flow cytometric techniques. An increase in CECs has been demonstrated after treatment with the vascular disrupting agent, ZD6126, and may represent a repair mechanism following vascular damage [97]. Treatment with sunitinib (cf. Chap. 14 or 15) in patients with metastatic GIST also appears to increase CECs, and has demonstrated a statistically significant correlation with clinical benefit [95]. Mature CEC were identified in this study using flow cytometry and were defined as CD45−/CD31+/P1H12+/CD133−, although there is not yet uniform agreement on the best method to identify these cells. In contrast, a study using bevacizumab in patients with rectal cancer demonstrated decreases in both CECs and EPCs (CD133+) [88], illustrating the complex nature of these markers and their response to therapy. At best it would be fair to say that for most circulating markers of angiogenesis, although PD effects consistent with proving drug action on the relevant physiological processes may be demonstrable, their significance, if any, for the patient largely remains obscure.

9.5 Choosing the Right Tissue for PD Studies: Circulating Tumor Cells and Tumor Cell-Derived Materials For hematological malignancies with a leukemic element, it is obvious that a simple blood sample can provide direct access to malignant cells for both PD readouts and as a response evaluation to any given intervention. However, even for solid tumors there is now the increasing potential to detect and interrogate circulating tumor cells (CTCs) or circulating tumor-derived material as PD biomarkers.

9  Pharmacodynamic Studies in Early Phase Drug Development

241

CTCs have been recognized for many years, but recently, robust technology to capture and quantify these cells has been licensed for a number of different solid tumors. The semiautomatic Veridex CellSearch Assay utilizes magnetic beads coated with anti-EpCAM to capture circulating epithelial cells from 7.5 ml blood samples. In theory, endothelial cells and blood cells should not be captured, but contaminating cells, particularly lymphocytes, are common. Really, the process is one of enrichment rather than of purification. Following enrichment by antiEpCAM, the cells are then fixed, permeabilized, and fluorescent-labeled with a cocktail of anti-CD45 (to identify white blood cells), DAPI (to delineate the nuclei), and anticytokeratin 8, 18, and 19 (to further define epithelial cells). The cells are then drawn into a viewing chamber where an image analysis program “counts” only those objects that meet the relevant cell and nuclear size/shape criteria, and which are also CD45 negative and cytokeratin positive. Although most of the CellSearch validation has been conducted on breast cancer, at least five cells per 7.5 ml blood are detectable in a significant proportion of many other common metastatic tumors including prostate, colorectal, and lung cancer [98]. In addition to generating absolute cell counts (essentially a variant on objective tumor response), that is being explored for its prognostic significance, the technology also permits additional labeling to be used to define the number of cells positive for other markers [99–101]. Utilizing the Veridex system to explore the presence of CTCs positive for IGF-1R, it was d­ emonstrated that CP-751,871, a monoclonal antibody directed against the extracellular domain of the IGF-1R, reduced both the total number of CTCs and the number of IGF-1R positive CTCs detectable in patients within early phase studies of the drug [102]. This example emphasizes the importance of taking into account the total CTC number and looking at relative change when considering any apparent PD effect from an intervention. For example, an increase in the average levels of a biomarker in response to a drug may reflect either a general PD upregulation or a specific deletion of cells with lower levels of the marker. Cell enrichment prior to fixation and additional labeling in a form suitable for alternative molecular biology assays is also possible. Detection of gene copy number alterations by FISH, for example, has also been described on CTCs captured via the Veridex system offering the potential to look for qualitative changes over time – either as a mechanism for individual CTC survival or overall acquired resistance at the point of progression [103]. How to accurately quantify proportional change in the biology of CTCs, as opposed to simple presence or absence of a marker of uncertain background incidence, remains to be determined. Following the same principle of anti-EpCAM capture, advances in microfluidic technology (so-called micro-post technology) may lead to a potentially even more sensitive method for CTC enrichment than the Veridex assay. Recently, this micropost technology was used to enrich the starting material for specific EGFR mutation detection in blood from patients with NSCLC [104]. Whether only whole CTCs, or also DNA associated with CTC fragments, was enriched is currently unclear. Using this technology, it was not only possible to detect mutant DNA from the CTC material more frequently than from plasma, but also to perform ­quantitative

242

D.R. Camidge et al.

assessments of changes in the amount of CTCs and, by using specific PCR probes, assess specific changes in their molecular nature while on EGFR directed therapy. However, although these new technologies are exciting, with enormous potential to look for PD effects, there are many issues still to be worked out, including the overall incidence of CTCs in different tumor types detectable by different technologies, and the inter- and intra-patient background variability of them such that we can interpret any changes in CTCs following an intervention appropriately. Also, even the simplest question – as to whether CTCs are truly representative of the underlying tumor on a molecular level, especially after the various manipulations of CTC capture – still remains to be answered [105].

9.6 Incorporation of PD Markers Within Early Phase Clinical Trials So far, we have explored the basic techniques for capturing material potentially suitable for PD biomarker assessments, the methodologies that may be used for such assessments, and some examples of where each has been used in practice. But now we must ask what PD biomarkers have really added to oncology drug development to date and can they add more?

9.6.1 Proof of Drug Mechanism Evidence of activity, in an early phase clinical study, on a specific molecule, at achievable pharmacokinetic exposures in humans, from a drug designed to hit that specific molecule constitutes “proof of drug mechanism.” This is the single most common success story for PD biomarkers to date. Although there are theoretical concerns about the exposures needed to hit the target molecule in a surrogate tissue and in the tumor, and whether the amount, nature, and context of the target is the same in both locations, surrogate tissues often offer the easiest route to proof of mechanism. Unless the anticipated PD effect is likely to be so obvious that there is no room for interpretation (e.g., the formation of monasters from aurora kinase inhibition – a phenomenon hardly ever seen in untreated samples), clarifying in advance the background variability of the PD biomarker of choice to ensure that false positives or negatives are minimized is essential [106]. Beware of the overoptimistic interpretation of PD results checked from multiple individuals, but only presented from a single patient who just happened to show PD change in the direction expected from the mechanism of action of the drug. The importance of establishing proof of mechanism, beyond the additional interest that such data generate in a purely scientific sense, is mostly risk management. If a new compound has completed phase I monotherapy studies with no clear clinical evidence of activity, demonstrating that at least the target was affected, reassures

9  Pharmacodynamic Studies in Early Phase Drug Development

243

the developer that the issues to address probably relate more to patient selection than to inadequate drug exposures. If there is clear evidence of clinical activity, the role of an early PD biomarker becomes relatively less important – “who cares if it hits the target – it works!” However, studying early PD biomarkers in this setting may still help support or deny the hypothesis that clinical activity has some correlation with effects on the proposed target pathway (cf. proof of concept) and help guide the search for susceptible groups of patients in the future. Because of the theoretical concerns about differences between malignant and surrogate tissues (Table 9.3), if malignant tissue itself can be utilized to show proof of mechanism, theoretically the risks of continued development of the compound are even further reduced. Often both surrogate and malignant tissues will be collected. In part, this is because surrogate tissues offer the likelihood of fuller data sampling. This is both in terms of number of patients with biopsies taken and number of biopsies taken within the same patient over a time course of exposure. These data are then ideally supported by sparse data derived from the tumor tissue. However, sampling both tissues also offers two other potential advantages. Firstly, because of possible target contextual differences and background variability differences between the tissues, the surrogate tissue may act as a relevant positive/ negative control for the PD assay (cf. Sect. 9.6.2). Secondly, and more speculatively, if it was possible to establish a clear drug exposure–response relationship in both tissues (cf. Sect. 9.6.3), in individual patients in the future it may be possible to titrate drug to the more easily accessible of the two tissues (cf. Sect. 9.6.5). Generally speaking, proof of mechanism is only important to establish once for each drug (Fig. 9.6a). Consequently, if it has been established within the first phase I study, there is little point in repeating it within other phase I studies or in subsequent phase II or III studies. Even if significant synergy is expected when more than one drug is combined, unless the PD biomarker of a drug is sufficiently subtle and the background variability sufficiently low, to reliably distinguish between different levels of target modification, there is also little point to include PD biomarkers within phase Ib combination studies after proof of mechanism in a phase I monotherapy study has been established. Utilizing the same logic (cf. Sect.  9.6.3), whether PD biomarkers are even included within the dose escalation period or only within the expanded cohort at the presumed recommended phase II dose (RP2D) of a phase I study remains a matter of debate. Unless an investigator has done a lot of background work to maximize the chances of his/her assay being able to determine an exposure–response relationship, saving PD biomarkers for the RP2D expanded cohort is probably the most economical and straightforward approach. Consequently, studies with biopsies that are optional in the dose escalation period and mandatory within the RP2D cohort (hence limiting the eligible population in the RP2D cohort to those patients who are amenable to additional invasive procedures and who have tissue accessible for biopsies) are becoming increasingly common (Fig.  9.6a). Mandatory biopsies are certainly permissible, advocating that drug development of targeted agents without a component of biological testing to better understand the effects of the drug does a poor service to the patient and scientific communities. Recent surveys among providers, institutional review boards (IRBs), and patients

244

D.R. Camidge et al.

a Phase I study with basic PD marker (anticipated poor ability to accurately distinguish subtle differences of effect)

Dose escalation Fixed dose cohort stage stage (e.g. RP2D expansion cohorts) First-in-class NOa YES b First-in-man NOa YES Phase I but c a,c YES NO not first-inman Combination a a NO NO studies a - Only if PD assay has sufficient dynamic range and low enough variability within the tissue of interest to reliably distinguish differences in extent of PD activity beyond presence/absence

Non-tolerated dose

Toxicity-based RP2D

b - added value in terms of risk management falls with increasing establishment of proof of mechanism/concept from other compounds in same class of drug c - only adds value in terms of risk management if proof of mechanism/concept not already established for drug

Dose escalation cohorts (optional/no PD markers)

Expanded cohort(s) at RP2D for additional safety information and exploration in specific tumor types (mandatory PD markers)

b Phase I study with advanced PD marker (anticipated good ability to accurately distinguish subtle differences of effect)

Lead on to randomized Phase II study comparing toxicity-based and PD-based RP2Ds (if different) (both efficacy and toxicity (but not PD) endpoints required)

Non-tolerated dose

Toxicity-based RP2D

Dose escalation cohorts (consider PD markers depending on cohort sizes)

versus

Expanded cohort(s) at RP2D and fraction(s) of RP2D to establish biological and toxicity-based RP2Ds (mandatory PD markers)

9  Pharmacodynamic Studies in Early Phase Drug Development

245

found that patients were more accepting of higher theoretical procedure-related risks than were providers and IRBs [107]. Most patients who have been biopsied recall a tendency to tolerate their biopsies well, and have little or no concern allowing their specimens to be tested for research purposes. One persistent area of controversy is whether PD biomarkers to establish proof of mechanism should only be used for agents that are first-in-class, or do they still add value to the development of a drug that is fourth or fifth or tenth to market in the same class? The short answer is that they do still add value to an individual drug, but if the class of agents is already well established, their relative contribution to risk management becomes less and they conceivably could be dispensed with, particularly if they appear to slow the trial process down or significantly increase its cost.

9.6.2 Proof of Concept Proof of concept also relates to risk management during drug development. Demonstrating that the drug affects elements beyond the drug target (ideally in addition to demonstrating PD effects at the level of the drug target), that are a step closer to a clinically relevant endpoint, reassures the drug developer that his or her initial theory – that the drug target is an important target – continues to be valid. While tumor shrinkage is the most robust proof of concept marker occurring in an

Fig.  9.6  Schema to display current (a) and future (b) potential uses of PD biomarkers within phase I oncology clinical trials. (a) Due to the majority of PD assays having insufficient dynamic range and excessive variability within the tissue of interest, PD biomarkers, especially those requiring invasive procedures (represented by the FNA needles in this schema), are not currently recommended during the dose escalation stage. However, as data on the presence/absence of change can always help to generate potentially useful hypotheses in terms of dosing or sensitivity/ resistance characteristics, simple PD assessments such as mechanism-based toxicities or sampling of freely offered accessible tissues may still be captured. For most relevant molecular PD assays, their use is recommended solely for the larger sample size available within cohorts treated at the toxicity-based predicted recommended phase 2 dose (RP2D). The added value of PD biomarkers even within a phase I study will vary depending on whether the drug under investigation is firstin-class, first-in-man, and whether these relate to monotherapy or to its use in combination therapy (see embedded table). PD biomarkers are not usually recommended for phase II or III studies. (b) In the event of an advanced PD biomarker being available, with good potential to reliably distinguish between subtle levels of effect, a different study design may be informative. Capturing PD assessments during the dose escalation stage, and ­particularly within the larger sample sizes of expanded cohorts at the RP2D and at dose levels that are fractions of the toxicity-based RP2D, a different, biologically based RP2D may also be determinable. In this situation, both the toxicity-based and biologically based RP2Ds could then be compared within a subsequent randomized phase II study utilizing both efficacy and toxicity endpoints to ultimately determine which dose of drug (if any) to take forward into phase III testing

246

D.R. Camidge et al.

early phase study, it is often a rare phenomenon, can be slow, and the specifics for categorizing responses are less than optimal and are poor at truly assessing disease stabilization. While advances in functional imaging and volumetric radiological measurements may improve matters for some cancers in the future, demonstration of PD changes at the cellular level – usually decreased markers of proliferation or increased markers of apoptosis – continue to be used as rapid proof of concept PD assessments in early phase studies. Beyond the usual issues of accurately determining background variability of the assay in the tissue of interest to most reliably interpret any PD changes, some preclinical assessment of the time to manifest effects downstream of the primary drug target should be made when considering the optimal time points postexposure to sample tissue for proof of concept markers. Because of the theoretical concerns about surrogate and malignant tissue differences, proof of concept, although demonstrable in surrogate tissues, is much better suited to direct tumor sampling. If sufficient work was put into determining the reliability of any given assay in surrogate tissues, in theory these tissues could offer fuller data sampling, in combination with drug target PD assessments, to permit exploration of the level and duration of target inhibition required to produce relevant downstream effects in humans. However, because of the significant differences in target and target context between neoplastic and non-neoplastic cells, this would only really generate a hypothesis that would then need to be confirmed in tumor tissues. The absence of expected changes in proof of concept markers can also be useful (provided the assay is sufficiently robust and/or relevant positive controls are included, for example, within a surrogate tissue to interpret the absence of change correctly). For example, a PD study of gefitinib in gastric cancer involved repeated tumor sampling via endoscopy. Although a month of exposure to gefitinib abolished phospho-EGFR signaling by IHC, there was no significant effect on downstream signaling in either the ras-raf-MEK-ERK (no change in phospho-ERK) or PI3-Kinase (no change in phospho-AKT) pathways, and no apparent clinical benefit from the treatment within the study [108]. What these data appear to relay is that the absence of activity in gastric cancer is not related to the absence of activity on the drug target at clinically achievable doses, but that the target is not generally relevant in the disease being studied. Intriguingly, within the same study, it was reported that specifically in those patients with elevated baseline EGFR activation (phospho-EGFR) in their tumors, gefitinib did reduce phospho-ERK levels, but not in the surrounding normal mucosa, providing important clues to a potentially sensitive subpopulation that could be explored more in the future. Tissue assessments related to proof of concept can also lead to better understanding of the relevant biology and the complex interplay between pathways. A series of recent, seemingly counter-intuitive, clinical observations have deepened our understanding of cancer. Patients treated with the rapamycin derivative, RAD001, showed unexpected increase in Akt activation [109], probably as a result of insulin receptor substrate-1-induced expression. The data that had also been documented in tumor samples from independent preclinical studies [50] suggested that downregulation of receptor tyrosine

9  Pharmacodynamic Studies in Early Phase Drug Development

247

kinase signaling is a frequent event in tumor cells with constitutive mTOR ­activation and that inhibition of mTOR releases additional traffic down this upregulated feedback loop. The existence of complex loops is also exemplified by the apparently paradoxical findings observed in a recently published clinical trial where upregulation of phosphorylated EGFR was observed after treatment with erlotinib in breast cancer patients [110]. Preclinical evidence that treatment with EGFR inhibitors in resistant models induced an upregulation of the EGFR protein following a classic enzymatic compensatory feedback loop [111] may help explain this clinical observation.

9.6.3 Dose/Regimen Selection In general, the use of PD biomarkers to select an appropriate dose or regimen of administration of drug to take forward into subsequent clinical studies is in its infancy. The field is currently plagued by inadequate biomarker preparatory work, inadequate technology, inappropriate trial design, and general nihilism based on, perhaps the not unsurprising, lack of early and easy successes. The theoretical value of a PD biomarker that is robust enough to demonstrate a clear dynamic range of effect from a drug intervention on human trial material includes: • Determination of a minimally effective biological dose – which if established even in surrogate tissues in healthy volunteers would save exposing patients with cancer in subsequent phase I studies to potentially pointless doses in the earliest dose escalation cohorts. • Determination of the magnitude of effect over a time course of exposure, which when correlated with pharmacokinetic (PK) information may inform the frequency of dosing. • Determination of a maximally effective biological dose to take forward that may be well below that of the maximally tolerated dose that may be minimizing unnecessary toxicity in patients, while preserving efficacy. • Determination of a maximally effective biological dose and correlating it with Cmax, AUC, or time above a threshold to take forward as a true target PK parameter to aim for in individual patients; and to use when considering the clinical relevance of any drug–drug interactions. • Determination of a minimally and maximally effective biological dose for monotherapy that may prove useful in choosing a starting dose for use in combination trials where there may be concerns relating to overlapping toxicities. If a maximally effective dose/regimen, sometimes called the optimally biologically effective dose (OBED), could be estimated through the use of PD biomarkers, it would then be important to perform, at least within the phase II setting, a randomized study comparing such a biologically determined dose with a more traditional toxicity-selected RP2D, utilizing both efficacy and toxicity endpoints in the final

248

D.R. Camidge et al.

analysis to confirm the dose/regimen to take forward to phase III (Fig.  9.6b). However, the basis of choosing a dose selected via PD endpoints to even consider taking through to a randomized study remains elusive. PD readouts relating to the effect of a rapamycin analog (CCI-779) on p70S6kinase activity in PBMCs revealed effects at all doses from 25 to 250 mg, but with no evidence of a dose response [82]. Using rapamycin-related PD effects in a different tissue, inhibition of phospho-P70S6Kinase utilizing H-scoring IHC in skin also failed to demonstrate a clear PK–PD relationship when correlated with PK exposures [9]. In the same study, a modified continuous reassessment method using real-time PD data from skin biopsies as the primary dose-estimation parameter was also explored [9]. For the purposes of dose escalation, PD effect was interpreted per patient as to whether there was at least 80% inhibition of phospho-P70S6Kinase from baseline. However, toxicity-based boundaries limited the escalation steps before PD readouts did. In order to have PD assays that would permit true dose selection, considerable work is required. All attempts to understand and to minimize the background variability of the biomarker, which will vary by biomarker and by tissue, reagent, and technique, must be made. Surrogate tissues offer the potential for larger numbers of samples and at least for PBMCs, material more easily amenable to molecular ­biological techniques with broad dynamic ranges such as ELISAs and Westerns. However, the levels of key target molecules may be far less than in tumors, and their type and context may also differ. Of note, in the rapamycin study, in contrast to the effects in normal tissues, FNA biopsies of tumors that demonstrated or did not demonstrate downregulation of phospho-P70S6Kinase staining appeared to distinguish between those patients who did or did not derive clinical benefit, respectively (Fig.  9.3). Malignant cell suspensions from third-space collections may offer amounts of tumor material in a format suitable for full wet laboratory investigations, but the increased variability of the assay readouts in these tissues and of the available amounts of material has to be addressed. IHC readouts – which are feasible on very small amounts of material and which may be acquired through repeated biopsies of tumor material – are the furthest advanced in terms of trying to establish dose–response relationships for PD endpoints. However, several issues exist with the use of IHC in this context. Firstly, it is mostly dose rather than exposure that has usually been correlated with PD effect, which is an additional abstraction. Secondly, the method of quantification of IHC signals is still very crude. Subjective assessments by pathologists are prone to considerable intra- and inter-observer variability. Categorical determinations make for less variability and simpler (yet less powerful) statistical comparisons, but may not truly reflect the underlying biology of the system. Whether recent advances in automated IHC quantitation, using various image analysis techniques, for example, the use of dual-labeling with different fluorescent markers to delineate regional masks that permit the assessment of specific cellular areas (e.g., membrane vs. cytoplasm vs. nucleus) to be counted separately, will advance PD readouts to the level whereby they may usefully contribute to dose-finding remains to be seen [112].

9  Pharmacodynamic Studies in Early Phase Drug Development

249

9.6.4 Lead/Backup Compound Selection Pharmaceutical companies commonly identify a drug target and then develop a series of different agents with the potential to interact with it. Choosing the compound from this stable of agents to take forward for full clinical development is often determined preclinically based on such issues as solubility, cost of preparation, preclinical efficacy, toxicity, and the potential for drug–drug interactions. However, if more than one compound were to enter clinical studies in patients or in healthy volunteers, even with single doses, the possibility that PD biomarkers could be used to identify the true lead compound early on has to be considered. Firstly, because of possible differences in tissue distribution (tumor vs. surrogate tissue) that may differ between compounds, the potential for surrogate tissues to inaccurately reflect the relevant site of action of the drug in the body means that they are probably too risky to be the basis for a conclusion regarding the level of activity in the tumor. Secondly, and perhaps more ­importantly, unless there are very dramatic differences – essential activity vs. no activity – on the drug target between the two drugs, currently available assays applied to tumor tissue are still unlikely to be subtle enough to distinguish relevant differences between similar agents.

9.6.5 Surrogate Marker of Clinical Benefit for Regulatory or Individual Patient Decision Making The strength of relationship between drug-induced changes in tumor size or in a recognized circulating tumor marker, such as PSA for prostate cancer, and endpoints considered acceptable by regulatory agencies, such as overall survival, or recently progression-free survival, to allow these earlier biomarkers to be considered as true surrogates of benefit takes many years and considerable data to establish [5, 6, 113]. Because of the potential for increased distance between drug target action and clinical benefit in surrogate tissues compared with malignant tissues, there does not seem to be any realistic role for surrogate tissue PD biomarkers in this context. Even for the tumor PD biomarkers discussed here, any use of them as surrogates of clinical benefit would only be conceivable many years in the future, if at all, and should only be explored once true clinical benefit from the drug in later phase studies has already been established. However, for individual patients, some aspects of PD biomarkers, beyond the well-established use of radiological and tumor marker responses, are already being explored for clinical decision making. None of the molecular PD biomarkers ­discussed in this chapter are yet being used as surrogates of clinical benefit or ­progression, but some mechanism-based toxicity endpoints are being used to at least explore adequate drug exposure in the clinic. Just as correlations between chemotherapy-induced myelosuppression and anticancer efficacy have been

250

D.R. Camidge et al.

reported, the absence of PD biomarker change in surrogate tissues – notable the absence of skin rash from EGFR directed therapies – has already been used to suggest that dosages of specific targeted drugs should be increased in such patients [78, 114].

9.7 Conclusions PD biomarkers developed over the last few years have generated considerable excitement in the field of oncology drug development. Their incorporation in early phase studies of novel agents, demonstrating proof of drug mechanism or proof of concept, is a useful risk management tool during drug development (Fig. 9.6). They can also add valuable general scientific interest to, and generate scientific insights within, early phase studies. However, to make them truly informative with regard to dose/regimen finding within a study, or conceivably later within individual patient decision making, remains problematic and will continue to require considerable preliminary work in each instance.

References 1. Yun CH, Mengwasser KE, Toms AV, et al: The T790M mutation in EGFR kinase causes drug resistance by increasing the affinity for ATP. Proc Natl Acad Sci USA 105:2070–5, 2008 2. Tao Y, Pinzi V, Bourhis J, et al: Mechanisms of disease: signaling of the insulin-like growth factor 1 receptor pathway – therapeutic perspectives in cancer. Nat Clin Pract Oncol 4: 591–602, 2007 3. Morgillo F, Kim WY, Kim ES, et al: Implication of the insulin-like growth factor-IR pathway in the resistance of non-small cell lung cancer cells to treatment with gefitinib. Clin Cancer Res 13:2795–803, 2007 4. Nahta R, Yuan LX, Zhang B, et al: Insulin-like growth factor-I receptor/human epidermal growth factor receptor 2 heterodimerization contributes to trastuzumab resistance of breast cancer cells. Cancer Res 65:11118–28, 2005 5. Louvet C, de Gramont A, Tournigand C, et al: Correlation between progression free survival and response rate in patients with metastatic colorectal carcinoma. Cancer 91:2033–8, 2001 6. Burzykowski T, Buyse M, Piccart-Gebhart MJ, et al: Evaluation of tumor response, disease control, progression-free survival, and time to progression as potential surrogate end points in metastatic breast cancer. J Clin Oncol 26:1987–92, 2008 7. Plecha DM, Goodwin DW, Rowland DY, et al: Liver biopsy: effects of biopsy needle caliber on bleeding and tissue recovery. Radiology 204:101–4, 1997 8. Knelson M, Haaga J, Lazarus H, et al: Computed tomography-guided retroperitoneal biopsies. J Clin Oncol 7:1169–73, 1989 9. Jimeno A, Rudek MA, Kulesza P, et al: Pharmacodynamic-guided modified continuous reassessment method-based, dose-finding study of rapamycin in adult patients with solid tumors. J Clin Oncol 26:4172–9, 2008 10. Rimm DL, Stastny JF, Rimm EB, et al: Comparison of the costs of fine-needle aspiration and open surgical biopsy as methods for obtaining a pathologic diagnosis. Cancer 81:51–6, 1997

9  Pharmacodynamic Studies in Early Phase Drug Development

251

11. Hidalgo M, Amador ML, Jimeno A, et al: Assessment of gefitinib- and CI-1040-mediated changes in epidermal growth factor receptor signaling in HuCCT-1 human cholangiocarcinoma by serial fine needle aspiration. Mol Cancer Ther 5:1895–903, 2006 12. Erickson RA, Sayage-Rabie L, Beissner RS: Factors predicting the number of EUS-guided fine-needle passes for diagnosis of pancreatic malignancies. Gastrointest Endosc 51:184–90, 2000 13. Jhala D, Eloubeidi M, Chhieng DC, et al: Fine needle aspiration biopsy of the islet cell tumor of pancreas: a comparison between computerized axial tomography and endoscopic ultrasoundguided fine needle aspiration biopsy. Ann Diagn Pathol 6:106–12, 2002 14. Tyan YC, Wu HY, Su WC, et al: Proteomic analysis of human pleural effusion. Proteomics 5:1062–74, 2005 15. Gortzak-Uzan L, Ignatchenko A, Evangelou AI, et al: A proteome resource of ovarian cancer ascites: integrated proteomic and bioinformatic analyses to identify putative biomarkers. J Proteome Res 7:339–51, 2008 16. Brown AP, Wendler DS, Camphausen KA, et al: Performing nondiagnostic research biopsies in irradiated tissue: a review of scientific, clinical, and ethical considerations. J Clin Oncol 26:3987–94, 2008 17. Tobkes AI, Nord HJ: Liver biopsy: review of methodology and complications. Dig Dis 13:267–74, 1995 18. Little AF, Ferris JV, Dodd GD, 3rd, et al: Image-guided percutaneous hepatic biopsy: effect of ascites on the complication rate. Radiology 199:79–83, 1996 19. Dowlati A, Haaga J, Remick SC, et al: Sequential tumor biopsies in early phase clinical trials of anticancer agents for pharmacodynamic evaluation. Clin Cancer Res 7:2971–6, 2001 20. Kulesza P, Eltoum IA: Endoscopic ultrasound-guided fine-needle aspiration: sampling, pitfalls, and quality management. Clin Gastroenterol Hepatol 5:1248–54, 2007 21. Wu JM, Fackler MJ, Halushka MK, et al: Heterogeneity of breast cancer metastases: comparison of therapeutic target expression and promoter methylation between primary tumors and their multifocal metastases. Clin Cancer Res 14:1938–46, 2008 22. Medeiros F, Rigl CT, Anderson GG, et  al: Tissue handling for genome-wide expression analysis: a review of the issues, evidence, and opportunities. Arch Pathol Lab Med 131:1805–16, 2007 23. van Maldegem F, de Wit M, Morsink F, et al: Effects of processing delay, formalin fixation, and immunohistochemistry on RNA Recovery From Formalin-fixed Paraffin-embedded Tissue Sections. Diagn Mol Pathol 17:51–8, 2008 24. Benchekroun M, DeGraw J, Gao J, et  al: Impact of fixative on recovery of mRNA from paraffin-embedded tissue. Diagn Mol Pathol 13:116–25, 2004 25. Ramaswamy S, Golub TR: DNA microarrays in clinical oncology. J Clin Oncol 20:1932–41, 2002 26. Tumor Analysis Best Practices Working Group: Expression profiling – best practices for data generation and interpretation in clinical trials. Nat Rev Genet 5:229–37, 2004 27. Micke P, Ohshima M, Tahmasebpoor S, et  al: Biobanking of fresh frozen tissue: RNA is stable in nonfixed surgical specimens. Lab Invest 86:202–11, 2006 28. Jochumsen KM, Tan Q, Dahlgaard J, et  al: RNA quality and gene expression analysis of ovarian tumor tissue undergoing repeated thaw-freezing. Exp Mol Pathol 82:95–102, 2007 29. Florell SR, Coffin CM, Holden JA, et al: Preservation of RNA for functional genomic studies: a multidisciplinary tumor bank protocol. Mod Pathol 14:116–28, 2001 30. Werner M, Chott A, Fabiano A, et al: Effect of formalin tissue fixation and processing on immunohistochemistry. Am J Surg Pathol 24:1016–9, 2000 31. Jimeno A, Kulesza P, Kincaid E, et  al: C-fos assessment as a marker of anti-epidermal growth factor receptor effect. Cancer Res 66:2385–90, 2006 32. Daneshmand M, Parolin DA, Hirte HW, et al: A pharmacodynamic study of the epidermal growth factor receptor tyrosine kinase inhibitor ZD1839 in metastatic colorectal cancer patients. Clin Cancer Res 9:2457–64, 2003

252

D.R. Camidge et al.

33. Appleton K, Mackay HJ, Judson I, et al: Phase I and pharmacodynamic trial of the DNA methyltransferase inhibitor decitabine and carboplatin in solid tumors. J Clin Oncol 25:4603–9, 2007 34. Garcia-Manero G, Kantarjian HM, Sanchez-Gonzalez B, et al: Phase 1/2 study of the combination of 5-aza-2¢-deoxycytidine with valproic acid in patients with leukemia. Blood 108:3271–9, 2006 35. Bittner M, Meltzer P, Chen Y, et al: Molecular classification of cutaneous malignant melanoma by gene expression profiling. Nature 406:536–40, 2000 36. Sorlie T, Perou CM, Tibshirani R, et al: Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Sci USA 98:10869–74, 2001 37. van de Vijver MJ, He YD, van’t Veer LJ, et al: A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med 347:1999–2009, 2002 38. Eschrich S, Yang I, Bloom G, et al: Molecular staging for survival prediction of colorectal cancer patients. J Clin Oncol 23:3526–35, 2005 39. Ramaswamy S, Tamayo P, Rifkin R, et  al: Multiclass cancer diagnosis using tumor gene expression signatures. Proc Natl Acad Sci USA 98:15149–54, 2001 40. Glinsky GV, Glinskii AB, Stephenson AJ, et al: Gene expression profiling predicts clinical outcome of prostate cancer. J Clin Invest 113:913–23, 2004 41. Modlich O, Prisack HB, Munnes M, et al: Predictors of primary breast cancers responsiveness to preoperative epirubicin/cyclophosphamide-based chemotherapy: translation of microarray data into clinically useful predictive signatures. J Transl Med 3:32, 2005 42. Potti A, Dressman HK, Bild A, et al: Genomic signatures to guide the use of chemotherapeutics. Nat Med 12:1294–300, 2006 43. Potti A, Mukherjee S, Petersen R, et al: A genomic strategy to refine prognosis in early-stage non-small-cell lung cancer. N Engl J Med 355:570–80, 2006 44. Chang JC, Wooten EC, Tsimelzon A, et al: Gene expression profiling for the prediction of therapeutic response to docetaxel in patients with breast cancer. Lancet 362:362–9, 2003 45. Ayers M, Symmans WF, Stec J, et al: Gene expression profiles predict complete pathologic response to neoadjuvant paclitaxel and fluorouracil, doxorubicin, and cyclophosphamide chemotherapy in breast cancer. J Clin Oncol 22:2284–93, 2004 46. Ferry DR, Anderson M, Beddard K, et  al: A phase II study of gefitinib monotherapy in advanced esophageal adenocarcinoma: evidence of gene expression, cellular, and clinical response. Clin Cancer Res 13:5869–75, 2007 47. Atkins D, Reiffen KA, Tegtmeier CL, et  al: Immunohistochemical detection of EGFR in paraffin-embedded tumor tissues: variation in staining intensity due to choice of fixative and storage time of tissue sections. J Histochem Cytochem 52:893–901, 2004 48. Cunningham D, Humblet Y, Siena S, et al: Cetuximab monotherapy and cetuximab plus irinotecan in irinotecan-refractory metastatic colorectal cancer. N Engl J Med 351:337–45, 2004 49. Slamon DJ, Leyland-Jones B, Shak S, et al: Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2. N Engl J Med 344:783–92, 2001 50. Jimeno A, Kulesza P, Wheelhouse J, et al: Dual EGFR and mTOR targeting in squamous cell carcinoma models, and development of early markers of efficacy. Br J Cancer 96:952–9, 2007 51. Messersmith W, Oppenheimer D, Peralba J, et al: Assessment of Epidermal Growth Factor Receptor (EGFR) signaling in paired colorectal cancer and normal colon tissue samples using computer-aided immunohistochemical analysis. Cancer Biol Ther 4:1381–6, 2005 52. Burtrum D, Zhu Z, Lu D, et  al: A fully human monoclonal antibody to the insulin-like growth factor I receptor blocks ligand-dependent signaling and inhibits human tumor growth in vivo. Cancer Res 63:8912–21, 2003 53. Jones RJ, Boyce T, Fennell M, et al: The impact of delay in cryo-fixation on biomarkers of Src tyrosine kinase activity in human breast and bladder cancers. Cancer Chemother Pharmacol 61:23–32, 2008

9  Pharmacodynamic Studies in Early Phase Drug Development

253

54. Hamilton M, Wolf JL, Rusk J, et al: Effects of smoking on the pharmacokinetics of erlotinib. Clin Cancer Res 12:2166–71, 2006 55. Petty WJ, Hughes AN, O’Brien MER, Chick JB, Rankin E, Woll P, Dunlop D, NIcolson M, Boinpally R, Price A: Overcoming a CYP1A1/1A2 mediated induction of metabolism by escalating erlotinib dose in current smokers – pharmacokinetic and exploratory survival analyses in patients with advanced non-small cell lung cancer. IASLC/ASCO ASTRO Multidisciplinary Symposium in Thoracic Oncology 2008: abstract 124 56. Camidge DR, Davies MJ, Laud PJ, et  al: Factors determining the optimal body site and method for obtaining punch biopsies of human skin as a tissue in which to assess pharmacodynamic and pharmacokinetic endpoints in drug development studies. Cancer Chemother Pharmacol 57:52–8, 2006 57. Salazar R TJ, Rojo F, et al: Dose-dependent inhibition of the EGFR and signaling pathways with the anti-EGFR monoclonal antibody (MAb) EMD 72000 administered every three weeks (q3w). A phase I pharmacokinetic/pharmacodynamic (PK/PD) study to define the optimal biological dose (OBD). Proceedings of the American Society of Clinical Oncology 2004: abstract 2002 58. Albanell J, Rojo F, Averbuch S, et  al: Pharmacodynamic studies of the epidermal growth factor receptor inhibitor ZD1839 in skin from cancer patients: histopathologic and molecular consequences of receptor inhibition. J Clin Oncol 20:110–24, 2002 59. Malik SN, Siu LL, Rowinsky EK, et  al: Pharmacodynamic evaluation of the epidermal growth factor receptor inhibitor OSI-774 in human epidermis of cancer patients. Clin Cancer Res 9:2478–86, 2003 60. Vanhoefer U, Tewes M, Rojo F, et al: Phase I study of the humanized antiepidermal growth factor receptor monoclonal antibody EMD72000 in patients with advanced solid tumors that express the epidermal growth factor receptor. J Clin Oncol 22:175–84, 2004 61. Folprecht G, Tabernero J, Kohne CH, et  al: Phase I pharmacokinetic/pharmacodynamic study of EKB-569, an irreversible inhibitor of the epidermal growth factor receptor tyrosine kinase, in combination with irinotecan, 5-fluorouracil, and leucovorin (FOLFIRI) in first-line treatment of patients with metastatic colorectal cancer. Clin Cancer Res 14:215–23, 2008 62. Alster TS, Tanzi EL: Hypertrophic scars and keloids: etiology and management. Am J Clin Dermatol 4:235–43, 2003 63. Akita RW, Sliwkowski MX: Preclinical studies with Erlotinib (Tarceva). Semin Oncol 30:15–24, 2003 64. Gho CG, Braun JE, Tilli CM, et al: Human follicular stem cells: their presence in plucked hair and follicular cell culture. Br J Dermatol 150:860–8, 2004 65. Moll I: Differential epithelial outgrowth of plucked and microdissected human hair follicles in explant culture. Arch Dermatol Res 288:604–10, 1996 66. Sperling LC: Hair anatomy for the clinician. J Am Acad Dermatol 25:1–17, 1991 67. Chamberlain AJ, Dawber RP: Methods of evaluating hair growth. Australas J Dermatol 44:10–8, 2003 68. Camidge DR, Randall KR, Foster JR, et  al: Plucked human hair as a tissue in which to assess pharmacodynamic end points during drug development studies. Br J Cancer 92:1837–41, 2005 69. Yap TA BD, Fong PC, et al: First in human phase I pharmacokinetic and pharmacodynamic study of KU-0059436 (Ku), a small molecule inhibitor of poly ADP-ribose polymerase (PARP) in cancer patients, including BRCA1/2 mutation carriers. Proceedings of the American Society of Clinical Oncology 2007: abstract 3529 70. Hirota M, Ito T, Okudela K, et al: Cell proliferation activity and the expression of cell cycle regulatory proteins in oral lichen planus. J Oral Pathol Med 31:204–12, 2002 71. Kurokawa H, Matsumoto S, Murata T, et  al: Immunohistochemical study of syndecan-1 down-regulation and the expression of p53 protein or Ki-67 antigen in oral leukoplakia with or without epithelial dysplasia. J Oral Pathol Med 32:513–21, 2003 72. Adjei AA, Mauer A, Bruzek L, et al: Phase II study of the farnesyl transferase inhibitor R115777 in patients with advanced non-small-cell lung cancer. J Clin Oncol 21:1760–6, 2003

254

D.R. Camidge et al.

73. Camidge DR, Pemberton MN, Growcott JW, et al: Assessing proliferation, cell-cycle arrest and apoptotic end points in human buccal punch biopsies for use as pharmacodynamic biomarkers in drug development. Br J Cancer 93:208–15, 2005 74. Camidge DR, Smethurst D, Growcott J, et al: A first-in-man phase I tolerability and pharmacokinetic study of the cyclin-dependent kinase-inhibitor AZD5438 in healthy male volunteers. Cancer Chemother Pharmacol 60:391–8, 2007 75. Camidge DR, Pemberton M, Growcott J, et  al: A phase I pharmacodynamic study of the effects of the cyclin-dependent kinase-inhibitor AZD5438 on cell cycle markers within the buccal mucosa, plucked scalp hairs and peripheral blood mononucleocytes of healthy male volunteers. Cancer Chemother Pharmacol 60:479–88, 2007 76. Jodrell DI, Egorin MJ, Canetta RM, et  al: Relationships between carboplatin exposure and tumor response and toxicity in patients with ovarian cancer. J Clin Oncol 10:520–8, 1992 77. Munzert G SS, Frost A, et al: A phase I study of two administration schedules of the Pololike kinase 1 inhibitor BI 2536 in patients with advanced solid tumors. Proceedings of the American Society of Clinical Oncology 2007: abstract 3069 78. Pallis AG, Agelaki S, Kakolyris S, et al: Chemotherapy-induced neutropenia as a prognostic factor in patients with advanced non-small cell lung cancer treated with front-line docetaxelgemcitabine chemotherapy. Lung Cancer 62(3):356–63, 2008 79. Dy GK, Bruzek LM, Croghan GA, et al: A phase I trial of the novel farnesyl protein transferase inhibitor, BMS-214662, in combination with paclitaxel and carboplatin in patients with advanced cancer. Clin Cancer Res 11:1877–83, 2005 80. Morris MJ, Tong WP, Cordon-Cardo C, et al: Phase I trial of BCL-2 antisense oligonucleotide (G3139) administered by continuous intravenous infusion in patients with advanced cancer. Clin Cancer Res 8:679–83, 2002 81. Adjei AA, Cohen RB, Franklin W, et al: Phase I pharmacokinetic and pharmacodynamic study of the oral, small-molecule mitogen-activated protein kinase kinase 1/2 inhibitor AZD6244 (ARRY-142886) in patients with advanced cancers. J Clin Oncol 26:2139–46, 2008 82. Peralba JM, DeGraffenried L, Friedrichs W, et al: Pharmacodynamic Evaluation of CCI-779, an Inhibitor of mTOR, in Cancer Patients. Clin Cancer Res 9:2887–92, 2003 83. Zhang H, Nimmer PM, Tahir SK, et al: Bcl-2 family proteins are essential for platelet survival. Cell Death Differ 14:943–51, 2007 84. Tse C, Shoemaker AR, Adickes J, et  al: ABT-263: a potent and orally bioavailable Bcl-2 family inhibitor. Cancer Res 68:3421–8, 2008 85. Roberts A GL, O’Connor OA, et al: Reduction in platelet counts as a mechanistic biomarker and guide for adaptive dose-escalation in phase I studies of the Bcl-2 family inhibitor ABT263. . ASCO Annual Meeting Proceedings 2008: abstract 3542 86. Hurwitz H, Fehrenbacher L, Novotny W, et al: Bevacizumab plus irinotecan, fluorouracil, and leucovorin for metastatic colorectal cancer. N Engl J Med 350:2335–42, 2004 87. Weidner N, Semple JP, Welch WR, et al: Tumor angiogenesis and metastasis – correlation in invasive breast carcinoma. N Engl J Med 324:1–8, 1991 88. Willett CG, Boucher Y, di Tomaso E, et al: Direct evidence that the VEGF-specific antibody bevacizumab has antivascular effects in human rectal cancer. Nat Med 10:145–7, 2004 89. Gordon MS, Margolin K, Talpaz M, et  al: Phase I safety and pharmacokinetic study of recombinant human anti-vascular endothelial growth factor in patients with advanced cancer. J Clin Oncol 19:843–50, 2001 90. Jayson GC, Mulatero C, Ranson M, et al: Phase I investigation of recombinant anti-human vascular endothelial growth factor antibody in patients with advanced cancer. Eur J Cancer 41:555–63, 2005 91. Loupakis F, Falcone A, Masi G, et al: Vascular endothelial growth factor levels in immunodepleted plasma of cancer patients as a possible pharmacodynamic marker for bevacizumab activity. J Clin Oncol 25:1816–8, 2007

9  Pharmacodynamic Studies in Early Phase Drug Development

255

92. Maurel J, Martin-Richard M, Conill C, et  al: Phase I trial of gefitinib with concurrent radiotherapy and fixed 2-h gemcitabine infusion, in locally advanced pancreatic cancer. Int J Radiat Oncol Biol Phys 66:1391–8, 2006 93. Hoekstra R, de Vos FY, Eskens FA, et al: Phase I safety, pharmacokinetic, and pharmacodynamic study of the thrombospondin-1-mimetic angiogenesis inhibitor ABT-510 in patients with advanced cancer. J Clin Oncol 23:5188–97, 2005 94. Rini BI, Michaelson MD, Rosenberg JE, et al: Antitumor activity and biomarker analysis of sunitinib in patients with bevacizumab-refractory metastatic renal cell carcinoma. J Clin Oncol 26:3743–8, 2008 95. Norden-Zfoni A, Desai J, Manola J, et al: Blood-based biomarkers of SU11248 activity and clinical outcome in patients with metastatic imatinib-resistant gastrointestinal stromal tumor. Clin Cancer Res 13:2643–50, 2007 96. Kendall RL, Thomas KA: Inhibition of vascular endothelial cell growth factor activity by an endogenously encoded soluble receptor. Proc Natl Acad Sci USA 90:10705–9, 1993 97. Beerepoot LV, Radema SA, Witteveen EO, et al: Phase I clinical evaluation of weekly administration of the novel vascular-targeting agent, ZD6126, in patients with solid tumors. J Clin Oncol 24:1491–8, 2006 98. Allard WJ, Matera J, Miller MC, et al: Tumor cells circulate in the peripheral blood of all major carcinomas but not in healthy subjects or patients with nonmalignant diseases. Clin Cancer Res 10:6897–904, 2004 99. Cristofanilli M, Budd GT, Ellis MJ, et al: Circulating tumor cells, disease progression, and survival in metastatic breast cancer. N Engl J Med 351:781–91, 2004 100. Olmos D, Arkenau HT, Ang JE, et al: Circulating tumour cell (CTC) counts as intermediate end points in castration-resistant prostate cancer (CRPC): a single-centre experience. Ann Oncol 20:27–33, 2009 101. Cohen SJ, Punt CJ, Iannotti N, et  al: Relationship of circulating tumor cells to tumor response, progression-free survival, and overall survival in patients with metastatic colorectal cancer. J Clin Oncol 26:3213–21, 2008 102. de Bono JS, Attard G, Adjei A, et  al: Potential applications for circulating tumor cells expressing the insulin-like growth factor-I receptor. Clin Cancer Res 13:3611–6, 2007 103. Shaffer DR, Leversha MA, Danila DC, et al: Circulating tumor cell analysis in patients with progressive castration-resistant prostate cancer. Clin Cancer Res 13:2023–9, 2007 104. Maheswaran S, Sequist LV, Nagrath S, et al: Detection of mutations in EGFR in circulating lung-cancer cells. N Engl J Med 359:366–77, 2008 105. Riethdorf S, Wikman H, Pantel K: Review: Biological relevance of disseminated tumor cells in cancer patients. Int J Cancer 123:1991–2006, 2008 106. Gautschi O, Heighway J, Mack PC, et  al: Aurora kinases as anticancer drug targets. Clin Cancer Res 14:1639–48, 2008 107. Agulnik M, Oza AM, Pond GR, et al: Impact and perceptions of mandatory tumor biopsies for correlative studies in clinical trials of novel anticancer agents. J Clin Oncol 24:4801–7, 2006 108. Rojo F, Tabernero J, Albanell J, et  al: Pharmacodynamic studies of gefitinib in tumor biopsy specimens from patients with advanced gastric carcinoma. J Clin Oncol 24:4309–16, 2006 109. O’Reilly KE, Rojo F, She QB, et al: mTOR inhibition induces upstream receptor tyrosine kinase signaling and activates Akt. Cancer Res 66:1500–8, 2006 110. Tan AR, Yang X, Hewitt SM, et al: Evaluation of biologic end points and pharmacokinetics in patients with metastatic breast cancer after treatment with erlotinib, an epidermal growth factor receptor tyrosine kinase inhibitor. J Clin Oncol 22:3080–90, 2004 111. Jimeno A, Rubio-Viqueira B, Amador ML, et al: Epidermal growth factor receptor dynamics influences response to epidermal growth factor receptor targeted agents. Cancer Res 65:3003–10, 2005

256

D.R. Camidge et al.

112. Zheng Z, Li X, Schell MJ, et al: Thymidylate synthase in situ protein expression and survival in stage I nonsmall-cell lung cancer. Cancer 112:2765–73, 2008 113. Lilja H, Ulmert D, Vickers AJ: Prostate-specific antigen and prostate cancer: prediction, detection and monitoring. Nat Rev Cancer 8:268–78, 2008 114. Wacker B, Nagrani T, Weinberg J, et al: Correlation between development of rash and efficacy in patients treated with the epidermal growth factor receptor tyrosine kinase inhibitor erlotinib in two large phase III studies. Clin Cancer Res 13:3913–21, 2007

Chapter 10

Prediction of Antitumor Response Fred R. Hirsch and Yu Shyr

10.1 Introduction A paradigm shift toward personalized medicine has occurred recently for many different types of cancer. Treatment of breast cancer, with individualized therapy introduced more than two decades ago, led the way. Today, the principle of personalized medicine applies for other major types of cancer, as well, including the number one cancer killer: lung cancer. Antitumor response may be predicted based on clinical parameters or through biomarker assessment. For patients with breast cancer, clinical factors have been used for many years to predict drug response and clinical outcome and to select patients for certain therapies. For example, a clinical factor of interest is menopausal status. In more recent years, selection of patients based on specific biomarkers has become more common, as specific targeted therapies have been developed. For example, with the development of antiestrogen therapy came the need to determine estrogen receptor status; with the development of anti-HER2 therapy (Herceptin®), the need to determine HER2 status. While biomarker status has been important for many years in malignant hematologic disorders as well as in breast cancer, today the principle is quickly entering clinical practice for several other solid tumor types. For example, we have seen that lung cancer patients with translocations of the ALK gene have a very high response rate (about 80%) to an ALK-fusion gene inhibitor [1]. Similarly, with the results of IPASS (Iressa versus Carboplatin/Paclitaxel Pan Asia Study), a large prospective study in which patients with epidermal growth factor receptor (EGFR) mutations fared significantly better with an EGFR tyrosine kinase inhibitor (e.g., gefitinib) than with chemotherapy, EGFR mutation testing has become a standard assessment for the determination of first-line therapy for patients with advanced nonsmall cell lung cancer (NSCLC) [2, 3]. F.R. Hirsch (*) Department of Medical Oncology, University of Colorado Cancer Center, Aurora, CO 80045, USA e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_10, © Springer Science+Business Media, LLC 2011

257

258

F.R. Hirsch and Y. Shyr

10.2 Tissue Collection Issues Moving toward individualized cancer therapy requires tissue collection. While, for many years, a diagnosis of malignant disease was sufficient for therapy decisions, today we must obtain sufficient tissues not only for histological diagnosis, but also for specific molecular profiling of the tumor. Thus, the patient and physician need to communicate in a different way to ensure the collection of an optimal tumor specimen. Tumor blocks (histology) are still the optimal specimen both for general histological diagnosis/subtyping and for molecular biomarker profiling. In many patients, however, particularly patients with advanced disease, the primary diagnosis is made by fine needle aspiration (cytology). Cytological specimens are suitable for some, but not all, molecular marker assessments. Thus, while clinicians and investigators continue to look for ways to assess biomarker status based on cytology and/ or peripheral blood, the acquisition of tumor tissue blocks should continue to have highest priority. Tissue procurement for biomarker assessment in multicenter studies can be a challenge. While some studies have achieved success rates of only 30–40% in obtaining sufficient tissue, other studies have been more successful, particularly after education at treatment sites. In a multicenter prospective phase II study performed by the University of Colorado and OSI Pharmaceuticals, in which advanced NSCLC patients were selected by EGFR biomarkers, the success rate for obtaining sufficient tumor tissue for biomarker studies was 80% of screened patients, with an average of 4 weekdays from the laboratory’s receiving tissue to the reporting of results [4]. Thus, with education, tissue collection for clinical trials can be successful. In routine clinical practice, collection of sufficient tissue might still be a challenge. With the increased tissue requirement for biomarker assessment, however, a change will inevitably be seen beyond clinical trials.

10.3 Selection of Appropriate Assay With the rapid development of biomarker-related therapies, an important question arises: which assay should be used for a particular biomarker assessment? Validation and standardization of assays are of crucial importance. The validation process, however, can be lengthy and complicated; in the meantime, new technologies continue to be developed. As discussed in above, biomarker-associated therapy in breast cancer has been a role model for the development of similar approaches for other solid tumors. Nevertheless, while HER2 determination was proposed and later validated in breast cancer more than a decade ago, controversy remains regarding the optimal methodology for HER2 determination [5]. Increased HER2 gene copy number detected by fluorescence in situ hybridization (FISH) has been a gold standard, but newer

10  Prediction of Antitumor Response

259

technologies have been introduced (e.g., silver in situ hybridization, SISH), and the role of immunohistochemistry (IHC) remains unclear [6–9]. For EGFR mutation screening, new technologies also have emerged in recent years. Traditional DNA sequencing technologies are laborious and time consuming; recently, more rapid and sensitive PCR-based mutation screening technologies (e.g., DXS Scorpion) have been developed. In addition, multigene detection technologies, such as Sequenom techniques, have been introduced [10, 11]. Likewise, we have seen technology developments for protein expression determination by IHC; more objective automated assessment methods (e.g., AQUA technology) are in clinical validation [12]. While technology develops very quickly, it remains important that the choice of biomarker assessment for therapeutic decisions be based on a thoroughly validated methodology with standardized parameters.

10.4 Centralized Laboratory Biomarker assay development and validation should be performed under standardized conditions. Thus, during the validation phase, assays should be performed by centralized laboratories with validated standard operating procedures (SOPs). When a predictive biomarker is ready for clinical practice, the predictive test must be performed in a laboratory certified under the Clinical Laboratory Improvement Amendments (CLIA) in the USA, or a similar certified laboratory elsewhere, if the treatment decision will be based on the predictive test. For some tests, a centralized laboratory with special expertise is preferable. Interpretation of test results from different platforms requires a high level of training, experience, and attention to detail. Essential to the maintenance of any diagnostic standard is the use of internal and external quality audit schemes. These provide a measure of individual laboratory performance and can direct selection of appropriate methodologies for diagnostic use. For example, for IHC, issues related to fixation, selection of different procedures for testing, and observer bias in the analysis pose key challenges to the wider application of the technology. Most studies comparing different IHC assay methodologies and quality controls have been performed for HER2 testing in breast cancer, and there is now considerable evidence that IHC performance is poorly controlled in the real world [5]. The ASCO-CAP guidelines draw attention to this with an alarming claim: “20% of HER2 assays performed in the field were incorrect” [13]. On the other hand, the latest external quality audits on FISH suggest that this technology is applied much more consistently. The United Kingdom National External Quality Assurance Scheme documents performance of diagnostic laboratories within the UK and across Europe and Asia, as well as including participants from the USA. While only 57–65% of participants using the DAKO HerceptTest for IHC determination of HER2 demonstrate acceptable performance, the rate of acceptable performance of FISH testing is consistently higher (89–96%) [5].

260

F.R. Hirsch and Y. Shyr

Studies evaluating inter-laboratory agreement for the HER2 FISH assay range in concordance from 92 to 99% [9]. Similarly, the College of American Pathologists (CAP) performed a proficiency testing survey for HER2 testing and found that 100% of the laboratories participating in the program correctly classified unknown samples using FISH [8]. Similar studies need to be undertaken for other biomarkers, as well.

10.5 Statistical Approaches to Quality Control Despite advances, the number of practical and clinically useful biomarkers for cancer diagnosis, prognosis, and disease prediction remains limited. Enthusiasm for high-dimensional data generated from high-throughput assays, including proteomic and genomic data, has diminished somewhat in recent years due to lack of reproducibility, procedural bias, and virtually no independent confirmation [14, 15]. Therefore, carefully designed experiments to assess repeatability, reproducibility, and quality control are essential for biomarker-based drug development. Repeatability refers to the consistency in measurements taken by a single person or instrument on the same sample (e.g., tissue or serum), under the same conditions. A measurement may be said to be repeatable when the variation is smaller than some agreed-upon limit. Reproducibility refers to the ability of an experiment to be precisely reproduced by another group or laboratory working independently, whether using the same samples or different samples from the same sample stream (e.g., same disease state). The primary objectives in performing a repeatability and reproducibility (R&R) study are to identify and quantify the absolute and relative contribution of each source of variation in data measurements, to ensure high-quality data. In quality control (QC), the goal is to reduce the variability in the system and, consequently, the variability in biomarker measurements. Several useful tools for assessing data quality, including coefficient of variation (CV), intraclass correlation coefficient (ICC), and variance component analysis are discussed in this section.

10.5.1 Coefficient of Variation For evaluating repeatability and reproducibility for biomarker-based drug development studies, the standard deviation has little interpretable meaning unless mean value is also reported. The coefficient of variation (CV) is a useful method to measure degree of variability in relation to mean value. CV is reported as a percentage and represented algebraically as:

s 100   %, x

10  Prediction of Antitumor Response

261

where s is the standard deviation and x is the mean value. CV is particularly useful when comparing datasets from different laboratories, dates of experimentation, operators, or machines. As an example of comparing two CVs, consider two datasets: dataset 1 consists of the integers 5, 6, 8, 10, 12, 14, and 15; dataset 2 consists of the integers 105, 106, 108, 110, 112, 114, and 115. Datasets 1 and 2 have the same sample standard deviation (s=3.87), but the sample mean is x = 10 for dataset 1 and x = 110 for dataset 2. The followings are the computations for CV:

 3.87  CVdataset1 = 100   % = 38.7,  10   3.87  CVdataset2 = 100   % = 3.52.  10  This tells us that the relative variation of the first dataset is approximately 11 times the relative variation of the second dataset, even though they have the same standard deviation. This quantity is useful for quality control purposes when data from several laboratories are compared, and it is also useful to compare within-laboratory variation from different experiment days. As a real-world example, Fig. 10.1a below (see Sect. 5.3) shows the CV for mass spectrometry peaks obtained on three different experiment days. Across the days, the mean CV is <5%, the largest CV is still <30%, and the CVs are consistent. These results indicate high-quality data, with low day-to-day variability.

10.5.2 Intraclass Correlation Coefficient and Variance Component Analysis Intraclass correlation coefficient (ICC) is used to assess the consistency between measures of the same variable (such as measures of the same gene or protein) under different conditions (such as different laboratories or experiment days). For example, in microarray experiments, we may use ICC to measure inter-lab/platform/operator reliability for two or more laboratories, platforms, or operators. ICC is represented algebraically as:

ICC =

2 σinterclass , 2 2 (σinterclass ) + σintraclass

2 2 where σinterclass is the interclass sample variance and σintraclass is the intraclass sample variance. Therefore, ICC measures the interclass reliability relative to the total variability of the ratios. For example, if the class is a technology platform, ICC gives the variance between platforms, divided by total variance. The interpretation

262

F.R. Hirsch and Y. Shyr

Fig. 10.1  Reproducibility analysis of the bevacizumab–erlotinib training dataset. (a) Coefficient of variation (CV) using 139 common peaks across all sera in the 3,000–20,000 m/z range. The results show low and comparable CV across all days, suggesting highly reproducible spectra. (b) Intraclass correlation coefficient (ICC), a measure of reliability, is 0.5192, suggesting that multiple measurements per patient are necessary to overcome the intra-class variation. (c) A summary of observed patient, day, and residual variance (a measure of variability distribution), for the 139 common peaks (reproduced from Salmon et al. [19])

of ICC is similar to that of kappa. ICC will approach 1.0 when there is no variance within class, indicating total variation in measurements is due solely to differences across the class (e.g., platform) variable. Alternatively, ICC may be used to measure the necessity of replication for the same tumor sample, if the intra-sample variation is large. Variance component analysis is an extension of ICC that allows researchers to study different sources of variation. It is most commonly used to determine the level at which variability is being introduced into an experiment. A typical biomarker-based drug development QC experiment might select several experimental conditions (e.g., days, machines, and operators), and then run replicate tests on each sample for different conditions. The goal is to determine the relative percentage of overall process variability that is being introduced at each level. Mixed or random

10  Prediction of Antitumor Response

263

effect model-based variance component analysis is the most useful tool to study sources of variance (e.g., day to day, machine to machine, operator to operator, patient to patient, disease status, and unknown factors).

10.5.3 Case Study Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF) has shown promise for biomarker discovery, potentially allowing the selection of patients who may benefit from specific therapies [16–19]. Investigators at Vanderbilt-Ingram Cancer Center used a MALDI-TOF proteomic algorithm developed from a phase I/II study of erlotinib–bevacizumab treated patients to predict the clinical outcome of patients treated with erlotinib alone [19]. Experimental variability and reproducibility (both analytical and biological) were first tested on 277 spectra generated from 37 available serum samples. To avoid systematic or procedural bias in the study, positions on the plates were randomized, and sample spectrometry was replicated a total of nine times on multiple days. CV was used as a measure of relative variability between expression levels of each peak or feature across samples and replicates. Variance component analysis was completed using a nested, random-effect model, testing the variability not only between patients, but also with the acquisition day factor nested within each patient. Intraclass correlation coefficient also was estimated to assess data quality. Figure 10.1 shows the distribution across peaks for CV, variance component analysis, and ICC. Despite the relatively small dataset, the low and consistent coefficient of variation among peaks and low day-to-day variance affirm the potential for reproducibility of proteomic approaches, when enough sample replicates (in this case, nine) are used to adjust for intrasample variability.

10.6 Statistical Design of Clinical Trials with a Predictive Marker Several biomarker-based trial designs have been introduced in the past several years; these include biomarker-adaptive threshold design [20] and adaptive signature design [21]. With many new designs available, selecting an optimal design is critical, to increase the likelihood of obtaining a meaningful result as well as reduce the cost of the study. Sample size estimation and power analysis are important issues when selecting a trial design. Sample size estimation for a traditional trial design and that for a biomarker-based design differ, due to the intra- and intersample variation seen in biomarker-based studies.

264

F.R. Hirsch and Y. Shyr

10.6.1 Biomarker-Adaptive Threshold Design Molecularly targeted anticancer agents benefit only a subset of treated patients. As a result, traditional randomized trials with broad eligibility may miss effective agents, due to dilution of the overall treatment effect. Biomarker-adaptive threshold design is a statistically rigorous phase III design for settings in which a putative biomarker to identify patients sensitive to the new agent is measured on a continuous or graded scale. The objective of the design is to determine whether (a) the experimental arm is better than the control arm for all randomly assigned patients, (b) the experimental arm is better than the control arm for a subset of patients defined by biomarker values greater than some value C0, or (c) the experimental arm is not better than the control arm. Two analysis procedures for the design were proposed by Jiang et al. [20]. In one procedure (procedure A), a test for treatment effect in all patients is conducted at a reduced significance level, a1. If the test is statistically significant, then for the purpose of formal valuation, the procedure is stopped and the null hypothesis of no treatment effect for randomized patients as a whole is rejected. Otherwise, an algorithm based on biomarker expression level is applied to test for treatment effect in a biomarker-defined subset of patients at a significance level of a2 = a−a1 (where a is typically 0.05, and a1 is the reduced significance level used to test for treatment effect in all patients, as defined above). Procedure B (which we do not discuss in detail here) is a generalization of procedure A, based on a more efficient approach to combining the overall and subset tests by incorporating the correlation structure of the test statistics between test statistics for testing the whole and a subset of the study population. See Jiang et al. [20] for more details of procedure B.

10.6.2 Adaptive Signature Design Genomic technologies, such as microarray analysis and single nucleotide polymorphism (SNP) genotyping, are powerful tools that hold great potential for identifying patients likely to benefit from a targeted agent. Due to the large number of genes available for analysis, however, interpretation of genomic data is complicated. Separation of reliable evidence from the random patterns inherent in highdimensional data requires specialized statistical methodology; such methodology is prospectively incorporated in the adaptive signature trial design [21]. This twostage trial design includes three components: (a) statistically valid identification, based on the first stage of the trial, of the subset of patients most likely to benefit from the new agent; (b) a properly powered test of overall treatment effect at the end of the trial using all randomized patients; and (c) a test of treatment effect for the subset identified in the first stage, but using only patients randomized in the remainder of the trial.

10  Prediction of Antitumor Response

265

In an adaptive signature study designed to accrue a total of N patients, N1 patients are accrued in stage 1, and the remaining N2 = N−N1 patients are accrued in stage 2. A key feature of the design is the development of a classifier that predicts whether a patient is more likely to benefit from the new treatment relative to the standard one. This classifier, developed using stage 1 patients only, is not used to restrict entry of patients during stage 2, but is prospectively applied to stage 2 patients to identify agent-sensitive patients. The final analysis consists of overall comparison of the two treatment arms using data from N = N1+N2 (a1), as well as comparison of the two arms in the selected subset of sensitive patients accrued during stage 2 (a2).

10.6.3 Power and Sample Size Analysis Power and sample size calculations for the classic two-arm parallel design with a continuous endpoint are usually based on a standard two-group comparison with Gaussian errors. Sample size estimation is represented algebraically as:

n=2

( z1−α / 2 + z1−β ) 2 δ2

,

where n is the number of subjects (e.g., patients and specimens) required for each group, z denotes the quantile from a Gaussian distribution, a is the significance level (often 0.05), and 1−b is power. The effect size, d, is defined as:

δ=

(µ 2 − µ1 ) , σ total

where m2 is the mean value of the endpoint for group 2, m1 is the mean for group 1, and mtotal is the pooled standard deviation across both groups. In studies with multiple observations per subject (subsamples), within-group variance (mtotal) can be decomposed into two parts: inter-subject (e.g., inter-patient and inter-specimen) variation and intra-subject variation. The within-group variance is represented algebraically as: 2 2 σ2total = σinter + σintra .

If m subsamples are acquired for each specimen, the effective total variance for each subject will be reduced and is described by:

σ2total =

2 2 σinter + σintra . m

Table  10.1 shows requirements of subject number (n) for selected values of subsample (1, 5, and 20), intra-subject variance (0.2, 0.5, and 1.0), and inter-subject

266

F.R. Hirsch and Y. Shyr Table 10.1  Required number of subjects (n) for selected values of subsample (m), intra-subject variance, and inter-subject variance Intra-subject variance 0.2 0.5 1.0 Inter-subject Inter-subject Inter-subject variance variance variance Subsample 0.2 0.5 1.0 0.2 0.5 1.0 0.2 0.5 1.0 Number (m) 1 5 20

6 4 4

11 9 8

19 17 16

11 5 4

16 10 9

24 17 16

19 7 4

24 11 9

32 19 17

variance (0.2, 0.5, and 1.0). Each calculation assumes a mean difference of 1  U, 80% power, and a 0.05 two-sided significance level. For each pair of variance component values, multiple combinations of subject number (n) and subsample number (m) satisfy these power and significance level requirements. For example, when inter-subject variance and intra-subject variance each equal to 0.5, a 1-U shift in mean could be detected using (a) n=16 and m=1, (b) n=10 and m=5, or (c) n=9 and m=20, all with 80% power and two-sided 0.05 significance level. With this type of information regarding the magnitude of variance components, experimental sampling plans may be designed to control the effects of inter- and intra-subject variations. For example, when intra-subject variability contributes substantially to total variation, a reduction in required sample size (e.g., number of patients) can be achieved by subsampling. In the case study discussed in Sect. 5.3 [19], the median ICC is 0.5, indicating that one-half of the total variation is attributable to interpatient sources and one-half to intra-patient sources. For an ICC of 0.5, the number of subjects is substantially reduced if subsampling is increased from 1 to 5 spectra per subject; however, further increasing to 20 spectra per subject results in little further reduction in n (see Table  10.1). In settings where the intra-subject variability is large and ICC is small, further reductions can be realized by additional subsampling. Because different choices of resource allocation to additional subjects vs. additional subsamples can achieve the same statistical power, efficient experimental design relies on balancing the complementary components of variance. In situations in which the cost of subject acquisition is expensive and the cost of incremental subsampling is cheap (e.g., microspotting), then the effects of within-group variance can be reduced by increasing the number of subsamples. Alternatively, if the cost of acquiring additional subsamples is time consuming or expensive, then more subjects may be required to achieve adequate statistical power. If estimates of inter- and intrasubject variances are available (e.g., from preliminary data), the usual sample size calculations can be easily adapted to accommodate subsampling.

10  Prediction of Antitumor Response

267

10.7 Analysis and Reporting of Studies with Predictive Markers 10.7.1 Class Prediction Class prediction methods can be used to examine the goodness-of-fit of a set of features identified through class comparison. There are two types of class prediction methods: those based on the “training” dataset and those based on the “test” dataset. For high-throughput data analysis, it is highly recommended that class prediction models be validated on test data, because they may easily overfit the training dataset, yielding optimistic estimates of accuracy. In addition, the sample size of the blinded/test dataset probably should be comparable to the sample size of the training dataset, if the sample size of the training dataset is very small. On the other hand, if the sample size of the training dataset is large, such as several hundred samples, it may be appropriate to use a smaller sample size for the testing dataset.

10.7.2 Compound Covariate Method Hedenfalk et al. [22] successfully applied the compound covariate method (CCM) [23] to class prediction analysis for BRCA1+ vs. BRCA1− breast cancer tumor types. A CCM-based predictor is built into two steps. First, a standard two-sample t test is performed to identify genes with significant differences (at level a; Hedenfalk et  al. used a = 0.0001) in log-expression ratios between the two tissue classes. Second, the log-expression ratios of differentially expressed genes are combined into a single compound covariate for each tissue sample; the compound covariate is used as the basis for class prediction. The compound covariate for tissue sample i is defined as

ci = ∑ j t j xij , where tj is the t-statistic for the two group comparison of classes with respect to gene j, xij is the log-expression ratio measured in tissue sample i for gene j, and the sum is over all differentially expressed genes. CCM reduces the data dimension from N × J to N × 1, where N is the total number of samples and J is the total number of genes. CCM may be viewed as the overall score of each tissue sample, which combines information for all important features from one statistical method.

268

F.R. Hirsch and Y. Shyr

10.7.3 Weighted Flexible Compound Covariate Method The weighted flexible compound covariate method (WFCCM) [16, 24, 25] is an extension of the CCM; where the CCM uses the t test only to identify genes with significant differential expression, WFCCM allows for the use of multiple statistical methods to select “winner” genes. Before WFCCM is applied, scores derived from different statistical methods should be given consistent signs. For example, signs for the t-statistic and significance analysis of microarrays (SAM) are always consistent, but weighted gene analysis (WGA) scores are always positive because the scoring system is based upon Euclidean distance. Therefore, WGA scores first must be multiplied by (−1) for all features that have negative SAM or t-statistic scores, when applying WFCCM using these three methods. The second step is to select winners based on all statistical methods. We may arbitrarily pick genes from each statistical method, such as the top 1% or top 100 features, or we may use some significance information to select features from the statistical methods, such as p value < 0.0001 for t-statistic, p value < 0.01 for REML-based mixed-effect model, or SAM > 3.5. In addition, we may use false discovery rate (FDR) or local FDR (fdr) as the cut-off for selecting winners. The WFCCM for tissue sample i is defined as

  WFCCM (i ) = ∑  ∑ (ST jkWk ) W j  xij , j  k  where xij is the log-expression ratio measured in tissue sample i for feature j; STjk is the standardized statistic/score of feature j, such as the standardized SAM score, for statistical analysis method k; and Wk is the weight of method k, which can be determined as

Wk = (1 − CCM misclassification ratek ), where CCM misclassification ratek is the misclassification rate of the CCM for statistical analysis method k for k = 1, …, K. Wj is the weight of feature j, which can be determined as

Wj = ∑ k

V jk K

,

where Vjk = 1, if the feature j is selected as a winner in method k, and Vjk = 0 otherwise. If feature j is selected by all methods, then Wj = 1. Wk and Wj also can be determined by other methods. For example, we may assign Wk = 1 for all K methods used in the variable selection stage if we believe that they perform equally well. We may also modify Wj as

  V jk  W j =  ∑ k  (1 − Info Score j ) . K   

10  Prediction of Antitumor Response

269

In this case, if feature j is selected by all methods and the Info Scorej = 0, then Wj = 1. WFCCM also reduces the data dimension from N × J to N × 1. WFCCM certainly may be viewed as the overall score of each tissue sample, which combines all information for all important features from several statistical methods.

10.7.4 Random Forest Algorithm and Neural Networks Random forest [26] is an algorithm used to select predictive biomarkers and estimate model prediction error. The random forest algorithm builds multiple classification trees by using bootstrap sampling of patient cases with random subsets of input variables as training sets and then estimates the prediction error by using the excluded subjects as test sets. Simultaneously, the relative importance of input variables is assessed by decreases in prediction accuracy and Gini distance (the probability that a pair of samples will be classified into the same leaf node) while permuting each input variable. Neural networks allow the building of regression models that predict a discrete outcome such as membership to a particular group based on a set of independent variables [27]. This supervised classification approach is based on a set of neurons or computing units that each performs a simple calculation using information from independent variables with the goal of collectively producing an output in the form of a prediction or classification. Feed-forward neural networks are generally used in classification and pattern recognition. These networks consist of computing units with one-way connections to other computing units. Computing inputs are organized into layers with connections from one layer to another. A neural network method for modeling the regression relationship between multiple-variable biomarker expression data and trial endpoints may be applied to data analysis.

10.7.5 Leave-One-Out Cross-Validated Class Prediction Model Cross-validation is a method for estimating generalization error based on resampling. Many classification methodologies do not have straightforward procedures for testing null vs. alternative hypotheses using p values or likelihood ratio tests. The primary measure of success for these methods is the error or misclassification rate [27]. Because many classification methodologies are susceptible to overfitting the data, it is common to base success on the predictive ability of models. This is accomplished through cross-validation. For example, in a procedure known as tenfold cross-validation, the data are divided into ten equal parts, and the model is developed using 9/10 of the data and tested for its ability to predict the remaining 1/10 of the data. This is repeated a total of ten times with ten training sets and ten test sets. A model that has a low misclassification rate (<5% per group) for prediction in 80% of the test sets may be considered statistically significant. When

270

F.R. Hirsch and Y. Shyr

sample size is relatively small, misclassification rate can be assessed using the leave-one-out cross-validation (LOOCV) class prediction method, a specific type of cross-validation. For example, using a Vanderbilt lung cancer SPORE matrixassisted laser desorption/ionization MS proteomic study [16], LOOCV was performed as follows: 1 . One tissue sample was selected and removed from the dataset. 2. WFCCM was applied to calculate the single compound covariate for each remaining tissue sample based on significant features, and the distance between the two tissue classes (e.g., tumor vs. normal) for the remaining tissue samples was calculated. 3. The removed tissue sample was classified based on the closeness of the distance of the two tissue classes, using, for example, the k-nearest neighbor approach, with k=2, or the midpoint of the means of the WFCCM for the two classes, as the threshold. 4. Steps 1 and 3 were repeated for each tissue sample. To determine whether the accuracy of predicting class membership of tissue samples, as measured by the number of correct classifications, was better than the accuracy that could be attained by random grouping of the tissue samples, 5,000 random datasets were created by permuting class labels among the tissue samples. Cross-validated class prediction was performed on the resulting datasets, and the percentage of permutations that resulted in as few or fewer misclassifications as for the original labeling of samples was reported. If less than 5% of permutations result in as few or fewer misclassifications, the accuracy of prediction is considered significant; therefore, this rate may be considered the “p value” for class prediction models. In the training dataset, WFCCM class-prediction models based on (a) 82, (b) 91, (c) 23, and (d) 20 differentially expressed MS peaks were found to classify (a) lung tumor vs. normal lung tissue, (b) primary NSCLC lung cancer vs. normal lung tissue, (c) primary NSCLC lung cancer vs. other type of lung tumor tissue, and “(d) adenocarcinoma vs. squamous cell carcinoma tissue, respectively. Table  10.2 shows the results. WGA, SAM, the Kruskal–Wallis test, Fisher’s exact test, and permutation t test were included in the compound covariate using cut-off points of 2.0, 3.5, p < 0.0001, p < 0.0001, and p < 0.0001, respectively. Wk = 1 was selected for all methods. WFCCM also was applied to a set of blinded test samples. Table 10.2 shows the results of the analyses. In general, the model performed reasonably well, with the average correct prediction rate in the blinded test dataset exceeding 93% except for the prediction of negative vs. positive nodal metastasis. The latter case is not surprising as only two differentially expressed MS peaks were selected for this class comparison. The lower bounds of the 95% confidence interval of the mean percentage of the correctly classified samples were all above 50%; therefore, the correct prediction rates in the blinded test datasets were not likely the result of chance. Figure 10.2 shows the results from a Vanderbilt lung cancer SPORE microarray study [25]. An agglomerative hierarchical clustering algorithm was applied to cluster adenocarcinomas and squamous cell carcinomas. The average linkage algorithm

20 20 12  2

Non-small cell lung cancer Adeno (13) vs. squamous (16) Adeno (13) vs. large (3) Squamous (16) vs. large (3) Nodal metastasis negative (25) vs. positive (7)

0 1 0 8

100 (88, 100)   93.8 (70, 99) 100 (82, 100)   75.0 (57, 88)

100 (92, 100) 100 (91, 100) 100 (91, 100)

0 0 0

82 91 23

<0.0001 <0.01 <0.01 <0.05 Percentage of correctly classified samples(95% CI)

0 (0) 0 (0) 0 (0) 5 (14.7)

<0.0001 <0.001 <0.001

No. of diff. expressed peaks No. of misclassified samples

20 20 12  2

Non-small cell lung cancer Adeno (14) vs. squamous (15) Adeno (14) vs. large (5) Squamous (15) vs. large (5) Nodal involvement negative (20) vs. positive (14)

0 (0) 0 (0) 0 (0)

Test dataset All samples Normal lung (6) vs. lung tumor (37) Normal lung (6) vs. primary NSCLC (32) Primary NSCLC (32) vs. other type of lung tumor (5)

82 91 23

Training dataset All samples Normal lung (8) vs. tumor (42) Normal lung (8) vs. primary NSCLC (34) Primary NSCLC (34) vs. other type lung tumor (7)

Prob. of random permutations No. of diff. expressed peaks No. of misclassified samples (%) with misclassifications

Table 10.2  WFCCM class prediction model in training and test datasets Classification (sample size)

10  Prediction of Antitumor Response 271

272

F.R. Hirsch and Y. Shyr

Fig. 10.2  Results of agglomerative hierarchical clustering of adenocarcinomas and squamous cell carcinomas

was applied to calculate the distance between clusters. All adenocarcinoma tissues clustered together, as did squamous cell tissues (see Fig.  10.2). The results look very promising, but these results may be used only to reconfirm that the selected genes perform differently in the two classes. Having a perfect or near-perfect cluster result was expected, given that cluster analysis was applied after selecting genes that performed differently using a supervised method. Thus, these results should not be applied to draw any class discovery conclusion.

10.8 Conclusions A variety of biomarkers may be associated with treatment outcome; though a number of such biomarkers have been identified, challenges remain in predicting antitumor response. For example, some biomarkers are heterogeneously distributed within a tumor. Due to this heterogeneous distribution, association with treatment outcome may be found to be weak or nonexistent. To address this issue, it is important that

10  Prediction of Antitumor Response

273

marker heterogeneity within and between individuals be estimated and used in the process of designing an appropriate study of association between biomarker and outcome [28], for example, through the use of subsampling, as discussed in Sect. 6.3. Another issue of concern is variation in biomarker expression based on ethnic differences, which might have a large impact on the sensitivity of a specific targeted drug from one region to another. A typical example is the difference in EGFR mutations between Caucasian and Asian populations. While only about 10% of Caucasians have these activating mutations, the mutations are found in 40–60% of NSCLC patients in Asian populations [2]. Similar studies comparing other ethnicities (i.e., African-American, Indian, etc.) will be important to study the association with drug response and outcome. A final issue is the question of whether primary tumors and metastases differ in biomarker expression. If so, overall response to a specific targeted agent may differ when studied in local-regional disease versus advanced disease. Clonal selection in the metastatic process is a possibility; indeed, for some biomarkers, a remarkable difference in expression has been reported when comparing primary tumors and metastases [29]. In summary, much progress has been made in the field of personalized medicine; nevertheless, challenges remain, and additional research will be required to address the many questions that remain unanswered. Acknowledgments  The authors wish to acknowledge Lynne Berry of the Cancer Biostatistics Center, Vanderbilt-Ingram Cancer Center (VICC), for her editorial suggestions. Y.S. wishes to acknowledge the support of the VICC Cancer Center Support Grant (5P30 CA068485-13).

References 1. Kwak EL, Camidge DR, Clark J, et al. Clinical activity observed in a phase I dose escalation trial of an oral c-met and ALK inhibitor, PF-02341066. J Clin Oncol 2009;27:148S(#3509). 2. Mok TS, Wu YL, Thongprasert S, et  al. Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma. N Engl J Med 2009;361(10):947–957. 3. Hirsch FR, Bunn PA, Jr. EGFR testing in lung cancer is ready for prime time. Lancet Oncol 2009;5:432–433. 4. Hirsch FR, Dziadziuszko R, Varella-Garcia M, et al. Randomized phase II study of erlotinib (E) or intercalated E with carboplatin/paclitaxel in chemonaive advanced NSCLC. Correlation of biomarker status and clinical benefit. J Clin Oncol 2009;27:413S(#8026). 5. Sauter G, Lee J, Barlett JMS, Slamon DJ, Press MF. Guidelines for human epidermal growth factor receptor 2 testing: biological and methodologic considerations. J Clin Oncol 2009;27:1323–1333. 6. Francis GD, Jones MA, Beadle GF, Stein SR. Bright-field in situ hybridization for HER2 gene amplification in breast cancer using tissue microarrays: correlation between chromogenic (CISH) and automated silver-enhanced (SISH) methods with patient outcome. Diagn Mol Pathol 2009;2:88–95. 7. Ellis IO, Bartlett J, Dowsett M, et al. Updated recommendations for HER2 testing in the UK. J Clin Pathol 2004;57:233–237. 8. Nagle RB, Tubbs RR, Roche PC, et al. Clinical laboratory assays for Her-2/neu amplification and overexpression: quality assurance, standardization, and proficiency testing. Arch Pathol Lab Med 2003;126:803–808.

274

F.R. Hirsch and Y. Shyr

9. Paik S, Bryant J, Tan-Chiu E, et al. Real-world performance of HER2 testing: National Surgical Adjuvant Breast and Bowel Project experience. J Natl Cancer Inst 2002;94:852–854. 10. Jimeno A, Messersmith WA, Hirsch FR, Franklin WA, Eckhardt SG. KRAS mutations and sensitivity to epidermal growth factor receptor inhibitors in colorectal cancer: practical application of patient selection. J Clin Oncol 2009;27:1130–1136. 11. Gabriel S, Ziaugra L, Tabbaa D. SNP genotyping using the Sequenom MassARRAY iPLEX platform. Curr Protoc Hum Genet 2009;60:2.12.1–2.12.18. 12. Moeder CB, Giltnane JM, Moulis SP, Rimm DL. Quantitative, fluorescence-based in-situ assessment of protein expression. Methods Mol Biol 2009;520:163–175. 13. Wolff A, Hammond M, Schwartz J, et al: American Society of Clinical Oncology/College of American Pathologists guideline recommendations for human epidermal growth factor receptor 2 testing in breast cancer. J Clin Oncol 2007;25:118–145. 14. Diamandis EP. Analysis of serum proteomic patterns for early cancer diagnosis: drawing attention to potential problems. J Natl Cancer Inst 2004;96:353–356. 15. Baggerly KA, Morris JS, Edmonson SR, Coombes KR. Signal in noise: evaluating reported reproducibility of serum proteomic tests for ovarian cancer. J Natl Cancer Inst 2005;97:307–309. 16. Yanagisawa K, Shyr Y, Xu BJ, et  al. Tumor proteomic patterns predict classification and tumor behavior in human non-small cell lung cancer. Lancet 2003;362(9382):433–439. 17. Schwartz SA, Weil RJ, Thompson RC, et  al. Proteomic-based prognosis of brain tumor patients using direct-tissue matrix-assisted laser desorption ionization mass spectrometry. Cancer Res 2005;65(17):7674–7681. 18. Yildiz PB, Shyr Y, Rahman JS, et  al. Diagnostic accuracy of MALDI mass spectrometry analysis of unfractionated serum in lung cancer. J Thorac Oncol 2007;2(10):893–901. 19. Salmon S, Chen H, Chen S, et  al. Classification by mass spectrometry can accurately and reliably predict outcome in patients with non-small cell lung cancer treated with erlotinibcontaining regimen. J Thorac Onol 2009;4(6):689–696. 20. Jiang W, Freidlin B, Simon R. Biomarker-adaptive threshold design: a procedure for evaluating treatment with possible biomarker-defined subsets. J Natl Cancer Inst 2007;99:1036–1043. 21. Freidlin B, Simon R. Adaptive signature design: an adaptive clinical trial design for generating and prospectively testing a gene expression signature for sensitive patients. Clin Cancer Res 2005;11(21):7872–7878. 22. Hedenfalk I, Duggan D, Chen Y, et al. Gene-expression profiles in hereditary breast cancer. N Engl J Med 2001;344(8):539–548. 23. Tukey JW. Tightening the clinical trial. Control Clin Trials 1999;14(4):266–285. 24. Shyr Y. Statistical strategies for analyzing the microarray data in human lung cancer. Lung Cancer 2003;41(2003):90–91. 25. Yamagata N, Shyr Y, Yanagisawa K, et al. A training-test approach to the molecular classification of resected non-small cell lung cancer. Clin Cancer Res 2003;9(13):4695–4704. 26. Breiman L. Random Forests. Machine Learning 2001;45(1):5–32. 27. Ripley BD. Pattern Recognition and Neural Networks. Cambridge, UK: Cambridge University Press, 1996. 28. Pintilie M, Iakovlev V, Fyles A, et al. Heterogeneity and power in clinical biomarker studies. J Clin Oncol 2009;27:1517–1521. 29. Daniele L, Cassoni P, Basillo E, et al. Epidermal growth factor receptor gene in primary tumor and metastatic sites from non-small cell lung cancer. J Thorac Oncol 2009;4(6):684–688.

Chapter 11

Imaging Studies in Anticancer Drug Development David A. Mankoff

11.1 Introduction 11.1.1 Overview of Imaging and Cancer Therapy The ability to assay tumor biologic features and the impact of drugs on tumor biology is fundamental to drug development. Advances in our ability to measure genomics, gene expression, protein expression, and cellular biology have led to a host of new targets for anticancer drug therapy. In translating new drugs into clinical trials and clinical practice, these same assays serve to identify patients most likely to benefit from specific anticancer treatments. As cancer therapy becomes more individualized and targeted, there is an increasing need to characterize tumors and identify therapeutic targets to select therapy most likely to be successful in treating the individual patient’s cancer. An example is the identification of HER2 overexpression to predict response to HER2-directed therapies such as trastuzumab and lapatinib [1]. There is a complementary need to assay cancer drug pharmacodynamics, namely the effect of a particular drug on the tumor, to determine whether or not the drug has “hit” the target and whether the drug is likely to be effective in slowing tumor growth and killing the cancer [2]. This is particular important in early drug trials as proof of mechanism and prediction of the likelihood of anticancer activity in patients. Thus far assays to identify cancer therapeutic targets or anticancer drug pharmacodynamics have been based upon in  vitro assay of tissue or blood samples. Advances in both technology and cancer science have led to the ability to perform noninvasive molecular assays. An example is the use of reporter genes whose expression results in the production of material such as green fluorescent protein or luciferase that can be detected without tissue sampling [3]. Another advance, applicable D.A. Mankoff (*) Department of Radiology, Seattle Cancer Care Alliance, G2-600, 825 Eastlake Avenue East, Seattle, WA 98109, USA e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_11, © Springer Science+Business Media, LLC 2011

275

276

D.A. Mankoff

to the entire range of biological systems from cell culture to humans, is functional and molecular imaging [4, 5], which is the focus of this review. Imaging has traditionally relied on structural and anatomic features to detect cancer and determine its extent [6]. This traditional form of imaging, often termed anatomic imaging, has made an important contribution to cancer care, and is widely used in the detection and staging of cancer patients using methods such as X-ray mammography and computed tomography (CT) [7]. More recently, imaging has expanded to include the ability to image regional biochemistry and molecular biology, often termed molecular imaging [4]. The focus for molecular imaging is not structure, but rather regional biology. Quantitative analysis is an important feature of this type of imaging, for example, the ability to measure regional tumor receptor expression [8, 9]. As such, molecular imaging can be considered an in vivo assay technique, capable of measuring regional tumor biology without perturbing it. This makes molecular imaging a unique tool for anticancer drug development, complementary to traditional assay methods.

11.1.2 Differences Between Imaging and Tissue/Blood Assays It is important to keep in perspective inherent differences in capabilities between tissue-based assays and in vivo assays using molecular imaging. Imaging is noninvasive and therefore better suited to serial assay. This is especially important in imaging specific drug pharmacodynamics and early tumor response. In addition, imaging typically surveys the entire animal or patient and therefore avoids sampling error that can occur for assays that require tissue sampling, especially when there is significant tumor heterogeneity. However, while sample-based methods can assay many different processes at once, for example, the expression of an array of genes [10], imaging can typically sample at most a few processes at the same time. Also, while it is possible to “batch” processing for many samples at the same time, imaging needs to be performed one subject at a time. Furthermore, the need for sophisticated equipment and imaging probes makes imaging typically more expensive than sample-based assays. These last two factors limit the number of subjects that can be studied by imaging compared with sample assay. In general, imaging methods are complementary to assay-based methods and best used to test novel drugs in the later stages of preclinical testing and early clinical trials, with more focused and limited use of imaging in laterstage drug trials.

11.1.3 Appropriate Roles It is also important to note that different approaches to imaging may be needed at different stages of drug testing. In early drug trials, it is important to gain knowledge about the mechanism of action and potential efficacy of the new drug. It may also be important to test imaging methods of response assessment for future use as endpoints in

11  Imaging Studies in Anticancer Drug Development

277

larger drug trials. Novel drugs may affect imaging probe metabolism and biodistribution in unexpected ways, confounding the interpretation of the imaging study. The differing approach to imaging for different trials is discussed in more detail below. 11.1.3.1 Early Drug Trials (Phase I/Early Phase II) In this setting, the imaging is not used to establish a clinical endpoint such as response or survival but rather it is to select patients appropriate for a particular targeted therapy and to confirm an effect of the drug on the tumor in patients. This setting will differ from late phase trials in the need for more detailed, likely kinetic, analysis of imaging studies to evaluate drug mechanisms and drug effect on the imaging probe in early evaluation of the drug in patients. This detailed imaging approach is possible in early trials, where patient numbers are smaller and trials occur in a limited number of centers, as compared to larger clinical trials or in a routine clinical setting, which involve more patients in a much larger number of centers. More complex analysis is often needed in early drug trials: (1) to obtain more detailed information of drug affect on tumor and normal tissues and (2) since the drug may affect imaging probe clearance and biodistribution and lead to misleading results for simple static uptake measures such as standardized uptake value (SUV) in positron emission tomography (PET) [11, 12]. In early drug trials, patient numbers will be small, and it may be feasible to use less widely available imaging probes, for example, the short-lived, but very useful, 11C (half-life = 20 min) for PET. 11C allows serial PET studies in the same imaging session and may therefore be very useful to measure tissue properties, such as drug transport, before and after administration of the drug being tested. This approach was recently used to study drug transport across the blood–brain barrier and the effect of P-glycoprotein transport inhibition [13]. 11.1.3.2 Late Phase II/III Drug Trials In later drug trials, molecular imaging may be particularly helpful as an early indicator of, or even potential surrogate endpoint for, drug response. Current clinical trials rely on tumor size as the endpoint for response evaluation [14]; however, it is likely that in  vivo molecular imaging measures will provide equally if not more helpful predictive endpoints. This may be especially true for targeted therapies, which are often cytostatic rather than cytotoxic, and therefore result in no appreciable change in tumor mass or size. For these trials, imaging approaches and probes must be widely available. For PET imaging in larger clinical trials, labeling with 18F, which can be shipped from regional cyclotrons, or possibly longer-lived positron emitters such as 64Cu or 124I, which can be shipped and even used for onsite chemistry, will most likely be needed. In addition, the need for shorter imaging times and robust data analysis will likely favor simple and/or static imaging measures versus the more complex dynamic acquisition and kinetic modeling possible in earlier trials. In phase II/III trials, since both drug and imaging application are

278

D.A. Mankoff

undergoing prospective clinical testing, it can be quite confusing to try to combine new treatment and new imaging treatment evaluation in a single clinical trial. The molecular imaging study will often need to be tested first as an exploratory endpoint or as a correlative measure in the drug therapy trial before it can be considered as an endpoint itself [6].

11.1.3.3 Clinical Drug Therapy Because it can quantify in vivo tumor biology changes over time, molecular imaging is likely to be very helpful to guide clinical drug therapy in established drugs. Here, the clinical trial focuses upon the imaging study itself, and not the therapy drug. The drug should have been previously tested with established indications for its use in cancer therapy. Well-designed studies should evaluate the ability of the imaging to guide the particular cancer therapy and ideally to improve patient outcome, such as survival. A particularly attractive design for this type of study is to randomize patients to use or not use the imaging study to guide patient treatment in order to test whether the use of the imaging improves outcome [15]. Here again, as in phase II/III trials, widely available imaging probes and simpler image acquisition and analysis schemes are likely to be favored.

11.2 Novel Imaging Methods for Drug Development: Overview of Imaging Modalities The imaging modalities most commonly used in molecular imaging are listed in Table 11.1. Common among these modalities is the ability to image functional and molecular tissue properties such as perfusion, metabolism, and receptor or oncogene expression [8, 16–23]. This section provides a brief description of each modality, along with its advantages and disadvantages.

11.2.1 Magnetic Resonance Imaging Magnetic resonance (MR) relies upon the interaction of atomic nuclei with radiofrequency signals in the presence of strong magnetic fields. MR imaging (MRI) offers high spatial resolution and functional contrast agents using magnetic elements such as Gd and Fe [18, 24]. MRI using nonspecific contrast agents such as Gd-DTPA has become an important part of clinical cancer care [25]. More detailed and quantitative approaches to dynamic contrast-enhanced (DCE) MRI have been increasingly used to examine tumor perfusion and capillary permeability as an indicator of tumor angiogenesis [26–29] and as a measure of response to antivascular therapy [30, 31].

11  Imaging Studies in Anticancer Drug Development Table 11.1  Functional and molecular imaging modalities Modality Advantages Magnetic resonance Magnetic resonance imaging High spatial resolution and (MRI) image detail Modest temporal resolution Magnetic resonance spectroscopy (MRS)

Radionuclide imaging Positron emission tomography (PET)

Single-photon emission tomography (SPECT)

Ultrasound, especially with contrast enhancement Optical imaging

Measures wide range of molecules Measures native molecules, no contrast needed Wide range of molecular imaging probes Imaging without perturbing biologic system Similar to PET Imaging probes more widely available Highly portable, inexpensive Molecular microbubble contrasts agents possible Highly portable, inexpensive High spatial resolution possible

279

Disadvantages Confined space Contrast designed limited by the need for magnetic atom Limited spatial resolution Challenging to get high-quality spectra in routine imaging Limited spatial resolution Some radiation exposure More limited range of molecular probes Less quantitatively accurate than PET Operator dependence Contrast agents confined to vascular space thus far Limited penetration from surface, limited to relatively superficial sites

More specific and targeted MRI agents have also been developed and undergone preliminary testing [32]; however, the range of possible molecular targets is somewhat constrained by the need to include a magnetically active atom such as Gd or Fe. Recent advances in pulse sequences and image acquisition have led to the ability to measure other tissue properties, such as water diffusion, which can provide information on cellularity and interstitial transport without the need for contrast [33, 34], and may provide an early indication of therapeutic efficacy [35, 36]. An advantage of MRI is its high spatial resolution and image quality, especially with increasing magnetic field strength, making it applicable to both small animals and patient imaging. Limitations include the cost of the imaging system and the thus far somewhat limited range of imaging probes that serve as MR contrast agents, although new approaches tested in preclinical models will provide increased capabilities for animal research and may be able to be translated to patient studies [32].

11.2.2 Magnetic Resonance Spectroscopy MR spectroscopy (MRS) takes advantage of the ability of nuclear magnetic ­resonance to identify specific chemical signatures and measures the regional

280

D.A. Mankoff

c­ oncentrations of biochemical species, using methods similar to those developed for basic chemical assays [17, 21, 37]. Much of the current work in patients uses hydrogen spectroscopy; however, spectroscopy for other biologically relevant nuclei such as phosphorus or sodium is also possible [21]. MRS can quantify the concentration of prevalent biochemical species without perturbing the system being imaged and without the need for imaging contrast administration. MRS has considerably more limited spatial resolution compared with MRI; however, recent advances in magnetic field strength and MRS technology have yielded the ability to generate 3D MRS concentration maps (MRS images, MRSI) with resolution on the order of 1  cm or less [17, 37, 38]. Recent studies suggest that changes in local metabolites with therapy may provide a very early indicator of cancer response [39, 40]. MRS has the advantage of being able to directly quantify molecular species without the need for contrast, with the disadvantages of more limited spatial resolution and the need for relatively high abundance to be able to reliably quantify regional biochemical concentration. It shares the need for relatively expensive equipment with MRI, and in fact, requires fairly high field strength, typically 3T or more to be effective for animal and patient studies, particularly outside of the brain [17, 37].

11.2.3 Radionuclide Imaging Radionuclide imaging relies on the use of imaging probes, typically termed radiopharmaceuticals, tagged with radioactive nuclei [8, 23, 41]. Position-sensitive radiation detectors identify emitted photons and generate images of regional radiopharmaceutical concentration. This imaging approach, sometimes also termed nuclear medicine, has traditionally relied on gamma emitters such as 99mTc or 131I to form images, and often termed single-photon emission computed tomography (SPECT). Somewhat more recently, advances in both instrumentation and radiochemistry have led to the ability to image positron-emitting nuclei, such as 11C and 18F, in a wide range of molecules in an approach known as PET [42]. Compared with SPECT, PET offers the potential for better spatial resolution, more accurate image quantification, and a wider range of possible imaging probes; however both PET and SPECT have made notable contributions to breast cancer clinical care and research [8, 23, 42, 43]. The chief advantage of radionuclide imaging is the ability to measure probe concentrations in nanomolar and even picomolar range, leading to the ability to measure even the most sensitive molecular processes without perturbing them. A wide range of radiopharmaceuticals has been developed to image diverse aspects of cancer biology [41]. Disadvantages include more limited spatial resolution and the need to produce and distribute relatively short-lived imaging probes. Recent development in dedicated imaging devices for small animals [44] and breast-specific imaging [45] has overcome some of the limitations in spatial resolution; however, inherent spatial resolution is less than for other methods such as CT or MRI. The combination of PET or SPECT with X-ray computed

11  Imaging Studies in Anticancer Drug Development

281

tomography (PET/CT or SPECT/CT) yields coregistered molecular and anatomic images and the opportunity to image molecular biology and anatomy simultaneously [46]. Radionuclide imaging probes and instrumentation are relatively expensive, with costs comparable to MRI.

11.2.4 Optical Imaging One of the oldest forms of imaging is optical imaging, using visible light to generate images. In many ways, optical imaging is the earliest form of cancer imaging, in the form of light microscopy. Recent advances in instrumentation, computational algorithms, and imaging probes have led to new capabilities in optical imaging of living organisms, including small animal models and patients [47–49]. A variety of optical methods have been developed that can yield in vivo images with high contrast, and in some cases, considerable detail, down to the microscopic level. Methods can measure regional biology such as vascularity and blood volume using inherent tissue optical properties [47–49], or take advantage of an ever increasing array or optical probes to image-specific molecular processes [50]. Low cost, portability, ease of use, and wide availability of imaging probes are key advantages of optical imaging. Its chief disadvantage is relatively limited tissue penetration. Thus while optical imaging has become an essential tool for animal research in cancer [48], its use in patients has been more limited. While promising early studies in some human tumors point toward future clinical application [47, 50], optical imaging has been mostly confined thus far to the preclinical setting, where it is an important tool for cancer research.

11.2.5 Ultrasound Ultrasound imaging works by using acoustical transducers to send and receive ultrasound frequency energy and generate three-dimensional images from either reflection or through transmission [16]. Conventional ultrasound provides highresolution anatomic detail, and ultrasound plays an important role in cancer diagnosis [51], and is particularly useful for directing tissue biopsy. Doppler technology also provides information on tumor vascularity and with the advent of microbubble contrast agents, tumor perfusion [52]. Recent advances in imaging technology [53] and the development of targeted microbubble contrast agents hold promise for molecular imaging [52]. The portability and relatively low cost of ultrasound make it an ideal tool for both animal and patient imaging, and the ability to measure molecular processes will make ultrasound a valuable tool for cancer research and possibly for drug development. Disadvantages include some operator dependence in image acquisition and interpretation, and some challenges in developing molecularly targeted microbubble contrast agents.

282

D.A. Mankoff

11.2.6 Other Imaging Other imaging modalities such as X-ray radiography and X-ray CT play an important role in structural imaging, but are more limited for molecular imaging. Dynamic contrast CT can be used to measure tissue perfusion, similar to DCEMRI, with the disadvantage of relatively high radiation exposure. Other techniques are also being investigated [54], but are at relatively early stages of development and not discussed in detail.

11.3 Imaging to Define Targets and Select Patients for Clinical Trials 11.3.1 Overview As the name would imply, targeted cancer treatment relies on the presence of therapeutic targets expressed to a greater extent in the tumor than in normal tissue. However, if the tumor does not express the target, the treatment is likely to fail. Therefore, successful development of targeted anticancer therapy relies upon the ability to determine the presence or absence of the target. Current approaches to target expression assay rely on the ability to measure the expression of specific gene products, typically proteins, in tissue samples obtained from biopsy. Examples include the expression of estrogen receptors (ERs), a target for endocrine therapy [55], and HER2, also increasingly a target of tumor-specific treatment in breast cancer and other tumors [1]. Molecular imaging has also been applied to measuring specific protein expression [9, 43]. Advantages of imaging relative to biopsy include imaging’s noninvasiveness, the ability to measure target expression in the entire disease burden and thus the ability to avoid sampling error that can occur with heterogeneous receptor expression, and the potential for serial studies of in vivo drug effects on the target. A very practical consideration is that imaging can assess expression at sites that are challenging to sample and assay, for example, bone metastases, where decalcification can make assay of tumor gene products challenging. Imaging protein expression, particularly tumor receptors, poses some unique challenges. For receptors, imaging results can be quite sensitive to the molecular quantity of the imaging probe needed to generate the image. Most receptors have high affinity for their ligands and are active at micromolar or nanomolar concentrations of the ligand. Even small molar quantities of the imaging agent may saturate the receptor and limit the ability to visualize receptor expression [56, 57]. For this reason, molecular imaging of tumor receptors has been most successful to date with radionuclide imaging, PET, and SPECT, where it is possible to generate images with nanomolar or picomolar amounts of the imaging probe. For larger molecules, like peptides and monoclonal antibodies, other labels suitable for optical, MR, and

11  Imaging Studies in Anticancer Drug Development

283

ultrasound imaging are possible [9]; however, for small-molecule receptor imaging agents, such as labeled steroids for steroid receptors, radionuclide imaging appears to be the only feasible approach.

11.3.2 Examples of Imaging Target Expression Examples of molecular imaging to identify target expression include PET ER imaging in breast cancer [57, 58], PET or MR imaging of 5-FU in GI cancers [59, 60], SPECT and PET imaging of HER2 expression in breast cancer [61, 62], and PET imaging of integrin expression as target for antiangiogenic therapy [63, 64] (see Table 11.2). The application of imaging to measuring ER expression is highlighted in more detail below. Although ER is not a novel cancer target, this example serves to illustrate how imaging can be helpful in identifying target expression and refining patient selection. Perhaps the earliest specific target in cancer therapy is the ER in breast cancer [65]. The majority of breast cancers express that ER and endocrine therapy has proved to be an important breast cancer treatment [66]. Although only 30–70% of patients whose tumors express ER benefit from endocrine therapy, benefit is rare in patients whose tumor do not express ER or the related progesterone receptor [67–72]. Assay of breast tumor biopsy material is a well-established standard of care for selecting breast cancer patients for endocrine therapy [73]. Preliminary studies using PET imaging of ER expression have shown promise for refining patient selection. A number of agents have been tested for PET ER imaging (reviewed in [57]). Work with 16-a-18F-fluro-17b-estradiol (FES) has been the most promising to date [74]. FES has binding characteristics that are similar to estradiol for both the ER and its transport protein sex hormone binding globulin (SHBG) [75]. Blood clearance curves and protein interactions of FES have been studied in humans and animals. FES is rapidly metabolized in the liver, largely to sulfate and glucuronidate conjugates of FES [76]. Typically in humans about 45% of 18F-FES in circulating plasma is bound to SHBG and is distributed between albumin and SHBG with equilibrium maintained under most circumstances [77, 78]. By 30  min after injection, blood clearance and washout of nonspecifically bound FES are sufficient to permit good-quality ER imaging [76]. The uptake of FES at the tumor site has been validated as a measure of ER expression against in vitro assay of biopsy material using both radioligand binding assays [79] and more recently against immunohistochemistry (IHC) [80]. Recent studies illustrate the use of 18F-fluorestradiol (FES) PET to image ER expression in metastatic breast cancer to identify the therapeutic target as a predictor of response to endocrine therapy (Fig.  11.1). Studies by Mortimer, Dehdashti, and colleagues [81] have shown that a high level of FES uptake in advanced tumors predicts a greater likelihood of response to tamoxifen. In another study, in patients with recurrent or metastatic breast cancer from ER primary tumors, many of whom had failed prior to endocrine therapy, FES PET

284

D.A. Mankoff

Table 11.2  Selecting examples of imaging targets relevant to drug development Imaging Target modality Technique or probe Tumor receptors 18 Estrogen receptor PET F-fluoroestradiol (FES) [75] SPECT 11-b-methoxy-17a-[123I]iodovinylestradiol [145] 18 Androgen receptor PET F-fluorodihydrotestosterone (FDHT) [146] 68 Epidermal growth factor PET Ga-DOTA-EGF [147]; 64Cu-DOTAreceptor (EGFR) cetuximab [148] 111 SPECT In-DTPA-EGF [149] Optical Cy5.5-EGF [150]; Cy5.5-cetuximab [151] 68 HER2/Neu PET Ga-DOTA-F(ab¢)2-herceptin [62] 111 SPECT In-DTPA-trastuzumab [61] Ultrasound PLA nanoparticle-trastuzumab [152] Optical RhodG-trastuzumab [153] MRI (Avidin-Gd)-biotinylated-anti-HER2/neu MoAb [154] 111 Somatostatin receptor SPECT In-octreotide [155] 68 PET Ga-DOTA-octreotide [156]; 64Cu-TETAoctreotide [157] 18 Hypoxia PET F-fluoromisonidazole [158]; 64Cu-ATSM [98] MRI BOLD MRI [96] Tumor Metabolism and Perfusion Glycolysis and energy metabolism Oxygen metabolism Lipid synthesis

PET MRS PET MRI PET

Cellular proliferation

MRS PET

Cell death

Tumor perfusion

Angiogenesis

Drug transport and metabolism P-glycoprotein

5-Fluorouracil

SPECT PET MRI MRI PET Ultrasound PET MRI Optical SPECT PET PET MRI

F-fluorodeoxyglucose (FDG) [159, 160] Phosphorus MRS [161] 15 O-O2 [162] BOLD MRI [96] 11 C-choline [163] ; 18F-fluorocholine [164]; 11 C-acetate [165] 1 H-MRS [166] 11 C-thmyidine [167]; 18F-fluorothymidine [116] 99m Tc-annexin V [121] 11 F-fluoroannexin [125, 168] Diffusion MRI [126] DCE-MRI [30] 15 O-water [158] Microbubble contrast imaging [52] 18 F-galacto-RGD [169] RGD-USPIO [170] RGD-Cy5.5 [171] 18

Tc-sestamibi (MIBI) [92] C-verapamil [91] ; 18F-fluoropaclitaxel [172] 18 F-5-FU [59] 18 F MRI [60] 99m 11

11  Imaging Studies in Anticancer Drug Development

285

Fig. 11.1  PET imaging of ER expression in breast cancer as a method of identifying the therapeutic target. Both patients shown in the figure had bone metastases arising from ER + primary tumors and both were treated with endocrine therapy. The top patient’s (a) pretherapy FDG and FES PET scans show FES uptake at all sites of active disease seen by FDG PET. A follow-up FDG PET scans shows response to therapy after starting an aromatase inhibitor. The lower patient (b) does not have FES uptake at the site of disease seen on FDG PET (arrow) and had subsequent disease progression on endocrine treatment, shown by the follow-up FDG PET (reprinted from [58])

identified a subset of patients with low or absent ER expression, none of whom responded to endocrine therapy [58]. This was approximately 30% of the overall population, and the use of FES as a predictive marker to select patients for treatment other than with hormonal therapy would have increased the response rate from approximately 25 to 50% [58]. These early example illustrates how the use of imaging can help select patient most likely to respond to drug therapy, even or established therapies. Imaging can play an important role in drug development, especially for targeted therapy, by restricting patient selection to those whose tumors clearly express the therapeutic target.

286

D.A. Mankoff

11.3.3 Imaging Resistance Factors Equally important to verifying that the therapeutic target is present is the need to identify potential factors mediating therapeutic resistance. Mechanisms may include factors that block drug transport, abrogate drug effect, or indicate the presence of functional pathways that may make the tumor insensitive to interruption of the chosen target. Examples include drug efflux transporters [82, 83], aberrant tumor perfusion leading to poor drug delivery [84, 85], and hypoxia as a factor mediating broad resistance to anticancer therapy [86, 87]. Some factors, for example, hypoxia and drug efflux transport involve functional in  vivo drug resistance mechanisms and may therefore be difficult to assay by in  vitro assay of biopsy material. Imaging may be particularly well suited to identifying functional drug resistance, and some notable examples are highlighted below. Aberrant tumor vasculature may limit the delivery of the drug to the tumor through a variety of mechanisms, including arterial-venous shunting and increased tumor interstitial pressure, limiting drug delivery from the capillaries through the interstitial space to the tumor cells [88]. Some images of recent studies using DCE MRI and other methods have highlighted this phenomenon and show how antivascular therapy can improve drug delivery. Batchelor showed in a small series of glioma patients treated with an antivascular agent that serial perfusion MRI indicated an early response to treatment, with more favorable conditions for drug delivery and lower tissue edema [89]. Another study using perfusion CT to evaluate vascular response showed similar results [90]. These early studies illustrate the potential for imaging to delineate factors important in drug delivery to the tumor and the effect of therapy designed to improve delivery. Even if the drug reaches the tumor cell, a lack of drug transport into the cell may limit drug efficacy. A number of approaches to imaging drug transport have been tested, many of them focusing on the membrane efflux pump, P-glycoprotein (P-gp) [91, 92]. Some approaches have used model substrates of drug transport as general probes of drug delivery, for example 99mTc-setamibi (MIBI) or 11 C-verapamil as a marker of p-gp transport [13, 91, 92]. Studies showed that tumors with rapid efflux of MIBI, as an indicator of p-gp activity, predicted poor response to chemotherapy where the principal agent (epirubicin) is a p-gp substrate [93]. Recent studies using 11C-verapamil showed that the pharmacologic inhibition of p-gp by cyclopsorin increased drug transport across the blood– brain barrier [13]. Other approaches have labeled the cancer drug itself to examine its delivery and uptake in the tumor as a predictor of response, for example [18F]-fluoropaclitaxel [94, 95]. Tumor hypoxia has been implicated as a factor mediating broad resistance to anticancer therapy through a variety of mechanism, including diminishing cell cycling and raising the threshold for cell death [87]. Imaging using either hypoxiaspecific probes for PET or hypoxia-specific MR methods has shown considerable promise for identifying regional tumor hypoxia [96, 97]. Several studies have shown that tumor hypoxia identified by PET predicts poor response and early

11  Imaging Studies in Anticancer Drug Development

287

relapse in a variety of tumors including cervical cancer, head and neck cancer, and brain tumors [98–100]. Hypoxia imaging may also be helpful to direct hypoxiaspecific therapy, using hypoxia as an anticancer target. A recent study of hypoxia imaging using 18F-fluoromsinidazole (FMISO) PET and the hypoxia-specific therapeutic agent, tirapazemine, in head and neck cancer yielded interesting results in this regard [101]. In this study comparing regimens with and without tirapazemine for advanced head and neck cancer, no benefit was seen in the general patient population. However, in the subset of patients who underwent FMISO PET, the presence of hypoxia determined by FMISO uptake was a significant predictor of tumor response.

11.4 Imaging to Assess Early Pharmacodynamics/Response 11.4.1 Overview An important aspect of drug development and early clinical testing is the ability to measure early drug effect on the tumor [2]. This may be important in proof of mechanism of action, and also in verifying that drugs which look favorable in preclinical testing are likely to be effective in treating human cancers. The ability to measure early pharmacodynamic measures by anatomic imaging and biopsy has been limited by a number of practical difficulties. The current approach to cancer response assessment relies on changes in tumor size [14], a relatively late and largely mechanism-independent response to anticancer therapy. Furthermore, size is an especially poor indicator of response to cytostatic treatments, where it can be difficult to discern disease stabilization from slow tumor growth and slow increase in tumor size. Serial biopsy may provide early assessment of response, for example, by assaying changes in tumor proliferation [102], and also mechanism-specific indications of drug pharmacodynamics [103]; however, biopsy is practically difficult and potentially morbid, and difficult to perform over the course of time to discern the timing of the onset of drug action. The measurement of early drug response is a task to which functional and molecular imaging is ideally suited.

11.4.2 Examples of Imaging Early Response Several recent studies have highlighted the ability of functional and molecular imaging to measure early response to anticancer agents. Much of the early work has relied upon downstream markers of tumor cell “health” to measure early drug effects and predict later tumor shrinkage and response by size criteria. The most widely studied cellular process is glycolysis, owing in part to the widespread availability of the PET glucose analog, 18F-fluorodeoxyglucose (FDG). The earliest

288

D.A. Mankoff

studies showed that FDG PET could identify tumor response after a single cycle of chemotherapy, long before size changes had occurred [104–108]. More recent studies have shown that an early decline in glycolysis may also occur in response to targeted therapies, such as imatinib, where declines in FDG uptake may be seen within 24–48 h after starting the drug [109]. Earlier studies suggest that an early decline in glycolysis accompanies treatment with other TKIs, including anti-EGFR agents and sunitinib [110], perhaps in advance of changes in cellular proliferation and cell death [111]. Some have suggested that FDG indicates tumor cell viability in cancer drug response evaluation; however, recent data for TKIs show that FDG tumor uptake precedes changes in cell death [111] and may increase again when the drug is removed. This suggests some caution is needed in interpreting a decline in FDG uptake as a decrease in the number of viable tumor cells. Another approach to early response evaluation uses MRS and relies on the fact that tumor cells have aberrant expression of certain membrane lipids, for example, choline, which can be quantified by MRS [112]. Early response to treatment results in a decrease of local choline concentration, measured by MRS, in some cases within 24 h of starting chemotherapy [39, 40], presumably as an indicator of tumor cellular dropout. Ongoing clinical trials in breast and other cancers are testing this hypothesis. Other cellular processes may be more specific to tumor response. Serial biopsy data have suggested that a decline in cellular proliferation is an early and robust indicator of tumor response [102]. Parallel results have been shown by PET cellular proliferation imaging. The earliest studies were performed using 11C-thymidine and showed large, early declines in thymidine retention in response to chemotherapy [113–115]. Changes in thymidine uptake were larger than changes in FDG uptake in the same patients [114] (Fig.  11.2). More recent studies have focused on the thymidine analog, 18F-flourothymidine (FLT), which has a longer isotope label halflife (110 versus 20 min) and fewer labeled metabolites to confound image interpretation [115, 116]. Early studies have shown the ability of serial FLT PET to measure early response to chemotherapy [117–119] with good precision and repeatability [117]. Besides an early decline in cell growth, effective treatments often lead to an early increase in cell death, typically by apoptosis [120]. Imaging directed at phosphytidylserine residues that normally reside on the intracellular membrane surface but that are translocated to the extracellular surface during apoptosis have been developed for apoptosis imaging. The SPECT agent 99mTc-annexin V has demonstrated the ability to image apoptosis in vivo, but use of this metallabeled agent was confounded by high background, including liver uptake [121]. In early studies using this agent in patients undergoing cancer treatments, the level of uptake in 99mTc-annexin V correlated with in vitro assay of apoptosis on biopsy material, but the level of uptake and target-to-background ratio were only modest [122]. Concern has been expressed that the relatively small number of cells undergoing apoptosis at any one time and the small time window to have access to phosphytidylserine moieties during the apoptotic

11  Imaging Studies in Anticancer Drug Development

289

Fig.  11.2  Images of a patient before and after one cycle of chemotherapy for small-cell lung cancer. Left side shows the images of pretherapy 18F-fluordoxyglucose (FDG) and 11C-thymidine (TdR) and right side shows the images after 1 week of therapy. Images show the lung tumor (arrowhead) and vertebral bone marrow metastases (arrow). While both tracers indicate a decline in tracer uptake in response to therapy, the decline is much greater in the thymidine images, confirmed by quantitative analysis. The patient went on to have a complete clinical response after several more cycles of chemotherapy. Some early marrow regeneration is seen in the vertebral body of the posttherapy thymidine image (adapted from [173])

process [123] may limit the widespread use of Annexin V-based imaging. However, these same considerations may provide an advantage for mechanistic studies of early response that are important in determining optimal timing in multiagent therapy. Annexin tracers labeled for use in PET will offer better image quality, better quantification, and the ability to measure smaller quantities of radiopharmaceutical, and have undergone preliminary validation in animals [124, 125]. Recent studies have suggested that diffusion MRI may provide an indirect measure of cell death and early response to treatment [34, 126]. In diffusion MRI, pulse sequences sensitive to the Brownian motion of water molecules provide an estimate of the apparent diffusion coefficient (ADC) [126]. Studies in preclinical models and early studies in humans show that successful cancer therapy is accompanied by an increase in ADC measured by diffusion MRI, where presumably tumor cell death leads to increased interstitial space and increased ADC. Ongoing trials are testing diffusion MRI as an early indicator of response [34, 35, 127].

290

D.A. Mankoff

11.4.3 Examples of Imaging Pharmacodynamic Effect The examples cited above demonstrated that imaging could quantify early antitumor effects by measuring changes in processes such as glycolysis and cellular proliferation that are downstream from the therapeutic target. This provides a valuable early measure of drug effect, but may not provide insight into the mechanism of action, especially for early studies translating preclinical results and seeking to establish proof of mechanism in patients. Some early examples of imaging to measure more specific pharmacodynamics are highlighted below. Perhaps the most widely studied use of imaging to measure pharmacodynamics has been in application to antivascular therapy. Here, imaging methods designed to measure tumor perfusion, largely DCE-MRI, have been tested as early indicators of specific response to antiangiogenic therapy. Studies have shown that tumor perfusion measured by DCE-MRI declines within days of starting antiangiogenic therapy [27, 89]. Early work using probes more specifically targeted to tumor neovasculature may offer advantages for more specifically indicating response to antiangiogenic therapy [63, 64]. Other studies have taken advantage of drug effect on molecular pathways to provide an early and specific indication of drug effect. An elegant study showed that a transient increase in thymidine retention, measured by PET, provided an indication of the effect of a thymidilate-synthase (TS) inhibitor, where the drug would be expected to transiently increase flux through the deNovo (salvage) pathway traced by thymidine [128]. Patients whose tumors demonstrated the transient increase in thymidine uptake after the TS inhibitor were shown to have a decline in tumor proliferation by Ki-67 assay. Imaging may provide unique opportunities to study drug–target interaction. Serial studies using FES PET in patients receiving tamoxifen showed that an early decline in FES uptake, indicating effective receptor blockade, was a predictor of subsequent response [81]. Another early study showed that the pure antiestrogen fulvestrant, a potent ER-blocker in preclinical studies, failed to completely block FES uptake in some patients, unlike comparable studies using tamoxifen, which showed complete tumor blockade in nearly all studies [129]. Elegant preclinical studies showed the ability to measure early changes in HER2 expression in response to HSP90 inhibitors, which are expected to lead to decreased HER2 expression in breast cancer [62]. These early examples demonstrate the unique capability of imaging to measure early effects on therapeutic target, which would provide valuable insights in early drug trials.

11.4.4 Imaging as a Surrogate Endpoint? An increasing trend in phase II trials of targeted anticancer therapy is the use of time-to-progression or progression-free survival, rather than objective response, as a primary endpoint. This pose challenges in the design and length of trials, ­especially for more indolent tumors. Early results with functional and molecular

11  Imaging Studies in Anticancer Drug Development

291

imaging, mostly applied to cytoxic chemotherapy, suggest that imaging may provide a reasonable surrogate endpoint for survival. Studies have shown, for example, that an absence of FDG uptake posttherapy predicts significantly better outcome than those with residual uptake for a number of tumor types, including lymphoma, lung cancer, and breast cancer [108, 130, 131]. Others have suggested that changes in FDG uptake early in the course of treatment predict relapse and survival [108, 130–132]. Recent data in breast cancer show that changes in perfusion measured by MRI or PET predict relapse and survival [133, 134], and prognostic information that is independent of tumor size changes and pathologic response [133]. These results suggest that functional and molecular imaging may provide alternate endpoints that predict downstream outcomes better than size-based response criteria; however, more study is needed, in particular in application to targeted agents.

11.5 Analysis and Reporting of Molecular Imaging Data 11.5.1 Standardization The increased sophistication and complexity of functional and molecular imaging techniques poses a challenge in obtaining consistent and reproducible results, especially in multicenter trials. Imaging research has yielded a variety of approaches to acquire and analyze functional and molecular imaging methods, each with its own unique approach, strengths, and weaknesses. This diversity makes for good imaging research and has led to significant advances in imaging methods, but poses a challenge to standardization in clinical trials. The first step in standardization is to standardize image acquisition methods. This includes the type of data collected, the rate of data sampling, and approach to image generation. In some instances, for example, FDG PET, this also includes a standardized approach to patient preparation for the imaging study [135]. Two recent consensus efforts have led to suggested standards for DCE-MRI and FDG PET, where the US NCI and other organization have helped consensus conferences to determine methods appropriate for clinical trials [135, 136]. This represents a significant step forward, but not all trials have conformed to these standards. Equally important is the standardization of image analysis and interpretation. For anatomic imaging and size-based criteria for response, the RECIST standard is widely accepted and used in clinical trials [14]. There have been some early attempts to generate similar standards for DCE-MRI and FDG PET [135–137]; however, there are no uniformly agreed upon criteria. One complication is that the expected magnitude and timing of response for functional and molecular imaging may vary considerably for different therapies and different tumor types. It may be necessary to conduct trials specifically designed to establish appropriate response endpoints based upon other outcomes such as survival. This type of study is now going on for several tumors types, including FDG PET in lymphoma studies [138].

292

D.A. Mankoff

Another important measure is the precision and repeatability of the imaging studies. This is best established using the test/retest paradigm, where serial imaging studies are performed without a therapeutic intervention simply to determine the repeatability of the test. This poses a challenge in cancer imaging, where patients are reluctant to forego therapy to complete such tests. Some studies have been conducted showing good precision for some tests, for example, FDG PET [139] and some early test of novel PET radiopharmaceuticals [117].

11.5.2 Approach to Imaging Analysis Functional and molecular imaging methods may acquire both spatially and temporally detailed imaging data that lend themselves to a variety of different approaches to image analysis to obtain measures of biologic and clinical relevance [140]. There are varying levels of sophistication, and complexity of image analysis can be tailored to the nature of the biologic question and the type of clinical trial. An illustrative example is the evaluation of FDG PET images. The standard clinical approach to FDG PET imaging is to inject the patient, wait for some fixed time (typically 60 min), and then to perform imaging at a single time point (static imaging) [135]. Regional FDG uptake at this single time point can be measured by the PET imaging process in absolute units (mCi/ml, for example) and then converted to a ratio of the injected dose per unit body weight. This metric is called the SUV [141], and it is widely used in PET clinical practice and clinical trials (Fig. 11.3). More detailed information on the regional tumor glucose metabolism can be obtained by dynamic imaging over time for a single imaging field and compartmental analysis of the resulting data to yield estimates of local FDG kinetics and thus of physiologic parameters such as glucose delivery and glycolytic rate [142]. While the simple, static measures such as SUV typically correlate well with more sophisticated measures such as metabolic rate, some studies have shown that the more detailed approaches have more precision in delineating response [143], particularly for lower levels of tumor uptake [144]. Recent studies, for example, in neoadjuvant chemotherapy of breast cancer [133], have shown that parameters obtained from kinetic analysis yield response and survival prediction not obtained from simple uptake measure such as SUV. This example illustrates the need to consider the question being addressed in early patient studies of novel cancer drugs and the need to consider a range of approaches in image analysis in these early studies.

11.6 Summary and Conclusions Advances in functional and molecular imaging have led to the ability to perform regional, noninvasive assays of cellular and molecular processes in patients. This provides information that is complementary to in  vitro assay of biopsy material,

11  Imaging Studies in Anticancer Drug Development

293

Fig. 11.3  Diagram illustrating PET quantification methods. Static uptake measures such as the SUV are most frequently used; however, dynamic imaging and kinetic modeling offer the greatest potential insight for quantifying in vivo cancer biological features

particularly in the ability to measure therapeutic target expression across the entire disease burden. Imaging is also particularly well suited to identifying functional drug resistance mechanisms leading to decreased drug delivery to tumor cells or abrogation of antitumor drug effect. Serial quantitative imaging is ideal for measuring early drug effect and for providing insights into drug mechanism of action in patients. However, a number of potential hurdles exist, limiting the application of advanced imaging to clinical drug testing thus far. The number of imaging devices and imaging probed has been limited in the past; however, the increasing use of MRI/MRS and PET in clinical practice has led to significantly increased availability. Diagnostic imaging probes are themselves considered experimental drugs, and the regulatory hurdles associated with the need to obtain approval for both therapeutic drug and the imaging probe introduce logistical challenges. Programs sponsored by the NCI support the generation of INDs for many of the imaging probes, making them more readily available for clinical trials. Finally, imaging, especially more advanced approaches, can be costly. Hopefully, investigators and sponsors will increasingly recognize the value of imaging for gaining insights into drug mechanism and efficacy early in the clinical trial process. The appropriate use of imaging can decrease the overall cost of drug development by more effectively identifying

294

D.A. Mankoff

those drugs likely to be successful in anticancer treatment in patients and quickly eliminating those destined for failure.

References 1. Slamon DJ, Leyland-Jones B, Shak S, Fuchs H, Paton V, Bajamonde A, et al. Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2. N Engl J Med 2001;344(11):783–92. 2. Ratain MJ, Schilsky RL, Conley BA, Egorin MJ. Pharmacodynamics in cancer therapy. J Clin Oncol 1990;8(10):1739–53. 3. Hutter H. Fluorescent reporter methods. Methods Mol Biol 2006;351:155–73. 4. Blasberg RG. Imaging update: new windows, new views. Clin Cancer Res 2007;13(12):3444–8. 5. Mankoff DA. A definition of molecular imaging. J Nucl Med 2007;48(6):18N, 21N. 6. Mankoff DA, O’Sullivan F, Barlow WE, Krohn KA. Molecular imaging research in the outcomes era: measuring outcomes for individualized cancer therapy. Acad Radiol 2007;14(4):398–405. 7. Husband JE. Monitoring tumor response. Eur Radiol 1996;6:775–85. 8. Benard F, Turcotte E. Imaging in breast cancer: single-photon computed tomography and positron-emission tomography. Breast Cancer Res 2005;7(4):153–62. 9. Mankoff DA, Link JM, Linden HM, Sundararajan L, Krohn KA. Tumor receptor imaging. J Nucl Med 2008;49(Suppl 2):149S–63S. 10. Welch DR. Microarrays bring new insights into understanding of breast cancer metastasis to bone. Breast Cancer Res 2004;6(2):61–4. 11. Hoekstra CJ, Paglianiti I, Hoekstra OS, Smit EF, Postmus PE, Teule GJ, et  al. Monitoring response to therapy in cancer using [18F]-2-fluoro-2-deoxy-D-glucose and positron emission tomography: an overview of different analytical methods. Eur J Nucl Med 2000;27:731–43. 12. Lammertsma AA. Measurement of tumor response using [18F]-2-fluoro-2-deoxy-D-glucose and positron-emission tomography. J Clin Pharmacol 2001;(Suppl):104S–106S. 13. Sasongko L, Link JM, Muzi M, Mankoff DA, Yang X, Collier AC, et  al. Imaging P-glycoprotein transport activity at the human blood-brain barrier with positron emission tomography. Clin Pharmacol Ther 2005;77(6):503–14. 14. Therasse P, Arbuck SG, Eisenhauer EA, Wanders J, Kaplan RS, Rubinstein L, et  al. New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst 2000;92(3):205–16. 15. Weber WA. Positron emission tomography as an imaging biomarker. J Clin Oncol 2006;24(20):3282–92. 16. Bloch SH, Dayton PA, Ferrara KW. Targeted imaging using ultrasound contrast agents. Progress and opportunities for clinical and research applications. IEEE Eng Med Biol Mag 2004;23(5):18–29. 17. Bolan PJ, Nelson MT, Yee D, Garwood M. Imaging in breast cancer: magnetic resonance spectroscopy. Breast Cancer Res 2005;7(4):149–52. 18. Lehman CD, Schnall MD. Imaging in breast cancer: magnetic resonance imaging. Breast Cancer Res 2005;7(5):215–9. 19. Mankoff D. Imaging in breast cancer – breast cancer imaging revisited. Breast Cancer Res 2005;7(6):276–8. 20. Berger F, Gambhir SS. Recent advances in imaging endogenous or transferred gene expression utilizing radionuclide technologies in living subjects: applications to breast cancer. Breast Cancer Res 2001;3(1):28–35.

11  Imaging Studies in Anticancer Drug Development

295

21. Gillies RJ, Morse DL. In vivo magnetic resonance spectroscopy in cancer. Annu Rev Biomed Eng 2005;7:287–326. 22. Leach MO. Magnetic resonance spectroscopy (MRS) in the investigation of cancer at The Royal Marsden Hospital and The Institute of Cancer Research. Phys Med Biol 2006;51(13):R61–82. 23. Siegel BA, Dehdashti F. Oncologic PET/CT: current status and controversies. Eur Radiol 2005;15(Suppl 4):D127–32. 24. Galbraith SM. MR in oncology drug development. NMR Biomed 2006;19(6):681–9. 25. Saslow D, Boetes C, Burke W, Harms S, Leach MO, Lehman CD, et al. American Cancer Society guidelines for breast screening with MRI as an adjunct to mammography. CA Cancer J Clin 2007;57(2):75–89. 26. Padhani AR, Hayes C, Assersohn L, Powles T, Makris A, Suckling J, et  al. Prediction of clinicopathologic response of breast cancer to primary chemotherapy at contrast-enhanced MR imaging: initial clinical results. Radiology 2006;239(2):361–74. 27. Wedam SB, Low JA, Yang SX, Chow CK, Choyke P, Danforth D, et al. Antiangiogenic and antitumor effects of bevacizumab in patients with inflammatory and locally advanced breast cancer. J Clin Oncol 2006;24(5):769–77. 28. Yankeelov TE, Lepage M, Chakravarthy A, Broome EE, Niermann KJ, Kelley MC, et  al. Integration of quantitative DCE-MRI and ADC mapping to monitor treatment response in human breast cancer: initial results. Magn Reson Imaging 2007;25(1):1–13. 29. Choyke PL, Knopp MV, Libutti SK. Special techniques for imaging blood flow to tumors. Cancer J 2002;8(2):109–18. 30. Leach MO, Brindle KM, Evelhoch JL, Griffiths JR, Horsman MR, Jackson A, et  al. Assessment of antiangiogenic and antivascular therapeutics using MRI: recommendations for appropriate methodology for clinical trials. Br J Radiol 2003;76(Spec No 1):S87–91. 31. Padhani AR, Leach MO. Antivascular cancer treatments: functional assessments by dynamic contrast-enhanced magnetic resonance imaging. Abdom Imaging 2005;30(3):324–41. 32. Sosnovik DE, Weissleder R. Emerging concepts in molecular MRI. Curr Opin Biotechnol 2007;18(1):4–10. 33. Morse DL, Galons JP, Payne CM, Jennings DL, Day S, Xia G, et al. MRI-measured water mobility increases in response to chemotherapy via multiple cell-death mechanisms. NMR Biomed 2007;20(6):602–14. 34. Theilmann RJ, Borders R, Trouard TP, Xia G, Outwater E, Ranger-Moore J, et al. Changes in water mobility measured by diffusion MRI predict response of metastatic breast cancer to chemotherapy. Neoplasia 2004;6(6):831–7. 35. Hamstra DA, Chenevert TL, Moffat BA, Johnson TD, Meyer CR, Mukherji SK, et  al. Evaluation of the functional diffusion map as an early biomarker of time-to-progression and overall survival in high-grade glioma. Proc Natl Acad Sci USA 2005;102(46):16759–64. 36. Stephen RM, Gillies RJ. Promise and progress for functional and molecular imaging of response to targeted therapies. Pharm Res 2007;24(6):1172–85. 37. Mountford C, Lean C, Malycha P, Russell P. Proton spectroscopy provides accurate pathology on biopsy and in vivo. J Magn Reson Imaging 2006;24(3):459–77. 38. McKnight TR, Noworolski SM, Vigneron DB, Nelson SJ. An automated technique for the quantitative assessment of 3D-MRSI data from patients with glioma. J Magn Reson Imaging 2001;13(2):167–77. 39. Meisamy S, Bolan PJ, Baker EH, Bliss RL, Gulbahce E, Everson LI, et  al. Neoadjuvant chemotherapy of locally advanced breast cancer: predicting response with in vivo (1)H MR spectroscopy – a pilot study at 4 T. Radiology 2004;233(2):424–31. 40. Murphy PS, Viviers L, Abson C, Rowland IJ, Brada M, Leach MO, et al. Monitoring temozolomide treatment of low-grade glioma with proton magnetic resonance spectroscopy. Br J Cancer 2004;90(4):781–6. 41. Mankoff DA, Eary JF, Link JM, Muzi M, Rajendran JG, Spence AM, et al. Tumor-specific positron emission tomography imaging in patients: [18F] fluorodeoxyglucose and beyond. Clin Cancer Res 2007;13(12):3460–9.

296

D.A. Mankoff

42. Mankoff DA, Eubank WB. Current and future use of positron emission tomography (PET) in breast cancer. J Mammary Gland Biol Neoplasia 2006;11(2):125–36. 43. Quon A, Gambhir SS. FDG-PET and beyond: molecular breast cancer imaging. J Clin Oncol 2005;23(8):1664–73. 44. Cherry SR. The 2006 Henry N. Wagner Lecture: Of mice and men (and positrons) – advances in PET imaging technology. J Nucl Med 2006;47(11):1735–45. 45. Rosen EL, Turkington TG, Soo MS, Baker JA, Coleman RE. Detection of primary breast carcinoma with a dedicated, large-field-of-view FDG PET mammography device: initial experience. Radiology 2005;234(2):527–34. 46. Alessio AM, Kinahan PE, Cheng PM, Vesselle H, Karp JS. PET/CT scanner instrumentation, challenges, and solutions. Radiol Clin North Am 2004;42(6):1017–32, vii. 47. Tromberg BJ, Cerussi A, Shah N, Compton M, Durkin A, Hsiang D, et al. Imaging in breast cancer: diffuse optics in breast cancer: detecting tumors in pre-menopausal women and monitoring neoadjuvant chemotherapy. Breast Cancer Res 2005;7(6):279–85. 48. Henriquez NV, van Overveld PG, Que I, Buijs JT, Bachelier R, Kaijzel EL, et al. Advances in optical imaging and novel model systems for cancer metastasis research. Clin Exp Metastasis 2007;24(8):699–705. 49. Kumar S, Richards-Kortum R. Optical molecular imaging agents for cancer diagnostics and therapeutics. Nanomedicine 2006;1(1):23–30. 50. Sokolov K, Nida D, Descour M, Lacy A, Levy M, Hall B, et al. Molecular optical imaging of therapeutic targets of cancer. Adv Cancer Res 2007;96:299–344. 51. Mendelson EB. Problem-solving ultrasound. Radiol Clin North Am 2004;42(5):909–18, vii. 52. Ferrara K, Pollard R, Borden M. Ultrasound microbubble contrast agents: fundamentals and application to gene and drug delivery. Annu Rev Biomed Eng 2007;9:415–47. 53. Huang SW, Kim K, Witte RS, Olafsson R, O’Donnell M. Inducing and imaging thermal strain using a single ultrasound linear array. IEEE Trans Ultrason Ferroelectr Freq Control 2007;54(9):1718–20. 54. Brenner RJ, Parisky Y. Alternative breast-imaging approaches. Radiol Clin North Am 2007;45(5):907–23, viii. 55. Jordan VC, Brodie AM. Development and evolution of therapies targeted to the estrogen receptor for the treatment and prevention of breast cancer. Steroids 2007;72(1):7–25. 56. Katzenellenbogen J. The pharmacology of steroid radiopharmaceuticals: specific and nonspecific binding and uptake selectivity. In: Nunn A, editor. Radiopharmaceuticals: chemistry and pharmacology. New York, NY: Marcel Dekker; 1992. pp. 297–331. 57. Katzenellenbogen JA, Welch MJ, Dehdashti F. The development of estrogen and progestin radiopharmaceuticals for imaging breast cancer. Anticancer Res 1997;17:1573–6. 58. Linden HM, Stekhova SA, Link JM, Gralow JR, Livingston RB, Ellis GK, et al. Quantitative fluoroestradiol positron emission tomography imaging predicts response to endocrine treatment in breast cancer. J Clin Oncol 2006;24(18):2793–9. 59. Dimitrakopoulou-Strauss A, Strauss LG, Schlag P, Hohenberger P, Mohler M, Oberdorfer F, et al. Fluorine-18-fluorouracil to predict therapy response in liver metastases from colorectal carcinoma. J Nucl Med 1998;39(7):1197–202. 60. Wolf W, Presant CA, Waluch V. 19F-MRS studies of fluorinated drugs in humans. Adv Drug Deliv Rev 2000;41(1):55–74. 61. Perik PJ, Lub-De Hooge MN, Gietema JA, van der Graaf WT, de Korte MA, Jonkman S, et al. Indium-111-labeled trastuzumab scintigraphy in patients with human epidermal growth factor receptor 2-positive metastatic breast cancer. J Clin Oncol 2006;24(15):2276–82. 62. Smith-Jones PM, Solit DB, Akhurst T, Afroze F, Rosen N, Larson SM. Imaging the pharmacodynamics of HER2 degradation in response to Hsp90 inhibitors. Nat Biotechnol 2004;22(6):701–6. 63. Beer AJ, Haubner R, Sarbia M, Goebel M, Luderschmidt S, Grosu AL, et al. Positron emission tomography using [18F]Galacto-RGD identifies the level of integrin alpha(v)beta3 expression in man. Clin Cancer Res 2006;12(13):3942–9. 64. Laking GR, West C, Buckley DL, Matthews J, Price PM. Imaging vascular physiology to monitor cancer treatment. Crit Rev Oncol Hematol 2006;58(2):95–113.

11  Imaging Studies in Anticancer Drug Development

297

65. Sledge GJ, McGuire W. Steroid hormone receptors in human breast cancer. Adv Cancer Res 1983;38:61–75. 66. Pujol P, Hilsenbeck SG, Chamness GC, Elledge RM. Rising levels of estrogen receptor in breast cancer over 2 decades. Cancer 1994;74(5):1601–6. 67. Briasoulis E, Karavasilis V, Kostadima L, Ignatiadis M, Fountzilas G, Pavlidis N. Metastatic breast carcinoma confined to bone: portrait of a clinical entity. Cancer 2004;101(7):1524–8. 68. Osborne CK, Yochmowitz MG, Knight WA, 3rd, McGuire WL. The value of estrogen and progesterone receptors in the treatment of breast cancer. Cancer 1980;46(12 Suppl):2884–8. 69. Bloom ND, Tobin EH, Schreibman B, Degenshein GA. The role of progesterone receptors in the management of advanced breast cancer. Cancer 1980;45(12):2992–7. 70. Mouridsen H, Gershanovich M, Sun Y, Perez-Carrion R, Boni C, Monnier A, et al. Superior efficacy of letrozole versus tamoxifen as first-line therapy for postmenopausal women with advanced breast cancer: results of a phase III study of the International Letrozole Breast Cancer Group. J Clin Oncol 2001;19(10):2596–606. 71. Nabholtz JM, Buzdar A, Pollak M, Harwin W, Burton G, Mangalik A, et al. Anastrozole is superior to tamoxifen as first-line therapy for advanced breast cancer in postmenopausal women: results of a North American multicenter randomized trial. Arimidex Study Group. J Clin Oncol 2000;18(22):3758–67. 72. Buzdar A, Douma J, Davidson N, Elledge R, Morgan M, Smith R, et al. Phase III, multicenter, double-blind, randomized study of letrozole, an aromatase inhibitor, for advanced breast cancer versus megestrol acetate. J Clin Oncol 2001;19(14):3357–66. 73. Fuqua SA. The role of estrogen receptors in breast cancer metastasis. J Mammary Gland Biol Neoplasia 2001;6(4):407–17. 74. Sundararajan L, Linden HM, Link JM, Krohn KA, Mankoff DA. 18F-Fluoroestradiol. Semin Nucl Med 2007;37(6):470–6. 75. Kiesewetter DO, Kilbourn MR, Landvatter SW, Heiman DF, Katzenellenbogen JA, Welch MJ. Preparation of four fluorine-18-labeled estrogens and their selective uptakes in target tissues of immature rats. J Nucl Med 1984;25(11):1212–21. 76. Mankoff DA, Tewson TJ, Eary JF. Analysis of blood clearance and labeled metabolites for the estrogen receptor tracer [F-18]-16 alpha-fluoroestradiol (FES). Nucl Med Biol 1997;24(4):341–8. 77. Mankoff DA, Peterson LM, Tewson TJ, Link JM, Gralow JR, Graham MM, et al. [18F]fluoroestradiol radiation dosimetry in human PET studies. J Nucl Med 2001;42(4):679–84. 78. Tewson TJ, Mankoff DA, Peterson LM, Woo I, Petra P. Interactions of 16alpha-[18F]-fluoroestradiol (FES) with sex steroid binding protein (SBP). Nucl Med Biol 1999;26(8):905–13. 79. Mintun MA, Welch MJ, Siegel BA, Mathias CJ, Brodack JW, McGuire AH, et  al. Breast cancer: PET imaging of estrogen receptors. Radiology 1988;169(1):45–8. 80. Peterson LM, Mankoff DA, Lawton T, Yagle K, Schubert EK, Stekhova S, et al. Quantitative imaging of estrogen receptor expression in breast cancer with PET and 18F-fluoroestradiol. J Nucl Med 2008;49(3):367–74. 81. Mortimer JE, Dehdashti F, Siegel BA, Trinkaus K, Katzenellenbogen JA, Welch MJ. Metabolic flare: indicator of hormone responsiveness in advanced breast cancer. J Clin Oncol 2001;19(11):2797–803. 82. Nakanishi T. Drug transporters as targets for cancer chemotherapy. Cancer Genomics Proteomics 2007;4(3):241–54. 83. Pauwels EK, Erba P, Mariani G, Gomes CM. Multidrug resistance in cancer: its mechanism and its modulation. Drug News Perspect 2007;20(6):371–7. 84. Fukumura D, Jain RK. Tumor microvasculature and microenvironment: targets for antiangiogenesis and normalization. Microvasc Res 2007;74(2–3):72–84. 85. Jain RK. Normalization of tumor vasculature: an emerging concept in antiangiogenic therapy. Science 2005;307(5706):58–62. 86. Sutherland R. Tumor hypoxia and gene expression. Acta Oncologica 1998;37:567–74. 87. Teicher BA. Hypoxia and drug resistance. Cancer Metastasis Rev 1994;13:139–68. 88. Jain RK. Haemodynamic and transport barriers to the treatment of solid tumors. Int J Radiat Biol 1991;60:85–100.

298

D.A. Mankoff

89. Batchelor TT, Sorensen AG, di Tomaso E, Zhang WT, Duda DG, Cohen KS, et al. AZD2171, a pan-VEGF receptor tyrosine kinase inhibitor, normalizes tumor vasculature and alleviates edema in glioblastoma patients. Cancer Cell 2007;11(1):83–95. 90. Willett CG, Boucher Y, di Tomaso E, Duda DG, Munn LL, Tong RT, et al. Direct evidence that the VEGF-specific antibody bevacizumab has antivascular effects in human rectal cancer. Nat Med 2004;10(2):145–7. 91. Hendrikse NH, de Vries EG, Eriks-Fluks L, van der Graaf WT, Hospers GA, Willemsen AT, et al. A new in vivo method to study P-glycoprotein transport in tumors and the blood-brain barrier. Cancer Res 1999;59(10):2411–6. 92. Piwnica-Worms D, Chiu ML, Budding M, Kronauge JF, Kramer RA, Croop JM. Functional imaging of multidrug-resistant P-glycoprotein with an organotechnetium complex. Cancer Res 1993;53(5):977–84. 93. Ciarmiello A, Vecchio SD, Silvestro P, Potenta M, Carriero M, Thomas R, et al. Tumor clearance of technetium 99m-sestamibi as a predictor of response to neoadjuvant chemotherapy for locally advanced breast cancer. J Clin Oncol 1998;16(5):1677–83. 94. Kurdziel KA, Kalen JD, Hirsch JI, Wilson JD, Agarwal R, Barrett D, et al. Imaging multidrug resistance with 4-[18F]fluoropaclitaxel. Nucl Med Biol 2007;34(7):823–31. 95. Hsueh WA, Kesner AL, Gangloff A, Pegram MD, Beryt M, Czernin J, et al. Predicting chemotherapy response to paclitaxel with 18F-Fluoropaclitaxel and PET. J Nucl Med 2006;47(12):1995–9. 96. Padhani AR, Krohn KA, Lewis JS, Alber M. Imaging oxygenation of human tumours. Eur Radiol 2007;17(4):861–72. 97. Rajendran JG, Krohn KA. Imaging hypoxia and angiogenesis in tumors. Radiol Clin North Am 2005;43(1):169–87. 98. Dehdashti F, Grigsby PW, Mintun MA, Lewis JS, Siegel BA, Welch MJ. Assessing tumor hypoxia in cervical cancer by positron emission tomography with 60Cu-ATSM: relationship to therapeutic response – a preliminary report. Int J Radiat Oncol Biol Phys 2003;55(5): 1233–8. 99. Rajendran JG, Schwartz DL, O’Sullivan J, Peterson LM, Ng P, Scharnhorst J, et al. Tumor hypoxia imaging with F-18 FMISO PET in head and neck cancer: value of pre-therapy FMISO uptake in predicting survival. Clin Cancer Res 2006;12:5435–41. 100. Spence AM, Muzi M, Swanson KR, O’Sullivan F, Rockhill JK, Rajendran JG, et al. Regional hypoxia in glioblastoma multiforme quantified with [18F]-fluoromisonidazole positron emission tomography before radiotherapy: correlation with time to progression and survival. Clin Cancer Res 2008;14(9):2623–30. 101. Rischin D, Hicks RJ, Fisher R, Binns D, Corry J, Porceddu S, et al. Prognostic significance of [18F]-misonidazole positron emission tomography-detected tumor hypoxia in patients with advanced head and neck cancer randomly assigned to chemoradiation with or without tirapazamine: a substudy of Trans-Tasman Radiation Oncology Group Study 98.02. J Clin Oncol 2006;24(13):2098–104. 102. Dowsett M, Smith IE, Ebbs SR, Dixon JM, Skene A, Griffith C, et al. Short-term changes in Ki-67 during neoadjuvant treatment of primary breast cancer with anastrozole or tamoxifen alone or combined correlate with recurrence-free survival. Clin Cancer Res 2005;11(2 Pt 2): 951s–8s. 103. Calvo E, Malik SN, Siu LL, Baillargeon GM, Irish J, Chin SF, et al. Assessment of erlotinib pharmacodynamics in tumors and skin of patients with head and neck cancer. Ann Oncol 2007;18(4):761–7. 104. Schelling M, Avril N, Nahrig J, Kuhn W, Romer W, Sattler D, et  al. Positron emission tomography using [18F] fluorodeoxyglucose for monitoring primary chemotherapy in breast cancer. J Clin Oncol 2000;18:1689–95. 105. Smith I, Welch A, Hutcheon A, Miller I, Payne S, Chilcott F, et al. Positron emission tomography using [18F]-fluorodeoxy-D-glucose to predict the pathologic response of breast cancer to primary chemotherapy. J Clin Oncol 2000;18:1676–88.

11  Imaging Studies in Anticancer Drug Development

299

106. Wahl RL, Zasadny K, Helvie M, et al. Metabolic monitoring of breast cancer chemohormonotherapy using positron emission tomography: initial evaluation. J Clin Oncol 1993;11:2101–11. 107. Weber WA. Use of PET for monitoring cancer therapy and for predicting outcome. J Nucl Med 2005;46(6):983–95. 108. Romer W, Hanauske A, Ziegler S, Thodtmann R, Weber W, Fuchs C, et al. Positron emission tomography in non-Hodgkin’s lymphoma: assessment of chemotherapy with fluorodeoxyglucose. Blood 1998;91:4464–71. 109. Stroobants S, Goeminne J, Seegers M, Dimitrijevic S, Dupont P, Nuyts J, et  al. 18FDGPositron emission tomography for the early prediction of response in advanced soft tissue sarcoma treated with imatinib mesylate (Glivec). Eur J Cancer 2003;39(14):2012–20. 110. Banzo I, Quirce R, Martinez-Rodriguez I, Jimenez-Bonilla J, Sainz-Esteban A, Barragan J, et al. F-18 FDG PET/CT assessment of gastrointestinal stromal tumor response to sunitinib malate therapy. Clin Nucl Med 2008;33(3):211–2. 111. Su H, Bodenstein C, Dumont RA, Seimbille Y, Dubinett S, Phelps ME, et al. Monitoring tumor glucose utilization by positron emission tomography for the prediction of treatment response to epidermal growth factor receptor kinase inhibitors. Clin Cancer Res 2006;12(19):5659–67. 112. Glunde K, Jacobs MA, Bhujwalla ZM. Choline metabolism in cancer: implications for diagnosis and therapy. Expert Rev Mol Diagn 2006;6(6):821–9. 113. Martiat P, Ferrant A, Labar D, Cogneau M, Bol A, Michel C, et al. In vivo measurement of carbon-11 thymidine uptake in non-Hodgkin’s lymphoma using positron emission tomography. J Nucl Med 1988;29(10):1633–7. 114. Shields AF, Mankoff DA, Link JM, Graham MM, Eary JF, Kozawa SM, et al. Carbon-11thymidine and FDG to measure therapy response. J Nucl Med 1998;39:1757–62. 115. Mankoff DA, Shields AF, Krohn KA. PET imaging of cellular proliferation. Radiol Clin North Am 2005;43(1):153–67. 116. Shields AF, Grierson JR, Dohmen BM, Machulla HJ, Stayanoff JC, Lawhorn-Crews JM, et al. Imaging proliferation in vivo with [F-18]FLT and positron emission tomography. Nat Med 1998;4(11):1334–6. 117. Kenny L, Coombes RC, Vigushin DM, Al-Nahhas A, Shousha S, Aboagye EO. Imaging early changes in proliferation at 1 week post chemotherapy: a pilot study in breast cancer patients with 3¢-deoxy-3¢-[18F]fluorothymidine positron emission tomography. Eur J Nucl Med Mol Imaging 2007;34(9):1339–47. 118. Pio BS, Park CK, Pietras R, Hsueh WA, Satyamurthy N, Pegram MD, et al. Usefulness of 3¢-[F-18]fluoro-3¢-deoxythymidine with positron emission tomography in predicting breast cancer response to therapy. Mol Imaging Biol 2006;8(1):36–42. 119. Wieder HA, Geinitz H, Rosenberg R, Lordick F, Becker K, Stahl A, et al. PET imaging with [18F]3¢-deoxy-3¢-fluorothymidine for prediction of response to neoadjuvant treatment in patients with rectal cancer. Eur J Nucl Med Mol Imaging 2007;34(6):878–83. 120. Hockenbery D. Defining apoptosis. Am J Pathol 1995;146(1):16–9. 121. Blankenberg F, Katsikis P, Tait J, Davis R, Naumovski L, Ohtsuki K, et al. Imaging of apoptosis (programmed cell death) with 99mTc annexin V. J Nucl Med 1999;40:184–191. 122. van de Wiele C, Lahorte C, Vermeersch H, Loose D, Mervillie K, Steinmetz ND, et  al. Quantitative tumor apoptosis imaging using technetium-99m-HYNIC annexin V single photon emission computed tomography. J Clin Oncol 2003;21(18):3483–7. 123. Mandl SJ, Mari C, Edinger M, Negrin RS, Tait JF, Contag CH, et al. Multi-modality imaging identifies key times for annexin V imaging as an early predictor of therapeutic outcome. Mol Imaging 2004;3(1):1–8. 124. Collingridge DR, Glaser M, Osman S, Barthel H, Hutchinson OC, Luthra SK, et al. In vitro selectivity, in vivo biodistribution and tumour uptake of annexin V radiolabelled with a positron emitting radioisotope. Br J Cancer 2003;89(7):1327–33. 125. Yagle KJ, Eary JF, Tait JF, Grierson JR, Link JM, Lewellen B, et al. Evaluation of 18F-annexin V as a PET imaging agent in an animal model of apoptosis. J Nucl Med 2005;46(4):658–66.

300

D.A. Mankoff

126. Hamstra DA, Rehemtulla A, Ross BD. Diffusion magnetic resonance imaging: a biomarker for treatment response in oncology. J Clin Oncol 2007;25(26):4104–9. 127. Lee KC, Moffat BA, Schott AF, Layman R, Ellingworth S, Juliar R, et al. Prospective early response imaging biomarker for neoadjuvant breast cancer chemotherapy. Clin Cancer Res 2007;13(2 Pt 1):443–50. 128. Wells P, Gunn RN, Alison M, Steel C, Golding M, Ranicar AS, et al. Assessment of proliferation in vivo using 2-[(11)C]thymidine positron emission tomography in advanced intraabdominal malignancies. Cancer Res 2002;62(20):5698–702. 129. Linden HM, Link JM, Stekhova S, Livingston RB, Gralow JR, Ellis GK, et  al. Serial 18 F-fluoroestradiol positron emission tomography (FES PET) measures estrogen receptor binding during endocrine therapy. Breast Cancer Res Treat 2005;94(S1):S237. 130. Cachin F, Prince HM, Hogg A, Ware RE, Hicks RJ. Powerful prognostic stratification by [18F]fluorodeoxyglucose positron emission tomography in patients with metastatic breast cancer treated with high-dose chemotherapy. J Clin Oncol 2006;24(19):3026–31. 131. Mac Manus MP, Hicks RJ, Ball DL, Kalff V, Matthews JP, Salminen E, et al. F-18 fluorodeoxyglucose positron emission tomography staging in radical radiotherapy candidates with nonsmall cell lung carcinoma: powerful correlation with survival and high impact on treatment. Cancer 2001;92(4):886–95. 132. Eary JF, O’Sullivan F, Powitan Y, Chandhury KR, Vernon C, Bruckner JD, et al. Sarcoma tumor FDG uptake measured by PET and patient outcome: a retrospective analysis. Eur J Nucl Med Mol Imaging 2002;29(9):1149–54. 133. Dunnwald LK, Gralow JR, Ellis GK, Livingston RB, Specht J, Doot RK, et  al. Tumor metabolism and blood flow changes by PET: relation to survival in patients with neoadjuvant chemotherapy for locally advanced breast cancer. J Clin Oncol 2008;26(27):4449–57. 134. Partridge SC, Gibbs JE, Lu Y, Esserman LJ, Tripathy D, Wolverton DS, et al. MRI measurements of breast tumor volume predict response to neoadjuvant chemotherapy and recurrence-free survival. AJR Am J Roentgenol 2005;184(6):1774–81. 135. Shankar LK, Hoffman JM, Bacharach S, Graham MM, Karp J, Lammertsma AA, et  al. Consensus recommendations for the use of 18F-FDG PET as an indicator of therapeutic response in patients in National Cancer Institute Trials. J Nucl Med 2006;47(6):1059–66. 136. Leach MO, Brindle KM, Evelhoch JL, Griffiths JR, Horsman MR, Jackson A, et  al. The assessment of antiangiogenic and antivascular therapies in early-stage clinical trials using magnetic resonance imaging: issues and recommendations. Br J Cancer 2005;92 (9):1599–610. 137. Young H, Baum R, Cremerius U, Herholz K, Hoekstra O, Lammertsma AA, et  al. Measurement of clinical and subclinical tumour response using [F-18]-fluorodeoxyglucose and positron emission tomography: review and 1999 EORTC recommendations. Eur J Cancer 1999;35:1773–82. 138. Juweid ME, Stroobants S, Hoekstra OS, Mottaghy FM, Dietlein M, Guermazi A, et al. Use of positron emission tomography for response assessment of lymphoma: consensus of the Imaging Subcommittee of International Harmonization Project in Lymphoma. J Clin Oncol 2007;25(5):571–8. 139. Weber WA, Ziegler SI, Thodtmann R, Hanauske AR, Schwaiger M. Reproducibility of metabolic measurements in malignant tumors using FDG PET. J Nucl Med 1999; 40(11):1771–7. 140. Mankoff DA, Muzi M, Zabib H. Quantitative analysis of nuclear oncologic images. In: Zabib H, editor. Quantitative analysis of nuclear medicine images. Hingham, MA: Springer; 2004. 141. Huang S-C. Anatomy of SUV. Nucl Med Biol 2000;27:643–6. 142. Mankoff DA, Muzi M, Krohn KA. Quantitative positron emission tomography imaging to measure tumor response to therapy: what is the best method? Mol Imaging Biol 2003;5(5):281–5. 143. Krak NC, Hoekstra OS, Lammertsma AA. Measuring response to chemotherapy in locally advanced breast cancer: methodological considerations. Eur J Nucl Med Mol Imaging 2004;31(Suppl 1):S103–11.

11  Imaging Studies in Anticancer Drug Development

301

144. Doot RK, Dunnwald LK, Schubert EK, Muzi M, Peterson LM, Kinahan PE, et al. Dynamic and static approaches to quantifying 18F-FDG uptake for measuring cancer response to therapy, including the effect of granulocyte CSF. J Nucl Med 2007;48(6):920–5. 145. Rijks LJ, Boer GJ, Endert E, de Bruin K, Janssen AG, van Royen EA. The Z-isomer of 11 beta-methoxy-17 alpha-[123I]iodovinylestradiol is a promising radioligand for estrogen receptor imaging in human breast cancer. Nucl Med Biol 1997;24(1):65–75. 146. Liu A, Carlson KE, Katzenellenbogen JA. Synthesis of high affinity fluorine-substituted ligands for the androgen receptor. Potential agents for imaging prostatic cancer by positron emission tomography. J Med Chem 1992;35(11):2113–29. 147. Velikyan I, Sundberg AL, Lindhe O, Hoglund AU, Eriksson O, Werner E, et al. Preparation and evaluation of (68)Ga-DOTA-hEGF for visualization of EGFR expression in malignant tumors. J Nucl Med 2005;46(11):1881–8. 148. Cai W, Chen K, He L, Cao Q, Koong A, Chen X. Quantitative PET of EGFR expression in xenograft-bearing mice using (64)Cu-labeled cetuximab, a chimeric anti-EGFR monoclonal antibody. Eur J Nucl Med Mol Imaging 2007;34(6):850–8. 149. Reilly RM, Chen P, Wang J, Scollard D, Cameron R, Vallis KA. Preclinical pharmacokinetic, biodistribution, toxicology, and dosimetry studies of 111In-DTPA-human epidermal growth factor: an auger electron-emitting radiotherapeutic agent for epidermal growth factor receptorpositive breast cancer. J Nucl Med 2006;47(6):1023–31. 150. Adams KE, Ke S, Kwon S, Liang F, Fan Z, Lu Y, et al. Comparison of visible and nearinfrared wavelength-excitable fluorescent dyes for molecular imaging of cancer. J Biomed Opt 2007;12(2):024017. 151. Rosenthal EL, Kulbersh BD, King T, Chaudhuri TR, Zinn KR. Use of fluorescent labeled anti-epidermal growth factor receptor antibody to image head and neck squamous cell carcinoma xenografts. Mol Cancer Ther 2007;6(4):1230–8. 152. Liu J, Li J, Rosol TJ, Pan X, Voorhees JL. Biodegradable nanoparticles for targeted ultrasound imaging of breast cancer cells in vitro. Phys Med Biol 2007;52(16):4739–47. 153. Koyama Y, Hama Y, Urano Y, Nguyen DM, Choyke PL, Kobayashi H. Spectral fluorescence molecular imaging of lung metastases targeting HER2/neu. Clin Cancer Res 2007;13(10):2936–45. 154. Artemov D, Mori N, Ravi R, Bhujwalla ZM. Magnetic resonance molecular imaging of the HER-2/neu receptor. Cancer Res 2003;63(11):2723–7. 155. Balon HR, Goldsmith SJ, Siegel BA, Silberstein EB, Krenning EP, Lang O, et al. Procedure guideline for somatostatin receptor scintigraphy with (111)In-pentetreotide. J Nucl Med 2001;42(7):1134–8. 156. Koukouraki S, Strauss LG, Georgoulias V, Eisenhut M, Haberkorn U, DimitrakopoulouStrauss A. Comparison of the pharmacokinetics of 68Ga-DOTATOC and [18F]FDG in patients with metastatic neuroendocrine tumours scheduled for 90Y-DOTATOC therapy. Eur J Nucl Med Mol Imaging 2006;33(10):1115–22. 157. Anderson CJ, Dehdashti F, Cutler PD, Schwarz SW, Laforest R, Bass LA, et al. 64Cu-TETAoctreotide as a PET imaging agent for patients with neuroendocrine tumors. J Nucl Med 2001;42(2):213–21. 158. Rajendran JG, Mankoff DA. Positron emission tomography imaging of hypoxia and blood flow in tumors. In: Shields AF, Price P, editors. Cancer drug discovery and development: in vivo imaging of cancer. Totowa, NJ: Humana Press; 2006. pp. 47–71. 159. Phelps M, Huang S, Hoffman E. Tomographic measurement of local cerebral glucose metabolic rate in humans with (18F)2-fluoro-2-deoxy-D-glucose: validation of method. Ann Neurol 1979;6(5):371. 160. Reivich M, Kuhl D, Wolf A, Greenberg J, Phelps M, Ido T, et al. The [18F]fluorodeoxyglucose method for the measurement of local cerebral glucose utilization in man. Circ Res 1979;44(1):127–37. 161. Arias-Mendoza F, Payne GS, Zakian KL, Schwarz AJ, Stubbs M, Stoyanova R, et al. In vivo 31 P MR spectral patterns and reproducibility in cancer patients studied in a multi-institutional trial. NMR Biomed 2006;19(4):504–12.

302

D.A. Mankoff

162. Beaney R, Jones T, Lammertsma A, McKenzie D, Halnan K. Positron emission tomography for in-vivo measurement of regional blood flow, oxygen utilisation, and blood volume in patients with breast carcinoma. The Lancet 1984; 1(8369):131–134. 163. de Jong IJ, Pruim J, Elsinga PH, Vaalburg W, Mensink HJ. Visualization of prostate cancer with 11C-choline positron emission tomography. Eur Urol 2002;42(1):18–23. 164. Price DT, Coleman RE, Liao RP, Robertson CN, Polascik TJ, DeGrado TR. Comparison of [18F]fluorocholine and [18F]fluorodeoxyglucose for positron emission tomography of androgen dependent and androgen independent prostate cancer. J Urol 2002;168(1):273–80. 165. Oyama N, Miller T, Dehdashti F, Siegel B, Fischer K, Michalski J, et al. 11C-acetate PET imaging of prostate cancer: detection of recurrent disease at PSA relapse. J Nucl Med 2003;44:549–555. 166. Glunde K, Ackerstaff E, Mori N, Jacobs MA, Bhujwalla ZM. Choline phospholipid metabolism in cancer: consequences for molecular pharmaceutical interventions. Mol Pharm 2006;3(5):496–506. 167. Vander Borght T, Labar D, Pauwels S, Lambotte L. Production of [2-11C]thymidine for quantification of cellular proliferation with PET. Int J Rad Appl Instrum [A] 1991;42(1):103–4. 168. Zijlstra S, Gunawan J, Burchert W. Synthesis and evaluation of a 18F-labelled recombinant annexin-V derivative, for identification and quantification of apoptotic cells with PET. Appl Radiat Isot 2003;58(2):201–7. 169. Haubner R, Wester HJ, Weber WA, Mang C, Ziegler SI, Goodman SL, et al. Noninvasive imaging of alpha(v)beta3 integrin expression using 18F-labeled RGD-containing glycopeptide and positron emission tomography. Cancer Res 2001;61(5):1781–5. 170. Johansson LO, Bjornerud A, Ahlstrom HK, Ladd DL, Fujii DK. A targeted contrast agent for magnetic resonance imaging of thrombus: implications of spatial resolution. J Magn Reson Imaging 2001;13(4):615–8. 171. Chen X, Conti PS, Moats RA. In vivo near-infrared fluorescence imaging of integrin alphavbeta3 in brain tumor xenografts. Cancer Res 2004;64(21):8009–14. 172. Kurziel KA, Kieswetter DO, Carson RE, Eckelman WC, Herscovitch P. Biodistribution, radiation dose estimates, and in vivo P-gp modulation studies of 18F-paclitaxel in nonhuman primates. J Nucl Med 2003;44:1330–9. 173. Mankoff DA, Dehdashti F, Shields AF. Characterizing tumors using metabolic imaging: PET imaging of cellular proliferation and steroid receptors. Neoplasia 2000;2(1–2):71–88.

Part IV

Chapter 12

Role of the US Food and Drug Administration in Cancer Drug Development Ann T. Farrell, Ramzi N. Dagher, and Richard Pazdur

12.1 Introduction The US Food and Drug Administration (FDA) oversees the development of agents to diagnose, cure, mitigate, treat, or prevent cancer. The FDA’s mission statement is “The FDA is responsible for protecting the public health by assuring the safety, efficacy, and security of human and veterinary drugs, biological products, medical devices, our nation’s food supply, cosmetics, and products that emit radiation. The FDA is also responsible for advancing the public health by helping to speed innovations that make medicines and foods more effective, safer, and more affordable; and helping the public get the accurate, science-based information they need to use medicines and foods to improve their health” [1]. The FDA accomplishes its mission through meetings with individual investigators and sponsors, review of investigational new drug (IND) and new drug applications (NDA), facilities inspections, approval of marketing and licensing applications, and the administration of grant programs. FDA’s role in cancer drug and biologic development derives from legislation. Table 12.1 provides a summary of landmark legislation. The Pure Food and Drug Act of 1906 established FDA’s initial role in drug development. The Act of 1906 prohibited interstate commerce of misbranded food and drugs. In 1938, the Federal Food, Drug, and Cosmetic Act (FDC) was passed in response to the 1937 sulfanilamide tragedy in which more than 100 people, mostly children, died when a highly toxic chemical analog of antifreeze was added to make an antibiotic elixir more palatable. Passage of this Act required that manufacturers provide evidence of safety before marketing.

A.T. Farrell (*) Division of Hematology Products, Office of Oncology Drug Products (OODP), Center for Drug Evaluation and Research, US Food and Drug Administration, Silver Spring, MD, 20993-0002, USA e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_12, © Springer Science+Business Media, LLC 2011

305

306

A.T. Farrell et al.

Table 12.1  Regulatory milestones Regulatory milestone or term Year Biologics Control Act 1902 Pure Food and Drug Act

1906

Food Drug and Cosmetic Act Kefauver–Harris Amendments

1938 1962

Subparts H/E Prescription Drug User Fee Act (PDUFA) Food and Drug Modernization Act (FDAMA) Best Pharmaceuticals for Children Act (BPCA) Pediatric Research Equity Act (PREA) Food and Drug Amendments Act (FDAAA)

1992 1992

NDA ODAC

1997 2002 2003 2007

FDA advisory committee meetings first held in 1963

Features Authorized manufacturing inspections Prohibited interstate commerce in misbranded foods and drugs Introduced safety requirements Strengthened safety provisions; introduced efficacy requirements Accelerated approval Fees for product applications Reauthorized PDUFA, several reforms, pediatric exclusivity Pediatric exclusivity and other initiatives Mandated pediatric studies under certain circumstances Reauthorized user fees, added additional safety provisions for drugs, biologics, pet and human foods New drug application Oncology drugs advisory committee

Subsequently, the recognition of thalidomide-related birth defects led to the passage of the 1962 Kefauver–Harris amendment, which required manufacturers to demonstrate efficacy as well as safety before a drug could be marketed. In addition, this amendment required FDA to assess the efficacy of all drugs introduced since 1938, instituted stricter control over drug trials (including a requirement that patients involved must give their informed consent), transferred from the Federal Trade Commission to the FDA regulation of prescription drug advertising, established good manufacturing practices by the drug industry, and granted the FDA greater powers to access company production and control records to verify those practices. This amendment mandated the submission of investigational new drug applications (INDs) prior to conducting research in humans. The Biologics Control Act of 1902 was passed in response to several tragic events in which deaths and injuries occurred after treatment with tainted biologic products. In 1901, 13 children died of tetanus after being injected with a diphtheria antitoxin made from the blood of a tetanus-infected horse. This tragedy stimulated legislation regulating the sale of biologics. The Act authorized the government to inspect manufacturing establishments, to determine whether products were properly labeled with the product name, the name, address, and license number of the manufacturer, and the expiration date of the product, and to determine the method of manufacture. Later legislation established provisions for biologic agents in the Public Health Service similar to those drugs.

12  Role of the US Food and Drug Administration in Cancer Drug Development

307

The FDA is organized into eight Centers or Offices. The two Centers involved in the development of agents to treat cancer are the Center for Drug Evaluation and Research (CDER) and the Center for Biologic Evaluation and Research (CBER). CDER has regulatory authority for drugs, monoclonal antibodies, and cytokines, whereas CBER has regulatory authority for gene therapy and vaccines. The Center for Devices and Radiologic Health (CDRH) has regulatory authority for devices to treat cancer and diagnostic tests. Within CDER one office, the Office of Oncology Drug Products has primary responsibility for review of submissions for agents to treat cancer. The office has two divisions devoted to the regulation of anticancer products, the Division of Biologic Oncology Products and the Division of Drug Oncology Products. Within CBER, the Division of Clinical Evaluation and Pharmacology Toxicology within the Office of Cellular, Tissue, and Gene Therapies has primary responsibility for review of submissions for vaccines and gene therapies to treat cancer.

12.2 Role in Premarketing Development 12.2.1 What is an IND Application? Federal law requires that a drug or biologic product be the subject of an approved marketing application before it is transported or distributed across state lines. Because during development of an anticancer agent, the agent will likely be shipped from manufacturing site to clinical investigators in one or more states, the investigator or pharmaceutical sponsor will need to seek an exemption through submission of an IND. In general, the legal requirement applies to almost all products (marketed and unmarketed) that are regulated by the FDA [2]. Certain exceptions exist, such as clinical investigations involving specific in vitro diagnostic tests such as blood grouping serum, reagent red blood cells, and antihuman globulin, drugs intended solely for tests in vitro or in laboratory research animals, certain studies involving use of a placebo, and certain bioavailability studies. Qualifying for an exception to the requirement of having an IND does not confer an exception from informed consent [3]. The Code of Federal Regulations (CFR) provides guidance regarding those clinical investigations of lawfully marketed products which do not require an IND. All five requirements must be met for a clinical investigation to be exempt from the requirement to submit an IND [3]. The five requirements are (1) the investigation is not intended to be reported to the FDA as a well-controlled study in support of a new indication for use nor intended to be used to support any other significant change in labeling for the product. (2) If the drug that is undergoing the investigation is lawfully marketed as a prescription product, the investigation is not intended to support a significant change in the advertising for the product. (3) The investigation does not involve a route of administration or dosage level or use in a patient ­population or other factor that significantly increases the risks (or decreases the

308

A.T. Farrell et al.

acceptability of the risks) associated with the use of the product. (4) The ­investigation is conducted in compliance with requirements for institutional review and informed consent regulations set forth in parts 56 and 50 [4, 5]. (5) The study is conducted in compliance with regulations governing promotion and charging for investigational drugs [6]. The Agency has provided additional guidance for cancer investigators in a publication entitled Guidance for Industry: IND Exemptions for Studies of Lawfully Marketed Drug or Biological Products for the Treatment of Cancer [7]. An IND can be held by a pharmaceutical sponsor, an investigator, an academic institution, or other entity such as a contract research organization (CRO). The required contents of an IND are outlined in the Code of Federal Regulations [8] and listed in Table 12.2. Prior to the submission of an initial proposal (NDA) to conduct a clinical study, a request can be made to meet with the Agency to get advice or reach an agreement on the format for the IND, the scope and design of planned Phase 1 clinical studies, the design of animal studies needed to support human clinical testing, and product characterization issues. In addition, sponsors can request to meet with the Agency at any time during development by submitting a request. While the staff cannot act as development consultants, they can provide regulatory and scientific advice to facilitate development. Meeting requests are usually submitted when the sponsor/investigator has a product that will be tested either nonclinically or clinically. Typically, meetings are held around certain timepoints in development: prior to clinical study, End-ofPhase 1, End-of Phase 2, and prior to submission of the NDA/Biologics Licensing Application (BLA). INDs can be submitted for any phase of clinical investigation. The regulations define study phases from phase 1 to phase 4. Currently, <10% of INDs for New Molecular Entities developed for oncologic use progress beyond the investigational stage to NDA submission. The FDA recognizes that the traditional approach to phase 1 development is resource intensive and the need exists to reduce resources expended on products unlikely to succeed and to move ahead more efficiently with promising candidates. To address these concerns, the FDA has issued a Guidance on Exploratory INDs. This guidance describes the CMC, clinical, and safety animal studies required to support phase 0 clinical trials. These trials are intended to encompass very limited human exposure without diagnostic or therapeutic intent. Phase 0 clinical studies may be designed to evaluate mechanism of action (MOA), pharmacokinetics, or selection of a lead candidate based on pharmacokinetics/ pharmacodynamics and biodistribution [9, 10]. Three major types of INDs and two major categories of INDs exist. The three types are the investigator IND, the Emergency Use IND, and the Treatment IND. An Investigator IND is submitted by a physician who both initiates and conducts an investigation, and under whose immediate direction the investigational drug is administered or dispensed. This type of IND may be a research IND which involves the study of an unapproved or an approved product for a new indication or in a new patient population. An Emergency Use IND is an IND that allows the FDA to

12  Role of the US Food and Drug Administration in Cancer Drug Development Table 12.2  Contents of an IND Regulatory Item requirement Cover Letter No Cover Sheet/Form 1571

Table of contents Introductory Statement and General Investigatory Plan

Investigator’s Brochure

Protocol(s)

Chemistry, manufacturing, and control information Pharmacology/toxicology

Previous human experience Additional information

309

Comments Provides information about the investigator/ sponsor intentions Yes By signing the investigator agrees to:  Commitment to not begin clinical investigations until an IND covering the investigations is in effect;  Commitment that an IRB complying with Part 56 will be responsible for the initial and continuing review of IND and that the investigator will report to the IRB any proposed changes  Commitment to conduct the investigation in accordance with the applicable regulations Yes Orients the submission Yes Contains the basic information necessary to understand the drug development plan including basic CMC, pharmacology/toxicology, previous human experience, whether the agent has been withdrawn from marketing, development plans Yes Includes summary information on CMC, pharmacology/toxicology, pharmacokinetics and disposition, previous human experience, description of expected risks and side effects, precautions or special monitoring necessary for use Yes Should contain: statement of objectives and purpose, name and qualifications of investigator and subinvestigators and IRB, name of research facilities, description of the design of the study including control groups, methods to minimize bias, method for determining the dose to be administered, planned maximum dosage, duration of individual patient exposure, description of the measurements and observations which will fulfill study objectives, description of clinical procedures, laboratory tests, or other measures to monitor the drug effects and minimize risk Yes Contains the drug substance, drug product and/ or placebo information, any labeling, an environmental analysis (waiver or assessment) Yes Contains pharmacology (including effects, mechanism of action, drug disposition), toxicology, toxicokinetics, genotoxicity Yes Contains summary of what is known about previous human effects Only if Contains information on drug dependence and abuse pertinent potential, pediatric studies, radioactive drugs

310

A.T. Farrell et al.

authorize use of an experimental drug in an emergency situation, which does not allow time for submission of an IND in accordance with 21 CFR 312.23 or 312.34. These regulations allow for the administration of drug to patients who do not meet the criteria of an existing study protocol, or if an approved study protocol does not exist. One example is the Single Patient IND. A Treatment IND is submitted for an experimental drug showing promise in clinical testing for serious or immediately life-threatening conditions while the final clinical work is conducted and the FDA review takes place. There are two major categories of INDs: commercial and research. A commercial IND is an IND submitted by a pharmaceutical company or sponsor whose primary goal is to eventually gain marketing approval. All other INDs are typically considered research INDs. On December 14, 2006, the Agency published a proposed rule to clarify procedures for allowing expanded access or emergency use and to revise the existing emergency use and treatment IND/protocol regulations. Publication of the final rule is expected to increase public awareness, utilization, and streamlining of the process for obtaining access to investigational agents. An IND may be submitted for any phase of investigation from phase 0 exploratory study to phase 4 clinical trials. The regulations stipulate certain elements are necessary for filing an IND. The elements are necessary to show that the agent is safe for clinical testing. The type and amount of data necessary may depend on the study phase, subject type (healthy volunteer or cancer patient), and underlying disease.

12.2.2 What is Needed for the IND Submission Table 12.2 provides the required contents for an IND. Below is a brief outline of what information is necessary to open an IND. The sponsor can request a pre-IND meeting with the Agency to get clarification about what a sponsor must provide in terms of data to open an IND by requesting a pre-IND meeting. 12.2.2.1 Chemistry, Manufacturing, and Control In each phase of study of the investigational product, sufficient information must be submitted to assure the proper identification, quality, purity, and strength. The amount of information needed to make that assurance will vary with the phase of the investigation, the proposed duration of the investigation, the dosage form, and the amount of information otherwise available [11]. For a phase 1 submission, the emphasis should be on the identification and control of the raw materials and the new drug substance. At this stage, final specifications for the drug substance and drug product are not expected. The amount of information to be submitted depends also on the scope of the proposed clinical investigation. For example, if very short-term studies

12  Role of the US Food and Drug Administration in Cancer Drug Development

311

are ­proposed, the amount of supporting stability data can be correspondingly limited. As development progresses and changes are made to the manufacturing process, the sponsor is required to submit the appropriate amendments to the IND. For an initial IND involving a new molecular entity or biological product, the Agency requests that all available chemistry, manufacturing, and control (CMC) information on the drug substance, drug product, or biological product be provided, including items such as certificates of analysis, impurity profiles, batch information, compatibility testing, microbiological testing, and stability. If the IND requires information on an investigational product that the investigator or sponsor does not own or cannot provide, then the sponsor may reference a Drug Master File (DMF) for the Agency to review. If an investigator’s protocol uses an unmarketed product being developed by a pharmaceutical sponsor, a letter of authorization from the pharmaceutical developer is required with the initial IND submission. If an investigator’s protocol uses a marketed product, then no letter of authorization is necessary. Questions regarding CMC issues can addressed in a pre-IND meeting with the Agency prior to submission.

12.2.2.2 Nonclinical Pharmacology/Toxicology The nonclinical safety evaluation ensures adequate characterization prior to clinical investigatory use. This evaluation includes toxic effects with respect to target organs, potential reversibility, dose dependence, and relationship to exposure. The amount of information required may depend on the product, stage of development, proposed indication, intended population, and the projected treatment duration. While some differences may exist regarding the development for biological products to treat cancer compared with drug products, most requirements are the same. This information is obtained from conducting toxicology, toxicokinetic, genotoxicity, carcinogenicity, and safety pharmacology studies. Ideally all nonclinical studies should be conducted under Good Laboratory Practices (GLP). If the studies are non-GLP, then the reasons for not proving GLP studies should be provided for review. The nonclinical studies determine the choice of an initial starting dose for clinical studies and the potential organ toxicities to be monitored in the clinical studies [12]. Any serious adverse events observed during nonclinical or clinical studies may warrant additional specific safety studies. At a minimum, the following preclinical studies are recommended for the initial IND filing: single-dose toxicity, repeat dose toxicity, and genotoxicity. However, genotoxicity studies may not be required at the IND filing for studies conducted in patients with advanced cancer. Additional preclinical studies may be required prior to IND filing depending on other factors mentioned above. The following paragraphs discuss basic principles for some of these nonclinical studies [13].

312

A.T. Farrell et al.

Toxicology In general, toxicology studies are done in the most relevant species. For small molecules, single acute dose toxicology is usually assessed in two mammalian species (generally, one rodent and one nonrodent species) [14–16]. The toxicology studies should use doses high enough to cause toxicity and assess clinical signs, body weight, food consumption, gross pathology, and histopathology. Repeat dose toxicity studies should use regimens similar to those planned for the clinical studies [17]. Ideally, the nonclinical studies will mirror the proposed clinical studies in terms of administration schedule, duration, and route of administration [14, 15]. If the product is to be administered subcutaneously, intramuscularly, dermally, or ophthalmologically, additional nonclinical local tolerance studies may be necessary [18]. For biological products (e.g., monoclonal antibodies), peptides, oligonucleotides, and biopharmaceutical products, a single study in the most relevant species may be most appropriate, using the species where there is some evidence that the drug is biologically active. Genotoxicity Drug development has used genotoxicity testing to predict carcinogenicity. Pharmaceutical agents that test positive in genotoxicity tests have the potential to be human carcinogens and/or mutagens. In general, in vitro and in vivo genotoxicity testing detects direct or indirect genetic damage and should be performed prior to the initiation of phase 1 studies [16, 19]. However, in clinical studies of patients with advanced disease, genetic testing may not be necessary prior to the initiation of phase 1 testing. The standard battery of genotoxic tests includes (1) a bacterial reverse mutation assay, (2)an in vitro test with cytogenetic evaluation of chromosomal damage, using mammalian cells or in vitro mouse lymphoma thymidine kinase (tk) assay, and (3) an in  vivo test for chromosomal damage, using rodent hematopoietic cells. Modifications to the standard battery testing can be made when drugs are excessively toxic to bacteria as this toxicity can interfere with interpretation. In general, for biotechnology-derived pharmaceuticals, genotoxicity studies are usually not needed. If all three of these studies have negative results, these results suggest that the agent is not genotoxic. However, the absence of genotoxicity does not imply that the absence of carcinogenicity. Carcinogenicity Carcinogenicity studies are used to identify tumorigenic potential in animals which may portend human risk. These studies are usually performed when the drug or biologic is expected to be administered on a regular basis over a substantial part of a patient’s life (e.g., more than 3–6 months to years). Other reasons are listed in

12  Role of the US Food and Drug Administration in Cancer Drug Development

313

Table 12.3  Potential reasons for carcinogenicity studies Anticipated use of the drug for 3–6 months or greater Potential carcinogenicity, e.g., carcinogenicity test results have been positive for other drugs in the product class Prolonged survival – intended patient population has a life expectancy greater than 2–3 years Exposure – an ophthalmologically or dermally applied product that may have extensive systemic exposure Similarity to a natural substance – product is similar to an endogenous substance given as replacement therapy

Table 12.3. If possible, these studies dose animals using the same administration route intended for humans [16]. These studies are not necessary for filing an IND but may be performed and submitted to the Agency later. Development of agents for neoadjuvant, adjuvant, or maintenance therapy when survival is anticipated to be several years will require carcinogenicity testing [20, 21]. Although standard carcinogenicity bioassays are generally inappropriate for biotechnology-derived products, assessment of carcinogenic potential may still be needed and a variety of approaches may be considered to evaluate this risk. Carcinogenicity studies are usually not required when the agent is unequivocally genotoxic. Typically, the requirement is for two long-term carcinogenicity rodent studies prior to the marketing of a new drug. However, deviations from this requirement have been accepted. Ssponsors seeking to deviate from this requirement should discuss their proposal with the Agency. Regulatory guidance exists to provide the study design, necessary monitoring, and required investigations [22–25]. Safety Pharmacology Safety pharmacology studies have three goals: (1) to identify undesirable effects of a drug on physiologic function, possibly relating to safety, (2) to evaluate adverse effects observed in toxicology and/or clinical studies, and (3) to investigate the mechanism of the observed or suspected adverse events [12, 26, 27]. The three areas crucial for investigation are those that effect essential functions (central nervous system, cardiovascular, and respiratory systems). In general, adverse events transiently effecting the hepatic, renal, or gastrointestinal system that do not cause irreversible effect may not warrant immediate study, except when there may be irreparable harm in a specific vulnerable patient population [12]. These nonclinical studies may include in vivo, ex vivo, and in vitro testing [27]. Ideally these studies are performed prior to phase 1 testing. Results from these studies do not abrogate further nonclinical or clinical testing. During pharmaceutical development, the potential for prolongation of the QT interval and to develop proarrthymias must be assessed through nonclinical and clinical testing. If the agent can be given to healthy volunteers, then a “Thorough QT/QTc study” to assess potential for prolongation should be performed [28]. If the agent cannot be given to healthy volunteers, then usually an alternative to the

314

A.T. Farrell et al.

“Thorough TQT study” should be discussed with the Agency because most cancer agents used in trials are dosed near the maximum tolerated dose, and controls cannot be used. Immunogenicity Since biotechnology-derived pharmaceuticals may be immunogenic and our ability to predict immunogenicity is not sufficient, key nonclinical testing prior to phase 1 includes an assessment of immunogenicity. Predicting whether an infusion reaction will occur is difficult; however, in vitro assays measuring cytokine release products may be predictive. Nonclinical toxicity testing rarely detects signs of an infusion reaction; however, if seen, it may be useful in determining an appropriate starting dose, infusion rate, and frequency schedule in phase 1/2 trials, or in recommending management of this syndrome. Tissue cross-reactivity studies may be used to identify unintended target tissues for therapeutic monoclonal antibodies. In addition, this testing may confirm the pharmacological relevance of the animal species selected for safety testing. Specific recommendations for performing an adequate tissue cross-reactivity studies are available [21].

12.2.3 How to Fulfill IND Requirements The sponsor can fulfill some IND requirements by (1) compiling and submitting any existing nonclinical data from past in  vitro laboratory or nonclinical studies with the compound; (2) compiling data from previous clinical testing or marketing of the drug in the USA or another country; or (3) undertaking new preclinical studies designed to provide the evidence necessary to support the safety of administering the compound to humans.

12.2.4 Responsibilities of the IND Holder The CFR defines the terms sponsor, investigator, and sponsor–investigator [29]. The sponsor is the individual who takes responsibility for and initiates a clinical investigation; the sponsor may be a person or entity, such as a pharmaceutical company. The investigator is the individual who conducts a clinical investigation. A sponsor–investigator is an individual who both initiates and conducts an investigation and under whose immediate direction the drug is administered or dispensed. One of the most important responsibilities of the IND sponsor/investigator is to ensure that the IND submission is complete and contains all the required items. Having a complete IND application decreases the possibility that the IND will be

12  Role of the US Food and Drug Administration in Cancer Drug Development

315

placed on clinical hold. After the initial submission of an IND, the sponsor must wait for 30 days before starting the clinical study. The 30-day time period is for Agency review of the IND. The IND sponsor/investigator has a number of responsibilities with respect to the IND. Subpart D of the IND regulations has identified these as key: selecting qualified investigators, providing the investigators with the information they need to conduct an investigation properly, ensuring proper monitoring of the investigation, ensuring that the investigation is conducted in accordance with the general investigational plan and protocols, and ensuring that the FDA, IRB, and other investigators are promptly informed of significant new adverse events or risks with respect to the drug [30]. Table  12.4 outlines the duties of sponsors.

Table 12.4  Responsibilities of an IND holder Category Subduties Selection of investigators and monitors

Informing investigators

Review of ongoing investigations

Record keeping and retention

Inspection of sponsor’s records and reports Assurance of drug disposition

Select investigators qualified by training and experience Control of drug – ship drug only to investigators participating in the investigation Obtain needed information from the investigator – 1572 form, protocol CV, financial disclosure, names and addresses of subinvestigators, facilities (clinical, research, IRB), commitments to conduct the study in accordance with the protocol and comply with appropriate regulations including those in regard to safety reporting, has read and understands the Investigator’s Brochure, an IRB is involved that complies with the regulations Select Monitors qualified by training and experience to monitor the progress Provide the investigator with current version of Investigator’s Brochure particularly with regard to adverse effects and safe use Monitor the progress of each study Monitor site investigator’s compliance with protocol including removal of noncompliant investigators Review and evaluate the evidence related to the safety and effectiveness, including determining whether the drug represents an unreasonable and significant risk to subjects Maintain records regarding drug shipment and disposition, maintain records regarding financial interests, maintain records for 2 years after a marketing application is approved, retain reserve samples for testing Permit FDA inspection Maintain records and assure the return or alternative disposition of unused drug

316

A.T. Farrell et al.

Twenty-one CFR 312.60 outlines investigator responsibilities, which are similar to those for sponsors regarding control of investigational drug, record keeping, reporting, assurance of IRB review, and FDA inspection of records and reports. 12.2.4.1 Required Safety Reporting One of the key responsibilities expected of the sponsor/investigator is compliance with the regulations regarding safety reporting including expedited reports. Investigators are expected to report any unexpected fatal or life-threatening adverse experience (AE) that is associated with use of the drug within 7 days [31]. Also, investigators are expected to report any AE that is associated with a drug that is both serious and unexpected. The phrase “associated with the use of the drug” is defined by the CFR as having a reasonable possibility that the experience may have been caused by the drug. A serious adverse experience is one that causes death, is life-threatening, causes hospitalization or prolongs hospitalization, or a persistent or significant disability/incapacity, or a congenital anomaly/birth defect. An unexpected adverse experience is one where the adverse experience would not be expected from reading the investigator’s brochure or if no investigator’s brochure is required or available, the specificity or severity of which is not consistent with risk information described elsewhere. The sponsor should follow up all adverse experiences as their association with the drug may change with additional information. Sponsor/investigators are required to report any findings from tests in laboratory animals that suggest a significant risk for human use (e.g., positive mutagenicity, carcinogenicity, or teratogenicity). Nonfatal or nonlife-threatening AEs may be reported within 15  days of initial notification of the sponsor. This information should be submitted on FDA Form 3500A or in a narrative format or if a foreign study on a CIOMS I form. Expedited safety information should also be submitted to the Institutional Review Board (IRB). In addition, the sponsor is expected to provide information on periodic safety under an IND. This information may be part of the required annual report. 12.2.4.2 Annual Report Sponsors/investigators are required to file an annual report within 60 days of the anniversary date that the IND application was accepted by the FDA. This annual report should include a brief report of the progress of the investigation: individual clinical study information (summary of each study’s status), summary of investigational drug information, concentrating on safety, deaths, dropouts, any new information about dose, response, bioavailability, nonclinical information and studies, any new significant chemistry, manufacturing, and control information, general investigational plan for the coming year, investigator brochure revisions, information on amendments to phase 1 protocols, information on foreign market developments, and any outstanding FDA business [32].

12  Role of the US Food and Drug Administration in Cancer Drug Development

317

12.2.4.3 Informed Consent An informed consent is necessary for most studies using investigational or marketed products conducted under an IND except for certain emergency situations [33, 34]. The CFR discusses these situations under which obtaining informed consent may not be feasible and the safeguards and procedures that should be in place for these emergency situations including IRB review. For most studies, investigators may involve a human being as a subject in research only after they have obtained the legally effective informed consent of the subject or the subject’s legally authorized representative [35]. The informed consent must provide the prospective subject or the representative sufficient opportunity to consider whether to participate and minimize the possibility of coercion or undue influence, in language that is understandable. The informed consent may not use exculpatory language which appears to waive rights of the subject or representative. The informed consent document must contain the information required by each of the eight basic elements of 21 CFR 50.25(a), and each of the six elements of 21 CFR 50.25(b) that is appropriate to the study. The eight basic and six additional elements are listed in Tables  12.5 and 12.6, respectively. IRBs have the final authority for ensuring the adequacy of the information in the informed consent document [36]. In most circumstances, the informed consent should be a written document. The consent document may be either in a long form with all the basic and additional elements or a short form stating that the required elements of informed consent have been presented orally to the subject or the subject’s legally authorized Table 12.5  Basic elements of informed consent A statement that the study involves research, an explanation of the purposes of the research and the expected duration of the subject’s participation, a description of the procedures to be followed, and identification of any procedures which are experimental A description of any reasonably foreseeable risks or discomforts to the subject A description of any benefits to the subject or to others which may reasonably be expected from the research A disclosure of appropriate alternative procedures or courses of treatment, if any, that might be advantageous to the subject A statement describing the extent, if any, to which confidentiality of records identifying the subject will be maintained and that notes the possibility that the Food and Drug Administration may inspect the records For research involving more than minimal risk, an explanation as to whether any compensation and an explanation as to whether any medical treatments are available if injury occurs and, if so, what they consist of, or where further information may be obtained An explanation of whom to contact for answers to pertinent questions about the research and research subjects’ rights, and whom to contact in the event of a research-related injury to the subject A statement that participation is voluntary, that refusal to participate will involve no penalty or loss of benefits to which the subject is otherwise entitled, and that the subject may discontinue participation at any time without penalty or loss of benefits to which the subject is otherwise entitled

318

A.T. Farrell et al.

Table 12.6  Additional elements of informed consent A statement that the particular treatment or procedure may involve risks to the subject (or to the embryo or fetus, if the subject is or may become pregnant) which are currently unforeseeable Anticipated circumstances under which the subject’s participation may be terminated by the investigator without regard to the subject’s consent Any additional costs to the subject that may result from participation in the research The consequences of a subjects’ decision to withdraw from the research and procedures for orderly termination of participation by the subject A statement that significant new findings developed during the course of the research which may relate to the subject’s willingness to continue participation will be provided to the subject The approximate number of subjects involved in the study

r­ epresentative. If the latter form is used, there should be a witness to the oral presentation and the IRB should have approved a written summary of what is to be said to the subject or the representative. The CFR gives additional details for signatures on these forms and who receives copies of these forms [37]. In addition, the CFR provides additional safeguards for clinical investigations in children by requiring IRB review for clinical studies, requirements for permission by parents or guardians or for assent by children, and for children who are wards [38]. 12.2.4.4 Charging to Recover Cost For studies conducted under an IND, the Agency permits charging to recover cost for marketed and unmarketed agents in certain circumstances. However, charging for an investigational drug in a clinical trial under an IND is not permitted without the prior written approval of FDA. The CFR states, “In requesting such approval, the sponsor shall provide a full written explanation of why charging is necessary in order for the sponsor to undertake or continue the clinical trial, e.g., why distribution of the drug to test subjects should not be considered part of the normal cost of doing business” [39]. Charging cannot constitute commercialization of the agent. The FDA plans to amend the regulations regarding charging for agents to provide more clarity later this year. 12.2.4.5 Clinical Trial Monitoring (Phase 1, 2 vs. 3) For all studies conducted under an IND, federal regulations require that a sponsor or the sponsor’s designee (e.g., CRO) monitor the progress of a clinical investigation. These monitoring functions/duties include: selection of a monitor, establishing written monitoring procedures, pre-investigation visits, periodic visits, review of subject records for compliance with regulations, and maintenance of a record of on-site visits to ensure compliance with procedures and monitoring responsibilities [30]. These procedures assure the adequate protection of the rights of human subjects and the safety of all subjects involved in clinical investigations. Additionally,

12  Role of the US Food and Drug Administration in Cancer Drug Development

319

the FDA may, as a result of the NDA or for other reasons, inspect the clinical trial and its sites, and this information may be part of the audit. 12.2.4.6 Data Monitoring Committees Historically, data monitoring committees (DMCs) have been a component of some clinical trials since at least the early 1960s. DMCs were initially used primarily in large randomized multicenter trials. Current FDA regulations do not require the use of DMCs in trials except for research studies in emergency settings conducted under 21 CFR 50.24(a)(7)(iv), in which the informed consent requirement may be waived. Several factors may be considered when determining whether or not to establish a DMC for a particular trial. These relate primarily to safety, practicality, and scientific validity and include the risk to trial participants, practicality of review, particularly if the study is long term, and assurance of scientific validity [40]. The Agency has a draft guidance, which discusses a DMC’s composition, functioning, record keeping, and data analyses.

12.2.5 What Are the FDA’s Ongoing Responsibilities? After the initial submission of an IND, the Agency reviews the entire submission (i.e., chemistry, pharmacology/toxicology, clinical pharmacology, and clinical protocol) within 30 days after receipt. If the Agency does not agree that the study protocol is “safe to proceed,” i.e., the protocol is safe for clinical investigation, the initial IND may be placed on Clinical Hold. Typically, the decision to place an IND on clinical hold occurs after one or more internal meetings and after attempts to reconcile the Agency’s concerns with the sponsor’s proposal and data have been unsuccessful. The clinical hold order is usually made by the Division Director with responsibility for review of the IND. If the IND is placed on clinical hold, the Agency will inform the investigator or sponsor by telecommunication. The Agency sends formal communication to investigators and sponsors informing them of the basis for the hold and notifying them of what they need to do to get off hold. For studies or INDs placed on clinical hold, the investigation may only resume after FDA has notified the sponsor that the investigation may proceed. Resumption of the affected investigation(s) will be authorized when the sponsor corrects the deficiency(ies) previously cited or otherwise satisfies the agency that the investigation(s) can proceed. Typically, the FDA responds in writing to the sponsor within 30-calendar days of receipt of the complete response to the deficiencies and a request to resume studies. A sponsor may not proceed with a clinical trial on which a clinical hold has been imposed until notified by FDA that the hold has been lifted. An IND may be placed on clinical hold at any time for the following reasons. The grounds for clinical hold are (1) unreasonable or significant risk of injury or illness to the subject, (2) lack of qualified investigators, (3) misleading erroneous

320

A.T. Farrell et al.

or incomplete investigator’s brochure, and (4) insufficient data submitted to evaluate the risk. Most commonly, an IND is placed on hold during the early phase of clinical testing, if the Agency becomes aware that the product exposes subjects to unreasonable or significant risk of injury or illness. During later stages of drug development such as phase 2 and 3 testing, an IND may be placed on clinical hold if the protocol is deficient in design to meet its stated objectives.

12.2.6 End of Phase 2 Meetings The Agency has end of phase 2 (EOP2) meetings with sponsors following completion of the phase 2 study. The meetings can be discipline-specific or multidisciplinary, depending upon need, and may include topics such as additional information to support a marketing application (e.g., CMC, additional safety pharmacology/ toxicology testing, need for carcinogenicity testing), phase 3 trial design, and statistical analysis. These meetings are crucial as decisions made at this time may affect the acceptability of the phase 3 development program and eventual approval of the investigational agent for marketing.

12.2.7 Special Protocol Assessments The SPA program is a special program wherein the Agency reviews certain protocols specifically to evaluate issues (e.g., design, conduct, and analysis) related to the adequacy to meet scientific and regulatory requirements. Currently, protocols eligible for SPA include animal carcinogenicity study protocols, final product stability protocols, and clinical protocols for phase 3 trials whose data will form the primary basis for an efficacy claim. If the Agency agrees with the sponsor’s proposal, the Agency is making an agreement that the design and planned analysis of a study adequately address objectives in support of a regulatory submission. An SPA agreement is not necessary for a sponsor to submit the NDA for approval. The final determinations of the adequacy of the data collected from the study(ies) conducted under an SPA agreement are made after a complete review of a marketing application. Approval decisions are based on the entire data in the application.

12.3 Role in Marketing and Postmarketing Considerations for the design of studies intended to support approval, including appropriate endpoints, filing of the NDA, and postmarketing activities are based on a long history of landmark legislative decisions and regulatory experience. Landmark events and relevant terms are outlined in Table 12.1. Specific topics are discussed below.

12  Role of the US Food and Drug Administration in Cancer Drug Development

321

12.3.1 NDA Classification and Content An application to market a drug is termed the NDA. A complete NDA submission includes relevant information regarding chemistry and manufacturing, preclinical pharmacology and toxicology, clinical pharmacology, and clinical studies including relevant statistical analyses. Applicants must submit financial disclosure information about investigators [41]. In addition, sponsors are required to submit proposed labeling that provides users with information on the appropriate use of the drug in the intended patient population. When the NDA is submitted, it is classified as either a standard review, which triggers a 10-month review clock, or a priority review, which translates into a 6-month review from the date of submission, according to the Prescription Drug User Fee Act (PDUFA). Applications representing a significant improvement over marketed products are assigned a priority status. All other applications receive a standard designation. The review classification status has no relation to whether an application will get accelerated approval or regular approval. Review of the NDA involves a multidisciplinary team of chemists, toxicologists, clinical pharmacologists, medical officers, statisticians, microbiologists, site inspectors, and project managers. Reviewers evaluate primary data, usually submitted in electronic datasets along with case report forms as needed. Reviewers verify analyses performed by the applicant and conduct additional analyses when appropriate. In the case of NDAs and some supplemental NDAs, field inspectors verify that information provided in case report forms and electronic datasets is supported by source data such as hospital charts. In all cases, ascertainment of the integrity of the manufacturing process includes manufacturing site inspections.

12.3.2 General Efficacy Requirements The requirement for demonstration of drug safety was introduced in 1938 with passage of the FDC. The requirement for demonstration of effectiveness prior to marketing in the USA was codified in the 1962 Kefauver–Harris Drug Amendments to the FDC. With regard to the quantity and quality of clinical evidence needed to support effectiveness, FDA has generally relied on the submission of results from more than one well-controlled and well-conducted clinical trial. However, FDA has outlined the characteristics of a single adequate and well-controlled study that could make it adequate to support an effectiveness claim. These include having (a) a large ­multicenter study, (b) consistency in findings across study subsets, (c) multiple studies in a single study (such as a factorial design), (d) persuasive evidence of an effect in multiple endpoints, (e) statistically very persuasive findings [42].

322 Table 12.7  Regular approval and accelerated approval Regular approval Since 1962 amendments to the FDC Must demonstrate clinical benefit All disease settings Comparison to available therapy not required

Variable postmarketing commitments

A.T. Farrell et al.

Accelerated approval Since 1992 subpart H Surrogate endpoint reasonably likely to predict benefit Serious or life-threatening diseases Must demonstrate an advantage over available therapy or an effect in a population with no available therapy Confirmation of clinical benefit in ongoing or subsequent trials required

In the 1980s and early 1990s, the need to expedite development and approval of drugs for life-threatening diseases such as HIV/AIDS and cancer was recognized. In 1992, subpart H was added to the NDA regulations, allowing for accelerated approval (AA) of drugs for serious or life-threatening diseases, if a drug ­demonstrates an advantage over available therapy. Approval is based on a surrogate endpoint “reasonably likely to predict clinical benefit.” The sponsor must study the drug further to demonstrate clinical benefit in subsequent or ongoing clinical trials [43]. Table 12.7 provides a comparison of regular approval and accelerated approval.

12.3.2.1 Efficacy Endpoints for Approval in Oncology The following discussion outlines FDA’s general perspective on the role of ­specific efficacy endpoints in clinical trials designed to demonstrate safety and effectiveness [44]. Before discussing specific endpoints, it is important to put these into the context of study designs used in the oncology setting. Although single-arm trials are often very useful in evaluating drug activity, they usually have several limitations in their ability to provide evidence to support safety and effectiveness, which include small sample sizes, lack of a concurrent comparator, and inability to evaluate time-toevent endpoints. Although randomized trials may require relatively larger sample sizes and longer periods of time for completion, they offer several advantages including the ability to examine time-to-event endpoints such as overall survival (OS) or progression-free survival (PFS), a better characterization of safety, ­especially for events that occur due to the disease process or from prior therapies, corroboration of a drug effect through secondary endpoints, and evaluation of symptom improvement and health-related quality of life, especially if blinded. Overall survival (OS), defined as the time from randomization to death from any cause, is a time-to-event endpoint that requires a randomized setting. This endpoint is preferred for many reasons. Survival is assessed daily and is readily documented

12  Role of the US Food and Drug Administration in Cancer Drug Development

323

through direct patient contact or verbal contact by phone. The date of death can be confirmed easily and is independent of causality. The objective nature of this endpoint allows for interpretation in an unblinded setting, which is often the case in oncology trials where blinding is not always feasible. However, OS may require a substantial sample size and significant time intervals for study completion. Another potential limitation of OS is confounding of study results by therapies given after progression. PFS, defined as the time from randomization to progression or death, also requires a randomized setting for evaluation. In comparison to OS, PFS may require a smaller sample size and a shorter time interval for study completion, depending on the disease setting. In addition, PFS is not confounded by therapies introduced after progression, whether in a cross-over setting or off-study. However, the more subjective nature of this endpoint necessitates strategies to minimize bias such as blinding and/or independent review of the progression endpoint by a blinded review committee. Objective response rate (ORR) and duration of response are often used in oncology trials for the evaluation of activity. From a regulatory standpoint, durable responses are used to support approval only in certain circumstances. For example, durable remissions observed in acute leukemias may support regular approval as they are associated with reductions in morbidity as well as with prolongation of survival or disease-free survival. Durable responses in cancer settings where skin lesions represent a substantial burden for patients may also support approval, such as in cutaneous T-cell lymphoma [45]. In some cases, durable ORR has been used to support accelerated approval in advanced refractory solid tumors, with the confirmation of benefit in randomized trials using less refractory populations [46]. Table  12.8 from the Agency’s guidance on clinical trial endpoints for cancer drug development summarizes when to consider particular endpoints, and the advantages and disadvantages of using them [44].

12.3.3 Postmarketing Considerations Frequently, FDA and NDA sponsors come to an agreement on relevant postmarketing studies to be conducted and/or completed after marketing approval and target dates for submission to the Agency. The designs of these studies depend on their goals, which may include evaluation of potential drug–drug interactions, effects of organ impairment on dosing, or further elucidation of factors contributing to known adverse reactions. After approval, applicants must submit postmarketing reports of adverse events and annual reports including any significant new information identified in the past year. Under the accelerated approval provisions, applicants are required to confirm clinical benefit by the submission of additional data from studies designed to evaluate efficacy. Preferably, these would be ongoing trials. Whether they are ongoing trials or trials to be initiated after marketing approval, the design of confirmatory studies should be discussed with FDA as early in development as possible. In oncology, two strategies have been observed. One strategy is to seek and obtain AA

Surrogate for accelerated Randomized studies essential approval or regular Blinding preferred approvala Blinded review recommended

Disease-free survival

Randomized blinded studies

Clinical benefit for regular approval

Symptom endpoints (patient-reported outcomes)

Table 12.8  A comparison of important cancer approval endpoints [44] Endpoint Regulatory evidence Study design Overall survival Clinical benefit for Randomized studies essential regular approval Blinding not essential

Smaller sample size and shorter follow-up necessary compared with survival studies

Patient perspective of direct clinical benefit

Advantages Universally accepted direct measure of benefit Easily measured Precisely measured

Disadvantages May involve larger studies May be affected by crossover therapy and sequential therapy Includes noncancer deaths Blinding is often difficult Data are frequently missing or incomplete Clinical significance of small changes is unknown Multiple analyses Lack of validated instruments Not statistically validated as surrogate for survival in all settings Not precisely measured; subject to assessment bias, particularly in open-label studies Definitions vary among studies

324 A.T. Farrell et al.

 

Objective response rate

Surrogate for accelerated Single-arm or randomized studies approval or regular can be used approvala Blinding preferred in comparative studies Blinded review recommended

Can be assessed in single-arm studies Assessed earlier and in smaller studies compared with survival studies Effect attributable to drug, not natural history Can be assessed in single-arm studies Durable complete responses can represent clinical benefit Assessed earlier and in smaller studies compared with survival studies Smaller sample size and shorter follow-up necessary compared with survival studies Measurement of stable disease included Not affected by crossover or subsequent therapies Generally based on objective and quantitative assessment

Not a direct measure of benefit Not a comprehensive measure of drug activity Only a subset of patients who benefit Complete response Surrogate for accelerated Single-arm or randomized studies Not a direct measure of approval or regular benefit in all cases can be used approvala Not a comprehensive Blinding preferred in comparative measure of drug studies activity Blinded review recommended Small subset of patients with benefit Progression-free survival Surrogate for accelerated Randomized studies essential Not statistically validated approval or regular (includes all deaths) as surrogate for survival Blinding preferred approvala or time to progression in all settings Blinded review recommended (deaths before Not precisely measured; progression censored) subject to assessment bias particularly in open-label studies Definitions vary among studies Frequent radiological or other assessments Involves balanced timing of assessments among treatment arms a Adequacy as a surrogate endpoint for accelerated approval or regular approval is highly dependent upon other factors such as effect size, effect duration, and benefits of other available therapy

  12  Role of the US Food and Drug Administration in Cancer Drug Development 325

326

A.T. Farrell et al.

based on OR data from single arm trials in refractory patients with confirmation of benefit in one or more randomized trials of less refractory populations, looking at time-to-event endpoints. Another strategy is to rely on one or more randomized trials for AA based on a surrogate endpoint and confirmation of benefit in the same trials. The second strategy has the advantages of confirmation of benefit in the same patient population as that for AA, facilitated completion of confirmatory trials, and the direct comparison to available therapy and evaluation of toxicity profile that a randomized trial provides [46]. With the passage of the FDAAA in 2007, several provisions were added to existing law, including those relevant to the postmarketing setting. Title VIII of FDAAA adds a requirement that additional information regarding clinical trials including trial results be submitted to the clinical trials data bank (http://www.ClinicalTrials. gov) previously established by the National Institutes of Health(NIH)/National Library of Medicine(NLM). Title IX provides FDA with enhanced authority with regard to postmarketing safety, including the authority to require submission and implementation of a risk evaluation and mitigation strategy (REMS), if FDA determines that the REMS is necessary to ensure that the benefits of a drug outweigh its risks. This determination may be made prior to marketing to apply in the marketing setting, or it may be made postmarketing due to emerging safety information [47].

12.4 Other Regulatory Considerations Throughout the Development Cycle 12.4.1 Agency Use of Consultants The Agency employs external expert consultants as Special Government Employees (SGEs) to provide independent scientific advice during the evaluation of regulated products and to aid the Agency in making decisions based on reasoned application of good science [48]. These SGEs must undergo a clearance process for conflict of interest each time their expertise is requested. The Agency may utilize these consultants at any time during the product development process. Use of SGEs is the primary means by which the FDA obtains independent scientific advice. Most commonly, they are utilized for Advisory Committees (ACs). 12.4.1.1 Oncology Drugs Advisory Committee Since 1977, the Agency has used ACs in their present form to provide independent scientific advice. AC meetings may be open to the public or closed if there are privacy concerns or confidentiality issues including discussion of commercial or trade secrets or law enforcement investigations. Key to understanding the ACs is an understanding of their role with respect to the FDA. First, ACs are independent with respect to influence by either the product sponsor or the FDA. Second, ACs

12  Role of the US Food and Drug Administration in Cancer Drug Development

327

provide “expert scientific advice” because the committee members are ­acknowledged experts in their respective fields. Third, the AC advises the FDA; however, it lacks the authority to make decisions to obligate the Agency. Fourth, the AC often addresses specific questions drafted by the Agency’s professional staff [48]. Typically, AC membership includes approximately ten recognized clinicians, researchers, and statisticians for the specific field. The membership should have a goal of ethnic, gender, geographic, and racial diversity. Additional experts are added as necessary for a particular meeting. Usually, these members are voting members. For Oncology Drugs Advisory Committee (ODAC), additional committee members include: an industry representative who addresses general issues for the pharmaceutical industry and does not represent a specific commercial sponsor, a patient representative, and a consumer representative. While the patient representative and the consumer representative are usually voting members, the industry representative is a nonvoting member [49]. Once the Agency decides that an ODAC meeting is necessary. The Agency typically informs the sponsor, publishes a Federal Register Notice inviting the public, drafts a briefing document, writes questions, and a prepares presentation. During the meeting, the committee members listen to presentations, public comments, ask questions of the sponsor and the FDA, and discuss issues central to Agency concern. While the SGEs may recommend a particular course of action, the Agency is not obligated to follow that advice.

12.4.2 Diagnostic Tests Increasing knowledge regarding cancer pathogenesis now provides more opportunities for the development of drugs that target specific disease pathways or components of a pathway. Understanding these pathways also provides opportunities and challenges in defining patient populations using molecular diagnostic techniques that extend beyond our current armamentarium that often is restricted to clinical and histologic criteria. Targeted subgroups may be useful in selecting patients who may be more likely to benefit from a specific therapy or who may be at greater risk to develop toxicities. Targeting a subgroup also has the potential advantage of requiring a smaller sample size, if the treatment effect is magnified, and it may help to redefine a disease based on molecular criteria. However, there are possible disadvantages to targeting a subpopulation. This approach may exclude patients who would benefit due to an unrecognized MOA. In addition, it may limit the potential commercial market. The development of assays to define subgroups should be established early in drug development. Several trial designs may be applied to address these issues. In general, a retrospective discovery process where all patients are treated with drug and evaluated for efficacy would be considered exploratory. In this scenario, a retrospective comparison suggesting that responders have a higher frequency of a particular biomarker than nonresponders would require further evaluation in prospective randomized trials [50]. Several prospective designs are outlined in Table 12.9.

328

A.T. Farrell et al.

Table 12.9  Targeting subgroups prospectively Category Design elements Prospective, screened Only patients who are target positive are enrolled and a comparison of drug versus control in a twoarm randomized setting is undertaken Prospective, stratified

Prospective, stratified, explicit study of unselected population

Comments Assumes no benefit in the marker-negative population; the target assay must be available at the time the drug is approved Allows an estimate of All patients are tested. efficacy and safety Both target negative and in marker-positive target positive patients and marker-negative are randomized to drug populations; allows versus control (stratified determination of randomization) utility of the target assay; could include interim analysis in the marker-negative subgroup with dropping of this arm, if there is inadequate evidence of efficacy Provides a comparison One group undergoes marker in a controlled testing with stratified environment where randomization as above and efficacy and safety another group is not tested in a marker-tested but simply randomized to population as well as drug versus control a broader population is possible; could be useful if the assay is not widely available and the therapy is relatively nontoxic

Regulatory review of assays is performed at FDA by the Center for Devices and Radiologic Health (CDRH). We encourage investigators and commercial sponsors to consult CDRH about assay development as early in development as possible, either during meetings with the Office of Oncology Drug Products or in separate discussions with CDRH.

12.4.3 Orphan Drug Program In January 1983, the Orphan Drug Act was signed into law with the intention of stimulating research, development, and approval of products that treat rare diseases [51]. “Orphan drug” is a product that treats a rare disease affecting fewer than

12  Role of the US Food and Drug Administration in Cancer Drug Development

329

200,000 people [52]. The Orphan Drug Act has provisions to achieve the goals of the Act. These provisions include (1) sponsors are granted 7  years of marketing exclusivity after approval of its orphan drug product, (2) sponsors also are granted tax incentives for clinical research they have undertaken, study design assistance for sponsors of drugs for rare diseases, and grant funding is available to defray costs of qualified clinical testing expenses incurred in connection with the development of orphan products. The regulations cited above outline the steps for applying for orphan drug status. The Orphan Drugs Product Development program has a website which gives further details regarding the grants program [53].

12.4.4 Pediatric Initiatives Similar to adult oncology drug development, three phases of clinical trials are usually conducted in the development of new treatments for pediatric malignancies. Phase I trials evaluate doses and schedules appropriate for further development. Phase II trials are designed to assess antitumor activity, and phase III trials explore new treatment approaches or additions/substitutions that may improve efficacy or reduce toxicity. However, the small pediatric oncology market compared with the adult oncology market makes pediatric drug development less financially attractive and likely contributes to the lack of pediatric labeling information for anticancer drugs. In order to encourage evaluation of new therapies in pediatric populations and submission of pediatric data to support labeling of commercially marketed drugs, the FDA has undertaken two initiatives that can be summarized as follows. A voluntary incentive program was outlined in the Food and Drug Modernization Act (FDAMA) of 1997. Under this program, a commercial sponsor may receive a 6-month extension of existing exclusivity by submitting data from pediatric studies in support of the NDA or a labeling supplement. The design of studies to be conducted and data to be submitted are described in a pediatric written request letter (WR) issued by FDA. The Best Pharmaceuticals for Children Act (BPCA) subsequently provided modifications to the program, including a mechanism for making summaries of medical and clinical pharmacology reviews available to the public [54]. It established an Office of Pediatric Therapeutics and defined the membership of a Pediatric Subcommittee of the Oncology Drugs Advisory Committee. Most recently, the Food and Drug Amendments Act (FDAAA) of 2007 provided further refinements [47]. It established an internal Pediatric Review Committee (PeRC) to provide oversight of pediatric activities within CDER and CBER. It mandated that all submissions in response to a written request receive priority review. It allowed the inclusion of preclinical studies as part of the terms of the WR (clinical studies would still be required). It mandated that WRs be made public. Finally, it mandated that labeling must include information from data submitted in response to the WR, regardless of the study results.

330

A.T. Farrell et al.

A mandatory program applies to indications where the disease in adults is similar to the adult disease. Originally described in the “Pediatric Rule” and subsequently legislated in the Pediatric Research Equity Act (PREA), this program allows the FDA to require sponsors to submit data from pediatric studies when an adult indication is granted, based on the specific drug and indication under review [55]. In order to avoid a potential delay in development and in access to treatments for lifethreatening diseases, a deferral for the submission of the pediatric data can be granted. This legislation applies to drugs and biologics. Products with orphan designation are excluded from this requirement. In 2007, refinements to PREA where included in the FDAAA legislation. These include a requirement that results of pediatric studies be included in the labeling and that pediatric adverse events be reviewed by the pediatric advisory committee. For both BPCA and PREA, the new sunset date is October 1, 2012.

References 1. FDA Mission statement at http://www.fda.gov/opacom/morechoices/mission.html (accessed 6/10/08) 2. Chapter 21 Code of Federal Regulations (CFR) 312 3. 21 CFR 312.2 4. 21 CFR 56 5. 21 CFR 50 6. 21 CFR 312.7 7. Guidance for Industry, IND Exemptions for Studies of Lawfully Marketed Drugs or Biological Products for the Treatment of Cancer, issued January 2004. http://www.fda.gov/cder/ Guidance/6036fnl.pdf (accessed 11/18/08) 8. 21 CFR 312.23 9. Guidance for Industry, Investigators, and Reviewers: Exploratory IND Studies, issued January 2006. http://www.fda.gov/CDER/guidance/7086fnl.htm (accessed 11/28/08) 10. Innovation or Stagnation, Challenge and Opportunity on the Critical Path to New Medical Products, March 2004 at http://www.fda.gov/oc/initiatives/criticalpath/whitepaper.html (accessed 6/17/08) 11. 21 CFR 312.23 (a)(7) 12. Farrell AT, Leighton J, Williams G, and Pazdur R. How Oncology Drug Development Differs from Other Fields. In Handbook of Anticancer Drug Development. Budman DR, Calvert AH, Rowinsky EK (eds). Lippincott Williams & Wilkens, Baltimore, MD, 2003. pp. 3–10 13. International Conference on Harmonization Draft Consensus Guideline Document S9 Nonclinical Evaluation for Anticancer Pharmaceuticals, November 2008. http://www.ich.org/ LOB/media/MEDIA4917.pdf (accessed 12/2/08) 14. International Conference on Harmonization Document M3 “Guidance for Industry: Nonclinical Safety Studies for the Conduct of Human Clinical Trials for Pharmaceuticals.” July 1997. http://www.fda.gov/CDER/guidance/1855fnl.pdf (accessed 11/28/08) 15. United States Department of Health and Human Services, Food and Drug Administration. “Single Dose Acute Toxicity Testing for Pharmaceuticals; Revised Guidance.” Federal Register 1996 August 26; 61(166): 43934–43935 16. DeGeorge JJ, Ahn CH, Andrews P, Brower M, Gorgio D, Goheer MA, Lee-Ham DY, McGuinn WD, Schmidt W, Sun CJ, and Tripathi S. Regulatory Considerations for Preclinical Development of Anticancer Drugs. Cancer Chemotherapy and Pharmacology 1998; 41: 173–185

12  Role of the US Food and Drug Administration in Cancer Drug Development

331

17. The European Agency for the Evaluation of Medicinal Products Committee for Proprietary Medicinal Products: Note for Guidance on Repeated Dose Toxicity. July 2000. http://www. emea.europa.eu/pdfs/human/swp/104299en.pdf (accessed 11/28/08) 18. The European Agency for the Evaluation of Medicinal Products Committee for Proprietary Medicinal Products: Note for Guidance on Non-clinical Local Tolerance Testing of Medicinal Products. March 2001. http://www.emea.europa.eu/pdfs/human/swp/214500en.pdf (accessed 11/28/08) 19. International Conference on Harmonization Document S2A Document “Guideline for Industry: Specific Aspects of Regulatory Genotoxicity Tests for Pharmaceuticals.” April 1996. http://www.fda.gov/cder/guidance/ichs2a.pdf (accessed 11/28/08) 20. International Conference on Harmonization Document S1A Document “Guideline for Industry: The Need for Long-term Rodent Carcinogenicity Studies of Pharmaceuticals.” March 1996. http://www.fda.gov/Cder/guidance/ichs1a.pdf (accessed 11/28/08) 21. International Conference on Harmonization Document S6 “Guidance for Industry Preclinical Safety Evaluation of Biotechnology-Derived Pharmaceuticals.” July 1997. http://www.fda. gov/cder/guidance/1859fnl.pdf (accessed 11/28/08) 22. International Conference on Harmonization Document S1B Document “Guidance for Industry: Testing for Carcinogenicity of Pharmaceuticals.” July 1997. http://www.fda.gov/ Cder/Guidance/1854fnl.pdf (accessed 11/28/08) 23. International Conference on Harmonization Document S1C Document “Guideline for Industry: Dose Selection for Carcinogenicity Studies of Pharmaceuticals.” March 1995. http:// www.ich.org/cache/compo/502-272-1.html (accessed 11/28/08) 24. International Conference on Harmonization Document S1C (R) Document “Guidance for Industry: Addendum to Dose Selection for Carcinogenicity Studies of Pharmaceuticals: Addition of a Limit Dose and Related Notes.” March 1995. http://www.ich.org/cache/ compo/502-272-1.html (accessed 11/28/08) 25. The European Agency for the Evaluation of Medicinal Products Committee for Proprietary Medicinal Products: Note for Guidance on Carcinogenic Potential. July 2002. http://www. emea.europa.eu/pdfs/human/swp/287700en.pdf (accessed 11/28/08) 26. International Conference on Harmonization Document S7A “Guidance for Industry: Safety Pharmacology Studies for Human Pharmaceuticals.” July 2001. http://www.fda.gov/Cder/ Guidance/4461fnl.pdf (accessed 11/28/08) 27. International Conference on Harmonization Document S7B “The Nonclinical Evaluation of the Potential for Delayed Ventricular Repolarization (QT Interval Prolongation) by Human Pharmaceuticals.” October 2005. http://www.fda.gov/cder/Guidance/6885fnl.htm (accessed 11/28/08) 28. International Conference on Harmonization Document E14 “The Clinical Evaluation of QT/ QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs.” May 2005. http://www.fda.gov/cder/guidance/6922fnl.htm (accessed 11/28/08) 29. 21 CFR 312.3 30. International Conference on Harmonization “Guidance for Industry E6 Good Clinical Practice Consolidated Guidance.” April 1996. http://www fda.gov/cder/guidance/959fnl.pdf (accessed 5/6/08) 31. 21CFR 312.32 32. 21CFR 312.33 33. 21CFR 50.23 34. 21CFR 50.24 35. 21CFR 50.20 36. FDA Informed Consent information at http://www.fda.gov/oc/ohrt/IRBs/informedconsent. html#general (accessed 6/17/08) 37. 21CFR 50.27 38. 21CFR 50. 50-56 39. 21CFR 312.7 (d)(1) 40. Guidance for Clinical Trial Sponsors on the Establishment and Operation of Clinical Trial Data Monitoring Committees, May 8, 2002. http://www.fda.gov/cber/gdlns/clindatmon.htm (accessed 5/6/08)

332

A.T. Farrell et al.

41. Guidance for Industry: Financial Disclosure by Clinical Investigators; issued March 2001. http://www.fda.gov/oc/guidance/financialdis.html (accessed 11/28/08) 42. Guidance for Industry: Providing Clinical Evidence of Effectiveness for Human Drug and Biological Products, May 1998. http://www.fda.gov/CDER/GUIDANCE/1397fnl.pdf (accessed 11/28/08) 43. 21 CFR, Parts 314.500 to 314.530 44. Guidance for Industry: Clinical Trial Endpoints for the Approval of Cancer Drugs and Biologics; issued May 2007. http://www.fda.gov/CbER/gdlns/clintrialend.htm (accessed 11/28/08) 45. Mann BS, Johnson JR, He K, Sridhara R, Abraham S, Booth BP, Verbois L, Morse D, Jee JM, Pope S, Harapanhalli R, Dagher R, Farrell A, Justice R, and Pazdur R. Vorinostat for Treatment of Cutaneous Manifestations of Advanced Primary Cutaneous T-Cell Lymphoma. Clinical Cancer Research 2007; 13(8): 2318–2322 46. Dagher R, Johnson J, Williams G, Keegan P, and Pazdur R. Accelerated Approval of Oncology Products: A Decade of Experience. Journal of the National Cancer Institute 2004; 96: 1500–1509 47. Food and Drug Administration Amendments Act of 2007, Public Law 110–85 48. Farrell AT, Papadouli I, Hori A, Harczy M, Harrison B, Asakura W, Marty M, Dagher R, and Pazdur R. The Advisory Process for Anticancer Drug Regulation: A Global Perspective. Annals of Oncology 2006 June; 17(6):889–996 49. Rettig RA, Earley LE, and Merrill RA (eds). Institute of Medicine: Food and Drug Administration Advisory Committees. National Academy Press, Washington, DC, 1992 50. Dagher RN and Pazdur R. Chapter 18: Clinical Trial Design and Regulatory Issues. In Antiangiogenic Cancer Therapy. Abbruzzese JL, Davis D, and Herbst R (eds). Taylor and Francis, Boca Raton, 2007 51. 21CFR 316 52. 21CFR 316.20 (b)(8)(i,ii) 53. Orphan Drugs Product Development at http://www.fda.gov/orphan/index.htm 54. Best Pharmaceuticals for Children Act, amended to the Federal Food, Drug and Cosmetic Act, Public Law 107–109, 1789, 2002 55. Pediatric Research Equity Act, Public Law 108–155, 2003

Part V

Chapter 13

Early Clinical Trials with Cytotoxic Agents M.J.A. de Jonge and Jaap Verweij

Abbreviations PD PK

Pharmacodynamics Pharmacokinetics

13.1 Introduction The principles of phase I studies in oncology have been extensively outlined in Chap. 5. The systemic treatment of cancer has long been based on cytotoxic drugs, and, more recently, it has been extended with the use of drugs aiming at a molecular aberration in the cancer cell. The principles of cytotoxic drug treatment itself are outlined elsewhere. But as a consequence of these principles, cytotoxic drugs are commonly given in so-called cycles. This, in turn, is due to the fact that the effect of cytotoxic drugs is limited not only to the cancer cell itself, but also affects normal cells in the body. The cytotoxic drug effect mainly has a differential between these normal cells and the cancer cell in the capability of these cells to recover. This means that we have to allow the normal cells to first recover from the drug effects, before a new dose can be given. This recovery phase thus dictates intervals between drug administrations. This contrasts with the daily administration of the newer noncytotoxic classes of drugs. For monoclonal antibodies, their long half-life results in long-term exposure after a single administration, which mimics the daily exposure resulting from daily administration of synthetic small molecules. So, the elimination half-life is the reason why treatment with monoclonal antibodies resembles the cyclic administration of cytotoxic agents. This cycling leads to important, albeit sometimes subtle, differences

M.J.A. de Jonge (*) Department of Medical Oncology, Erasmus University Medical Center (Daniel den Hoed Kliniek), P.O. Box 5201, 3008 AE, Rotterdam, The Netherlands e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_13, © Springer Science+Business Media, LLC 2011

335

336

M.J.A. de Jonge and J. Verweij

Table 13.1  Major differences in study design between cytotoxic and noncytotoxic agents Cytotoxics Noncytotoxics Exposure Intermittent Continuous Recommended dose phase II Highest possible Optimal biological Correlative studies PK (PD) PD (PK) Progression free rate Efficacy Response rate Growth modulation index Progression free rate Early progression rate Growth modulation index Early progression rate PK pharmacokinetics, PD pharmacodynamics

in protocol descriptions in early clinical studies. This chapter highlights the areas where trials in the early development of cytotoxics may be different (Table 13.1).

13.2 Starting Dose and Schedule of Administration The same principles apply to the determination of the starting dose for cytotoxic drugs and other drugs. The essence is a balance between consideration of the desired effects and the possible side effects, and the minimization of the risk for patients [1]. However, for drug administration schedules, there are differences. For cytotoxic drugs, the principle derived from their mechanism of action, and described above, is that there is an effect on both the cancer cell as well as some normal cells. The differential is usually that the effect on the cancer cell is more extensive, and/or the capability for recovery is better in the normal cells. It is beyond the scope of this chapter to further explain this. Yet, the fact that recovery is important means that some time without drug exposure has to be built in. Terms such as drug interval or drug holiday are used to describe this. Classical administration schedules of cytotoxic drugs, based on the above principles, are one administration per 3–4 weeks in which case the cycle length also is defined as lasting 3 or 4 weeks. An example is the administration of docetaxel once every 3 weeks [2]. Weekly administrations are also commonly applied, mostly followed by one or multiple weeks without drug administration, in which case the total cycle length is defined by the period of drug administration plus the period without this, prior to repetition. For example, if the drug is given weekly twice, but skipped in the third week, the cycle length would be 3 weeks (2 + 1), or if 4 weekly administrations are followed by 2 weeks without administration, this would determine the cycle length at 6  weeks (4 + 2). Examples of weekly regimens are gemcitabine administered weekly for 3 out of every 4 weeks or docetaxel administered for 4 out of every 6 weeks [3, 4]. Another frequently explored schedule is a short-term or prolonged daily administrations, especially applied for agents with short half-lives with a time-dependent mechanism of action needing prolonged exposure to fully exhibit their action. Depending on the pharmacologic characteristics of the agent involved one could for instance administer it 3 or 5 days in a row, followed by 18 or 16 days without treatment respectively, both resulting in cycle lengths of 21 days and thus 3 weeks. For

13  Early Clinical Trials with Cytotoxic Agents

337

instance, topotecan is one of the agents that has been developed for patients with ovarian cancer in a regimen administering drug daily over 30  min on days 1–5, every 3 weeks [5]. To enable the best possible choice of schedules for clinical testing, we have to base this on similar experiments in animals. When doing so, we have to take into account the fact that drug-elimination half-lives in animals, as well as tumor growth characteristics in animal models, are frequently considerably different from those in human beings. Yet, it is possible to use the information from the animal models for this purpose.

13.3 Dose Escalation Methods The way how to escalate doses in phase I studies involving cytotoxic drugs in essence is not different from the methods used for other drugs. The reader is thus referred to Chap. 5 for information on this topic. The side effects expected are based on the information from animal models. But, for the sake of completeness, one should always also consider dose-limiting toxicity definitions for damage to organ systems that are not expected to be affected by the drug. Given their mechanism of action, cytotoxic drugs are still dosed at the highest possible dose where tolerance is still considered acceptable. This means that the study protocol should specify the definitions of so-called dose-limiting toxicity. In this respect, again the fact that there will be a recovery phase should be taken into account. This is different from studies with other classes of agents where drug administration is mostly daily and as a consequence no recovery of side effects is built in. This has an impact on the maximal grade of toxicity that is considered to be acceptable during a defined treatment period. Because a recovery period is essential for cytotoxic agents , this will also imply that more severe toxicity can be accepted for short periods of time. However, as targeted agents tend to be administered orally on a continuous basis, the definition of toxicity which is considered dose-limiting should be adjusted accordingly. For these drugs, less severe toxicity, for instance grade 2 gastrointestinal toxicity, might even be considered doselimiting. With the newer agents involved, it has become questionable if one should still continue to aim for the maximal tolerated dose (MTD), or rather aim for something such as the optimal biological dose (OBD). But, this discussion, in particular, is not relevant to the development of cytotoxic agents. Related to the definition of dose-limiting side effects is the definition of the dose for further studies. There are subtle differences in the definition of the so-called maximal tolerated dose (MTD). This is the dose that either is recommended for further studies (US definition), or the dose that in principle is leading to unacceptable, yet reversible side effects (European definition). Whatever the definition, in phase I studies with cytotoxic agents, one basically aims to identify the dose with unacceptable, yet reversible toxicity, and then recommends the dose below that dose, as assessed in the phase I study. In the following section, the US definition of MTD, basically indicating the dose(s) recommended for further studies, is used.

338

M.J.A. de Jonge and J. Verweij

13.4 Correlative Studies in Clinical Trials with Cytotoxic Agents Correlative studies in a clinical study are aiming to describe either the pharmacokinetics (simplified as describing what the body does to drug) or the pharmacodynamics (simplified as describing what the drug does to the body). The desired effects of tumor size reduction, or even better prolonged survival, are long-term effects that can take a long time before being achieved. Particularly to speed up drug development and the drug-approval process, investigators have been searching for so-called surrogate markers of activity, which in essence are also pharmacodynamic parameters. It is of crucial importance in general to include pharmacokinetic assessment into phase I study of any agent, but given its side-effect profiles, this certainly holds for cytotoxics. In general, the phase I study programme is the time and place where the pharmacokinetics of the cytotoxic drug should become understood [6]. The pharmacokinetic profile may lead one to consider changes in dose or schedule of drug administration [7, 8]. In order to enable this, it is of utmost importance to ensure that the data become available in a real-time way, in other words, while the study is running and decisions need to be made [6]. Certainly for cytotoxic drugs, a nonlinear pharmacokinetic behaviour would be reason for extra caution as, in theory, this increases the safety risks of the participating patients. As a rule of thumb, one could say that the turn-around time for pharmacokinetic data should be no more than the interval of dose-escalation decisions. Pharmacokinetics may also be important in the explanation of metabolism and elimination, both of which are elements that guide decisions. Obtaining information on pharmacodynamic parameters, the read out of drug effect, may be very helpful in explaining the drug’s mechanism of action but will likely be less important to guiding the performance of the study [6, 9, 10]. In other words, the biomarker assessment that has become an integral (but possible overestimated) part of phase I studies, with agents targeting specific molecular aberrations, is less important to guide the development of a cytotoxic. This is mainly related to the lack of cell specificity of the cytotoxic agents. As a consequence of their mechanism of action and lack of cell specificity, cytotoxics will all cause in some degree myelosuppression and/or mucositis. Although not sufficient for antitumor activity, the lack of mechanism-based toxicity will caution further development of a cytotoxic drug.

13.5 Clinical Trials Combining Cytotoxic Agents When combining cytotoxic drugs in phase I studies, a few complexities arise that require attention in the protocol.

13  Early Clinical Trials with Cytotoxic Agents

339

13.5.1 Starting Dose(s) Most often, a new drug is added to a drug of which the details, particularly the side effect and activity profiles, are already well known. If the “older” drug has known antitumor activity, compromising on the dose required for activity is obviously illogical. In this scenario, one starts using either the standard single agent dose of the known cytotoxic or a slightly lower dose still known to yield activity but ­inducing less toxicity, and adds the new agent. The choice of the dose of the new agents is dependent on the side-effect profile obtained from single-agent phase I studies, the recommended dose from those studies, and the possible overlap in toxicity from both of the agents in the combination [11–14].

13.5.2 Dose Escalation A second important consideration is how and which drug to escalate. There are two possible scenarios. One would be to first escalate the known agent to the standard dose, if dosing has started below that standard dose. Once this standard dose is reached, the dose escalation proceeds by escalating the dose of the new agent, until the MTD is reached. The alternative is to start by escalating the dose of the new agent and subsequently escalating the dose of the known agent. Finally, a complex scenario of escalating both in alternating fashion is conceivable. As a consequence, it is also conceivable that multiple MTDs of a combination are determined within the context of a single study (Fig. 13.1), and as we will not know the relevance of dose or dose intensity of each of the drugs in the combination, this identification of multiple MTDs suggests that subsequent randomized phase II studies should explore the different recommended dosages (Fig. 13.2). A similar design could also be used for identification of the best schedule of administration in case of a single agent [15]. 100 90 80 70 60 50 40 30 20 10 0

A

B

C

A

B DRUG

Fig. 13.1  Multiple MTDs conceivable

C

A

B

C

340 Fig. 13.2  Multiple conceivable MTDs invite to a randomized phase II study to find the best combination dose

M.J.A. de Jonge and J. Verweij

STANDARD R A N D O M I Z E

MTD 1

MTD 2

13.5.3 DLT Definition A major problem for the phase I combination study with cytotoxics is the definition of DLT, or, even more, the contribution of each of the agents to this side effect observed. As a consequence, the type of side effect as well as its rate of occurrence has to be carefully balanced against the drugs involved in the combination under study. This means that it is conceivable that the commonly used criteria for determining MTD may or may not apply in a phase I study with a combination of cytotoxics. For instance, if we combine a new agent with docetaxel, a drug that is known to produce a single-agent febrile neutropenia rate of 20–25% per cycle and 25–35% per patient [16], one would have to assess chance observations of neutropenic fever against real consequences of a pharmacodynamic drug–drug interaction in the combination involved. Also, one would have to carefully consider if the rule of accepting only one DLT out of six patients (16%) to determine the recommended dose for further study would apply. The latter percentage would, in this example, be lower than the current practice of docetaxel and would, thus, inevitably lead to a required dose reduction. This would not make sense as, for the single agent ­docetaxel, we would then be accepting more toxicity than for the combination under study. So, specific rules will have to be defined by protocol, and these definitions would have to be based on the specifics of the side effects anticipated. And, in general, it may be advantageous to use six patient cohorts in the dose escalation part of the phase I study combination, and 12 in the expansion cohort. Another way to try to rule out a chance observation is to introduce a randomization in the phase I study as proposed by Ratain (personal communication). For each three patient cohort, one extra (fourth) patient could be added. The extra patient could be selected by randomization and would be scheduled to receive the standard dose of a combination or the agent to which the new drug is added. Patients randomized to this standard could be pooled throughout the study and serve as an internal control for assessing the contribution of toxicity.

13  Early Clinical Trials with Cytotoxic Agents

341

In defining the DLT, it is also very important to take the type of side effect into account. Alopecia in this respect may, for instance, be judged differently from thrombocytopenia.

13.5.4 Pharmacokinetics and Drug–Drug Interaction The relevance of obtaining pharmacokinetic information during the course of a phase I study has already been discussed and also applies to the phase I study of a combination. An extra element of developing a combination is the fact that one also has to investigate potential drug–drug interactions (DDI) at the pharmacokinetic level [12–14]. The drug–drug interaction study can frequently be integrated into the first phase I study of the combination [6]. Again, obtaining the PK information in a realtime way is important as it could guide changes in study design and study performance. The DDI could be exploited and thus be considered positive, or result in unacceptable side effects and be considered negative. DDI might also depend on the scheduling of the drugs, which should be evaluated during the phase I study.

13.6 Early Efficacy-Based Trials of Cytotoxic Agents Phase II studies are designed to evaluate the activity of anticancer agents and, to be more precise, to exclude inefficient drugs from further development. In recent years, there has been a tendency to include an expansion cohort at the end of a phase I study. Although such a cohort might provide additional safety data, these cohorts are too small and the patient population too heterogeneous to replace a formal phase II study. In most phase II studies with cytotoxic agents, the activity is evaluated using the Response Criteria In Solid Tumors (RECIST) [17, 18]. These criteria evaluate onedimensional tumor measurement in a selection of target lesions with a minimum size. Criteria in overall changes in tumor dimensions are set to define tumor response. In several tumor types, RECIST are more difficult to apply, even with cytotoxic drug trials. For instance, Byrne et al. proposed in mesothelioma to use the longest perpendicular diameter of the pleural mass to chest wall or mediastinum measured at three different levels on CT scans [19]. Responses defined according to these modified criteria resulted in longer survival and improved pulmonary function. The evaluation of responses to treatment in brain tumors is based on major changes in tumor size on the enhanced CT or MRI scan. These changes are interpreted in light of steroid requirements and neurological findings [20]. As steroids by themselves improve signs and symptoms, maintain clinical improvements for prolonged periods, and even substantially decrease the size of some malignant gliomas on CT scans, their use must be considered in response assessment. Prostate cancer metastases are also well known to be difficult to measure by imaging. Most

342

M.J.A. de Jonge and J. Verweij

metastatic patients present with bone lesions only, while truly measurable lesions according to RECIST are only present in approximately 30–50% of all patients [7]. In most phase II studies in prostate cancer, other means to evaluate benefit of treatment are incorporated, including PSA response, pain response, and quality of life assessments [21]. Historically, response rate has been used as the most important end point in phase II trials. And this certainly applies for cytotoxic agents. The advantage of this approach is that response is relatively easily measurable. The disadvantages, however, are that response rate does not take into account the duration of response and the percentage of patients that achieve durable stable disease on treatment. For example, the average response rate on DTIC in malignant melanoma is approximately 20%, however, with a duration of only 10  weeks [22]. Thus, this agent should not be considered active in melanoma. In several tumor types, it was noted that, even with cytotoxic agents, the prognosis of patients achieving an objective tumor response and patients with durable disease stabilization was comparable, indicating that achieving disease stabilization might be valuable for the patient and also be an indication of antitumor activity of the drug. However, several cytotoxic drugs have been discarded only because of a low response rate. In addition, in phase III studies, the final proof of efficacy of a drug, response rate is no longer relevant and the primary end point becomes progression free or overall survival. An alternative end point to consider in phase II studies might be the “growth modulation index”. It is based on the assumption that the new agent should change the natural course of the disease, which could also apply to cytotoxic agents. The index is defined as the ratio of the time to progression (TTP) on the current treatment (TTP2, the agent on study) and the TTP of the previous treatment period (TTP1, conventional treatment). It is postulated that a 33% increase of TTP is proof of activity. Therefore, a ratio >1.33 is considered to be indicative of the new drug having antitumor activity [23]. Possible limitations to this approach could be the frequent lack of prestudy tumor-measurement data, as well as the fact that the increase of 33% in TTP is chosen arbitrarily. Another alternative end point in phase II clinical testing could be assessment of the progression free rate (PFR) [24]. When using this end point, one would have to define the appropriate target for progression-free survival indicative of activity or inactivity for the tumor type under investigation, based on historical data. Also, the time point at which the PFR will be assessed is of importance. For slowly growing tumors, determining the PFR at 4 months, for example, can be misleading as in this time frame, the natural course of the disease might not produce a measurable progression. Also, stable disease cannot be considered as evidence of treatment activity if no documented data on disease progression was available prior to start of treatment. Thus, only truly progressive patients should be entered in a phase II trial where PFR is an end point. As an example, based on outcomes of former trials, the Soft Tissue Sarcoma group of the EORTC established reference values for the PFR for patients with metastatic soft tissue sarcoma, both patients pretreated with cytotoxic drugs and chemo-naïve patients providing cut-off values (with a standard

13  Early Clinical Trials with Cytotoxic Agents

343

error of approximately 5%) that can be used in the statistical evaluation of future phase II trials [25]. Drugs that show a high proportion of early disease progression can hardly be of importance for further testing. Low rates of early disease progression, on the other hand, may be a sign of anticancer drug activity, even if this is accompanied by disease stabilization and a limited response rate. In this specific phase II design, both response rate and early progression rate are incorporated in decision rules to define a potentially active or inactive drug [26–28]. Again, dependent on tumor type, stage, and pretreatment, different reference matrices need to be set. When screening for antitumor activity of a combination of agents, certainly when one (or more) of the involved agents has known antitumor activity, one is faced with the problem of having to take into account this background activity when assessing the combination of a given combination. This could be done either by performing a single arm study with a very large number of patients or by incorporating a randomization into the study design [29]. The single arm study by definition will be unable to exclude even unintended selection bias and thus by definition is not appropriately interpretable for outcome. As a consequence, we prefer the use of randomization in the phase II development of a combination of agents. This takes away the possibility of selection bias, and frequently even limits the required number of patients as compared to the nonconclusive single arm study. Obviously, when using the randomized phase II study design, one should realize that the power of the observation is still limited and that such a design can never serve as an underpowered phase III study. The outcome of the study should enable correct statistical estimates for the design of the subsequent phase III trial.

13.7 Conclusions Early clinical trials in the drug development of cytotoxic agents bear great resemblance to those of drugs with other mechanisms of action, but also harbour differences. These differences can be found in the cycling aspect of drug administration, the various schedules of drug administration, the way drug doses are escalated and dose-limiting toxicities are defined, and the way activity is assessed. In designing these studies, the specifics of the cytotoxic agent will have to be taken into account.

References 1. Arbuck SG: Workshop on phase I study design. Ninth NCI/EORTC New Drug Development Symposium, Amsterdam, March 12, 1996. Ann Oncol 1996; 7: 567–573. 2. Chan S, Friedrichs K, Noel D, Pintér T, Van Belle S, Vorobiof D, Duarte R, Gil Gil M, Bodrogi I, Murray E, Yelle L, von Minckwitz G, Korec S, Simmonds P, Buzzi F, González Mancha R, Richardson G, Walpole E, Ronzoni M, Murawsky M, Alakl M, Riva A, Crown J:

344

M.J.A. de Jonge and J. Verweij

Prospective randomized trial of docetaxel versus doxorubicin in patients with metastatic breast cancer. J Clin Oncol 1999; 17: 2341–2354. 3. Anderson H, Lund B, Bach F, Thatcher N, Walling J, Hansen HH: Single-agent activity of weekly gemcitabine in advanced non-small-cell lung cancer: a phase II study. J Clin Oncol 1994; 12: 1821–1826. 4. Burstein HJ, Manola J, Younger J, Parker LM, Bunnell CA, Scheib R, Matulonis UA, Garber JE, Clarke KD, Shulman LN, Winer E: Docetaxel administered on a weekly basis for metastatic breast cancer. J Clin Oncol 2000; 18: 1212–1219. 5. Gore M, ten Bokkel Huinink W, Carmichael J, Gordon A, Davidson N, Coleman R, Spaczynski M, Héron JF, Bolis G, Malmström H, Malfetano J, Scarabelli C, Vennin P, Ross G, Fields SZ: Clinical evidence for topotecan-paclitaxel non-cross-resistance in ovarian cancer. J Clin Oncol 2001; 19: 1893–1900. 6. Verweij J: “No risk, no fun”: challenges for the oncology phase I clinical trial time-­ performance. Eur J Cancer 2008; 44(17): 2600–2607. 7. LoRusso PM, Rudin CM, Borad MJ, Vernillet L, Darbonne WC, Mackey H, DiMartino JF, de Sauvage F, Low JA, Von Hoff DD: A first-in-human, first-in-class, phase (ph) I study of systemic Hedgehog (Hh) pathway antagonist, GDC-0449, in patients (pts) with advanced solid tumors. J Clin Oncol 2008; 26(suppl): 3516. 8. Soepenberg O, Sparreboom A, de Jonge MJ, Planting AS, de Heus G, Loos WJ, Hartman CM, Bowden C, Verweij J: Real-time pharmacokinetics guiding clinical decisions: phase I study of a weekly schedule of liposome encapsulated paclitaxel in patients with solid tumours. Eur J Cancer 2004; 40: 681–688. 9. El-Maraghi LH, Eisenhauer EA: Review of phase II trial designs used in studies of molecular targeted agents: outcomes and predictors of success in phase III. J Clin Oncol 2008; 26: 1346–1354. 10. Goulart BH, Clark JW, Pien HH, Roberts TG, Finkelstein SN, Chabner BA: Trends in the use and role of biomarkers phase I oncology trials. Clin Cancer Res 2007; 13: 6719–6726. 11. Rowinsky EK, Kaufmann SH, Baker SD, Grochow LB, Chen TL, Peereboom D, Bowling MK, Sartorius SE, Ettinger DS, Forastiere AA, Donehower RC: Sequences of topotecan and cisplatin: phase I, pharmacologic, and in vitro studies to examine sequence dependence. J Clin Oncol 1996; 14: 3074–3084. 12. De Jonge MJA, Sparreboom A, Gelderblom H, Planting AS, van der Burg MEL, Loos WJ, Brouwer E, van Beurden V, Mantel MA, Doyle E, Hearn S, Verweij J: Phase I and pharmacological study of oral topotecan and intravenous cisplatin: sequence dependent hematologic side effects. J Clin Oncol 2000; 18: 2104–2115. 13. De Jonge MJA, Sparrreboom A, Planting AS, van der Burg MEL, de Boer-Dennert MM, ter Steeg J, Jacques C, Verweij J: Phase I study of 3-weekly schedule of irinotecan combined with cisplatin in patients with advanced solid tumors. J Clin Oncol 2000; 18: 187–194. 14. Smorenburg CH, Sparreboom A, Bontenbal M, Verweij J: Combination chemotherapy of taxanes and antimetabolites: its use and limitations. Eur J Cancer 2001; 37(18): 2310–2323. 15. Van Cutsum E, Findlay M, Osterwalder B, Kocha W, Dalley D, Pazdur R, Cassidy J, Dirix L, Twelves C, Allman D, Seitz J-F, Scholmerich J, Burger HU, Verweij J: Capecitabine, an oral fluoropyrimidine carbamate with substantial activity in advanced colorectal cancer: results of a randomized phase II study. J Clin Oncol 2000; 18(6): 1337–1345. 16. Engels FK, Verweij J: Docetaxel administration schedule: from fever to tears? A review of randomised studies. Eur J Cancer 2005; 41: 1117–1126. 17. Therasse P, Arbuck SG, Eisenhauer EA et  al.: New guidelines to evaluate the response to treatment in solid tumours. J Natl Cancer Inst 2000; 92: 205–216. 18. Therasse P, Eisenhauer EA, Verweij J: RECIST revisited: a review of validation studies on tumour assessment. Eur J Cancer 2006; 42: 1031–1039. 19. Byrne MJ, Nowak AK: Modified RECIST criteria for assessment of response in malignant pleural mesothelioma. Ann Oncol 2004; 15: 257–260. 20. MacDonald DR, Cascino TL, Schold SC, Cairncross JG: Response criteria for phase II studies of supratentorial malignant glioma. J Clin Oncol 1990; 8: 1277–1280.

13  Early Clinical Trials with Cytotoxic Agents

345

21. Scher HI, Michael JM, Kelly WK et al.: Prostate cancer clinical trails endpoints: “RECIST”ïng a step backwards. Clin Cancer Res 2005; 11: 5223–5232. 22. Bajetta E, Di Leo A, Zampino MG et al.: Multicenter randomized trial of dacarbazine alone or in combination with two different doses and schedules of interferon alfa-2a in the treatment of advanced melanoma. J Clin Oncol 1994; 12: 806–811. 23. Mick R, Crowley J, Caroll R et al.: Phase II clinical trial design for noncytotoxic anticancer agents for which time to disease progression is the primary endpoint. Control Clin Trials 2000; 21: 343–359. 24. Korn E, Arbuck S, Pluda J et  al.: Clinical trial designs for cytostatic agents: are new approaches needed? J Clin Oncol 2001; 19: 265–272. 25. Van Glabbeke M, Verweij J, Judson I et al.: Progression-free rate as the principal end-point for phase II trials in soft-tissue sarcomas. Eur J Cancer 2002; 38: 543–549. 26. Van Oosterom AT: Progression arrest. In: Pinedo H, Verweij J (eds) Clinical Management of Soft Tissue Sarcomas. Martinus Nijhoff Publishers, Boston, MA, 1986; pp. 131–138. 27. Zee B, Melnychuk D, Dancey J et  al.: Multinomial phase II cancer trials incorporating response and early progression. J Bioph Stat 1999; 9: 351–363. 28. Dent S, Zee B, Dancey J et al.: Application of a new multinomial phase II stopping rule using response and early progression. J Clin Oncol 2001; 19: 785–791. 29. The Protocol Review Committee, the Data Center, the Research and Treatment Division, the New Drug Development Office: European Organization for Research and Treatment of Cancer: phase II trials in the EORTC. Eur J Cancer 1997; 33: 1361–1363.

Chapter 14

Challenges and Successes in Developing Effective Anti-angiogenic Agents Laura Q.M. Chow and S. Gail Eckhardt

14.1 Introduction Although the concept of tumors having a distinct capillary network and blood supply was first described by Virchow in 1863, and intensively studied by Goldman in 1907 [1], many decades passed before Judith Folkman discovered that angiogenesis and vascularization were critical to tumor growth and metastasis in 1971 [2–4]. In 1973, endothelial cells derived from human umbilical veins were first successfully cultured [5], leading to angiogenic bioassay, in vivo model development, and the discovery of angiogenic growth factors [6]. The potential inhibition of angiogenesis as anti-cancer therapy elicited an explosion of research in the 1970s; however, there would be many difficulties, failures, and delays before translating this new class of anti-cancer agents to clinical trials. In the late 1980s, the anti-angiogenic properties of low-dose interferon were used in treating hemangiomas and angioblastomas [7], and thalidomide’s anti-angiogenic properties became beneficial in treating multiple myelomas in 1999 [8]. In 2003, the anti-angiogenic anti-vascular endothelial growth factor (VEGF) monoclonal antibody, bevacizumab, demonstrated improved progression-free survival (PFS) benefits in advanced renal cell carcinoma (RCC) patients in an earlyphase trial [9]. Bevacizumab was the first anti-angiogenic agent approved by the United States Food and Drug Administration (FDA) in 2004 for its improvement in overall survival (OS) when administered in conjunction with chemotherapy in advanced colorectal carcinoma patients in a multicenter phase III trial [10]. Subsequently, multi-targeted anti-angiogenic receptor tyrosine kinase (RTK) agents, sunitinib and sorafenib, were approved for their efficacy in advanced RCC [11]. Current approved anti-angiogenic agents are ­summarized in Table  14.1 and select anti-angiogenic agents in oncologic clinical development are shown in Table 14.2.

L.Q.M. Chow (*) Division of Medical Oncology, Department of Medicine, University of Washington, 825 East lake Avenue East (SCCA) MS: G4-940, Campus Box 358081, Seattle, Washington, USA 98109-1023 e-mail: [email protected]; [email protected]; [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_14, © Springer Science+Business Media, LLC 2011

347

Table 14.1  Current US Federal Drug and Administration approved anti-angiogenic agents for oncologic use [123] Drug Mechanism Combined with Indication Bevacizumab Humanized antibody to Intravenous 5-FU-based First-line treatment of (Avastin™) VEGF chemotherapy metastatic carcinoma of the colon and rectum Bevacizumab (Avastin™) Humanized antibody to VEGF Intravenous 5-FU-based Second line treatment of chemotherapy metastatic carcinoma of the colon and rectum 1st line recurrent/locally Bevacizumab (Avastin™) Humanized antibody to VEGF Platinum based advanced/metastatic lung chemo – paclitaxel cancer and carboplatin Bevacizumab (Avastin™) Humanized antibody to VEGF Paclitaxel Locally recurrent or metastatic breast cancer Sorafenib (Nexavar®) Alone Advanced renal Multi-targeted anti-angiogenic cell carcinoma VEGF/PDGFR raf kinase tyrosine kinase inhibitor Sorafenib (Nexavar®) Alone Advanced unresectable Multi-targeted anti-angiogenic hepatocellular VEGF/PDGFR raf kinase carcinoma Tyrosine kinase inhibitor Sunitinib (Sutent®) Alone Advanced renal Multi-targeted anti-angiogenic cell carcinoma VEGF/PDGFR tyrosine kinase inhibitor Sunitinib (Sutent®) Alone Imatinib-refractory GIST Multi-targeted anti-angiogenic VEGF/PDGFR tyrosine kinase inhibitor Thalidomide Alone For advanced newly Non-specific anti-angiogenic (Thalidomid®) diagnosed multiple agent with activity against myeloma PDGFR May 2006

January 2006

January 2006

November 2007

December 2005

February 2008

October 2006

June 2006

US FDA approval date February 2004

348 L.Q.M. Chow and S.G. Eckhardt

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

349

Table 14.2  Anti-angiogenic monoclonal antibodies and multi-targeted tyrosine kinase agents in clinical development [66, 75, 125, 212] Phase of clinical Monoclonal antibodies Mechanism of action development Bevacizumab Monoclonal antibody Phase III/IV (Genentech/Roche) to VEGF-A VEGF-trap (Regeneron/Bayer) Anti-VEGF protein/antibody Phase III binding VEGF-A IMC-1121B (Imclone) Monoclonal antibody Phase II binding VEGFR-2 HuMV833 (Protein Design Anti-VEGF antibody Phase I Labs) Multi-targeted antiangiogenic agent Imatinib (Novartis) SU11248, sunitinib (Pfizer) BAY439006, sorafenib (Bayer/Onyx) PTK787/ZK222584, valatanib, (Schering AG, Novartis) AG-013736, axitinib (Pfizer) AZD6474, vandetanib (AstraZeneca) GW786034 (Glaxo-Smith-Kline) AZD2171, cediranib (AstraZeneca) AEE788 (Novartis) BIBF1120 (Boehinger) BMS-582664 (BMS) AMG-706 (Amgen) CHIR-258 (Novartis) BAY-579352 (Bayer) XL-999 (Exelixis) XL-820 (Exelixis) XL-647 (Exelixis)

Mechanism of action – inhibition against C-kit, BCR/Abl, and PDGFR VEGFR, PDGFR, FGFR, KIT, RET receptors, and FLT-3 VEGF, PDGFR, FGF, and Raf kinase VEGF, PDGFR, and FGFR

Phase of clinical development Phase III/IV Phase III/IV

VEGFR and PDGFR, VEGFR2, VEGFR3, and EGFR

Phase III Phase III

VEGFR VEGFR-1/2, PDGFR-b, c-Kit, and Flt-4 VEGFR/EGFR VEGFR, PDGFR, and FGFR VEGFR/FGFR VEGFR, PDGFR, KIT, and RET receptors VEGFR, FGFR, PDGFR and c-KIT, FLT-3 receptors VEGFR2 and PDGFR FGFR, VEGFR and PDGFR RTKs and FLT-3 receptors VEGFR, PDGFR, c-KIT receptors HER2, EGFR, VEGFR, and EphB4 receptors

Phase II Phase III

Phase III/IV Phase III

Phase I/II Phase II Phase I Phase II Phase I/II Preclinical Phase II Phase I Phase II

Currently, over 50 anti-angiogenic agents are being developed world-wide as anti-cancer therapy, amongst other uses, including treatment of macular degeneration, diabetic retinopathy, cardiac disease, arthritis, and psoriasis [12–16]. Angiogenic concepts, preclinical and clinical pitfalls and successes, and ongoing questions in anti-angiogenic drug development will be discussed and explored.

350

L.Q.M. Chow and S.G. Eckhardt

14.2 Angiogenesis and Its Mediators Angiogenesis refers to the development of new vessels from existing blood vessels. In normal conditions, such as wound healing and pregnancy, this process is tightly regulated; however, in malignancy, this unregulated process increases tumor growth and metastases [17]. An “angiogenic switch” occurs when pro-angiogenic factors outweigh anti-angiogenic factors, activating small 1–2 mm tumor cell conglomerates to produce their own vasculature and grow beyond 2  mm to form rapidly expanding masses [17]. Pro-angiogenic growth factor secretion ensues, and there is activation of lytic enzymes in the extracellular matrix (ECM) to degrade the basement membrane (BM) and the underlying interstitium, as shown in Fig. 14.1 [18]. Cellular movement for new blood vessel formation ensues and inflammatory cells in the immune system are activated to elicit their effects on tumor cells, blood vessels, and surrounding stroma [19]. Hypoxia and oncogenes increase pro-angiogenic growth factor gene expression and growth factor release [19]. Angiogenesis involves complex interactions between growth factors, the ECM, the immune system, endogenous mediators, and endothelial cells. Common stimulators and inhibitors of angiogenesis are summarized in Table 14.3.

Fig. 14.1  Key mediators involved in degradation and reformation of the vascular basement membrane (VBM) in angiogenesis (sourced from [18])

Canstatin

Anti-angiogenic antithrombin III and prothrombin kringle-2 Platelet factor 4 (PF4)

Receptors for matrix macromolecules and proteinases

Interleukin 4, 12, and 18

Interferons -agc

Stabilizes vessels by stimulating extracellular Endostatin matrix production Potent inhibitor of endothelial cell growth Fibronectin fragment in vitro but angiogenic in vivo Stimulates the synthesis of platelet-activating factor (PAF) by monocytes and endothelial cells Chemokine that stimulates angiogenesis Heparanases

Macrophage chemoattractant Stimulates arteriogenesis protein (MCP-1)

Integrins avb3, avb5

Interleukin-8 (IL-8)

Growth factor with role pathologic angiogenesis

Placental growth factor (PIGF) Fibroblast growth factor (FGF), Hepatocyte growth factor (HGF) Transforming growth factor beta (TGF-b) Tumor necrosis factor (TNF)

Growth factors in stimulating angiogenesis and arteriogenesis

Growth factor recruiting smooth muscle cells

Platelet derived growth factor – beta (PDGF-b)

Table 14.3  Selected activators and inhibitors of angiogenesis and mechanisms of action [339, 348, 349] Activators/stimulators of angiogenesis Mechanisms of action Inhibitors of angiogenesis Vascular endothelial Growth factor stimulating angiogenesis, Angiostatin and related growth factor (VEGF) permeability, and lymphangiogenesis plasminogen kringles

Degrade heparan sulfate glycosaminoglycan in the basement membrane, resulting in the loss of BM integrity, and release of angiogenic and growth-promoting factors Cytokines/chemokines inhibiting endothelial migration IFN-a downregulates bFGF Cytokines inhibiting endothelial migration Downregulate bFGF (continued)

Heparin-binding chemokine which inhibits binding of bFGF and VEGF Protein with potent inhibitory effects on the proliferation and growth of endothelial cells Collagen XIII fragment which inhibits endothelial survival and migration Inhibits growth of endothelial cells

Mechanism of action Proteolytic fragments of plasminogen which inhibit endothelial migration and survival Fragments of hemostatic factors suppress endothelial growth

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents 351

Ephrins

Heparanases

VE-cadherin

Matrix metalloproteases (MMPs)

Nitric oxide synthase

Angiopoietin-1 and Tie2 receptor

Leptin

Beta-estradiol

Angiogenin

Table 14.3  (continued) Activators/stimulators of angiogenesis

Ribonuclease that binds to actin on the surface of endothelial cells, it is endocytosed and translocated to the nucleus It promotes endothelial invasiveness and induces vascularization of normal and malignant tissues Regulates cytokine release from macrophages Functional estrogen receptors are essential for the augmentation of bFGF-induced angiogenesis Induces endothelial cell proliferation and expression of matrix metalloproteinases in vivo and in vitro Stabilizes vessels/vessel walls by tightening endothelial smooth muscle and inhibiting permeability Nitric oxide and prostaglandins stimulate angiogenesis and vasodilatation Matrix remodeling with liberation of FGF and VEGF from matrix, activation of TGF-b, and generation of angiostatin Endothelial junctional molecules essential for endothelial survival effect Degrade heparan sulfate glycosaminoglycan in basement membrane, resulting in the loss of integrity, and release of angiogenic and growth-promoting factors Regulate arterial/venous specification

Mechanisms of action

Stabilizes nascent vessels by preventing matrix dissolution

Plasminogen activator inhibitor-1 (PAI)

SPARC Osteopontin

Thrombospondins 1–2

Vasostatin and calreticulin

Retinoids

Tissue inhibitors of MMP (TIMPs) Angiopoietin-2

Extracellular matrix proteins which inhibit endothelial migration, growth, adhesion, and survival Inhibits binding and activity of VEGF Inhibits integrin signaling

Antagonist of Ang 1 – induces vessel regression and destabilizes vessels before sprouting Causes non-specific decrease in tumor induced angiogenesis Inhibit endothelial growth and endothelial cell proliferation

Suppress pathological angiogenesis

Soluble VEGFR1 and NRP-1 Decoy receptors for VEGF-B and PDGF

Pigment epithelium derived Inhibits growth factor-induced growth factor (PEDF) angiogenesis in microvascular endothelial cells Prolactin 16 kDa fragment Inhibits bFGF and VEGF

Mechanism of action

Inhibitors of angiogenesis

352 L.Q.M. Chow and S.G. Eckhardt

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

353

14.2.1 Angiogenic Growth Factors In order for the angiogenic switch to occur, angiogenic growth factors such as VEGF, platelet derived growth factor (PDGF), fibroblast growth factor (FGF), hepatocyte growth factor (HGF), transforming growth factor (TGF), tumor necrosis factor (TNF), interleukin-8 (IL-8), angiogenin, angiopoietins, and Tie receptor ligands, amongst many others, need to be activated [19]. The role, interactions, and activity of many of these growth factors in angiogenesis are still being explored. The most potent and most studied growth factor for endothelial cell proliferation and angiogenesis is VEGF [20]. VEGF expression is elevated in many malignant tumors – including breast, colorectal, lung, prostate, and RCC [21–23] and excessive levels correlate with increased microvascular density, cancer recurrence, and decreased survival [14, 24, 25]. The VEGF receptors VEGFR-1 (Flt-1), VEGFR-2 (Flk1 or KDR), and VEGFR-3 (Flt4) are predominantly found on endothelial cells. They have RTK activity promoting intracellular signaling, ­leading to downstream effects on tumor growth promotion, angiogenesis, and metastasis [24]. PDGF receptors a and b are both over-expressed on many solid tumors, on pericytes (smooth muscle cells which mechanically support vasculature) in the ECM, and on fibroblasts in the tumor stoma and neovasculature [26]. These receptors are upregulated during tumor progression to directly stimulate the growth and proliferation of pericytes and fibroblasts surrounding endothelial cells, and PDGF also upregulates VEGF expression to further promote angiogenesis [26–28]. Basic fibroblast growth factor (bFGF) and its receptor also directly promote tumor growth, angiogenesis, and indirectly induce VEGF expression [12, 29, 30]. Therefore, inhibition of these growth factors and their pathways results in profound anti-angiogenic effects.

14.2.2 The Extracellular Matrix Endothelial cells must move through the BM, interstitium, and the various layers in the ECM before they can form new blood vessels at the tumor site. Once angiogenic growth factors are secreted, lytic enzymes such as the matrix metalloproteinases (MMP), the serine protease urokinases, and the endoglycosidases are produced to digest various components of the ECM to allow cellular invasion [31]. MMPs, overexpressed by invasive tumor cells, modulate cell adhesion and generate chemotactic ECM degradation products for endothelial cells [31]. The MMPs, zinc-dependent enzymes which are secreted as inactive pro-enzymatic forms to be cleaved and activated by ECM substrates, include collagenases (MMP-1, MMP-8, and MMP-13), stromelysins (MMP-3, MMP-10, and MMP-7), and elastases (MMP-12) [31, 32]. Tissue inhibitors of metalloproteinases (TIMPs) regulate and inactivate MMPs by forming tight non-covalent associations with their active sites [33]. Urokinase enzymes are also regulated by growth factors, oncogenes, and specific plasminogen activator

354

L.Q.M. Chow and S.G. Eckhardt

inhibitors (PAIs) [34]. High levels and high expression of urokinase contribute to an aggressive invasive phenotype and are poor prognostic indicators in several cancers [35]. These enzymes interact with specific cell-surface receptors to localize enzymatic activity to the cell surface and activate enzymes such as plasminogen [32]. Heparan sulfate (HS) is a main component of the ECM and BM, and it is a cell-surface receptor for angiogenic factors such as bFGF, aFGF, and VEGF; therefore, inhibition of the HS-degrading enzymes can have potent effects on tumor blood vessel growth, cell invasion, migration, adhesion, differentiation, and proliferation [36, 37]. The endoglycosidase enzyme, heparanase, degrades HS glycosaminoglycan to result in BM integrity loss and release of HS-bound angiogenic factors, and induces endothelial cell migration via activation of the protein kinase B/Akt signaling pathway [38]. Heparanase mRNA is overexpressed in many cancers including head and neck tumors [39], pancreatic tumors [40], and hepatocellular carcinoma (HCC) [41].

14.2.3 The Immune System Inflammatory and immune mediators such as cytokines, and T- and B cells may elicit either an anti-tumor response or promote tumor growth by increasing growth factor production and facilitating dissemination, angiogenesis, and tissue destruction. Infiltrating leukocytes such as inflammatory macrophages, polymorphonuclear neutrophils (PMNs), and granulocytes release inflammatory pro-angiogenic factors that promote tumor growth and the release of angiogenic enzymes such as MMPs [42]. The tumor itself secretes chemotactic cytokines to recruit tumor-associated macrophages from circulating peripheral blood monocytes, which then produce more chemokines and angiogenic growth factors, recruiting more endothelial cells and activating new infiltrating cells in a powerful self-amplifying angiogenic cascade [43]. Interferons (INFs) have an immodulatory and angiogenesis-inhibiting effect. They suppress bFGF gene expression and production to inhibit endothelial and metastatic cell migration and invasion [42, 44]. Low dose INFa significantly decreases IL-8, reduces uPA activity, and inhibits MMP-9 protein expression and activity [29]. Interleukins can also inhibit or stimulate angiogenesis: IL-8 is an angiogenic factor in cancer progression, IL-1 is involved in tumor growth and inflammation, IL-4 inhibits bFGF induced angiogenesis, and IL-12 and IL-18 cytokines induce INFg and inhibit angiogenesis through inhibiting downstream cytokinases and FGF-stimulated endothelial cell proliferation, respectively [43, 45–47].

14.2.4 Endothelial Cells and Endogenous Mediators of Angiogenesis Tumor endothelial cells divide up to 50 times faster than normal endothelial cells, and prominently overexpress integrin avb3 and VEGF receptors [48]. Integrins are

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

355

cell-adhesion receptors which bind a number of matrix proteins and MMP-2, localizing matrix degradation to the endothelial surface and modulating endothelial cell survival [49–52]. Integrin avb3, undetectable on normal endothelium, is highly upregulated in cytokine or tumor-induced angiogenesis [49–52]. Multiple endogenous mediators of angiogenesis, found in the blood of cancer patients, interact with these endothelial cells, the immune system, and ECM in a complex manner. Angiostatin, a fragment derived from the proteolytic cleavage of plasminogen, binds to endothelial cell surface ATP-synthase to trigger apoptosis, and binds to integrin avb3 to inhibit plasmin binding and plasmin-induced endothelial cell migration [53, 54]. Angiostatin also directly inhibits PMN and monocyte recruitment and migration to target inflammatory cell-mediated angiogenesis [53, 54]. Endostatin is a collagen-derived angiogenesis inhibitor which interferes with FGF-2 signal transduction, inhibits endothelial cell motility, induces apoptosis, and causes G1 arrest of endothelial cell signaling via direct interaction with VEGF RTKs [55, 56]. It downregulates genes involved in the immediate-early response, cell cycle progression and apoptosis regulation, and it downregulates stress response kinase pathways, endothelial cell growth factor receptors, mitogenic factors, MMPs, and cell structure and adhesion components [55, 56]. Thrombospondin glycoproteins also mediate cell–cell and cell–matrix interactions and suppress tumor growth through anti-angiogenic effects [57]. Thrombospondin 1 specifically increases Fas-ligand induced apoptosis, inhibits VEGF mobilization from the ECM, and suppresses endothelial cell migration [57, 58]. There are a multitude of different methods of blocking angiogenesis through inhibition of growth factors and ECM enzymes, modulating the immune system, inhibition of endothelial cell receptors, and finally through changing the balance of endogenous angiogenic mediators. Through in vitro and in vivo assays, anti-angiogenic compounds could be tested for their effects on angiogenesis with varying success.

14.3 Preclincal Aspects: Assessing Anti-angiogenic Activity and Determining the Starting Dose and Schedule of Administration Due to the multitude of potential angiogenic targets, it was exceptionally challenging to select the most relevant targets, and equally difficult to predict which compounds would lead to successful clinical outcomes. With the development of preclinical in  vitro assays and in  vivo models of angiogenesis, the hypothesis of angiogenesis inhibition could be tested and potential compounds could be screened. In addition to screening potential targets and compounds by trial and error for antiangiogenic activity, the formulation, chemistry, and pharmacokinetics (PK) of the investigational agent must be favorable for further preclinical and clinical development.

356

L.Q.M. Chow and S.G. Eckhardt

14.3.1 Preclinical Screening Assays and Models Assessing Anti-angiogenic Activity Angiogenesis is a complex process involving multiple signaling pathways, mediators, and receptors. Some angiogenesis assays assess inhibition of one step, whereas, others can assess inhibition of multiple processes. In vitro, the most commonly used models are assays that assess endothelial proliferation, migration or tube formation using counting, thymidine incorporation, or immunohistochemical staining techniques [59]. The Boyden chamber endothelial cellmigration assay uses endothelial cells grown on Matrigel™ (Becton Dickinson, Bedford, MA) as a matrix below which an endothelial cell chemoattractant is introduced. Tube and vessel assays performed using Matrigel™ appear to be the best way to assess vascular formation [59, 60]. Other effective proliferation and endothelial growth assays using matrices include rat aortic ring and human saphenous vein assays [61]. Migration and proliferation assays are relatively reproducible and easy to perform; however, reagents, culture conditions, cell lines, and experimental techniques vary substantially between researchers making comparison difficult [61]. Furthermore, endothelial cells in vitro inhibition may not predict in vivo response; therefore, more than one preclinical model should be used. To accurately reflect anti-angiogenic activity, in  vivo assays should be performed in conjunction with in  vitro studies. In vitro assays are cheaper and faster, but provide less information; whereas, in vivo assays provide physiologic assessment and information, but are time-consuming and expensive [61]. The cornea micropocket assay is the simplest in vivo model that allows for frequent observations. This assay involves implantation of a sustained-release pellet containing tumor cells or pro-angiogenic peptides onto an avascular cornea to induce an angiogenic response, and then anti-angiogenic substances are tested by directly incorporating them into the pellet or by systemic administration [61]. The chick chorioallantoic membrane (CAM) assay is inexpensive and commonly used. Fertilized eggs are incubated, and the CAM is either removed from the shell entirely or the CAM is exposed in the shell through a small window. Antiangiogenic compounds are implanted into the growing CAM at the peak of angiogenesis and assessed for their ability to inhibit capillary growth [62]. Tumors can be implanted on the CAM to directly show tumor-associated angiogenesis. Despite its efficiency, the CAM is difficult to quantify and a non-specific inflammatory response is generated by the implanted material [61]. Disc and sponge angiogenesis assays use polyvinyl alcohol foam material which is covered on both sides by Millipore filters, to suspend a test agent or impregnate tumor cells, for subcutaneous (SC) implantation. Angiogenesis is measured as the disc or sponge becomes vascularized subcutaneously and subsequently resected for analysis. The Matrigel™ plug assay involves a tumor extract of BM proteins injected SC into mice that becomes surrounded by granulation tissue and

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

357

supports an intense angiogenic response when supplemented by angiogenic growth factors. Anti-angiogenic test compounds or tumor cells can be added to Matrigel™ in a manner similar to the disc and sponge assays, and angiogenic responses can be measured quantitatively using a radioactive tracer or by assessment of the hemoglobin content of the gel. Zebra fish as a whole animal model can also be used for screening small molecules – these embryos are transparent allowing direct and microscopic assessment of blood vessel development. Fluorescent-labeled markers can be used to identify endothelial cells and other cell types in this model. Table  14.4 summarizes some of the advantages and disadvantages of the major in vivo angiogenesis assays [61]. Unfortunately, criteria to determine what truly constitutes anti-angiogenic activity are poorly defined and controversial. Generally, anti-angiogenic agents should kill endothelial cells at lower concentrations than that required to induce similar effects in cancer cells in vitro, inhibit endothelial cell function and tubule formation, interfere with a specific known portion of the angiogenic cascade, and demonstrate in  vivo evidence of inhibiting angiogenesis with minimal weight loss, toxicity and toxic deaths at relevant doses [62, 63]. Interestingly, many current cytotoxic agents would satisfy these criteria and demonstrate anti-angiogenic activity in the above assays [63]. Surprisingly, the optimal in vivo model to determine angiogenic anti-tumor activity has not been established. In vivo murine angiogenic models are limited in predicting human angiogenesis as they differ greatly from human models: murine models are faster growing with high endothelial cell growth rates and poorer pericyte coverage, and variations in tumor burden greatly influence treatment results [63]. As angiogenesis is heterogeneous with intra- and intertumoral angiogenic variability, large sample numbers are required [59, 63]. As tumor excision is needed to assess angiogenesis, serial assessments are not possible. The site of tumor growth and the appropriate host microenvironment are important determinants of anti-angiogenic therapeutic efficacy [59]. Human tumor implantation onto xenograft animal models itself stimulates growth of murine vessels – a factor which may need to be overcome by grafting human foreskin onto immunodeficient mice and injecting tumor cells into the foreskin graft [61]. Therapy that is effective at one site might not be appropriate at another site, and major response differences have been observed between  orthotropic and heterotopic implants of the same tumor. Orthotopi­ cally xenografted tumor models may be more physiologically relevant [59]. Transgenic or spontaneous-arising tumor models may also be more useful angiogenic models as they rapidly grow with high proportions of new blood vessels [59]. In summary, interpretation of preclinical studies to predict clinical activity poses challenges. Despite the ability of preclinical cellular assays and in vivo animal studies to screen for anti-angiogenic and anti-tumor activity, they do not necessarily predict clinical activity due to the multifactorial nature of human angiogenesis in vivo.

358

L.Q.M. Chow and S.G. Eckhardt

Table 14.4  Advantages and disadvantages of the major in vivo angiogenesis assaysa In vivo assay Advantages Disadvantages Normal cornea is avascular, Corneal micropocket New vessels are easily therefore angiogenesis is assay identified – shows atypical angiogenesis by sprouting Cornea is not a highly relevant and uses an immunologically site for tumor growth privileged site before Technically demanding vascularization especially in mice and Able to use in rats, rabbits, and traumatic technique mice (mammals) Ethically questionable and Permits non-invasive expensive observation and long term Exposure to oxygen on surface monitoring affects angiogenesis Quantitative Non-specific inflammatory response with some compounds Very sensitive to increases in Technically simple and Chick embryo oxygen tension inexpensive chorioallantoic Visualization of new vessels Allows non-invasive observation membrane assay can be difficult Suitable for mammalian (CAM) Non-mammalian xenografts Embryonic Suitable for large-scale Non-specific inflammatory screening reactions common Angiogenesis by sprouting and Drugs requiring metabolic intussusceptive growth activation cannot be Accelerates or suppresses assessed angiogenesis up to 10–11 days Time consuming and Mesentery angiogenic Adult tissue is vascularized technically demanding assay without significant Mice less suitable for physiologic angiogenesis, quantitative angiogenesis angiogenesis occurs by analysis than rats but sprouting rats require 10× greater Quantitative and allows quantities of test agents dose–response studies than mice for microvessel spatial Does not allow real-time extension, density and observations number and length of microvessel segments and sprouts in rats Minimal trauma inflicted on test tissue Test tissue is visceral – which is a frequent site of tumor metastasis Suitable for measurement of growth factor-induced signaling in intact microvessels, Suitable for intra-vital microscopy and molecular-activity studies (continued)

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents Table 14.4  (continued) In vivo assay Advantages Matrigel plug

Technically simple and suitable for large-scale screening Rapid quantitative analysis in chambers

Sponge/matrix

Technically easy and inexpensive Well tolerated Time course of response can be recorded Suitable for study of tumor angiogenesis

Disc assay

Technically easy, assesses wound healing and angiogenesis Quantitative analyses

Zebra fish

Intact whole animal and technically simple Allows genetic analysis of vessel development Truly quantitative Large numbers of animals available for statistical analysis Relatively fast assay and suitable for large-scale screening a  Table modified and adapted from [61]

359

Disadvantages Matrigel is not chemically defined Difficult to make plugs uniform in 3-D shape (except in chambers) Analysis in plugs is time consuming, expensive Expensive Not responsive to VEGF in chambers Subcutaneous tissue is not a highly relevant site for tumor growth Time consuming and can result in encapsulation by granulation tissue Sponge composition varies therefore inter-experimental comparisons are difficult Variable retention of test compound within implant Subcutaneous tissue is not a highly relevant site for tumor growth Animals need to be kept individually Encapsulated by granulation tissue Subcutaneous tissue is not a highly relevant site for tumor growth Non-mammalian Embryonic Expensive to maintain in breeding condition

14.3.2 Determining the Starting Dose and Schedule of Administration Once potent anti-angiogenic agents are identified through preclinical screening assays, the starting dose in phase I clinical trials is determined similarly to that of other targeted agents. This strategy involves assessment of the minimal doses in the ­nanomolar or micromolar range that inhibit therapeutic targets in vitro, assessment

360

L.Q.M. Chow and S.G. Eckhardt

of bioavailability and the ability to achieve these doses in animal studies in vivo, assessment of tumor inhibition or regression in xenograft or other human tumor animal models, and determination of an active yet minimally toxic, non-lethal dose through animal toxicology studies [64, 65]. From preclinical studies, the ideal starting non-toxic dose in humans should be far below the toxic and lethal doses in animals, without being below anti-angiogenic target inhibition and tumor inhibition/regression levels. The schedule of administration is determined through assessment of PK, pharmacodynamics (PD), and recovery from toxicity similar to other agents. Many preclinical studies demonstrate that angiogenesis inhibitors are most effective when administered by a dose and schedule that maintains a constant therapeutic concentration of the inhibitor in the circulation; unlike cytotoxic drugs which are administered at their maximum tolerated doses (MTD), followed by off-therapy intervals to recover from toxicity [63, 66]. The dosing schedule should allow for chronic administration and pharmacokinetically show sustained inhibition of antiangiogenic targets.

14.4 Clinical Aspects: Dose Escalation and Toxicity 14.4.1 Phase I Clinical Trial Methods and Design The main goals of phase I trials are to determine the optimal dose of new agents by dose escalation and toxicity assessments in small groups of patients. Starting doses, dosing schedules, and potential toxicities are determined from preclinical models. Traditional designs with cytotoxic agents define the MTD through dose-limiting toxicity (DLT), since anti-cancer activity is assumed to correlate with toxicity. As anti-angiogenic agents may be more selective and produce different toxicities than cytotoxic agents, dose escalation to the MTD is not the ideal approach [63]. Serum concentrations of anti-angiogenic agents can exceed well above 50% of the minimum inhibitory concentration (IC50) without reaching traditionally defined DLTs – particularly hematologic [67]. In addition, anti-angiogenic toxicity may be produced through different mechanisms than those producing therapeutic effects, and may not predict therapeutic effects [68]. Furthermore, rapid dose escalation of anti-angiogenic agents may be necessary, assuming a more shallow dose–response curve [62]. Determining the optimal biologic dose (OBD) for target inhibition would be a better approach; however, as objective tumor responses are not commonly seen in phase I, surrogate markers strongly correlating with anti-angiogenic drug effect on the tumor target are required [69]. Unfortunately, surrogate markers have yet to be validated [70, 71]. As long-term administration and continuous dosing approaches may be more favorable for anti-angiogenic activity, rather than cyclical administration, toxicities can be acute or delayed. Typical phase I dose escalation designs determining MTD due to toxicity in the first cycle of therapy may not reflect long term toxicity or

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

361

tolerability – both of which should be taken into consideration when determining the recommended phase II dose (RPTD). PK dose finding, correlation with PD target inhibition, and scheduling to achieve concentrations to maintain maximum target inhibition in vivo are essential in determining an active and safe therapeutic level. The tumor type and growth rate of different tumors can be differentially affected by the anti-angiogenic dose with more rapidly growing tumors possibly requiring a higher dosage; therefore, better PK modeling may be in order when designing these trials [72, 73].

14.4.2 Anti-angiogenic Agents and Toxicity Although anti-angiogenic therapies were initially assumed to be less toxic than conventional chemotherapy, chronic administration of these agents could cause severe non-hematologic toxicities affecting quality of life (QOL). The role of angiogenesis in organogenesis, embryonic development, wound healing, diabetic retinopathy, rheumatoid arthritis, psoriasis, as well as tumorigenesis and metastasis, needs to be taken into account when looking at toxicity and designing clinical trials [42]. The PK characteristics, particularly tissue penetration and distribution, of anti-angiogenic agents are important: small molecule RTK agents have smaller molecular weights (<500  kDa) and are more likely than monoclonal antibodies to cross the blood–brain barrier [74]. Therefore, these agents are more likely to interfere with VEGF in the central nervous system [74]. It is unknown if the side effects of fatigue or headache may be related to this fact [66, 75]. Small molecule RTK agents have less specificity and inhibit a larger number of targets, whereas monoclonal antibodies are more selective, which also affects the toxicity profile. In clinical trials, toxicities of VEGF-targeting anti-angiogenic agents were predominantly asthenia, fatigue, nausea, and headache [66]. Hypertension and proteinuria were common with VEGF-binding antibodies and multi-targeted RTK agents [66, 76]. Over 30% of patients treated with bevacizumab developed drug-related hypertension, and more than 40% of patients developed asymptomatic proteinuria with rarely observed drug-related nephrotic syndrome [77, 78] Arterial thromboembolism rates were increased in patients >65 years of age with coronary artery disease or with risk factors for cerebral vascular accidents [79]. Bleeding was ­common and mild – usually skin and nose bleeds; however, pulmonary hemorrhage occurred in some patients with advanced central squamous cell carcinomas (SCC) treated with bevacizumab [80, 81]. As these patients were subsequently excluded from clinical trials of anti-angiogenics, the real effectiveness of these agents in this patient population is unknown. Thromboembolic and bleeding phenomena were idiosyncratic, unpredictable, and not dose-related in studies of anti-angiogenic agents [82]. Rare unpredictable toxicities such as gastrointestinal perforation appeared in 3–11% of patients treated with bevacizumab in colon cancer and ­ovarian cancer trials [79, 82, 83]. Multi-targeted RTK agents demonstrate similar

362

L.Q.M. Chow and S.G. Eckhardt

side effects to bevacizumab, including prominent hypertension, bleeding, poor wound healing, and also increased gastrointestinal perforations [84]. Hoarseness, mucositis, nausea, diarrhea, skin changes, rash, fatigue, hypothyroidism, elevated transaminases, palmar-plantar erythrodysesthesia (PPE), and myelosuppression were more commonly reported with multi-targeted RTK agents [66, 85]. Rarely, cardiac failure, intracerebral hemorrhage (in patients with brain metastases), and pancreatitis have been observed [86–91]. Fatigue, rash, PPE, and diarrhea were dose-related, commonly limiting dose-escalation and chronic dosing with ­anti-angiogenic multi-targeted RTK agents [66, 92]. The slope and steepness of the dose–response curve, the therapeutic index, and the full toxicity profile of these agents are key issues to be considered when designing clinical trials and escalating doses for long-term administration. Improved clinical study design and understanding of the anti-angiogenic mechanisms of action are needed in conjunction with better PK, PD, and biologic and radiologic markers when moving ahead to further develop these agents, and to find an efficacious and biologically active dose, while avoiding excessive toxicity.

14.5 Efficacy-Oriented Anti-angiogenic Single Agent Clinical Trial Design Anti-angiogenic agents successfully induce tumor regressions in animal models; however, the preclinical success has not translated well to the clinic. Antiangiogenics can be administered to mice with low tumor burdens – usually 1–2 mm tumors, whereas, patients with such small tumors or low disease burden are usually not eligible for clinical trials or do not have relevant disease on conventional imaging [67]. In addition, xenograft models have new active forming tumor blood vessels which differ greatly from established clinically apparent human tumors that have older, mature, differentiated vessels [67, 73, 93]. In mature vasculature, pericyte formation is increased in the endothelium, which may protect endothelial cells from apoptosis [94, 95]. Early-stage tumors that are rapidly dividing and vascularizing with VEGF as their more prominent angiogenic factor may be more sensitive to anti-angiogenics [22]. In early stage breast cancer, VEGF is the only angiogenic agent produced; however, during progression, PIGF, PDGF, bFGF, TGF-b, pleiotrophin, and other angiogenic factors become important [96, 97]. As tumors are comprised of heterogenic cells secreting numerous angiogenic factors, blocking a single angiogenic protein may not be successful, and resistance may occur with the overwhelming secretion of other pro-angiogenic factors – particularly in late stages of disease with high tumor burdens [64]. Resistance may also be due to selection of tumor cells with mutations resistant to hypoxia or metabolic starvation, upregulation of endothelial cells with anti-apoptotic genes, recruitment of pericytes to promote endothelial cell survival, and secretion of alternative survival, growth, and motility factors by PMNs, macrophages, cytokines, and fibroblasts in the stoma [98, 99].

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

363

In traditional clinical trial designs for efficacy, late stage advanced cancers are assessed first due to cost, ethics, feasibility and time; however, this setting is not optimal to determine anti-angiogenic effectiveness. Targeting early-stage tumors with anti-angiogenics as adjuvant therapy would be a more successful approach – particularly in patients with a high likelihood of tumor recurrence post radical surgery [67]. Patients with metastatic disease may benefit from anti-angiogenic agents as maintenance therapies after achieving a response to chemotherapy [67]. Moreover, patient therapy can be tailored by assessing growth factor expression in the individual’s tumor tissue, considering combining anti-angiogenic therapies, or using a multi-targeted approach depending on the stage and extent of disease [71]. While it is feasible to enroll patients with many different tumor types into trials with cytotoxic agents as these agents have non-specific hematologic toxicities and anti-cancer effects, animal studies indicate that different tumors respond to antiangiogenic therapy with substantial heterogeneity [73]. Perhaps, developing these compounds in a homogenous patient population early would yield more clinical success [73]. Phase II clinical trials should demonstrate the clinical activity, toxicity, and biologic modulation of the target with tumor response. Unfortunately, markers of drug activity or efficacy have not been validated and require further exploration before use in determining the OBD [62, 71]. As anti-angiogenic therapies may be slower than cytotoxic therapy in inducing tumor regression and tend to inhibit tumor growth rather than show dramatic regressions, endpoints such as SD, TTP, PFS or OS may be more relevant than the response rate (RR) [68, 70]. Phase II studies using historical controls and assessing for tumor response without a placebo control arm are not optimal, as tumor responses are less common. Therefore, studies of chemotherapy combinations randomized to placebo or the experimental anti-angiogenic agent are better in determining efficacy [68]. In rarer diseases, where this strategy is not possible, non-randomized studies would still have merit [68, 100]. It is challenging to identify the most promising drugs from early clinical trials. Important factors to consider include toxicity, QOL, tumor response, TTP, OS, and clinical benefit with adequate exploration of markers of efficacy, before further clinical development. Phase III clinical studies are designed to prove increased clinical benefit of a new agent or combined schedule when compared to standard therapy using endpoints of OS, TTP, and QOL. Confirmed indicators of molecular target expression would be ideal to select patients for these trials; however, these still need to be validated [73]. Patient selection will play a role in the clinical developmental success of anti-angiogenic agents, and molecular characterization of their tumors could determine whether specific patient subsets will benefit from anti-angiogenic agents. For example, tumor cells positive for wild type p53 mutations may have an increased response to anti-angiogenic therapy when compared to those null for p53 [101]. Null p53 cells appear to be more capable of adapting to hypoxic conditions induced by anti-angiogenics [101]. Patients with loss of heterozygosity or mutated von Hippel–Lindau (VHL) tumor suppressor genes (i.e. RCC, hemangioblastomas, pheochromocytomas, and pancreatic islet cell tumors) would be more susceptible

364

L.Q.M. Chow and S.G. Eckhardt

to developing highly vascular tumors, which would be good candidates for antiangiogenic therapy [102–104]. This selection strategy has been clinically successful in RCC where VHL mutations affect the ability of VHL protein to downregulate hypoxia inducible factor 1-alpha (HIF-1a) and decrease VEGF levels [104]. In addition, VEGF has been demonstrated to induce vascular permeability in murine models of lung and ovarian models, and high VEGF levels have been observed from pleural and ascitic fluids of patients [105, 106]. Therefore, anti-angiogenic therapy can potentially decrease malignant effusions to improve QOL. These data indicate that there may be predictive biomarkers that should be validated in randomized studies. As anti-angiogenic therapies are known to be additive or synergistic with chemotherapy, hormonal therapy, and radiotherapy, combination approaches should be considered early in development [75]. Additionally, these new agents would be best developed with good biologic and preclinical rationale in malignancies where there is an unmet clinical need. Anti-angiogenic therapy in less chemosensitive tumors such as RCC, HCC, and gastrointestinal stromal tumors (GIST), has already proved to be highly successful [107–109].

14.6 Specific Anti-angiogenic Agents in Clinical Trials 14.6.1 Matrix Metalloproteinase Inhibitors As MMP molecules are over-expressed in cancer and high levels are associated with a poor prognosis, matrix metalloproteinase inhibitors (MMPIs) were among the first anti-angiogenic compounds to be explored [31, 32, 110]. Despite their preclinical efficacy in inhibiting ECM degradation, blocking angiogenesis and tumor growth, and preventing cancer-cell invasion and metastases, MMPIs were not clinically effective [110–113]. Clinical development of batimastat (BB94) and marimastat (BB2516), an orally bioavailable equivalent of BB94, was abandoned. Batimastat was poorly soluble requiring intraperitoneal or intrapleural administration [111, 112], and marimastat caused severe toxicities of fatigue and inflammatory polyarthritis [114]. Other MMPI’s such as BAY12-9566, prinomastat (AG3340), COL-3, neovastat (AE-941), BMS-275291, and CGS27023A also demonstrated substantial toxicity, and disappointingly, a lack of efficacy [42, 111, 112]. As a result, enthusiasm to further develop non-specific MMPIs waned [111]. MMPs degrade basal membranes and ECM proteins releasing both proangiogenic factors and inhibitors of angiogenesis; therefore, inhibiting MMPs may not produce net anti-tumor effects in  vivo [33, 111, 115]. As MMPs are necessary for the homeostasis and turnover of normal tissues, non-specific MMP inhibition caused severe side-effects, especially in joints and muscles [111, 112]. More pre-clinical studies and better clinical modeling of in vivo MMPs and their angiogenic effects could have predicted some of these issues and prevented some

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

365

of these side-effects [111, 112]. These agents entered clinical testing too early without adequate understanding of the preclinical activity and toxicity. Moreover, early MMPI trials were limited by lack of a biological marker of activity [111, 112]. A recent phase I study of the novel specific MMPI, S-3304, inhibiting MMP-2 and MMP-9 without inhibiting MMP-1, MMP-3, or MMP-7, demonstrated good tolerability without musculoskeletal toxicity, and determined that film in-situ zymography (FIZ) of gelatinolytic activity could potentially serve as a surrogate biomarker of MMP activity [116].

14.6.2 Non-selective Anti-angiogenic Treatments Thalidomide, initially used as a sedative and an anti-inflammatory agent, demonstrated anti-angiogenic effects by inhibiting VEGF secretion, endothelial cell migration, adhesion, and capillary tube formulation [8, 117]. Thalidomide clinically demonstrated anti-angiogenic and immunomodulatory effects by decreasing serum VEGF and bFGF levels with therapy, decreasing TNFa and other cytokines, and reducing circulating endothelial cells [117–119]. In 1999, thalidomide was found to be active in reducing myeloma protein serum levels and Bence-Jones protein urine levels in multiple myeloma patients [8, 120]. Since that time, a metaanalysis of 4,144 patients in nine phase III randomized controlled trials (RCTs) with induction thalidomide demonstrated that thalidomide improved OS when added to standard, non-transplantation myeloma therapy (HR 0.67; 95% CI 0.56– 0.81) [120, 121]. Three RCTs assessing maintenance thalidomide post-autologous transplantation demonstrated a significant overall survival benefit in meta-analysis (HR 0.49, 95% CI 0.32–0.74) [120–122]. Based on promising results, thalidomide was approved by the US FDA on July 20th, 2006 for the treatment of multiple myeloma [123]. Despite the effectiveness of thalidomide in multiple myeloma, the exact clinically significant anti-angiogenic mechanisms of thalidomide have not been fully elucidated. No consistent correlation between clinical effect and biomarkers such as angiogenic growth factors or cytokines has been confirmed in these clinical trials [118, 119, 121, 124]. Thalidomide is currently being explored as a single agent and in combination in a number of trials [125].

14.6.3 Antibody Therapies Against VEGF The first successful anti-angiogenic approach involved targeting VEGF with antibodies [75]. Antibodies, produced through hybridoma technology in mice, are easily isolated, but expensive to produce [102, 126]. They bind to their target with high affinity and specificity, and they have long half-lives requiring IV administration on a bi- or tri-weekly basis [102, 126]. Bevacizumab, a murine monoclonal antibody binding VEGF-A ligand, was the first anti-angiogenic agent approved for clinical

366

L.Q.M. Chow and S.G. Eckhardt

use, in combination with chemotherapy [127]. The VEGFR2 antibody, IMC-1121B, is being explored in phase II and III trials alone and in combination in various tumor types [128]. VEGF-Trap is an engineered decoy-soluble receptor, a fusion of proteins from domains of VEGFR1 and VEGFR2 with immunoglobulin IgG Fc regions, binding all VEGF isoforms and PDGF [129]. It has higher affinity blocking VEGF-A and VEGF-B, and abolishes mature vessels in established xenografts and lung micrometa stase’s [129, 130]. This agent is undergoing phase III evaluation in a number of tumors – particularly NSCLC [131, 132]. Bevacizumab binds VEGF-A ligand with a half-life of 17–21 days, and changes the regulatory balance to favor anti-angiogenesis by reducing tumor vascular permeability, interstitial pressure, and oxygen and nutrient delivery to cancer cells [133]. Preclinical anti-angiogenic and anti-tumor activity were observed, and it was well-tolerated in phase I studies without DLTs or chemotherapy interactions [83, 130, 133, 134]. As tumor suppressor gene mutations in VHL lead to VEGF oversecretion in clear cell RCC patients, bevacizumab compared to placebo in previously IL-2-treated RCC patients, demonstrated a significantly prolonged PFS benefit of 4.8 versus 2.5  months (p < 0.001) [9]. In a phase III trial of IFN-2a combined with either bevacizumab or placebo in metastatic RCC patients, there was a statistically improved PFS (10.2 versus 5.4 months, p < 0.001), objective RRs (31% versus 12%, p < 0.0001), and a trend towards improved OS (p = 0.067) favoring bevacizumab over placebo [135]. As there was limited activity of bevacizumab as a single agent, except for some activity in RCC, bevacizumab was combined early with chemotherapy [136]. Synergistically, bevacizumab in combination decreased the number of blood vessels, normalized their function, and promoted chemotherapy accumulation in tumor tissue [83, 99]. The optimal dosing of bevacizumab in combination with chemotherapy is unknown and merits further exploration. The relationship between the MTD and the OBD is unclear for anti-angiogenic agents: toxicities can be low and the MTD can often greatly exceed the OBD. The OBD may differ for different tumor types, for those with different growth rates, and for those tumors previously treated with many lines of therapy [137]. Bevacizumab had a linear PK profile in phase I studies: doses ³0.3  mg/kg ­completely suppressed free serum VEGF and doses >1 mg/kg produced serum levels in the preclinical maximal inhibitory target range of ³10 ug/mL for at least 14 days [138, 139]. Exploration of doses from 3 to 20 mg/kg every 2–3 weeks in the phase II setting demonstrated a dose–response relationship where higher doses were more ­efficacious than lower doses in advanced NSCLC (15  mg/kg versus 7.4  mg/kg in ­combination with paclitaxel and carboplatin) [80, 81] and RCC (10 mg/kg was more active than 3 mg/kg as a single agent) [9]. However, in breast and colon cancer trials, the dose–response relationship was unclear [137]. In the phase II study with 104 advanced colorectal patients treated with 5-fluorouracil (5-FU), leucovorin, and bevacizumab at 5 mg/kg or 10 mg/kg every 2 weeks, the response (40% versus 24%) and survival benefit (21.5 months versus 16.1 months,) favored the low-dose group [140]. It was felt that perhaps the low dose improved chemotherapy delivery and anti-tumor effect; whereas, the higher dose may have resulted in vascular collapse and limited drug ­delivery to the tumor [140]. Although bevacizumab increased the PFS compared

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

367

to chemotherapy alone in colorectal carcinoma patients, improvements in PFS were not observed past 24 months, indicating that perhaps resistance can develop to antiangiogenic agents [10, 140–142]. Maintenance regimens and different dosing regimens are being explored to decrease resistance and maximize benefit. In February 2004, the US FDA approved the first anti-angiogenic agent in combination with chemotherapy based on evidence from the phase III double-blind placebo-controlled study which randomized untreated metastatic colorectal carcinoma patients to bevacizumab or placebo in combination with IFL (irinotecan/5FU/leucovorin) chemotherapy which confirmed improved RRs of 44.8% versus 34.8% (p = 0.004), a prolonged PFS of 10.6 versus 6.2 months (p < 0.001), and a OS benefit of 20.3 versus 15.6  months favoring bevacizumab [10, 123, 141, 142]. Grade 3 hypertension was more common with bevacizumab therapy, without major increases in hemorrhage, thromboembolism, and proteinuria [10, 141–143]. Rare but serious fatal gastrointestinal perforations were observed in 1.5% of patients [10, 79]. Microdissection of optional archival tumor tissue for DNA sequence analysis of k-ras, b-raf, and p53 mutational status or P53 expression from 295 patients treated with IFL +/− bevacizumab in this study could not predict which patients were likely to respond to bevacizumab, and there was no significant statistical relationship between mutational status and the increased survival associated with bevacizumab [144]. The phase III ECOG E3200 trial further confirmed the benefit of bevacizumab in previously-treated advanced colorectal cancer patients treated with FOLFOX-4 (oxaliplatin/5-FU,leucovorin) with/without bevacizumab [145, 146]. Bevacizumab significantly improved the RR (21.8% versus 9.2% p < 0.001), PFS (7.4 versus 5.5 months p < 0.001), and OS (12.5 versus 10.7 months p = 0.002) [145, 146] Another phase III 2 × 2 study of bevacizumab combined with FOLFOX-4 or XELOX (oxaliplatin and capecitabine) in untreated metastatic patients, demonstrated an improved median PFS of 10.4 months versus 8.1 months (HR = 0.63, p < 0.0001) favoring bevacizumab over placebo [147, 148]. Due to the high activity in the metastatic setting, bevacizumab is being assessed in phase III trials in the adjuvant high risk stage II and III setting in combination with modified FOLFOX-6 in the NSABP C08 study, with FOLFOX-4 or XELOX in the AVANT study, and with FOLFOX in the E5202 trial in high risk stage II patients [149, 150]. Several of these studies are ongoing and examining tissue and biomarkers in an effort to find predictive markers of response to anti-angiogenics [125, 143]. Bevacizumab (7.5 mg/kg or 15 mg/kg) with paclitaxel (200 mg/m2) and carboplatin (AUC 6) in 99 advanced/recurrent untreated NSCLC patients demonstrated an improved median TTP (7.4 versus 4.2 months, p = 0.023) and RR (31.5% versus 18.8%) at a dose of 15 mg/kg, when compared to chemotherapy alone in the phase II setting [80, 151]. However, hemoptysis and severe pulmonary hemorrhage in patients with central cavitating/necrotic SCC tumors close to major blood vessels were observed [80, 151]. On the basis of this study, the phase III ECOG 4599 study included 878 non-SCC patients randomized to carboplatin and paclitaxel, with or without bevacizumab at 15  mg/kg q3 weeks for up to six cycles, followed by ­maintenance bevacizumab therapy until PD or tolerable toxicity up to 1 year [80]. There was a significant advantage for the addition of bevacizumab in improved OS

368

L.Q.M. Chow and S.G. Eckhardt

(12.5 versus 10.2 months p = 0.007), PFS (6.4 versus 4.5 months p < 0.0001), and RR (27.2% versus 10% p < 0.0001) which led to this combination being approved in October 2006 by the US FDA as standard of care for first-line advanced nonSCC NSCLC patients [80, 123, 152]. Severe hemoptysis in 1.9% versus 0.2% was observed with bevacizumab, with 1.2% of the pulmonary hemorrhages being fatal [80]. Based on the promising efficacy, further combination therapies with bevacizumab in NSCLC are being assessed. The clinical successes in anti-angiogenesis in the first and second-line settings for colorectal and lung cancer highlight the concept that tumors become less dependent on VEGF in the later stages of their growth when angiogenesis involves many additional factors [149, 153]. The optimal time to give anti-angiogenic therapy appears to be early in the course of metastatic disease, or adjuvantly [64]. Bevacizumab combinations failed to demonstrate a significant benefit in third-line metastatic colorectal carcinoma, and a phase III study of heavily pre-treated metastatic breast cancer patients treated with capecitabine with/without bevacizumab (15  mg/kg q 3  weeks) only demonstrated an improved RR (19.8% versus 9.1% p = 0.001) without any significant improvements in PFS or OS with the addition of bevacizumab [154–156]. Conversely, bevacizumab at 10  mg/kg administered on day 1 and 15, with paclitaxel at 90 mg/m2 on days 1, 8, and 15 on a 28-day cycle in previously untreated metastatic breast cancer demonstrated statistically significant improvements in RR (36.9% versus 21.2% and p < 0.001) and in PFS (11.8 versus 5.9 months, HR 0.60, and p < 0.001), without an improvement in OS, over chemotherapy alone [157]. Controversially, the FDA approved bevacizumab for this indication on the basis of the PFS benefit [123, 158].

14.6.4 Multi-targeted Receptor Tyrosine Kinase Inhibitors Receptor tyrosine kinases (RTKs) are cell-surface/transmembrane proteins with extracellular ligand binding domains and intracellular catalytic domains transducing extracellular signals to the cytoplasm [159]. Ligand binding induces RTKdimerization, resulting in cytoplasmic domain autophosphorylation, and stimulation of multiple cytoplasmic pathways and signaling molecules which affect DNA synthesis, cell division, growth, progression, migration, differentiation, and death [159]. Sunitinib and sorafenib were the first multi-targeted RTK inhibitors (RTKI) to demonstrate clinical efficacy. Table  14.5 shows common angiogenic RTKIs with a comparison of their preclinical activity. RTKIs appear more active as single agents than bevacizumab in certain tumors, likely due to their ability to inhibit multiple targets in different oncologic pathways [27, 160, 161]. In fact, synergy is shown with PDGFR and VEGFR: PDGFR inhibition disrupts mature and growing blood vessels, better by dissociating endothelial cells from surrounding pericytes and rendering endothelial cells more sensitive to anti-VEGF therapy [27, 28, 161, 162]. The use of unique designs such as the randomized discontinuation trial (RDT) was also a successful approach for assessing the disease-stabilizing activity of these agents.

Cediranib (AZD 2171) [359–361] IC50 (uMol/L)

Axitinib (AG013736) [207, 273, 274, 362] IC50 (uMol/L)

Sorafenib (BAY 43-9006) [163, 166] IC50 (uMol/L)

Sunitinib (SU11248) [161, 181, 182, 363–366] IC50 (uMol/L)

0.077a 1.6a 0.005a 0.0012b – – VEGFR1 – 0.010a a c a a b a VEGFR2 – 0.047 –0.007 0.037 0.04 0.0005 , <0.001 0.00025b 0.030b–0.090a 0.004b VEGFR3 – 0.030a 0.62–0.66a 0.11a £0.003a 0.00029b 0.020a–0.100d – PDGFR-a 0.10–1.0e 0.071a – – 0.036a–0.005f – – 0.069b PDGFR-b 0.10b 0.084a 0.58a 1.1a 0.005a–0.008f 0.0016b 0.057a–0.080c 0.039b c-KIT 0.1g 0.074a 0.73a >20a 0.002a–0.001g 0.0017b 0.068a 0.001–0.01g h a i c a Flt-3 10.0 – – – >1 –>10 – 0.020 –0.058 0.008–0.014j FGFR1 – 0.72b – 3.6a 0.026a – 0.58a 0.88b EGFR – – – 0.5a 1.6a–1.1k – >100a >10.0a l a a c-met 37 – – >10 – >100 4.0a a a IGFR-1R – – – >200 – – >100 2.4a CSF-IR – – 1.4a – 0.21b – – 0.05–0.10b Raf-1 – – – – – – 0.006a – a Non-cellular IC50’s determined from biochemical assays (including those to inhibit recombinant tyrosine kinase, receptor phosphorylation activity using 96-well ELISA enzyme assays, scintillation proximity-based assays, and other assays) b Cellular IC50s in NIH-3T3 cells and serum starved human umbilical vascular endothelial cells (HUVEC’s), determined from inhibition of receptor phosphorylation assays and ligand-dependent cell kinase assays c Cellular IC50 assays using HaoSMC human smooth muscle cell lines d Cellular IC50 assays using HEK-293 human embryonic kidney cell lines e Cellular IC50 on CSOC-272 serious papillary ovarian tumor line expressing PDGFR-a f Cellular IC50 on MG63 cell line g Stem-cell factor stimulated KIT phosphorylation in NCI-H526 small cell lung cancer cell line h Cellular IC50 on myeloid M1 cells i Cellular IC50 on MonoMac6 cell line j Cellular IC50 proliferation assay with MV4:11 and OCL-AML5 mutant, and wild-type AML cell lines respectively k Cellular IC50 on KB cell line l Cellular IC50 on TT cells, a human medullary thyroid cell line

Table 14.5  Receptor tyrosine kinase activity of several RTKIs in development [92] Vandetanib Vatalanib (ZD6474) (PTZ787/ Imatinib (STI- Pazopanib ZK222584) [356, [206, 209, 571) [350– (GW-786034) Receptor 358] IC50 353] IC50 [164, 354, 355] 357] IC50 tyrosine IC50 (uMol/L) kinase agents (uMol/L) (uMol/L) (uMol/L)

370

L.Q.M. Chow and S.G. Eckhardt

14.6.4.1 Sorafenib Sorafenib (BAY43-9006, Nexavar; Bayer pharmaceuticals, Onyx pharmaceuticals) is a small molecule inhibitor of the Raf serine/threonine kinases (Raf-1, wild-type B-Raf, and b-raf V600E), as well as the anti-angiogenic VEGFR 1,2,3, PDGFR-b, Flt-3, c-KIT, and FGF-1 RTKs, and p38 kinases [163–167]. Sorafenib was discovered through a screen of Raf-kinase inhibitors, and later found to have anti-­ angiogenic activity with inhibition of neovascularization and evidence of tumor inhibition/regressions in xenograft models [163–167]. Phase I trials of sorafenib determined a well-tolerated continuous dose of 400  mg BID producing serum ­levels exceeding the minimum inhibitory concentrations to inhibit Raf, VEGFR, and PDGFR RTK activity [167–172]. Severe toxicities were uncommon. Stable disease and rare objective responses were seen in a variety of tumors – including RCC, HCC, and ovarian carcinoma [167–172]. A phase II RDT design enriching patients with SD was studied in 502 patients with advanced tumors (including 202 RCC patients) [173, 174]. Patients were treated with sorafenib for 12 weeks and those with modified WHO criteria of >25% tumor regression continued on drug, and those who had >25% PD were taken off study [174]. Patients who did not meet the criteria for the 25% reduction or progression were deemed SD and randomized in a placebo-controlled manner to either continuing or discontinuing sorafenib [174]. Tumor regressions were observed in 73 patients, and 66 patients with SD were randomized to sorafenib or placebo. At 24 weeks, a significant number of sorafenib-treated patients were progression-free: 59% versus 18% respectively (p = 0.007) with a 6-week longer median PFS from randomization when compared to placebo (p = 0.009) [174]. Although trial results were significant and the study design allowed the rapid identification of an active therapy, responses were low. A subsequent large multi-centered phase III trial randomized 905 metastatic clear cell RCC patients who had progression on one prior systemic therapy, to receive sorafenib or placebo, without crossover [175]. Although, only a 2% PR (RECIST criteria) was observed in the 335 patients receiving sorafenib, 78% demonstrated SD with minor responses, and PFS was significantly prolonged to 24 weeks versus 12 weeks (p < 0.000001) in the sorafenib arm over placebo [175]. After 367 events, 48% of the placebo patients had crossed over to receive sorafenib, and updated results showed improvements in OS of 19.3 versus 15.9 months (p = 0.015, HR 0.77) favoring sorafenib [175]. Sorafenib was approved by the US FDA in December 2005 for the treatment of advanced/metastatic RCC [107, 176–178]. Sorafenib is currently being compared to INF-a in treatment-naïve RCC [125]. Sorafenib is also being investigated as a single agent, in combination with chemotherapy, and as adjuvant or maintenance therapy in various tumor types [125]. A phase II study of sorafenib in 137 patients with inoperable untreated HCC patients (Child–Pugh liver function class A) demonstrated substantial activity with at least 3 PR and 46 SD >16 weeks, a median TTP of 4.2 months, and an OS of 9.2 months [108, 109]. The subsequent phase III double-blind randomized placebocontrolled SHARP trial in 602 untreated advanced inoperable HCC patients was stopped early due to superior OS in the sorafenib-treated group: based on 321 deaths

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

371

(sorafenib n = 143; placebo n = 178), the HR 0.69 (95% CI: 0.55, 0.87; p = 0.0006) for OS favored sorafenib, representing a 44% improvement in OS (10.7 versus 7.9 months,), with improvements in median TTP (5.5 versus 2.8 months) HR 0.58 (95% CI: 0.45, 0.74; p = 0.000007) and disease control rate (43% versus 32%), ­compared to placebo [179]. These highly significant results led to the US FDA approval of sorafenib in the treatment of HCC on November 16, 2007 [123]. 14.6.4.2 Sunitinib Development of the anti-angiogenic multi-targeted RTKIs, SU5416 (semaxanib, Sugen Inc.) and SU668 (Sugen Inc), was abandoned in favor of sunitinib (SU11248, Sutent, Pfizer Ltd) due to its superior preclinical activity and favorable PK profile [180–182]. Sunitinib is a rationally designed inhibitor with micro to nanomolar range potency against VEGFR-1, -2 and PDGFR-a, -b, stem cell factor receptor, c-KIT, FGFR-1, and the FLT3-ligand receptor. It has high solubility, stability in serum, good bioavailability for oral administration, and high efficacy in producing growth ­inhibition and tumor regressions in a variety of xenograft models [92, 183, 184]. In early phase trials, daily dosing with 50 mg reached target plasma concentrations of >50 ng/mL for preclinical PDGFR and VEGFR RTK inhibition, and DLTs of fatigue, asthenia, and thrombocytopenia at 75 mg established the RPTD of 50 mg on discontinuous schedules (2/1, 2/2, or 4/2) [92, 184–187]. Objective responses were observed in patients with thyroid cancer, metastatic RCC, neuroendocrine cancer, GIST, ­sarcoma, unknown primary adenocarcinoma, NSCLC and in melanoma, forming the basis for phase II/III clinical development [92, 184, 188]. Two single-arm phase II trials of sunitinib at 50 mg/day on a 4/2 discontinuous schedule demonstrated high RRs of 39–40%, SD > 3 months in 23–27% of patients, a median TTP of 8.7 months and an OS of 16.4 months in refractory RCC patients [92, 189–192]. These results led to the accelerated FDA approval of sunitinib for the treatment of advanced RCC [92, 192, 193]. A randomized phase III trial ­comparing sunitinib to INF-a at 9 million units SC three times weekly, in 375 untreated RCC patients demonstrated a highly significant early median PFS benefit (47.3 versus 24.9  weeks, HR 0.394, p < 0.000001) and RR (24.8% versus 4.9% p < 0.000001) favoring sunitinib over INF-a, further confirming the clinical activity of sunitinib in RCC [194, 195]. In comparison to other anti-angiogenic agents in RCC, bevacizumab (10 mg/kg), in a randomized phase II study in advanced RCC demonstrated a increased TTP of 4.8 versus 2.5 months, (p < 0.001), a 10% RR, but no survival benefit over placebo [9, 107, 196]. Sorafenib demonstrated a superior PFS and an OS benefit, with a high predominance of SD; but only a 2% RR [160, 175, 197]. Why higher tumor regressions were observed with sunitinib compared to sorafenib merits further investigation. AG-013736, an oral anti-angiogenic RTKI similar to sunitinib, also demonstrated high RRs with 24 PR (46%) and 21 SD (40%), without reaching the TTP after 12–18 months of follow-up in 52 evaluable patients [198, 199]. The predominant clinical benefit appears to be disease ­stabilization leading to a longer PFS in all these agents [200]. However, without

372

L.Q.M. Chow and S.G. Eckhardt

direct comparisons of these drugs, it is difficult to assess the superiority of one agent over another. As single agents, multi-targeted RTKIs appear more active than bevacizumab alone, likely due to their potency against a broad range of RTKs [107, 160, 196, 201]. Preclinical potencies of the RTKIs are described in Table 14.5. In early phase studies in imatinib-refractory (or intolerant) GIST patients, sunitinib demonstrated tumor regressions and a predominance of sustained SD leading to improvements in TTP, which were felt to be due to its anti-angiogenic effects, in addition to its effects on c-KIT [92, 184, 202, 203]. A double-blind, placebocontrolled, randomized phase III trial of sunitinib (50  mg on a 4/2 schedule) ­confirmed its efficacy by exceeding its primary endpoint with an almost fourfold increase in the TTP of 27.3 weeks in the sunitinib group compared to 6.4 weeks in the placebo group [204]. Despite the crossover of 59 patients on the placebo arm to sunitinib, there was a survival benefit for the sunitinib group with a HR of 0.491 (95% CI 0.290–0.831 p = 0.007), which led to FDA approval of sunitinib for this indication [92, 204, 205]. Sunitinib has also shown promising single agent activity in phase II studies of in breast cancer, colorectal carcinoma, neuroendocrine ­carcinomas and NSCLC [92]. Many other promising studies of sunitinib alone and in combination with chemotherapy, and radiation are ongoing in a variety of tumor types [178]. 14.6.4.3 Other Anti-angiogenic Multi-targeted Receptor Tyrosine Kinase Agents The anti-angiogenic multi-targeted RTK agents in clinical development and their preclinical potencies, respectively, are shown in Tables 14.2 and 14.5. Vandetanib (AZD6474), a RTK inhibiting VEGFR, EGFR, and RET dependent pathways, demonstrated improvements in PFS in advanced NSCLC [206–209]. It is currently being explored in the phase III setting alone and in combination with chemotherapy [206–209]. As bevacizumab was so active with 5-FU therapy, in colorectal carcinoma, it was puzzling that the multi-targeted RTKIs SU5416 and vatalanib (PTK787/ZK222584) failed to demonstrate a survival benefit in phase III trials in combination with 5-FU-based therapy [65, 210–215]. Despite dynamic contrast enhanced (DCE)-Magnetic resonance imaging (MRI), evidence of anti-angiogenic activity with vatalanib, the CONFIRM-1 trial of vatalanib with FOLFOX versus FOLFOX alone in the first-line metastatic setting demonstrated only a small increase in PFS that did not reach statistical significance (HR-0.88, p = 0.118) [210, 213] and the CONFIRM-2 trial of the same regimen in 855 irinotecan-refractory colorectal carcinoma patients, also demonstrated a slight increase in PFS with ­vatalanib of 1.4  months (5.5  month versus 4.1  month, HR = 0.83, and p = 0.026), without any improvements in OS [211]. Perhaps the short half-life of vatalanib did not maintain sufficient sustained levels in the serum or PDGFR-b targeting interfered with vascular normalization by blocking perivascular cell recruitment and possibly impeded chemotherapy delivery [216, 217]. The utility of DCE-MRI to predict clinical benefit is also brought into question. Thus far, it is not clear whether

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

373

the effects observed on DCE-MRI by anti-angiogenic agents indicate anything beyond an impact on intratumoral blood flow dynamics [210, 218–222]. Many other anti-angiogenic multi-targeted RTKIs do show promise as single agents and in combination chemotherapy, and are being explored with radiation therapy or in trimodality regimens.

14.6.5 Other Approaches to Angiogenesis Inhibition There are many mediators and inhibitors of angiogenesis, as shown in Table 14.3, with many others that are yet to be discovered or understood. Clearly, the interactions involving angiogenesis in vivo are very complex. We still do not understand the exact mechanisms behind the clinical success of some agents and the failure of  others. Furthermore, we do not know how to select patients for treatment or ­understand the clinical or biological factors underlying response or resistance [73, 75]. In addition to monoclonal antibodies and multi-targeted RTKIs against growth factors of VEGF and PDGF, new anti-angiogenic agents under exploration include: inhibitors to PIFG, FGF, Tie-2, heparanase, integrin and cadherin adhesion  molecules, intracellular ­signaling molecules such as ILK, FAK, mTOR and Src, chemokines and their receptors (CXCL12/CXCR4), vascular guidance and regulators of differentiation such as DLL4/notch, slit/robo/netrin/neogenin and eph/ephrin, Id regulators of transcription, and soluble high affinity cytokine receptors [34, 64, 70, 73, 75, 223, 224]. Peptide, antibody and small molecule inhibitors are under development, as are aptamers (synthetic nucleic acid ligands binding to protein targets), anti-sense oligonucleotides, short interfering RNA that hybridize to specific mRNA sequences inducing mRNA degradation or inhibition of translation, vascular disrupting agents, and viral gene therapy [224]. Exploration of peptide therapy in modulating endogenous compounds and/or receptors is merited as they are easy and inexpensive to produce [224]. However, they require IV administration due to their short half-lives [224]. Peptides have been developed to inhibit endothelial cell surface receptors, integrins, cadherin adhesion molecules, and other anti-angiogenic small molecules [224]. The list in development is extensive and an anticipated multitude of new anti-angiogenic agents will be in future clinical development. Discussed subsequently are a few of the more promising unique agents being developed. 14.6.5.1 Inhibitors of Endogenous Compounds or Receptors More than 40 endogenous natural inhibitors of angiogenesis are known and are being evaluated as therapeutic agents [34]. Early clinical trials with angiostatin, endostatin, thrombospondin, and 2-ME demonstrate limited efficacy [42, 223]. These agents are difficult to manufacture and produce due to stability and solubility issues, and susceptibility to angiogenesis inhibitors can vary by tumor stage [42, 223]. ABT-510 is a peptide mimetic of thrombospondin-1 which is being developed

374

L.Q.M. Chow and S.G. Eckhardt

as an angiogenesis inhibitor [225]. Aminopeptidase N (APN or CD13) is expressed selectively only on endothelial cells undergoing active angiogenesis as it is involved in the cleavage of active peptides such as enkephalins, endorphins, and angiotensins [34]. Anti-APN antibodies and the APN antagonist, bestatin, are in early clinical development due to promising anti-angiogenic effects and tumor regressions in breast cancer models [226]. Integrins mediate signal transduction and cellular attachment to the ECM on growing endothelial and tumor cells and on cancer-associated, but not mature endothelia [223]. Peptide inhibitors and anti-avb3 integrin antibodies block tumor angiogenesis to promote tumor regression in animal models [50]. Abegrin (formerly Vitaxin) is a humanized anti-avb3 integrin antibody in phase II clinical trials for advanced metastatic melanoma and androgen-independent prostate cancer with bone metastases [34, 223]. Cilengitide (EMD 121974), a cyclized RGD pentapeptide blocking integrin avb3 -and avb5-mediated endothelial cell attachment and migration, is a well-tolerated anti-angiogenic agent being explored in the phase I/II setting in glioblastoma multiforme with radiation, and in lymphoma, prostate cancer, and melanoma [227]. PI-88 (Progen Ltd, Darra, Queensland) is a highly reproducible sulfonated ­oligosaccharide mixture that inhibits heparanase cleavage of HS and competes for HS-binding of FGF and VEGF with potent anti-angiogenic and anti-metastatic effects in preclinical models [228–230]. Dose-limiting immune-related thrombocytopenia was observed with limited biologic activity in the IV formulation; whereas, the SC formulation was well tolerated up to 250 mg/day with linear PK and antitumor activity – particularly in melanoma [231–233]. The activated partial thromboplastin time (APTT) serves as a surrogate marker of PI-88 activity [232, 233]. Phase II data demonstrated that PI-88 at 160  mg SC administered to 172 HCC Taiwanese patients following curative liver resection, improved the disease-free rate by 25% at 48 weeks with a 78% improvement disease-free survival, meriting FDA Fast Track status in September 2007 [131, 178, 234]. A confirmatory phase III PI-88 adjuvant trial with an aim of 600 HCC patients worldwide began enrollment, but was halted by the company due to low accrual once sorafenib was approved for HCC in 2008 [125, 235, 236]. 14.6.5.2 Small-Molecule Vascular Targeting Agents Anti-angiogenic agents affect new blood vessels formation, whereas, vascular targeting agents disrupt mature tumor vasculature causing vascular structures inside established solid tumors to collapse, decreasing blood, oxygenation and nutrient supplies, there by causing secondary hemorrhagic necrosis [237]. These agents do not have specific targets nor do they select out particular tumors; instead, they preferentially affect differences in tumor vasculature. Tumor vasculature has greater proliferation, fragility, more irregular intercellular openings and overlap, higher vascular permeability, and internal fluid pressure than normal vasculature [237, 238]. Vascular targeting agents showed promise in early clinical trials and are being further developed. Auristatin (TZT-1027) is a synthetic cytotoxic pentapeptide dolastatin derivative

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

375

which inhibits microtubule assembly by interacting with ­tubulin in the vinca alkaloid binding domain to collapse of tumor vasculature with  potent cytotoxic and antivascular effects. Adverse effects included nausea, vomiting, diarrhea and fatigue [238]. DLTs were neutropenia, and arm infusion pain [238]. As a heavily pre-treated liposarcoma patient had a PR lasting >54 weeks in a phase I clinical trial of 17 patients, this drug is being further explored in phase II in advanced soft tissue sarcoma [238, 239]. Combretastatin A-4 phosphate (CA-4-P) is a prodrug of CA-4 which binds to tubulin, causing microtubule depolymerization, complete inhibition of tumor blood flow, and selectively resulting in tumor necrosis; however, it has not increased OS due to viable cells in the peripheral rims of tumors contributing to tumor regrowth [240–242]. In several phase I trials of IV CA-4-P alone, anti-tumor activity with a complete response (CR) in an anaplastic thyroid cancer patient, a PR in a metastatic soft-tissue sarcoma patient, and a minor response in adrenocortical carcinoma were observed [243–246]. There were toxicities of hypertension, fatigue, pain, diarrhea, hypotension, vomiting and dyspnea and DLTs of reversible ataxia, neuropathy, and vasovagal syncope [238, 243–246]. In combination with chemotherapy, prolonged SD has been observed; therefore, phase II trials are further exploring CA-4-P [237, 247]. DMXAA (5,6-Dimethylxanthenone-4-acetic acid) is a small molecule flavonoid. It activates tumor-associated macrophages to release chemokines and inflammatory cytokines, including TNFa and CD8+ T cells to potently produce rapid tumor blood flow decreases and subsequent necrosis without a viable peripheral rim within 24 h in a variety of solid tumor mouse models [248– 250]. Two unconfirmed PRs, and 28 patients with SD were observed in a phase I trial involving 109 patients; [251] therefore, phase II combination studies are underway with DMXAA and taxanes in metastatic prostate cancer, and with carboplatin and paclitaxel for lung and ovarian cancers [248]. 14.6.5.3 Viral Gene Therapy Oncolytic viruses can inhibit angiogenesis by carrying genes that stimulate the production of angiogenesis inhibitors, through direct targeting of endothelial cells, or through triggering inflammation to mediate vascular shutdown [252]. Cancer cells are genetically instable and prone to resistance, as opposed to endothelial cells which are genetically normal, stable, and less likely to carry mutations for ­resistance [17]. Gene therapy mediated through oncolytic viruses may potentially better target endothelial cells and stimulate endogenous angiogenesis inhibitor production [253]. Adenovirus dl922/947 is a replication-competent oncolytic virus which has successfully mediated endostatin delivery and inhibited mammary carcinoma growth in SV40 T-antigen transgenic mice [254]. When this virus was combined with a ­non-replicating adenovirus encoding a soluble VEGF Ad Flk1-Fc receptor, anti-­ angiogenic anti-tumor effects in colon and prostate xenograft models were increased [255–257]. A replication competent oncolytic herpes virus expressing IL-12 (NV1042) also demonstrated substantial tumor regression in a murine model of SCC. IL-12-induced stimulation of T helper cells, cytotoxic T ­lymphocytes and

376

L.Q.M. Chow and S.G. Eckhardt

natural killer cells, and induction of INF-g mediated anti-angiogenic effects – as evidenced by decreased microvessel density and Matrigel plug assays [258]. Herpes simplex virus-1 (HSV-1) G207 directly inhibits angiogenesis by infecting endothelial cells with subsequent inhibition of in  vitro tube and vessel formation in the Matrigel assay [259], an effect which is blocked by acyclovir. Anti-tumor activity is enhanced when low-dose G207 is combined with chemotherapy with in rhabdomyosarcoma and NSCLC, or with erlotinib in malignant peripheral nerve sheath tumors xenograft models [259–261]. Furthermore, using a G207 backbone as a new vector to incorporate genes for potent angiogenesis inhibitors such as the endostatinangiogstatin fusion protein AE618 gene, human umbilical vein endothelial cell growth and NSCLC xenograft growth could be inhibited with potent oncolytic and anti-angiogenic effects [262]. Oncolytic virus infection can induce inflammation which triggers endothelial dysfunction, microvascular thrombosis, and blood clot formation leading to hypoperfusion, ischemia, and tumor cell death [252, 263]. Vesicular stomatitis virus (VSV), systemically administered, was rapidly, extensively and selectively amplified in CT-26 tumor cells surrounding tumor neovasculature in immunecompetent BALB/c mice to trigger inflammation, tumor cell death, and loss of tumor perfusion [252]. Oncolytic viruses are more active when combined with anti-angiogenic or vascular targeting agents as these agents shut down the blood supply, decrease nutrient and oxygen supplies, slow down tumor growth and trap virus in the tumor, allowing viral spread and preventing viral loss in the general circulation [252]. Furthermore, combination approaches may help attack the ­viable tumor rim, ­commonly observed post oncolytic-virus infection and vasculartargeting therapy (as shown in Fig. 14.2), to prevent tumor outgrowth [264].

Fig. 14.2  Tumor with a viable rim (V) and central necrosis (N) following exposure to a vascular disrupting agent or oncolytic virus therapy (sourced from [347]). Persistent cellular proliferation and intact vessels are observed associated with tumor outgrowth typical of treatment with vascular disrupting agents and various oncolytic virus therapies [252, 347]

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

377

14.7 Combination Therapy with Anti-angiogenics 14.7.1 Anti-angiogenics and Chemotherapy As anti-angiogenic therapy alone uncommonly induces tumor regression, it is maximized in combination with agents that do induce regression [265]. Many preclinical studies have demonstrated additive effects or synergy when anti-angiogenic agents are combined with chemotherapy or radiation therapy, with superior antitumor effects when all modalities are combined, compared to single or dual therapy [68, 265, 266]. Anti-angiogenic agents change tumor vasculature and morphology [267, 268]. Tumor vessels that survive therapy have increased permeability and reduced interstitial pressure which lead to increased perfusion, facilitation of chemotherapy diffusion and delivery, reduced intratumoral hypoxia, and enhanced oxygen pressure in the tumor mass [269]. In particular, anti-VEGF agents increase vessel permeability and extravasation of cytotoxic agents, and block autocrine and paracrine stimulation to reduce secretion of endothelial and tumor growth factors, respectively [270]. PDGFR antagonists improve chemotherapy uptake and thereby enhance anti-tumor effects [28]. Synergistic effects with multi-targeted anti-angiogenic RTK agents may also be due to both indirect effects on endothelial cells and direct effects on tumor cells [27, 161]. There is a short critical window by which tumor vessel normalization occurs to increase tumor cell exposure to cytotoxic drugs; therefore the dosing, timing, and sequencing of chemotherapy with anti-angiogenics is very important to therapeutic response [99, 267, 268]. As previously discussed, variations in bevacizumab dosing in combination with chemotherapy can affect the clinical response [140]. As opposed to high doses in the preclinical setting, low doses of sunitinib improve temozolomide tumor concentrations in clinical trials [271]. The optimal timing is yet to be determined and may differ with the type of anti-angiogenic therapy [268]. In animal studies, inhibiting antibodies to VEGF and VEGFR improve tumor blood flow and drug delivery in a transient “normalization window”; but chronic dosing and inhibition reduce tumor blood perfusion and increase tumor hypoxia in animal studies – indicating that intermittent dosing may be more advantageous [272]. Multi-targeted anti-angiogenic RTK agents may improve tumor chemotherapy drug uptake; however, they predominantly decrease drug uptake and increase tumor hypoxia preclinically [99, 273, 274]. Although these multi-targeted RTK agents were more active as single-agents than bevacizumab in certain tumor types, in combination with chemotherapy, they did not demonstrate the significantly improved survival benefits seen with bevacizumab – likely due to unknown effects on tumor drug uptake and the “normalization window” [99, 211]. Preclinical studies have shown that anti-angiogenic efficacy is enhanced when combined with frequent continuous dosing of chemotherapy at low sub-cytotoxic dosages i.e. metronomic dosing [72, 275]. The shorter intervals between treatments decrease the chance of endothelial and tumor recovery and resistance, therefore, increasing apoptosis and tumor growth suppression [72, 275]. In addition, lower

378

L.Q.M. Chow and S.G. Eckhardt

doses decrease myelosuppression due to decreased bone marrow progenitor and hematopoietic stem cell damage [73, 276]. As endothelial cells proliferate at a lower rate than tumor cells, they are less affected by traditional chemotherapy and  more affected by metronomic dosing [277, 278]. Metronomic approaches of ­chemotherapy with anti-angiogenic agents are currently being explored in breast cancer and other tumor types [275, 277, 278]. Overall, combining anti-angiogenic therapies with chemotherapy has been very successful, and further combinations with radiation therapy, chemoradiation and other targeted therapies are currently underway [136, 278].

14.7.2 Anti-angiogenics and Radiation Therapy Hypoxia and hypoxia-inducible factor-1a overexpression is associated with increased microvessel density and/or VEGF expression in a large number of tumor types, correlating with lower radiation response, cancer progression, and increased mortality risk [279, 280]. Tumors recurring after radiation therapy have a higher propensity to metastasize due to increased hypoxic cell fractions, hypoxia-induced upregulation of metastasis, and promotion of hypoxia-induced neoangiogenesis gene products [281]. Radiation induces transient tumor hypoxia, stimulating VEGF production, VEGFR-2 expression, and upregulating the endothelial cell nitric oxide pathway to promote tumor angiogenesis [282]. Radiation-induced increases in VEGF levels protect endothelial cells from radiation-induced cytotoxicity and vascular damage, serving as a paracrine ­proliferative stimulus to promote tumor growth outside the radiation field [283]. Therefore, VEGF inhibition during fractionated radiation therapy antagonizes the increased hypoxia and causes tumor growth delay and apoptosis [284]. ­Anti-angiogenic therapy combined with radiation enhances radiation-induced cell death, due to normalization of tumor vasculature and increased oxygenation [267, 280]. Sequencing and timing of anti-angiogenic agents and radiation therapy around the vascular normalization process are crucial for optimization of therapy, and this process differs for each agent and each tumor type [280]. The vascular targeting agent, ZD6126, showed high tumor cell kill when administered 24 h prior, or more than one hour after radiation therapy; however, it was not as effective if given one hour before radiation therapy due to acute ZD6126-induced tumor hypoxia [280, 285]. Similarly, the VEGFR2-specific monoclonal antibody DC101 administered concurrently with radiation in a xenograft model demonstrated additive tumor growth delay [280, 286]. Conversely, radiation administered 4–6  days after DC101 treatment demonstrated synergistic growth delay, correlating with the time of maximum tumor oxygenation and pericyte vessel coverage [280, 286]. Surprisingly, maximal tumor responses were observed when bevacizumab was administered post-radiation therapy [287]. The anti-angiogenic multi-targeted RTK agents vandetanib, sunitinib and vatalanib were each

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

379

c­ ombined concurrently with radiation, and greater anti-tumor effects were seen in combination than singly [284, 288–291]. Decreased tumor perfusion, impaired re-oxygenation between fractions of radiation therapy, tumor growth delay, and tumor cell apoptosis were evident [284, 288–291]. The enhanced anti-tumor effects of sunitinib with radiation were independent of sequencing; however, sunitinib maintenance therapy post-radiation was effective in preventing tumor regrowth [289, 290, 292]. Vandetanib when administered post-radiation therapy, further enhanced efficacy in NSCLC; however, sequencing did not affect tumor response in a colorectal cancer model [280, 288, 291]. Although combining radiation therapy with anti-angiogenic agents led to additive to synergistic anti-tumor effects, increases in radiation-induced toxicity to normal tissue were observed. The myocardial damage and reduction in ejection fraction observed with sunitinib and imatinib may be potentiated by radiation therapy – particularly near the heart [87, 192, 280, 293]. PDGF and TGF-b play a role in radiation pneumonitis [28], and VEGF is elevated in chronic obstructive pulmonary disease (COPD) [294, 295]. PDGF inhibitors can potentially reduce radiation fibrosis, and anti-angiogenic RTKIs can also cause alveolar cell apoptosis and worsening COPD [28, 280, 294–297]. Radiation therapy may worsen bevacizumab-induced non-healing erosion of the nasal septum [298], and its effects on pulmonary hemorrhage in SCC patients treated with bevacizumab and chemotherapy are unknown [80, 81]. The toxicity to the bowel observed with anti-angiogenic agents is increased in combination with radiation therapy: acute ischemic colitis, and obstruction and perforation have been reported when bevacizumab was administered after pelvic radiotherapy in patients with colorectal carcinoma [77, 299]. These effects are likely due to VEGF: anti-VEGFR2 antibodies administered one month after whole-body irradiation in mice resulted in high fatal bowel toxicity [286]. As VEGF is also involved in neuroprotection and repair of neuronal cells, neurotoxicity can potentially worsen with combined therapy [300]. Although increased efficacy can be observed with radiation therapy, anti-angiogenics, and potential chemotherapy, there needs to be great caution with the design of these trials in terms of the optimal sequencing and timing of these combined therapies to maximize efficacy and minimize toxicity.

14.8 Correlative Anti-angiogenic Studies As anti-angiogenic agents act primarily on blood vessels, their effects on the tumor are most often indirect. Clinical outcomes with anti-angiogenic drugs generally translate into SD and improved TTP, rather than tumor regression or response. Therefore, tumor regression via standard RECIST or WHO criteria is not a good measure of anti-angiogenic efficacy, and surrogate markers of antiangiogenic effects are needed [301]. The ideal surrogate marker of activity should be non-invasive, readily available, affordable, robust, sensitive, reproducible, and specific for tumor-associated angiogenesis [70, 71, 302, 303].

380

L.Q.M. Chow and S.G. Eckhardt

Table  14.6  Surrogate biomarkers and radiologic markers of anti-angiogenic activity (modified from [70, 71]) Markers Technique Comments Molecular markers Circulating growth factors such as Elisa, WB proteomics, and Prognostic but low VEGF, FGF, and MMP’s antibody chips predictive power Elisa, WB proteomics, and Limited to “known” Circulating endothelial cell antibody chips molecules molecules such as VCAM1, L-selectin, and ESM1 Prognostic and predictive Promising to identify Circulating proteins and fragments Elisa, WB proteomics, and antibody chips novel molecules such as endostatin and in serum or tumstatin microdissected tumor tissue Cellular markers Immunohistochemistry Microvascular density (MVD) assessing for CD31/CD34 positive vessels Circulating endothelial cells such Flow cytometry as CD45-, CD31+, CD146, CD133+, CD34, CD144+, and KDR+

Prognostic but not predictive Requires repeat biopsies Promising but complex Non-standardized assay procedures

Radiologic markers Depends on the Molecular targeting and imaging Target antibody and peptide specificity and coupled with a tracer and MRI, efficacy of targeting CT or SPECT detection Does not allow dynamic measurements Allows measurement of Vascular blood flow imagingDynamic contrast enhanced blood flow, blood based techniques (DCE)-MRI or CT, volume, mean transit PET, Power Doppler, time, p02, pH, and contrast-enhanced extravascular volume, ultrasound and able to do serial assessments to see effects of therapy Promising, inexpensive, and safe method to measure tumor blood flow in animals and patients No standardization of protocols or technique

It should also correlate closely with anti-angiogenic and anti-tumor efficacy in different tumor types at various tumor stages [70, 71, 302, 303]. Selected molecular, cellular, and radiologic markers under investigation are summarized in Table 14.6.

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

381

14.8.1 Biologic Markers of Angiogenesis Biologic markers of angiogenesis are being explored, but none have been validated due to variable and inconsistent results across different disease stages, tumor types and therapies [71, 301]. Tumor biopsies demonstrating high microvascular density (MVD) were originally thought to be predictive of recurrence, metastases, and decreased survival in a variety of cancers, including prostate, bladder, cervical, esophageal, gastric, colorectal cancers, and melanoma [304]. Despite requiring cumbersome repeat biopsies, MVD has not been clinically validated as a predictive marker [304]. This discrepancy may be due to the fact that MVD indicates only the presence of high numbers of microvessels on a high power field, which does not determine the tumor’s dependence on blood vessels for growth [304, 305]. Assessment of circulating angiogenesis-associated growth factors in serum has also been ineffective as predictive or PD markers of activity [62]. Serum VEGF, FGF2, or HGF levels were initially thought to be prognostic and correlating with disease progression in breast, head and neck, colorectal, ovarian, lung, esophageal, and gastric cancers [62, 306–309]. Unfortunately, there are a multitude of angiogenic and antiangiogenic factors with varying binding capabilities in the blood. Furthermore, angiogenic growth factor serum levels fluctuate rapidly with time due to multiple sites of production, storage, and complex cross regulation [310]. Clinically, circulating growth factors have not predicted response: instead of decreasing with anti-angiogenic therapy, they appear to increase with anti-angiogenic therapy or chemotherapy as a positive feedback response to tumor necrosis and hypoxia [71, 311]. Cell surface molecules released by active angiogenesis and proliferating endothelial cells could potentially be surrogate markers – these include cell adhesion molecules, growth factor receptors, and ECM fragments [312]. Soluble VEGF-r1/Flt- was observed in patients with colorectal and breast cancer, but not from healthy patients [313–315]. Increasing serum VCAM1 levels correlate with disease progression and activity in breast cancer [309, 316, 317]. Other circulating molecules such as sTie2, sVEGFR1, VEGFR2, sVAM, and sE-selectin measured pre and post therapy did not demonstrate any obvious changes [71, 309, 316, 317]. Clearly, there are discrepancies in the use of cell surface molecules as biomarkers of angiogenic activity, and further examination and research are required. Circulating endothelial cells (CEC) in the peripheral blood and bone marrow derived circulating endothelial cell progenitors (CECP) were observed to increase in the blood of breast and lymphoma patients over that of healthy patients [71, 318]. Mature CECs are CD45 negative, but may be easily detected with a CD146 marker, along with other mature endothelial markers such as von Willebrand factor and vascular endothelial cadherin [318–320]. CECPs can be identified by hematopoetic stem cell progenitor markers such as CD34 and CD133 [48, 321–323]. Marker frequency can be measured by flow-cytometry. It may increase with the degree of growth-factor induced angiogenesis and may decrease in frequency with angiogenic inhibition [320, 322, 323]. In fact, CECPs may be recruited to tumor sites in

382

L.Q.M. Chow and S.G. Eckhardt

response to chemotherapy [321, 324]. CECs and CECPs, with more investigation, may be potentially useful biomarkers assessing anti-angiogenic response pre and post therapy; however, it is not clear whether validated assays can be developed for widespread use. Other assessments of markers including proteomics analysis to monitor tumor angiogenesis with tumor interstitial fluid sampling, analysis of angiogenic proteins, and gene expression profiling are early in exploration and merit further study [325, 326].

14.8.2 Radiologic Markers of Angiogenesis There are many radiological imaging techniques such as DCE, MRI, computer tomography (CT), ultrasound, doppler ultrasound, position emission tomography (PET), and single photo emission computer tomography (SPECT) that can be used to monitor blood flow, blood volume, transit time, permeability, pH and, pO2 to assess tumor perfusion and drug effects in animals and patient studies [70, 305]. Their use is not fully validated and they are still being assessed experimentally with a need for standardization of protocols, a determination of specificity, efficacy, and the possibility of dynamic measurements. Phase I and II studies assessing MRI and PET preliminarily results indicate that these radiologic imaging techniques reflect changes in vascular permeability, volume fraction, or metabolism after therapy, but they do not necessarily predict the clinical efficacy of anti-angiogenic agents [219, 270]. DCE-MRI generally uses gadolinium as a marker of extravasation, and utilizes PK time-intensity curves analyses to quantify the degree and changes in tumor vascularization pre and post therapy [70, 218, 327]. Despite over thirty DCE-MRI studies of anti-angiogenic and vascular targeting agents reported in early phase studies, early correlations with drug efficacy have only been reported with a few agents – vatalanib, bevacizumab, axitinib, cediranib, and CA-4-P [328]. The use of DCE-MRI as an indicator of clinical outcome has not been confirmed [221, 328]. Difficulties of DCE-MRI include technical standardization, variability of enhancement kinetics and Ktrans measurements [220, 329], poor correlation with PK models, low reproducibility, and lack of validation with biomarkers of angiogenesis [219, 221, 222, 330, 331]. In a phase I study of advanced colorectal cancer patients treated with vatalanib, there was a significant reduction in DCE-MRI parameters after administration of vatalanib, with a significant relationship between reduction of contrast enhancement and tumor regression [210]; however, despite DCE-MRI changes, no survival benefit was observed with the addition of vatalanib in the phase III setting of vatalanib with FOLFOX-4 versus FOLFOX-4 alone [213]. PET and SPECT can be used to assess blood flow, blood volume, and vascular permeability – especially with the use of specific radiolabeled molecules [71]. A variety of radioactive nucleotides can monitor blood flow within tumors such as

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

383

radioactive forms of water labeled with 15O, 94mTc-labeled erythrocytes ­measure blood volume, and radiolabeled carbon monoxide 11CO irreversibly binds red blood cells to measure vascular volume [270]. Although not a direct measure of biological activity, 18F-deoxyglucose (FDG)-PET is a validated marker to monitor cellular metabolism and determine metabolic changes to therapy [332–334]. Both SPECT and PET can monitor anti-angiogenic drug PK – 99mTc-labeled endostatin can determine its tissue distribution and correlate with response when administered alone or in combination with paclitaxel in rat xenografts [335]. An example of the utility of PET was demonstrated in a study of advanced solid tumor patients treated with sunitinib at 50  mg on Schedule 4/2: 13 patients (45%) demonstrated >20% reduction in the FDG standard uptake value (SUV), correlating with anti-tumor activity assessed by conventional radiological imaging; however, tumor re-growth and increased PET metabolism were evident during the off-therapy period between cycles [336]. Using O15-H2O PET to image the tumor blood flow in 55 patients, sunitinib treatment produced changes in plasma VEGFR and soluble VEGFR2 levels, which appeared to be associated with decreases in blood flow, a fall in the O15-H2O SUV in the tumor, correlating with a decrease in FDG-SUV on parallel standard FDG-PET imaging [337]. Synthesis of other novel PET tracers and imaging agents, such as [18F] sunitinib, may facilitate the ability to more directly assess the distribution and biological activity of these agents in  vivo [338]. PET and SPECT are promising, but still in early exploration as radiologic markers of antiangiogenic activity.

14.8.3 Mechanism-Based Toxicities and Clinical Markers of Anti-angiogenic Effects Hypertension, bleeding, thrombotic events, and proteinuria were the most frequent VEGF and VEGFR inhibition mechanism-based toxicities observed with bevacizumab and other anti-VEGF therapies in the phase II and III setting [68, 83, 339]. Hypertension may be dose-related, is more frequently occurring in patients treated with bevacizumab doses ³10  mg/kg and is very common in VEGF-inhibiting therapy [78, 83, 133]. The mechanisms behind hypertension are not well elucidated and may include vascular rarefaction, endothelial dysfunction, altered nitric oxide metabolism, and aberrant neurohormones [76]. Abnormal hemostasis with bleeding and clotting is due to inhibition of VEGF’s effects on regulating the vascular endothelium [82]. VEGF inhibition decreases endothelial renewal and repair, therefore, increasing the bleeding tendency in response to trauma [82]. VEGF inhibition also leads to endothelial dysfunction and defects in the vascular lining which expose subendothelial collagen and increase the frequency of thrombosis [82]. Proteinuria and nephrotic syndrome can occur when the mitogenic effects of VEGF on glomerular endothelial cells and glomerular repair are inhibited [340–342]. These mechanism-based toxicities could potentially be considered as PD markers of anti-angiogenic drug effect.

384

L.Q.M. Chow and S.G. Eckhardt

Clinical observations may also be used as markers of anti-angiogenic drug effects. Hair depigmentation, a characteristic sign of KIT RTK inhibition, was commonly observed in sunitinib-treated patients [343]. Erythema of the face, scalp, and upper thorax are observed in patients after 1–2  weeks of sorafenib and improve when therapy is continued [176, 344, 345]. Skin changes of PPE, observed with many of the anti-angiogenic multi-targeted RTKS, were due to dose-dependent effects on VEGFR3 and FLT-3 [176, 344, 345]. Although these receptors were not observed on keratinocytes, these cells were significantly morphologically changed on histological examination [344, 345]. Signs of possible VEGFR inhibition also included non-thrombotic/embolic painless distal subungual splinter hemorrhages that occur after 2–4  weeks of therapy with sunitinib, sorafenib, and other multitargeted RTK agents [346].

14.9 Conclusion The successful development of anti-angiogenic agents is one of the greatest oncologic milestones. This class of agents has had a long and difficult clinical development fraught with trial and tribulation before finally achieving clinical efficacy. Although these agents have demonstrated modest single-agent efficacy, patient selection will be important to determine who will respond to this novel class of agents. Combination therapy with chemotherapy demonstrates strong preclinical evidence of additive activity or synergy, and clinical trials have demonstrated that combination therapy can improve response and survival in certain tumor types. Antiangiogenic agents improve chemotherapy delivery to the tumor; however, this may not be the only mechanism for anti-tumor activity and OS. The selectivity of these agents may play a role in their activity – more selective agents such as bevacizumab appear to be more active in combination with chemotherapy, whereas, less selective multi-targeted RTK anti-angiogenic agents appear to be more active as single agents. The selectivity of these agents and their clinical activity is intriguing and merits further investigation. This new class of agents is not free of toxicities and establishment of tolerable doses for chronic prolonged administration is difficult, as well as the management of these toxicities in the general oncologic population, outside of the clinical trial setting. In combination, whether vertical concurrent therapy versus horizontal sequenced therapy will confer less toxicity is unknown, and requires further exploration. Greater understanding regarding mechanism-based toxicities and other non-specific toxicities will be vital for development of this class of agents. There needs to be identification of clinical scenarios where anti-angiogenic agents may be particularly useful or confer the best responses – such as in the adjuvant setting or as maintenance therapy. Clinical trials need to be intelligently designed in regards to patient selection, toxicity, PK, PD, reflective endpoints, and correlative surrogate markers, as these elements are essential to moving forward. Despite many new anti-angiogenic agents demonstrating established efficacy and effectiveness, to

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

385

further predict response and select out patient for treatment with these agents, one must go back to the bench-side to develop better PD markers of activity, and predictive markers of response and survival for early development. Although the last few decades of research have brought these agents to the clinical setting and established their activity, clearly much more information and research is needed in regards to further effectively develop anti-angiogenic agents.

References 1. Eichhorn ME, Strieth S, Dellian M: Anti-vascular tumor therapy: recent advances, pitfalls and clinical perspectives. Drug Resist Updat 7:125–38, 2004 2. Folkman J: Tumor angiogenesis: a possible control point in tumor growth. Ann Intern Med 82:96–100, 1975 3. Folkman J: Anti-angiogenesis: new concept for therapy of solid tumors. Ann Surg 175:409– 16, 1972 4. Folkman J: Tumor angiogenesis: therapeutic implications. N Engl J Med 285:1182–6, 1971 5. Jaffe EA, Nachman RL, Becker CG, et al: Culture of human endothelial cells derived from umbilical veins. Identification by morphologic and immunologic criteria. J Clin Invest 52:2745–56, 1973 6. Folkman J, Merler E, Abernathy C, et al: Isolation of a tumor factor responsible for angiogenesis. J Exp Med 133:275–88, 1971 7. Chang E, Boyd A, Nelson CC, et  al: Successful treatment of infantile hemangiomas with interferon-alpha-2b. J Pediatr Hematol Oncol 19:237–44, 1997 8. Adlard JW: Thalidomide in the treatment of cancer. Anticancer Drugs 11:787–91, 2000 9. Yang JC, Haworth L, Sherry RM, et  al: A randomized trial of bevacizumab, an anti-vascular endothelial growth factor antibody, for metastatic renal cancer. N Engl J Med 349:427–34, 2003 10. Hurwitz H, Fehrenbacher L, Novotny W, et al: Bevacizumab plus irinotecan, fluorouracil, and leucovorin for metastatic colorectal cancer. N Engl J Med 350:2335–42, 2004 11. Srinivasan R, Armstrong AJ, Dahut W, et al: Anti-angiogenic therapy in renal cell cancer. BJU Int 99:1296–300, 2007 12. Bisacchi D, Benelli R, Vanzetto C, et  al: Anti-angiogenesis and angioprevention: mechanisms, problems and perspectives. Cancer Detect Prev 27:229–38, 2003 13. De Bandt M, Ben Mahdi MH, Ollivier V, et  al: Blockade of vascular endothelial growth factor receptor I (VEGF-RI), but not VEGF-RII, suppresses joint destruction in the K/BxN model of rheumatoid arthritis. J Immunol 171:4853–9, 2003 14. Kowanetz M, Ferrara N: Vascular endothelial growth factor signaling pathways: therapeutic perspective. Clin Cancer Res 12:5018–22, 2006 15. Malemud CJ: Growth hormone, VEGF and FGF: involvement in rheumatoid arthritis. Clin Chim Acta 375:10–9, 2007 16. Carmeliet P, Jain RK: Angiogenesis in cancer and other diseases. Nature 407:249–57, 2000 17. Folkman J, Hanahan D: Switch to the angiogenic phenotype during tumorigenesis. Princess Takamatsu Symp 22:339–47, 1991 18. Kalluri R: Basement membranes: structure, assembly and role in tumor angiogenesis. Nat Rev Cancer 3:422–33, 2003 19. Ferrara N, Gerber HP, LeCouter J: The biology of VEGF and its receptors. Nat Med 9:669– 76, 2003 20. Parikh AA, Ellis LM: The vascular endothelial growth factor family and its receptors. Hematol Oncol Clin North Am 18:951–71, vii, 2004

386

L.Q.M. Chow and S.G. Eckhardt

21. Rosen LS: Clinical experience with angiogenesis signaling inhibitors: focus on vascular endothelial growth factor (VEGF) blockers. Cancer Control 9:36–44, 2002 22. Dvorak HF: Vascular permeability factor/vascular endothelial growth factor: a critical cytokine in tumor angiogenesis and a potential target for diagnosis and therapy. J Clin Oncol 20:4368–80, 2002 23. Nagy JA, Vasile E, Feng D, et al: Vascular permeability factor/vascular endothelial growth factor induces lymphangiogenesis as well as angiogenesis. J Exp Med 196:1497–506, 2002 24. Malik AK, Gerber HP: Targeting VEGf ligands and receptors in cancer. Targets 2:48–57, 2003 25. Ahmed SI, Thomas AL, Steward WP: Vascular endothelial growth factor (VEGF) inhibition by small molecules. J Chemother 16 Suppl 4:59–63, 2004 26. Sundberg C, Ljungstrom M, Lindmark G, et  al: Microvascular pericytes express plateletderived growth factor-beta receptors in human healing wounds and colorectal adenocarcinoma. Am J Pathol 143:1377–88, 1993 27. Bergers G, Song S, Meyer-Morse N, et al: Benefits of targeting both pericytes and endothelial cells in the tumor vasculature with kinase inhibitors. J Clin Invest 111:1287–95, 2003 28. Pietras K, Rubin K, Sjoblom T, et al: Inhibition of PDGF receptor signaling in tumor stroma enhances antitumor effect of chemotherapy. Cancer Res 62:5476–84, 2002 29. Dinney CP, Bielenberg DR, Perrotte P, et  al: Inhibition of basic fibroblast growth factor expression, angiogenesis, and growth of human bladder carcinoma in mice by systemic interferon-alpha administration. Cancer Res 58:808–14, 1998 30. Giavazzi R, Sennino B, Coltrini D, et  al: Distinct role of fibroblast growth factor-2 and vascular endothelial growth factor on tumor growth and angiogenesis. Am J Pathol 162:1913–26, 2003 31. Sternlicht MD, Werb Z: How matrix metalloproteinases regulate cell behavior. Annu Rev Cell Dev Biol 17:463–516, 2001 32. Rabbani SA: Metalloproteases and urokinase in angiogenesis and tumor progression. In Vivo 12:135–42, 1998 33. Mannello F, Gazzanelli G: Tissue inhibitors of metalloproteinases and programmed cell death: conundrums, controversies and potential implications. Apoptosis 6:479–82, 2001 34. Nyberg P, Xie L, Kalluri R: Endogenous inhibitors of angiogenesis. Cancer Res 65:3967–79, 2005 35. Rabbani SA, Xing RH: Role of urokinase (uPA) and its receptor (uPAR) in invasion and metastasis of hormone-dependent malignancies. Int J Oncol 12:911–20, 1998 36. El-Assal ON, Yamanoi A, Ono T, et al: The clinicopathological significance of heparanase and basic fibroblast growth factor expressions in hepatocellular carcinoma. Clin Cancer Res 7:1299–305, 2001 37. Gutterman JU: Cytokine therapeutics: lessons from interferon alpha. Proc Natl Acad Sci U S A 91:1198–205, 1994 38. Turnbull J, Powell A, Guimond S: Heparan sulfate: decoding a dynamic multifunctional cell regulator. Trends Cell Biol 11:75–82, 2001 39. Simizu S, Ishida K, Wierzba MK, et al: Expression of heparanase in human tumor cell lines and human head and neck tumors. Cancer Lett 193:83–9, 2003 40. Koliopanos A, Friess H, Kleeff J, et  al: Heparanase expression in primary and metastatic pancreatic cancer. Cancer Res 61:4655–9, 2001 41. Xiao Y, Kleeff J, Shi X, et al: Heparanase expression in hepatocellular carcinoma and the cirrhotic liver. Hepatol Res 26:192–8, 2003 42. Liekens S, De Clercq E, Neyts J: Angiogenesis: regulators and clinical applications. Biochem Pharmacol 61:253–70, 2001 43. Scapini P, Lapinet-Vera JA, Gasperini S, et al: The neutrophil as a cellular source of chemokines. Immunol Rev 177:195–203, 2000 44. Dong Z, Greene G, Pettaway C, et  al: Suppression of angiogenesis, tumorigenicity, and metastasis by human prostate cancer cells engineered to produce interferon-beta. Cancer Res 59:872–9, 1999 45. Scapini P, Laudanna C, Pinardi C, et al: Neutrophils produce biologically active macrophage inflammatory protein-3alpha (MIP-3alpha)/CCL20 and MIP-3beta/CCL19. Eur J Immunol 31:1981–8, 2001

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

387

46. Benelli R, Morini M, Carrozzino F, et al: Neutrophils as a key cellular target for angiostatin: implications for regulation of angiogenesis and inflammation. FASEB J 16:267–9, 2002 47. Brunda MJ, Luistro L, Rumennik L, et al: Interleukin-12: murine models of a potent antitumor agent. Ann N Y Acad Sci 795:266–74, 1996 48. Goon PK, Lip GY, Boos CJ, et al: Circulating endothelial cells, endothelial progenitor cells, and endothelial microparticles in cancer. Neoplasia 8:79–88, 2006 49. Buerkle MA, Pahernik SA, Sutter A, et al: Inhibition of the alpha-nu integrins with a cyclic RGD peptide impairs angiogenesis, growth and metastasis of solid tumors in  vivo. Br J Cancer 86:788–95, 2002 50. Cai W, Chen X: Anti-angiogenic cancer therapy based on integrin alphavbeta3 antagonism. Anticancer Agents Med Chem 6:407–28, 2006 51. Hynes RO, Lively JC, McCarty JH, et al: The diverse roles of integrins and their ligands in angiogenesis. Cold Spring Harb Symp Quant Biol 67:143–53, 2002 52. Hynes RO, Bader BL, Hodivala-Dilke K: Integrins in vascular development. Braz J Med Biol Res 32:501–10, 1999 53. Tarui T, Majumdar M, Miles LA, et al: Plasmin-induced migration of endothelial cells. A potential target for the anti-angiogenic action of angiostatin. J Biol Chem 277:33564–70, 2002 54. Tarui T, Miles LA, Takada Y: Specific interaction of angiostatin with integrin alpha(v)beta(3) in endothelial cells. J Biol Chem 276:39562–8, 2001 55. Wickstrom SA, Alitalo K, Keski-Oja J: Endostatin signaling and regulation of endothelial cell-matrix interactions. Adv Cancer Res 94:197–229, 2005 56. Wickstrom SA, Alitalo K, Keski-Oja J: Endostatin associates with integrin alpha5beta1 and caveolin-1, and activates Src via a tyrosyl phosphatase-dependent pathway in human endothelial cells. Cancer Res 62:5580–9, 2002 57. Lawler J, Detmar M: Tumor progression: the effects of thrombospondin-1 and -2. Int J Biochem Cell Biol 36:1038–45, 2004 58. Volpert OV, Zaichuk T, Zhou W, et al: Inducer-stimulated Fas targets activated endothelium for destruction by anti-angiogenic thrombospondin-1 and pigment epithelium-derived factor. Nat Med 8:349–57, 2002 59. Kruger EA, Duray PH, Price DK, et al: Approaches to preclinical screening of antiangiogenic agents. Semin Oncol 28:570–6, 2001 60. Nakamura T, Matsumoto K: Angiogenesis inhibitors: from laboratory to clinical application. Biochem Biophys Res Commun 333:28991, 2005 61. Norrby K: In vivo models of angiogenesis. J Cell Mol Med 10:588–612, 2006 62. Jubb AM, Oates AJ, Holden S, et al: Predicting benefit from anti-angiogenic agents in malignancy. Nat Rev Cancer 6:626–35, 2006 63. Miller KD, Sweeney CJ, Sledge GW, Jr.: Redefining the target: chemotherapeutics as antiangiogenics. J Clin Oncol 19:1195–206, 2001 64. Rosen L: Antiangiogenic strategies and agents in clinical trials. Oncologist 5 Suppl 1:20–7, 2000 65. Carter SK: Clinical strategy for the development of angiogenesis inhibitors. Oncologist 5 Suppl 1:51-4, 2000 66. Eskens FA, Verweij J: The clinical toxicity profile of vascular endothelial growth factor (VEGF) and vascular endothelial growth factor receptor (VEGFR) targeting angiogenesis inhibitors; a review. Eur J Cancer 42:3127–39, 2006 67. Fox E, Curt GA, Balis FM: Clinical trial design for target-based therapy. Oncologist 7:401–9, 2002 68. Deplanque G, Harris AL: Anti-angiogenic agents: clinical trial design and therapies in development. Eur J Cancer 36:1713–24, 2000 69. Emmenegger U, Kerbel RS: A dynamic de-escalating dosing strategy to determine the optimal biological dose for antiangiogenic drugs. Clin Cancer Res 11:7589–92, 2005 70. Ruegg C, Mutter N: Anti-angiogenic therapies in cancer: achievements and open questions. Bull Cancer 94:753–62, 2007 71. Ruegg C, Meuwly JY, Driscoll R, et al: The quest for surrogate markers of angiogenesis: a paradigm for translational research in tumor angiogenesis and anti-angiogenesis trials. Curr Mol Med 3:673–91, 2003

388

L.Q.M. Chow and S.G. Eckhardt

72. Kerbel RS, Klement G, Pritchard KI, et  al: Continuous low-dose anti-angiogenic/ metronomic chemotherapy: from the research laboratory into the oncology clinic. Ann Oncol 13:12–5, 2002 73. Kerbel R, Folkman J: Clinical translation of angiogenesis inhibitors. Nat Rev Cancer 2: 727–39, 2002 74. Madhusudan S, Harris AL: Drug inhibition of angiogenesis. Curr Opin Pharmacol 2:403–14, 2002 75. Eskens FA: Angiogenesis inhibitors in clinical development; where are we now and where are we going? Br J Cancer 90:1–7, 2004 76. Izzedine H, Ederhy S, Goldwasser F, et  al: Management of hypertension in angiogenesis inhibitor-treated patients. Ann Oncol 18:1121–1122, 2009 77. Saif MW, Mehra R: Incidence and management of bevacizumab-related toxicities in colorectal cancer. Expert Opin Drug Saf 5:553–66, 2006 78. Pande AU, Lombardo JC, Fakih M, et al: Bevacizumab (BV) induced hypertension (HT): a manageable toxicity. J Clin Oncol, 2006 ASCO Annual Meeting Proceedings Part I. Vol. 24. Abstract No: 13539, 2006 79. Saif MW, Elfiky A, Salem RR: Gastrointestinal perforation due to bevacizumab in colorectal cancer. Ann Surg Oncol 14:1860–9, 2007 80. Sandler A, Gray R, Perry MC, et al: Paclitaxel-carboplatin alone or with bevacizumab for non-small-cell lung cancer. N Engl J Med 355:2542–50, 2006 81. Sandler A: Bevacizumab in non small cell lung cancer. Clin Cancer Res 13:s4613–6, 2007 82. Kilickap S, Abali H, Celik I: Bevacizumab, bleeding, thrombosis, and warfarin. J Clin Oncol 21:35–42; author reply 35–43, 2003 83. Zondor SD, Medina PJ: Bevacizumab: an angiogenesis inhibitor with efficacy in colorectal and other malignancies. Ann Pharmacother 38:1258–64, 2004 84. Takimoto CH, Awada A: Safety and anti-tumor activity of sorafenib (Nexavar) in combination with other anti-cancer agents: a review of clinical trials. Cancer Chemother Pharmacol 61:535–48, 2008 85. van Hinsbergh VW, Collen A, Koolwijk P: Angiogenesis and anti-angiogenesis: perspectives for the treatment of solid tumors. Ann Oncol 10 Suppl 4:60–3, 1999 86. Telli ML, Witteles RM, Fisher GA, et al: Cardiotoxicity associated with the cancer therapeutic agent sunitinib malate. Ann Oncol 19(9):1613–1618, 2008 87. Chu TF, Rupnick MA, Kerkela R, et  al: Cardiotoxicity associated with tyrosine kinase inhibitor sunitinib. Lancet 370:2011–9, 2007 88. Joensuu H: Cardiac toxicity of sunitinib. Lancet 370:1978–80, 2007 89. Pouessel D, Culine S: High frequency of intracerebral hemorrhage in metastatic renal carcinoma patients with brain metastases treated with tyrosine kinase inhibitors targeting the vascular endothelial growth factor receptor. Eur Urol 53:376–81, 2008 90. Li M, Srinivas S: Acute pancreatitis associated with sorafenib. South Med J 100:909–11, 2007 91. Amar S, Wu KJ, Tan WW: Sorafenib-induced pancreatitis. Mayo Clin Proc 82:521, 2007 92. Chow LQ, Eckhardt SG: Sunitinib: from rational design to clinical efficacy. J Clin Oncol 25:884–96, 2007 93. Bergers G, Benjamin LE: Tumorigenesis and the angiogenic switch. Nat Rev Cancer 3:401– 10, 2003 94. Bergers G, Song S: The role of pericytes in blood-vessel formation and maintenance. Neuro Oncol 7:452–64, 2005 95. Liang WC, Wu X, Peale FV, et al: Cross-species vascular endothelial growth factor (VEGF)blocking antibodies completely inhibit the growth of human tumor xenografts and measure the contribution of stromal VEGF. J Biol Chem 281:951–61, 2006 96. Ferrara N: VEGF as a therapeutic target in cancer. Oncology 69 Suppl 3:11–6, 2005 97. Relf M, LeJeune S, Scott PA, et al: Expression of the angiogenic factors vascular endothelial cell growth factor, acidic and basic fibroblast growth factor, tumor growth factor beta-1, platelet-derived endothelial cell growth factor, placenta growth factor, and pleiotrophin in human primary breast cancer and its relation to angiogenesis. Cancer Res 57:963–9, 1997

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

389

98. Bergers G, Hanahan D: Modes of resistance to anti-angiogenic therapy. Nat Rev Cancer 8:592–603, 2008 99. Ma J, Waxman DJ: Combination of antiangiogenesis with chemotherapy for more effective cancer treatment. Mol Cancer Ther 7:3670–84, 2008 100. McCarty MF, Liu W, Fan F, et al: Promises and pitfalls of anti-angiogenic therapy in clinical trials. Trends Mol Med 9:53–8, 2003 101. Yu JL, Rak JW, Coomber BL, et al: Effect of p53 status on tumor response to antiangiogenic therapy. Science 295:1526–8, 2002 102. Rini BI, Small EJ: Biology and clinical development of vascular endothelial growth factortargeted therapy in renal cell carcinoma. J Clin Oncol 23:1028–43, 2005 103. Turner KJ, Moore JW, Jones A, et al: Expression of hypoxia-inducible factors in human renal cancer: relationship to angiogenesis and to the von Hippel-Lindau gene mutation. Cancer Res 62:2957–61, 2002 104. Rosa DD, Ismael G, Lago LD, et  al: Molecular-targeted therapies: lessons from years of clinical development. Cancer Treat Rev 34:61–80, 2008 105. Yano S, Shinohara H, Herbst RS, et al: Production of experimental malignant pleural effusions is dependent on invasion of the pleura and expression of vascular endothelial growth factor/ vascular permeability factor by human lung cancer cells. Am J Pathol 157:1893–903, 2000 106. Zebrowski BK, Yano S, Liu W, et al: Vascular endothelial growth factor levels and induction of permeability in malignant pleural effusions. Clin Cancer Res 5:3364–8, 1999 107. Patel PH, Chaganti RS, Motzer RJ: Targeted therapy for metastatic renal cell carcinoma. Br J Cancer 94:614–9, 2006 108. Simpson D, Keating GM: Sorafenib: in hepatocellular carcinoma. Drugs 68:251–8, 2008 109. Zhu AX: Development of sorafenib and other molecularly targeted agents in hepatocellular carcinoma. Cancer 112:250–9, 2008 110. Egeblad M, Werb Z: New functions for the matrix metalloproteinases in cancer progression. Nat Rev Cancer 2:161–74, 2002 111. Coussens LM, Fingleton B, Matrisian LM: Matrix metalloproteinase inhibitors and cancer: trials and tribulations. Science 295:2387–92, 2002 112. Zucker S, Cao J, Chen WT: Critical appraisal of the use of matrix metalloproteinase inhibitors in cancer treatment. Oncogene 19:6642–50, 2000 113. Stetler-Stevenson WG: Matrix metalloproteinases in angiogenesis: a moving target for therapeutic intervention. J Clin Invest 103:1237–41, 1999 114. Sparano JA, Bernardo P, Stephenson P, et al: Randomized phase III trial of marimastat versus placebo in patients with metastatic breast cancer who have responding or stable disease after first-line chemotherapy: Eastern Cooperative Oncology Group trial E2196. J Clin Oncol 22:4683–90, 2004 115. Zucker S, Hymowitz M, Conner C, et  al: Measurement of matrix metalloproteinases and tissue inhibitors of metalloproteinases in blood and tissues. Clinical and experimental applications. Ann N Y Acad Sci 878:212–27, 1999 116. Chiappori AA, Eckhardt SG, Bukowski R, et al: A phase I pharmacokinetic and pharmacodynamic study of s-3304, a novel matrix metalloproteinase inhibitor, in patients with advanced and refractory solid tumors. Clin Cancer Res 13:2091–9, 2007 117. Komorowski J, Jerczynska H, Siejka A, et al: Effect of thalidomide affecting VEGF secretion, cell migration, adhesion and capillary tube formation of human endothelial EA.hy 926 cells. Life Sci 78:2558–63, 2006 118. Mileshkin L, Honemann D, Gambell P, et  al: Patients with multiple myeloma treated with thalidomide: evaluation of clinical parameters, cytokines,angiogenic markers, mast cells and marrow CD57+ cytotoxic T cells as predictors of outcome. Haematologica 92:1075–82, 2007 119. Cibeira MT, Rozman M, Segarra M, et al: Bone marrow angiogenesis and angiogenic factors in multiple myeloma treated with novel agents. Cytokine 41:244–53, 2008 120. von Lilienfeld-Toal M, Hahn-Ast C, Furkert K, et al: A systematic review of phase ii trials of thalidomide/dexamethasone combination therapy in patients with relapsed or refractory multiple myeloma. Eur J Haematol 81:247–252, 2008

390

L.Q.M. Chow and S.G. Eckhardt

121. Prince HM, Schenkel B, Mileshkin L: An analysis of clinical trials assessing the efficacy and safety of single-agent thalidomide in patients with relapsed or refractory multiple myeloma. Leuk Lymphoma 48:46–55, 2007 122. Zomas A, Anagnostopoulos N, Dimopoulos MA: Successful treatment of multiple myeloma relapsing after high-dose therapy and autologous transplantation with thalidomide as a single agent. Bone Marrow Transplant 25:1319–20, 2000 123. FDA: http://www.accessdata.fda.gov/scripts/cder/drugsatfda/index.cfm, US Food and Drug Administration, 2009 124. Rosinol L, Cibeira MT, Segarra M, et  al: Response to thalidomide in multiple myeloma: impact of angiogenic factors. Cytokine 26:145–8, 2004 125. NCI/CTEP: NCI Clinical Trials Information Site, United States National Institutes of HealthBethesda, MD, USA, NIH, 2009, pp http://www.cancer.gov/clinicaltrials 126. Wang Y, Fei D, Vanderlaan M, et al: Biological activity of bevacizumab, a humanized antiVEGF antibody in vitro. Angiogenesis 7:335–45, 2004 127. Chen HX: Expanding the clinical development of bevacizumab. Oncologist 9 Suppl 1:27–35, 2004 128. Youssoufian H, Hicklin DJ, Rowinsky EK: Review: monoclonal antibodies to the vascular endothelial growth factor receptor-2 in cancer therapy. Clin Cancer Res 13:5544s-5548s, 2007 129. Holash J, Davis S, Papadopoulos N, et al: VEGF-Trap: a VEGF blocker with potent antitumor effects. Proc Natl Acad Sci U S A 99:11393–8, 2002 130. Bergsland EK: Update on clinical trials targeting vascular endothelial growth factor in cancer. Am J Health Syst Pharm 61:S12–20, 2004 131. Aita M, Fasola G, Defferrari C, et al: Targeting the VEGF pathway: antiangiogenic strategies in the treatment of non-small cell lung cancer. Crit Rev Oncol Hematol 68:183–96, 2008 132. Riely GJ, Miller VA: Vascular endothelial growth factor trap in non small cell lung cancer. Clin Cancer Res 13:s4623–7, 2007 133. Ignoffo RJ: Overview of bevacizumab: a new cancer therapeutic strategy targeting vascular endothelial growth factor. Am J Health Syst Pharm 61:S21–6, 2004 134. Ranieri G, Patruno R, Ruggieri E, et al: Vascular endothelial growth factor (VEGF) as a target of bevacizumab in cancer: from the biology to the clinic. Curr Med Chem 13:1845–57, 2006 135. Escudier B, Pluzanska A, Koralewski P, et al: Bevacizumab plus interferon alfa-2a for treatment of metastatic renal cell carcinoma: a randomised, double-blind phase III trial. Lancet 370:2103–11, 2007 136. de Castro Junior G, Puglisi F, de Azambuja E, et al: Angiogenesis and cancer: a cross-talk between basic science and clinical trials (the “do ut des” paradigm). Crit Rev Oncol Hematol 59:40–50, 2006 137. Bergsland E, Dickler MN: Maximizing the potential of bevacizumab in cancer treatment. Oncologist 9 Suppl 1:36-42, 2004 138. Gordon MS, Margolin K, Talpaz M, et  al: Phase I safety and pharmacokinetic study of recombinant human anti-vascular endothelial growth factor in patients with advanced cancer. J Clin Oncol 19:843–50, 2001 139. Lin YS, Nguyen C, Mendoza JL, et al: Preclinical pharmacokinetics, interspecies scaling, and tissue distribution of a humanized monoclonal antibody against vascular endothelial growth factor. J Pharmacol Exp Ther 288:371–8, 1999 140. Kabbinavar F, Hurwitz HI, Fehrenbacher L, et  al: Phase II, randomized trial comparing bevacizumab plus fluorouracil (FU)/leucovorin (LV) with FU/LV alone in patients with metastatic colorectal cancer. J Clin Oncol 21:60–5, 2003 141. Hurwitz H, Kabbinavar F: Bevacizumab combined with standard fluoropyrimidine-based chemotherapy regimens to treat colorectal cancer. Oncology 69 Suppl 3:17–24, 2005 142. Hurwitz HI, Fehrenbacher L, Hainsworth JD, et al: Bevacizumab in combination with fluorouracil and leucovorin: an active regimen for first-line metastatic colorectal cancer. J Clin Oncol 23:3502–8, 2005 143. Hurwitz HI: New agents in colon cancer. Clin Adv Hematol Oncol 1:404–5, 2003

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

391

144. Ince WL, Jubb AM, Holden SN, et al: Association of k-ras, b-raf, and p53 status with the treatment effect of bevacizumab. J Natl Cancer Inst 97:981–9, 2005 145. Giantonio B, Catalano PJ, Meropol NJ, et al: High dose bevacizumab (antiVEGF) in combination with FOLFOX4 improves survival in patients with previously treated advanced colorectal cancer: results from the Eastern Cooperative Oncology Group (ECOG) study E3200. J Clin Oncol, Hollywood FL. ASCO Gastrointestinal Cancers Symposium January 27–29, 2005 146. Giantonio BJ, Catalano PJ, Meropol NJ, et  al: Bevacizumab in combination with oxaliplatin, fluorouracil, and leucovorin (FOLFOX4) for previously treated metastatic colorectal cancer: results from the Eastern Cooperative Oncology Group Study E3200. J Clin Oncol 25:1539–44, 2007 147. Saltz LB, Clarke S, Diaz-Rubio E, et al: Bevacizumab in combination with oxaliplatin-based chemotherapy as first-line therapy in metastatic colorectal cancer: a randomized phase III study. J Clin Oncol 26:2013–9, 2008 148. Cassidy J, Clarke S, Diaz-Rubio E, et al: Randomized phase III study of capecitabine plus oxaliplatin compared with fluorouracil/folinic acid plus oxaliplatin as first-line therapy for metastatic colorectal cancer. J Clin Oncol 26:2006–12, 2008 149. Marshall J: The role of bevacizumab as first-line therapy for colon cancer. Semin Oncol 32:S43–7, 2005 150. de Gramont A, Tournigand C, Andre T, et al: Targeted agents for adjuvant therapy of colon cancer. Semin Oncol 33:S42–5, 2006 151. Johnson DH, Fehrenbacher L, Novotny WF, et  al: Randomized phase II trial comparing bevacizumab plus carboplatin and paclitaxel with carboplatin and paclitaxel alone in previously untreated locally advanced or metastatic non-small-cell lung cancer. J Clin Oncol 22:2184–91, 2004 152. Gutierrez M, Giaccone G: Antiangiogenic therapy in nonsmall cell lung cancer. Curr Opin Oncol 20:176–82, 2008 153. Laskin JJ, Sandler AB: First-line treatment for advanced non-small-cell lung cancer. Oncology (Williston Park) 19:1671–6; discussion 1678–80, 2005 154. Miller KD, Chap LI, Holmes FA, et al: Randomized phase III trial of capecitabine compared with bevacizumab plus capecitabine in patients with previously treated metastatic breast cancer. J Clin Oncol 23:792–9, 2005 155. Giantonio BJ: Bevacizumab in the treatment of metastatic colorectal cancer (mCRC) in second- and third-line settings. Semin Oncol 33:S15–8, 2006 156. Gundgaard MG, Soerensen JB, Ehrnrooth E: Third-line therapy for metastatic colorectal cancer. Cancer Chemother Pharmacol 61:1–13, 2008 157. Miller KD: E2100: a phase III trial of paclitaxel versus paclitaxel/bevacizumab for metastatic breast cancer. Clin Breast Cancer 3:421–2, 2003 158. Sachdev JC, Jahanzeb M: Evolution of bevacizumab-based therapy in the management of breast cancer. Clin Breast Cancer 8:402–10, 2008 159. Pawson T: Regulation and targets of receptor tyrosine kinases. Eur J Cancer 38 Suppl 5:S3–10, 2002 160. Favaro JP, George DJ: Targeted therapy in renal cell carcinoma. Expert Opin Investig Drugs 14:1251–8, 2005 161. Potapova O, Laird AD, Nannini MA, et al: Contribution of individual targets to the antitumor efficacy of the multitargeted receptor tyrosine kinase inhibitor SU11248. Mol Cancer Ther 5:1280–9, 2006 162. Xu L, Tong R, Cochran DM, et al: Blocking platelet-derived growth factor-D/platelet-derived growth factor receptor beta signaling inhibits human renal cell carcinoma progression in an orthotopic mouse model. Cancer Res 65:5711–9, 2005 163. Wilhelm SM, Carter C, Tang L, et al: BAY 43-9006 exhibits broad spectrum oral antitumor activity and targets the RAF/MEK/ERK pathway and receptor tyrosine kinases involved in tumor progression and angiogenesis. Cancer Res 64:7099–109, 2004 164. Wakelee HA, Schiller JH: Targeting angiogenesis with vascular endothelial growth factor receptor small-molecule inhibitors: novel agents with potential in lung cancer. Clin Lung Cancer 7 Suppl 1:S31–8, 2005

392

L.Q.M. Chow and S.G. Eckhardt

165. Tong FK, Chow S, Hedley D: Pharmacodynamic monitoring of BAY 43-9006 (Sorafenib) in phase I clinical trials involving solid tumor and AML/MDS patients, using flow cytometry to monitor activation of the ERK pathway in peripheral blood cells. Cytometry B Clin Cytom 70:107–14, 2006 166. Wilhelm S, Chien DS: BAY 43-9006: preclinical data. Curr Pharm Des 8:2255–7, 2002 167. Hotte SJ, Hirte HW: BAY 43-9006: early clinical data in patients with advanced solid malignancies. Curr Pharm Des 8:2249–53, 2002 168. Moore M, Hirte HW, Siu L, et al: Phase I study to determine the safety and pharmacokinetics of the novel Raf kinase and VEGFR inhibitor BAY 43-9006, administered for 28 days on/7 days off in patients with advanced, refractory solid tumors. Ann Oncol 16:1688–94, 2005 169. Awada A, Hendlisz A, Gil T, et al: Phase I safety and pharmacokinetics of BAY 43-9006 administered for 21 days on/7 days off in patients with advanced, refractory solid tumors. Br J Cancer 92:1855–61, 2005 170. Strumberg D, Richly H, Hilger RA, et al: Phase I clinical and pharmacokinetic study of the Novel Raf kinase and vascular endothelial growth factor receptor inhibitor BAY 43-9006 in patients with advanced refractory solid tumors. J Clin Oncol 23:965–72, 2005 171. Richly H, Kupsch P, Passage K, et al: A phase I clinical and pharmacokinetic study of the Raf kinase inhibitor (RKI) BAY 43-9006 administered in combination with doxorubicin in patients with solid tumors. Int J Clin Pharmacol Ther 41:620–1, 2003 172. Strumberg D, Voliotis D, Moeller JG, et al: Results of phase I pharmacokinetic and pharmacodynamic studies of the Raf kinase inhibitor BAY 43-9006 in patients with solid tumors. Int J Clin Pharmacol Ther 40:580–1, 2002 173. Jain L, Venitz J, Figg WD: Randomized discontinuation trial of sorafenib (BAY 43-9006). Cancer Biol Ther 5:1270–2, 2006 174. Ratain MJ, Eisen T, Stadler WM, et al: Phase II placebo-controlled randomized discontinuation trial of sorafenib in patients with metastatic renal cell carcinoma. J Clin Oncol 24:2505–12, 2006 175. Escudier B, Szczylik C, Eisen T, et  al: Randomized phase III trial of the Raf kinase and VEGFR inhibitor sorafenib (BAY 43-9006) in patients with advanced renal cell carcinoma (RCC). Proc Am Soc Clin Oncol, Orlando, FL. Abstract No: LBA4510 ASCO, 2005 176. Rini BI: Sorafenib. Expert Opin Pharmacother 7:453–61, 2006 177. Hahn O, Stadler W: Sorafenib. Curr Opin Oncol 18:615–21, 2006 178. Bayes M, Rabasseda X, Prous JR: Gateways to clinical trials. Methods Find Exp Clin Pharmacol 29:467–509, 2007 179. Llovet J, Ricci S, Mazzaferro V, et al: Sorafenib improves survival in advanced Hepatocellular Carcinoma (HCC): Results of a Phase III randomized placebo-controlled trial (SHARP trial). J Clin Oncol, ASCO Annual Meeting Proceedings Part I. Vol 25, (June 20 Supplement), 2007 180. Sun L, Liang C, Shirazian S, et al: Discovery of 5-[5-fluoro-2-oxo-1,2- dihydroindol-(3Z)ylidenemethyl]-2,4- dimethyl-1H-pyrrole-3-carboxylic acid (2-diethylaminoethyl)amide, a novel tyrosine kinase inhibitor targeting vascular endothelial and platelet-derived growth factor receptor tyrosine kinase. J Med Chem 46:1116–9, 2003 181. O’Farrell AM, Abrams TJ, Yuen HA, et al: SU11248 is a novel FLT3 tyrosine kinase inhibitor with potent activity in vitro and in vivo. Blood 101:3597–605, 2003 182. Mendel DB, Laird AD, Xin X, Li G, Schreck RE, Carver J, Louie SG, Sukbuntherng J, Plise E, Kelsey S, Scigalla P, Cherrington JM: Development of a preclinical pharmacokinetic/ pharmacodynamic relationship for the angiogenesis inhibitor SU11248, a selective inhibitor of VEGF and PDGF receptor tyrosine kinases in clinical development. Proc Am Soc Clin Oncol. Abstract No: 94, 2002 183. Osusky KL, Hallahan DE, Fu A, et  al: The receptor tyrosine kinase inhibitor SU11248 impedes endothelial cell migration, tubule formation, and blood vessel formation in vivo, but has little effect on existing tumor vessels. Angiogenesis 7:225–33, 2004 184. Faivre S, Delbaldo C, Vera K, et  al: Safety, pharmacokinetic, and antitumor activity of SU11248, a novel oral multitarget tyrosine kinase inhibitor, in patients with cancer. J Clin Oncol 24:25–35, 2006

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

393

185. Fiedler W, Serve H, Dohner H, et al: A phase 1 study of SU11248 in the treatment of patients with refractory or resistant acute myeloid leukemia (AML) or not amenable to conventional therapy for the disease. Blood 105:986–93, 2005 186. O’Farrell AM, Foran JM, Fiedler W, et al: An innovative phase I clinical study demonstrates inhibition of FLT3 phosphorylation by SU11248 in acute myeloid leukemia patients. Clin Cancer Res 9:5465–76, 2003 187. Raymond E, Faivre S, Vera K, Delbaldo C, Robert C, Spatz A, Bello C, Brega N, Scigalla P, Armand JP: Final results of a phase I and pharmacokinetic study of SU11248, a novel multi-target tyrosine kinase inhibitor, in patients with advanced cancers. Proc Am Soc Clin Oncol, ASCO. pp 192. Abstract No: 767, 2003 188. Rosen L, Mulay M, Long J, Wittner J, Brown J, Martino A-M, Bello CL, Walter S, Scigalla P, Zhu J: Phase I trial of SU011248, a novel tyrosine kinase inhibitor in advanced solid tumors. Proc Am Soc Clin Oncol, Vol. 22, p 191. Abstract No: 765, 2003 189. De Mulder PH, Roigas J, Gillessen S, et al: A phase II study of sunitinib administered in a continuous daily regimen in patients with cytokine-refractory metastatic renal cell carcinoma (mRCC). J Clin Oncol, 2006 ASCO Annual Meeting Proceedings Part I. Vol 24. Abstract No: 4529, 2006 190. Motzer RJ, Rini BI, Michaelson MD, et al: Phase 2 trials of SU11248 show antitumor activity in second-line therapy for patients with metastatic renal cell carcinoma (RCC). Proc Am Soc Clin Oncol, Orlando, FL. Abstract No: 4508 ASCO, 2005 191. Motzer RJ, Mazumdar M, Bacik J, et al: Effect of cytokine therapy on survival for patients with advanced renal cell carcinoma. J Clin Oncol 18:1928–35, 2000 192. Motzer R. J., Michaelson M. D., Redman B. G., et al: Activity of SU11248, a multitargeted inhibitor of vascular endothelial growth factor receptor and platelet-derived growth factor receptor, in patients with metastatic renal cell carcinoma. J Clin Oncol 24:16–24, 2006 193. Motzer RJ, Michaelson MD, Rosenberg J, et al: Sunitinib efficacy against advanced renal cell carcinoma. J Urol 178:1883–7, 2007 194. Motzer RJ, Hutson TE, Tomczak P, et  al: Phase III randomized trial of sunitinib malate (SU11248) versus interferon-alfa (IFN-a) as first-line systemic therapy for patients with metastatic renal cell carcinoma (mRCC). J Clin Oncol, ASCO Annual Meeting Proceedings Part I. Vol. 24, No. 18S (June 20 Supplement). Late Breaking Abstract No: 3, 2006 195. Motzer RJ, Hutson TE, Tomczak P, et al: Sunitinib versus interferon alfa in metastatic renalcell carcinoma. N Engl J Med 356:115–24, 2007 196. van Spronsen DJ, Mulders PF, De Mulder PH: Novel treatments for metastatic renal cell carcinoma. Crit Rev Oncol Hematol 55:177–91, 2005 197. Tamaskar I, Shaheen P, Wood L, et al: Antitumor effects of sorafenib and sunitinib in patients (pts) with metastatic renal cell carcinoma (mRCC) who had prior therapy with anti-­ angiogenic agents. J Clin Oncol, 2006 ASCO Annual Meeting Proceedings Part I. Vol. 24. Abstract No: 4597, 2006 198. Rini BI: SU11248 and AG013736: current data and future trials in renal cell carcinoma. Clin Genitourin Cancer 4:175–80, 2005 199. Rixe O, Bukowski RM, Michaelson MD, et al: Axitinib treatment in patients with cytokinerefractory metastatic renal-cell cancer: a phase II study. Lancet Oncol 8:975–84, 2007 200. Motzer RJ, Basch E: Targeted drugs for metastatic renal cell carcinoma. Lancet 370: 2071–3, 2007 201. Rini BI: VEGF-targeted therapy in metastatic renal cell carcinoma. Oncologist 10:191–7, 2005 202. George S, Casali PG, Blay J, et al: Phase II study of sunitinib administered in a continuous daily dosing regimen in patients (pts) with advanced GIST. J Clin Oncol, ASCO Annual Meeting Proceedings Part I. Vol 24, No. 18S (June 20 Supplement). Abstract No: 9532, 2006 203. Heinrich MC, Maki RG, Corless CL, et  al: Sunitinib (SU) response in imatinib-resistant (IM-R) GIST correlates with KIT and PDGFRA mutation status. J Clin Oncol, ASCO Annual Meeting Proceedings Part I. Vol 24, No. 18S (June 20 Supplement). Abstract No: 9502, 2006

394

L.Q.M. Chow and S.G. Eckhardt

204. Casali PG, Garrett R, Blackstein ME, et al: Updated results from a phase III trial of sunitinib in GIST patients (pts) for whom imatinib (IM) therapy has failed due to resistance or intolerance. J Clin Oncol, 2006 ASCO Annual Meeting Proceedings Part I. Vol 24, No. 18S (June 20 Supplement). Abstract No: 9513, 2006 205. Demetri GD, van Oosterom AT, Blackstein M, Garrett C, Shah M, Heinrich M, McArthur G, Judson I, Baum CM, Casali PG: Phase 3, multicenter, randomized, double-blind, placebocontrolled trial of SU11248 in patients (pts) following failure of imatinib for metastatic GIST. Proc Am Soc Clin Oncol, Orlando, FL. Abstract No: 4000 ASCO, 2005 206. Wedge SR, Ogilvie DJ, Dukes M, et al: ZD6474 inhibits vascular endothelial growth factor signaling, angiogenesis, and tumor growth following oral administration. Cancer Res 62:4645–55, 2002 207. Lee D, Heymach JV: Emerging antiangiogenic agents in lung cancer. Clin Lung Cancer 7:304–8, 2006 208. Hanrahan EO, Heymach JV: Vascular endothelial growth factor receptor tyrosine kinase inhibitors vandetanib (ZD6474) and AZD2171 in lung cancer. Clin Cancer Res 13:s4617–22, 2007 209. Herbst RS, Heymach JV, O’Reilly MS, et  al: Vandetanib (ZD6474): an orally available receptor tyrosine kinase inhibitor that selectively targets pathways critical for tumor growth and angiogenesis. Expert Opin Investig Drugs 16:239–49, 2007 210. Morgan B, Thomas AL, Drevs J, et  al: Dynamic contrast-enhanced magnetic resonance imaging as a biomarker for the pharmacological response of PTK787/ZK 222584, an inhibitor of the vascular endothelial growth factor receptor tyrosine kinases, in patients with advanced colorectal cancer and liver metastases: results from two phase I studies. J Clin Oncol 21:3955–64, 2003 211. Koehne C Bajetta E, Lin E et al: Final results of CONFIRM-2 @: a multinational randomized, double blind, phase III study in patients with previously treated metastatic adenocarcinoma of the colon or rectum receiving FOLFOX4 and PTK787/ZK22584 or placebo. (CONFIRM-2) Proc Am Soc Clin Oncol 4033a, 2007 212. Zakarija A, Soff G: Update on angiogenesis inhibitors. Curr Opin Oncol 17:578–83, 2005 213. Tyagi P: Vatalanib (PTK787/ZK 222584) in combination with FOLFOX4 versus FOLFOX4 alone as first-line treatment for colorectal cancer: preliminary results from the CONFIRM-1 trial. Clin Colorectal Cancer 5:24–6, 2005 214. Drevs J, Muller-Driver R, Wittig C, et al: PTK787/ZK 222584, a specific vascular endothelial growth factor-receptor tyrosine kinase inhibitor, affects the anatomy of the tumor vascular bed and the functional vascular properties as detected by dynamic enhanced magnetic resonance imaging. Cancer Res 62:4015–22, 2002 215. Mendel DB, Laird AD, Smolich BD, et al: Development of SU5416, a selective small molecule inhibitor of VEGF receptor tyrosine kinase activity, as an anti-angiogenesis agent. Anticancer Drug Des 15:29–41, 2000 216. Arora A, Scholar EM: Role of tyrosine kinase inhibitors in cancer therapy. J Pharmacol Exp Ther 315(3):971–9, 2005 217. Prat A, Casado E, Cortes J: New approaches in angiogenic targeting for colorectal cancer. World J Gastroenterol 13:5857–66, 2007 218. Furman-Haran E, Schechtman E, Kelcz F, et al: Magnetic resonance imaging reveals functional diversity of the vasculature in benign and malignant breast lesions. Cancer 104:708– 18, 2005 219. Laking GR, West C, Buckley DL, et al: Imaging vascular physiology to monitor cancer treatment. Crit Rev Oncol Hematol 58:95–113, 2006 220. Kershaw LE, Buckley DL: Precision in measurements of perfusion and microvascular permeability with T1-weighted dynamic contrast-enhanced MRI. Magn Reson Med 56:986–92, 2006 221. O’Connor JP, Jackson A, Parker GJ, et al: DCE-MRI biomarkers in the clinical evaluation of antiangiogenic and vascular disrupting agents. Br J Cancer 96:189–95, 2007 222. Rehman S, Jayson GC: Molecular imaging of antiangiogenic agents. Oncologist 10:92–103, 2005 223. Cao Y: Endogenous angiogenesis inhibitors and their therapeutic implications. Int J Biochem Cell Biol 33:357–69, 2001

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

395

224. Ruegg C, Hasmim M, Lejeune FJ, et al: Antiangiogenic peptides and proteins: from experimental tools to clinical drugs. Biochim Biophys Acta 1765:155–77, 2006 225. Westphal JR: Technology evaluation: ABT-510, Abbott. Curr Opin Mol Ther 6:451–7, 2004 226. Pasqualini R, Koivunen E, Kain R, et al: Aminopeptidase N is a receptor for tumor-homing peptides and a target for inhibiting angiogenesis. Cancer Res 60:722–7, 2000 227. Hariharan S, Gustafson D, Holden S, et al: Assessment of the biological and pharmacological effects of the alpha nu beta3 and alpha nu beta5 integrin receptor antagonist, cilengitide (EMD 121974), in patients with advanced solid tumors. Ann Oncol 18:1400–7, 2007 228. Parish CR, Freeman C, Brown KJ, et  al: Identification of sulfated oligosaccharide-based inhibitors of tumor growth and metastasis using novel in vitro assays for angiogenesis and heparanase activity. Cancer Res 59:3433–41, 1999 229. Progen, Industries, Ltd: Investigator’s brochure on PI-88. Progen Industries Ltd, Darra, Queensland, Australia. July 2004. (unpublished) Progen Industries Ltd, 2004 230. Pavlakis N, Parish C, Freeman C, et al: The heparanase inhibitor PI-88 reduces tumor growth in two rat mammary adenocarcinoma models, demonstrating schedule dependency and possible synergy with cisplatin. (abstr). Proc Am Assoc Cancer Res, 2000 231. Rosenthal MA, Rischin D, McArthur G, et al: Treatment with the novel anti-angiogenic agent PI-88 is associated with immune-mediated thrombocytopenia. Ann Oncol 13:770–6, 2002 232. Basche M, Gustafson DL, Holden SN, et al: A phase I biological and pharmacologic study of the heparanase inhibitor PI-88 in patients with advanced solid tumors. Clin Cancer Res 12:5471–80, 2006 233. Karoli T, Liu L, Fairweather JK, et al: Synthesis, biological activity, and preliminary pharmacokinetic evaluation of analogues of a phosphosulfomannan angiogenesis inhibitor (PI-88). J Med Chem 48:8229–36, 2005 234. Ferro V, Dredge K, Liu L, et al: PI-88 and novel heparan sulfate mimetics inhibit angiogenesis. Semin Thromb Hemost 33:557–68, 2007 235. Bushell-Embling D, McDonald K: Progen abandons PI-88: massive share price plunge as Progen abandons lead compound trial, in Scientist AL (ed): Medical News (January 28, 2009). Australia, 2009, pp http://www.biotechnews.com.au/index.php/id;513818088 236. Progen: Progen Pharmaceuticals Pipeline, in http://www.progen.com.au/pipeline/default. aspx (ed), 2009 237. Thorpe PE: Vascular targeting agents as cancer therapeutics. Clin Cancer Res 10:415–27, 2004 238. Lippert JW, 3rd: Vascular disrupting agents. Bioorg Med Chem 15:605–15, 2007 239. Banerjee S, Wang Z, Mohammad M, et al: Efficacy of selected natural products as therapeutic agents against cancer. J Nat Prod 71:492–6, 2008 240. Maxwell RJ, Wilson J, Prise VE, et al: Evaluation of the anti-vascular effects of combretastatin in rodent tumors by dynamic contrast enhanced MRI. NMR Biomed 15:89–98, 2002 241. Tozer GM, Kanthou C, Parkins CS, et al: The biology of the combretastatins as tumor vascular targeting agents. Int J Exp Pathol 83:21–38, 2002 242. Kanthou C, Tozer GM: The tumor vascular targeting agent combretastatin A-4-phosphate induces reorganization of the actin cytoskeleton and early membrane blebbing in human endothelial cells. Blood 99:2060–9, 2002 243. West CM, Price P: Combretastatin A4 phosphate. Anticancer Drugs 15:179–87, 2004 244. Stevenson JP, Rosen M, Sun W, et al: Phase I trial of the antivascular agent combretastatin A4 phosphate on a 5-day schedule to patients with cancer: magnetic resonance imaging evidence for altered tumor blood flow. J Clin Oncol 21:4428–38, 2003 245. Dowlati A, Robertson K, Cooney M, et al: A phase I pharmacokinetic and translational study of the novel vascular targeting agent combretastatin a-4 phosphate on a single-dose intravenous schedule in patients with advanced cancer. Cancer Res 62:3408–16, 2002 246. Rustin GJ, Galbraith SM, Anderson H, et al: Phase I clinical trial of weekly combretastatin A4 phosphate: clinical and pharmacokinetic results. J Clin Oncol 21:2815–22, 2003 247. Tozer GM, Kanthou C, Baguley BC: Disrupting tumor blood vessels. Nat Rev Cancer 5:423–35, 2005 248. Thotathil Z, Jameson MB: Early experience with novel immunomodulators for cancer treatment. Expert Opin Investig Drugs 16:1391–403, 2007

396

L.Q.M. Chow and S.G. Eckhardt

249. Zhao L, Ching LM, Kestell P, et al: Improvement of the antitumor activity of intraperitoneally and orally administered 5,6-dimethylxanthenone-4-acetic acid by optimal scheduling. Clin Cancer Res 9:6545–50, 2003 250. Zhao L, Ching LM, Kestell P, et  al: The antitumor activity of 5,6-dimethylxanthenone-4acetic acid (DMXAA) in TNF receptor-1 knockout mice. Br J Cancer 87:465–70, 2002 251. Jameson MB, Thompson PI, Baguley BC, et al: Clinical aspects of a phase I trial of 5,6-dimethylxanthenone-4-acetic acid (DMXAA), a novel antivascular agent. Br J Cancer 88:1844–50, 2003 252. Breitbach CJ, Paterson JM, Lemay CG, et al: Targeted inflammation during oncolytic virus therapy severely compromises tumor blood flow. Mol Ther 15:1686–93, 2007 253. Hermiston TW, Kuhn I: Armed therapeutic viruses: strategies and challenges to arming oncolytic viruses with therapeutic genes. Cancer Gene Ther 9:1022–35, 2002 254. Feldman AL, Restifo NP, Alexander HR, et al: Antiangiogenic gene therapy of cancer utilizing a recombinant adenovirus to elevate systemic endostatin levels in mice. Cancer Res 60:1503–6, 2000 255. Feldman AL, Libutti SK: Progress in antiangiogenic gene therapy of cancer. Cancer 89:1181–94, 2000 256. Calvo A, Feldman AL, Libutti SK, et al: Adenovirus-mediated endostatin delivery results in inhibition of mammary gland tumor growth in C3(1)/SV40 T-antigen transgenic mice. Cancer Res 62:3934–8, 2002 257. Thorne SH, Tam BY, Kirn DH, et al: Selective intratumoral amplification of an antiangiogenic vector by an oncolytic virus produces enhanced antivascular and anti-tumor efficacy. Mol Ther 13:938–46, 2006 258. Wong RJ, Chan MK, Yu Z, et  al: Angiogenesis inhibition by an oncolytic herpes virus expressing interleukin 12. Clin Cancer Res 10:4509–16, 2004 259. Cinatl J, Jr., Michaelis M, Driever PH, et al: Multimutated herpes simplex virus g207 is a potent inhibitor of angiogenesis. Neoplasia 6:725–35, 2004 260. Mahller YY, Vaikunth SS, Currier MA, et  al: Oncolytic HSV and erlotinib inhibit tumor growth and angiogenesis in a novel malignant peripheral nerve sheath tumor xenograft model. Mol Ther 15:279–86, 2007 261. Yang CT, Lin YC, Lin CL, et al: Oncolytic herpesvirus with secretable angiostatic proteins in the treatment of human lung cancer cells. Anticancer Res 25:2049–54, 2005 262. Bennett JJ, Malhotra S, Wong RJ, et al: Interleukin 12 secretion enhances antitumor efficacy of oncolytic herpes simplex viral therapy for colorectal cancer. Ann Surg 233:819–26, 2001 263. Bell JC: Oncolytic viruses: what’s next? Curr Cancer Drug Targets 7:127–31, 2007 264. Siemann DW, Shi W: Efficacy of combined antiangiogenic and vascular disrupting agents in treatment of solid tumors. Int J Radiat Oncol Biol Phys 60:1233–40, 2004 265. Huber PE, Bischof M, Jenne J, et al: Trimodal cancer treatment: beneficial effects of combined antiangiogenesis, radiation, and chemotherapy. Cancer Res 65:3643–55, 2005 266. Siemann DW, Chaplin DJ, Horsman MR: Vascular-targeting therapies for treatment of malignant disease. Cancer 100:2491–9, 2004 267. Jain RK, Finn AV, Kolodgie FD, et al: Antiangiogenic therapy for normalization of atherosclerotic plaque vasculature: a potential strategy for plaque stabilization. Nat Clin Pract Cardiovasc Med 4:491–502, 2007 268. Jain RK: Normalization of tumor vasculature: an emerging concept in antiangiogenic therapy. Science 307:58–62, 2005 269. Jain RK: Antiangiogenic therapy for cancer: current and emerging concepts. Oncology (Williston Park) 19:7–16, 2005 270. Gasparini G, Longo R, Toi M, et  al: Angiogenic inhibitors: a new therapeutic strategy in oncology. Nat Clin Pract Oncol 2:562–77, 2005 271. Zhou Q, Guo P, Gallo JM: Impact of angiogenesis inhibition by sunitinib on tumor distribution of temozolomide. Clin Cancer Res 14:1540–9, 2008 272. Franco M, Man S, Chen L, et al: Targeted anti-vascular endothelial growth factor receptor-2 therapy leads to short-term and long-term impairment of vascular function and increase in tumor hypoxia. Cancer Res 66:3639–48, 2006

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

397

273. Ma J, Waxman DJ: Dominant effect of antiangiogenesis in combination therapy involving cyclophosphamide and axitinib. Clin Cancer Res 15:578–88, 2009 274. Ma J, Waxman DJ: Modulation of the antitumor activity of metronomic cyclophosphamide by the angiogenesis inhibitor axitinib. Mol Cancer Ther 7:79–89, 2008 275. Bertolini F, Paul S, Mancuso P, et al: Maximum tolerable dose and low-dose metronomic chemotherapy have opposite effects on the mobilization and viability of circulating endothelial progenitor cells. Cancer Res 63:4342–6, 2003 276. Kerbel RS, Kamen BA: The anti-angiogenic basis of metronomic chemotherapy. Nat Rev Cancer 4:423–36, 2004 277. Pietras K, Hanahan D: A multitargeted, metronomic, and maximum-tolerated dose “chemoswitch” regimen is antiangiogenic, producing objective responses and survival benefit in a mouse model of cancer. J Clin Oncol 23:939–52, 2005 278. Gasparini G: Metronomic scheduling: the future of chemotherapy? Lancet Oncol 2:733–40, 2001 279. Hirota K, Semenza GL: Regulation of angiogenesis by hypoxia-inducible factor 1. Crit Rev Oncol Hematol 59:15–26, 2006 280. Senan S, Smit EF: Design of clinical trials of radiation combined with antiangiogenic therapy. Oncologist 12:465–77, 2007 281. Geng L, Donnelly E, McMahon G, et al: Inhibition of vascular endothelial growth factor receptor signaling leads to reversal of tumor resistance to radiotherapy. Cancer Res 61:2413–9, 2001 282. Sonveaux P, Brouet A, Havaux X, et  al: Irradiation-induced angiogenesis through the upregulation of the nitric oxide pathway: implications for tumor radiotherapy. Cancer Res 63:1012–9, 2003 283. Kermani P, Leclerc G, Martel R, et al: Effect of ionizing radiation on thymidine uptake, differentiation, and VEGFR2 receptor expression in endothelial cells: the role of VEGF(165). Int J Radiat Oncol Biol Phys 50:213–20, 2001 284. Riesterer O, Honer M, Jochum W, et  al: Ionizing radiation antagonizes tumor hypoxia induced by antiangiogenic treatment. Clin Cancer Res 12:3518–24, 2006 285. Hoang T, Huang S, Armstrong E, et al: Augmentation of radiation response with the vascular targeting agent ZD6126. Int J Radiat Oncol Biol Phys 64:1458–65, 2006 286. Kozin SV, Boucher Y, Hicklin DJ, et  al: Vascular endothelial growth factor receptor-2blocking antibody potentiates radiation-induced long-term control of human tumor xenografts. Cancer Res 61:39–44, 2001 287. Dings RP, Loren M, Heun H, et  al: Scheduling of radiation with angiogenesis inhibitors anginex and Avastin improves therapeutic outcome via vessel normalization. Clin Cancer Res 13:3395–402, 2007 288. Brazelle WD, Shi W, Siemann DW: VEGF-associated tyrosine kinase inhibition increases the tumor response to single and fractionated dose radiotherapy. Int J Radiat Oncol Biol Phys 65:836–41, 2006 289. Ning S, Laird D, Cherrington JM, et  al: The antiangiogenic agents SU5416 and SU6668 increase the antitumor effects of fractionated irradiation. Radiat Res 157:45–51, 2002 290. Schueneman AJ, Himmelfarb E, Geng L, et al: SU11248 maintenance therapy prevents tumor regrowth after fractionated irradiation of murine tumor models. Cancer Res 63:4009–16, 2003 291. Williams KJ, Telfer BA, Brave S, et al: ZD6474, a potent inhibitor of vascular endothelial growth factor signaling, combined with radiotherapy: schedule-dependent enhancement of antitumor activity. Clin Cancer Res 10:8587–93, 2004 292. Marzola P, Degrassi A, Calderan L, et al: Early antiangiogenic activity of SU11248 evaluated in  vivo by dynamic contrast-enhanced magnetic resonance imaging in an experimental model of colon carcinoma. Clin Cancer Res 11:5827–32, 2005 293. Kerkela R, Grazette L, Yacobi R, et al: Cardiotoxicity of the cancer therapeutic agent imatinib mesylate. Nat Med 12:908–16, 2006 294. Kierszniewska-Stepien D, Pietras T, Gorski P, et al: Serum vascular endothelial growth factor and its receptor level in patients with chronic obstructive pulmonary disease. Eur Cytokine Netw 17:75–9, 2006

398

L.Q.M. Chow and S.G. Eckhardt

295. Santos S, Peinado VI, Ramirez J, et al: Enhanced expression of vascular endothelial growth factor in pulmonary arteries of smokers and patients with moderate chronic obstructive pulmonary disease. Am J Respir Crit Care Med 167:1250–6, 2003 296. Kasahara Y, Tuder RM, Taraseviciene-Stewart L, et al: Inhibition of VEGF receptors causes lung cell apoptosis and emphysema. J Clin Invest 106:1311–9, 2000 297. Li M, Jendrossek V, Belka C: The role of PDGF in radiation oncology. Radiat Oncol 2:5, 2007 298. Fakih MG, Lombardo JC: Bevacizumab-induced nasal septum perforation. Oncologist 11:85–6, 2006 299. Lordick F, Geinitz H, Theisen J, et al: Increased risk of ischemic bowel complications during treatment with bevacizumab after pelvic irradiation: report of three cases. Int J Radiat Oncol Biol Phys 64:1295–8, 2006 300. Zachary I: Neuroprotective role of vascular endothelial growth factor: signalling mechanisms, biological function, and therapeutic potential. Neurosignals 14:207–21, 2005 301. Bertolini F, Mancuso P, Shaked Y, et al: Molecular and cellular biomarkers for angiogenesis in clinical oncology. Drug Discov Today 12:806–12, 2007 302. Therasse P, Arbuck SG, Eisenhauer EA, et al: New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst 92:205–16, 2000 303. Shaked Y, Bocci G, Munoz R, et al: Cellular and molecular surrogate markers to monitor targeted and non-targeted antiangiogenic drug activity and determine optimal biologic dose. Curr Cancer Drug Targets 5:551–9, 2005 304. Hlatky L, Hahnfeldt P, Folkman J: Clinical application of antiangiogenic therapy: microvessel density, what it does and doesn’t tell us. J Natl Cancer Inst 94:883–93, 2002 305. Tozer GM: Measuring tumor vascular response to antivascular and antiangiogenic drugs. Br J Radiol 76 Spec No 1:S23–35, 2003 306. Salven P, Manpaa H, Orpana A, et  al: Serum vascular endothelial growth factor is often elevated in disseminated cancer. Clin Cancer Res 3:647–51, 1997 307. Seon BK, Takahashi N, Haba A, et al: Angiogenesis and metastasis marker of human tumors. Rinsho Byori 49:1005–13, 2001 308. Heer K, Kumar H, Read JR, et al: Serum vascular endothelial growth factor in breast cancer: its relation with cancer type and estrogen receptor status. Clin Cancer Res 7:3491–4, 2001 309. Byrne GJ, Bundred NJ: Surrogate markers of tumoral angiogenesis. Int J Biol Markers 15:334–9, 2000 310. Taipale J, Keski-Oja J: Growth factors in the extracellular matrix. FASEB J 11:51–9, 1997 311. Zaman K, Driscoll R, Hahn D, et al: Monitoring multiple angiogenesis-related molecules in the blood of cancer patients shows a correlation between VEGF-A and MMP-9 levels before treatment and divergent changes after surgical vs. conservative therapy. Int J Cancer 118:755–64, 2006 312. Kuroi K, Toi M: Circulating angiogenesis regulators in cancer patients. Int J Biol Markers 16:5–26, 2001 313. Mysliwiec P, Piotrowski Z, Zalewski B, et al: Plasma VEGF-A and its soluble receptor R1 correlate with the clinical stage of colorectal cancer. Rocz Akad Med Bialymst 49 Suppl 1:85–7, 2004 314. Kumar H, Heer K, Greenman J, et al: Soluble FLT-1 is detectable in the sera of colorectal and breast cancer patients. Anticancer Res 22:1877–80, 2002 315. Toi M, Bando H, Ogawa T, et al: Significance of vascular endothelial growth factor (VEGF)/ soluble VEGF receptor-1 relationship in breast cancer. Int J Cancer 98:14–8, 2002 316. Byrne GJ, McDowell G, Agarawal R, et  al: Serum vascular endothelial growth factor in breast cancer. Anticancer Res 27:3481–7, 2007 317. Byrne GJ, Hayden KE, McDowell G, et  al: Angiogenic characteristics of circulating and tumoral thrombospondin-1 in breast cancer. Int J Oncol 31:1127–32, 2007 318. Lin Y, Weisdorf DJ, Solovey A, et al: Origins of circulating endothelial cells and endothelial outgrowth from blood. J Clin Invest 105:71–7, 2000

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

399

319. Norden-Zfoni A, Manola J, Desai J, et al: Levels of circulating endothelial cells (CECs) and monocytes as pharmacodynamic markers of SU11248 activity in patients (pts) with metastatic imatinib-resistant GIST. Proc Am Soc Clin Oncol, 2005 ASCO Annual Meeting, Orlando, FL. Abstract No: 9036, 2005 320. Bertolini F, Mancuso P, Kerbel RS: Circulating endothelial progenitor cells. N Engl J Med 353:2613–6; author reply 2613–6, 2005 321. Shaked Y, Bertolini F, Emmenegger U, et al: On the origin and nature of elevated levels of circulating endothelial cells after treatment with a vascular disrupting agent. J Clin Oncol 24:4040; author reply 4040–1, 2006 322. Goon PK, Watson T, Shantsila E, et al: Standardization of circulating endothelial cell enumeration by the use of human umbilical vein endothelial cells. J Thromb Haemost 5:870–2, 2007 323. Boos CJ, Goon PK, Lip GY: Circulating endothelial progenitor cells. N Engl J Med 353:2613–6; author reply 2613–6, 2005 324. Shaked Y, Ciarrocchi A, Franco M, et al: Therapy-induced acute recruitment of circulating endothelial progenitor cells to tumors. Science 313:1785–7, 2006 325. Celis JE, Celis P, Palsdottir H, et  al: Proteomic strategies to reveal tumor heterogeneity among urothelial papillomas. Mol Cell Proteomics 1:269–79, 2002 326. Celis JE, Gromov P, Cabezon T, et  al: Proteomic characterization of the interstitial fluid perfusing the breast tumor microenvironment: a novel resource for biomarker and therapeutic target discovery. Mol Cell Proteomics 3:327–44, 2004 327. Abramovitch R, Dafni H, Smouha E, et al: In vivo prediction of vascular susceptibility to vascular susceptibility endothelial growth factor withdrawal: magnetic resonance imaging of C6 rat glioma in nude mice. Cancer Res 59:5012–6, 1999 328. Jackson A, O’Connor JP, Parker GJ, et al: Imaging tumor vascular heterogeneity and angiogenesis using dynamic contrast-enhanced magnetic resonance imaging. Clin Cancer Res 13:3449–59, 2007 329. Buckley DL: Uncertainty in the analysis of tracer kinetics using dynamic contrast-enhanced T1-weighted MRI. Magn Reson Med 47:601–6, 2002 330. Leach MO, Brindle KM, Evelhoch JL, et al: Assessment of antiangiogenic and antivascular therapeutics using MRI: recommendations for appropriate methodology for clinical trials. Br J Radiol 76 Spec No 1:S87–91, 2003 331. Leach MO, Brindle KM, Evelhoch JL, et al: The assessment of antiangiogenic and antivascular therapies in early-stage clinical trials using magnetic resonance imaging: issues and recommendations. Br J Cancer 92:1599–610, 2005 332. Chen X, Park R, Khankaldyyan V, et  al: Longitudinal microPET imaging of brain tumor growth with F-18-labeled RGD peptide. Mol Imaging Biol 8:9–15, 2006 333. Schirner M, Menrad A, Stephens A, et al: Molecular imaging of tumor angiogenesis. Ann N Y Acad Sci 1014:67–75, 2004 334. McDonald DM, Choyke PL: Imaging of angiogenesis: from microscope to clinic. Nat Med 9:713–25, 2003 335. Yang DJ, Kim KD, Schechter NR, et al: Assessment of antiangiogenic effect using 99mTcEC-endostatin. Cancer Biother Radiopharm 17:233–45, 2002 336. Toner GC, Mitchell PL, De Boer R, et al: PET imaging study of SU11248 in patients with advanced malignancies. Proc Am Soc Clin Oncol, Vol. 22, p 191. Abstract No: 767, 2003 337. Scott A, Mitchell P, O’Keefe G, et al: Tumor perfusion as assessed by Oxygen-15-Water PET imaging during treatment with SU011248 in patients with advanced malignancies, EORTC/ AACR/NCI Molecular Targets Meeting. Philadelphia, PN, 2005 338. Wang JQ, Miller KD, Sledge GW, et al: Synthesis of [18F]SU11248, a new potential PET tracer for imaging cancer tyrosine kinase. Bioorg Med Chem Lett 15:4380–4, 2005 339. Pandya NM, Dhalla NS, Santani DD: Angiogenesis – a new target for future therapy. Vascul Pharmacol 44:265–74, 2006 340. Halimi JM, Azizi M, Bobrie G, et al: [Vascular and renal effects of anti-angiogenic therapy]. Nephrol Ther 4:602–15, 2008

400

L.Q.M. Chow and S.G. Eckhardt

341. Ostendorf T, Kunter U, Eitner F, et  al: VEGF mediates glomerular endothelial repair. J Clin Invest 104:913–23, 1999 342. Izzedine H: [Angiogenesis inhibitor therapies: focus on hypertension and kidney toxicity]. Bull Cancer 94:981–6, 2007 343. Moss KG, Toner GC, Cherrington JM, et  al: Hair depigmentation is a biological readout for pharmacological inhibition of KIT in mice and humans. J Pharmacol Exp Ther 307:476–80, 2003 344. Autier J, Mateus C, Wechsler J, et al: [Cutaneous side effects of sorafenib and sunitinib]. Ann Dermatol Venereol 135:148–53; quiz 147, 154, 2008 345. Robert C, Soria JC, Spatz A, et al: Cutaneous side-effects of kinase inhibitors and blocking antibodies. Lancet Oncol 6:491–500, 2005 346. Robert C, Faivre S, Raymond E, et al: Subungual splinter hemorrhages: a clinical window to inhibition of vascular endothelial growth factor receptors? Ann Intern Med 143:313–4, 2005 347. Hinnen P, Eskens FA: Vascular disrupting agents in clinical development. Br J Cancer 96:1159–65, 2007 348. Conway EM, Collen D, Carmeliet P: Molecular mechanisms of blood vessel growth. Cardiovasc Res 49:507–21, 2001 349. Muehlbauer PM: Anti-angiogenesis in cancer therapy. Semin Oncol Nurs 19:180–92, 2003 350. Buchdunger E, Cioffi CL, Law N, et al: Abl protein-tyrosine kinase inhibitor STI571 inhibits in vitro signal transduction mediated by c-kit and platelet-derived growth factor receptors. J Pharmacol Exp Ther 295:139–45, 2000 351. Krystal GW, Honsawek S, Litz J, et al: The selective tyrosine kinase inhibitor STI571 inhibits small cell lung cancer growth. Clin Cancer Res 6:3319–26, 2000 352. Matei D, Chang DD, Jeng MH: Imatinib mesylate (Gleevec) inhibits ovarian cancer cell growth through a mechanism dependent on platelet-derived growth factor receptor alpha and Akt inactivation. Clin Cancer Res 10:681–90, 2004 353. Skinner MA, Safford SD, Freemerman AJ: RET tyrosine kinase and medullary thyroid cells are unaffected by clinical doses of STI571. Anticancer Res 23:3601–6, 2003 354. NCI/CTEP: NCI/CTEP Mass Solicitation for Phase I and II Clinical Trials Involving GW786034. Glaxo-Smith-Kline Product Information. GW786034, 2005 355. Sonpavde G, Hutson TE: Pazopanib: a novel multitargeted tyrosine kinase inhibitor. Curr Oncol Rep 9:115–9, 2007 356. Wood JM, Bold G, Buchdunger E, et al: PTK787/ZK 222584, a novel and potent inhibitor of vascular endothelial growth factor receptor tyrosine kinases, impairs vascular endothelial growth factor-induced responses and tumor growth after oral administration. Cancer Res 60:2178–89, 2000 357. Drevs J, Hofmann I, Hugenschmidt H, et  al: Effects of PTK787/ZK 222584, a specific inhibitor of vascular endothelial growth factor receptor tyrosine kinases, on primary tumor, metastasis, vessel density, and blood flow in a murine renal cell carcinoma model. Cancer Res 60:4819–24, 2000 358. Ciardiello F, Bianco R, Caputo R, et al: Antitumor activity of ZD6474, a vascular endothelial growth factor receptor tyrosine kinase inhibitor, in human cancer cells with acquired resistance to antiepidermal growth factor receptor therapy. Clin Cancer Res 10:784–93, 2004 359. Wedge SR, Kendrew J, Hennequin LF, et al: AZD2171: a highly potent, orally bioavailable, vascular endothelial growth factor receptor-2 tyrosine kinase inhibitor for the treatment of cancer. Cancer Res 65:4389–400, 2005 360. Gomez-Rivera F, Santillan-Gomez AA, Younes MN, et  al: The tyrosine kinase inhibitor, AZD2171, inhibits vascular endothelial growth factor receptor signaling and growth of anaplastic thyroid cancer in an orthotopic nude mouse model. Clin Cancer Res 13:4519–27, 2007 361. Takeda M, Arao T, Yokote H, et al: AZD2171 shows potent antitumor activity against gastric cancer over-expressing fibroblast growth factor receptor 2/keratinocyte growth factor receptor. Clin Cancer Res 13:3051–7, 2007 362. Wilmes LJ, Pallavicini MG, Fleming LM, et  al: AG-013736, a novel inhibitor of VEGF receptor tyrosine kinases, inhibits breast cancer growth and decreases vascular permeability as detected by dynamic contrast-enhanced magnetic resonance imaging. Magn Reson Imaging 25:319–27, 2007

14  Challenges and Successes in Developing Effective Anti-angiogenic Agents

401

363. Abrams TJ, Lee LB, Murray LJ, et al: SU11248 inhibits KIT and platelet-derived growth factor receptor beta in preclinical models of human small cell lung cancer. Mol Cancer Ther 2:471–8, 2003 364. Abrams TJ, Murray LJ, Pesenti E, et al: Preclinical evaluation of the tyrosine kinase inhibitor SU11248 as a single agent and in combination with “standard of care” therapeutic agents for the treatment of breast cancer. Mol Cancer Ther 2:1011–21, 2003 365. Murray LJ, Abrams TJ, Long KR, et  al: SU11248 inhibits tumor growth and CSF-1Rdependent osteolysis in an experimental breast cancer bone metastasis model. Clin Exp Metastasis 20:757–66, 2003 366. Mendel DB, Laird AD, Xin X, et al: In vivo antitumor activity of SU11248, a novel tyrosine kinase inhibitor targeting vascular endothelial growth factor and platelet-derived growth factor receptors: determination of a pharmacokinetic/pharmacodynamic relationship. Clin Cancer Res 9:327–37, 2003

Chapter 15

Targeted Therapeutics in Cancer Treatment Colin D. Weekes and Manuel Hidalgo

15.1 Introduction Historically, chemotherapy used for the treatment of malignancy was restricted to cytotoxic agents. Perturbation of DNA synthesis and the events regulating cell division are the primary targets of traditional cytotoxic drugs, as outlined in a previous chapter. Unfortunately, the events regulating cell division are not specific to cancer cells; therefore, these medications result in broad range of toxic side effects due to damage of normal cells. The narrow therapeutic window of traditional cytotoxic drugs is quite troublesome, given that palliation is the primary goal of oncology therapy. Recently, oncology therapy has migrated to the use of small molecules targeting intracellular events specific to tumor cells. The era of “targeted therapy” was heralded by the approval of the antibodies Rituximab and Trastuzumab for the treatment of relapsed or refractory low-grade follicular B-cell non-Hodgkin’s lymphoma and Her2/Neu-positive metastatic breast cancer, respectively [1, 2]. The use of small molecule inhibitors of intracellular pathways as a therapeutic principle was validated by the efficacy of imatinib mesylate for the treatment of bcr/abl-positive chronic myelogenous leukemia and gastrointestinal stromal tumors (GIST) [3, 4]. The development of small molecule inhibitors for cancer therapy has paralleled the scientific understanding of cellular processes regulating oncogenesis. The accumulation of genetic alterations effecting cellular differentiation and proliferation as well as apoptosis represents the critical cellular events mediating cancer cell initiation and survival. Cell proliferation and differentiation are regulated by a number of hormones, growth factors, and cytokines. These molecules interact with cellular receptors and communicate with the nucleus through a network of intracellular signaling pathways. In cancer cells, key components of these pathways may be altered through gene amplification resulting in overexpression or mutation, acting M. Hidalgo (*) Department of Oncology, School of Medicine, CEU San Pablo University, Madrid, Spain; Centro Integral Oncológico Clara Campal (CIOCC), Madrid, Spain; Gastrointestinal Clinical Research Unit, Centro Integral Investigaciones Oncológicas (CNIO), C/ Melchor Fernández Almagro 3, 28029 Madrid, Spain e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_15, © Springer Science+Business Media, LLC 2011

403

404

C.D. Weekes and M. Hidalgo

as oncogenes that lead to dysregulated cell signaling and proliferation, metastasis, and inhibition of apoptosis. Loss of heterozygosity (LOH) and the resultant silencing of tumor-suppressor genes either by mutation or epigenetic modification repress the negative regulation of cellular processes to promote oncogenesis. The components of abnormal signaling pathways specific to neoplastic cells, represent potential selective targets for new anticancer therapies. These potential targets include ligands, cellular receptors, intracellular second messengers, and nuclear transcription factors. Additional targets for small molecule inhibitors include proteins involved in epigenetic modification, DNA repair, metabolic pathways, protein transport, and protein degradation. This chapter is not meant to be all inclusive; however, will focus on demonstrating how various cellular processes regulating oncogenesis can be targeted for molecular cancer therapy. Table 15.1 summarizes the key novel targets and provides a limited listing of some of the agents currently in development against these targets. Additionally, this chapter will provide insight into the challenges of incorporating molecularly targeted therapy into the armamentarium of general oncology practice either as single agents or in combination with other molecularly targeted therapy or cytotoxic chemotherapy.

15.2 Therapeutic Technologies The age of molecular therapy has been heralded by the development of new forms of therapeutic compounds. These include monoclonal antibodies, kinase inhibitors, peptides, and nucleic acid-based technologies. The advent of technologies to mass produce such compounds has made it feasible to target extracellular proteins, cell surface receptors, and intracellular processes in an attempt to inhibit tumor growth and metastasis in patients.

15.2.1 Antibodies The advent of hybridoma technology in 1975 facilitated the therapeutic use of monoclonal antibodies [5]. Antibody therapy offers the potential for target-specific inhibition and immune-mediated tumor suppression. Monoclonal antibodies are composed of a heavy chain (a, d, e, g, m) and light chain (Kappa (k) and lambda (l)). Antigen recognition occurs at the cluster determining region (CDR) contained within the variable region of the antibody amino terminus. The carboxy terminus constant region binds to the Fc-receptor (FcR) on immune cells. Currently, only chimeric, humanized, and fully human antibodies are in clinical investigation. Chimeric antibodies possess murine variable regions fused to a human constant region. In contrast, humanized antibodies are comprised of murine CDR sequences grafted onto a human antibody. Lastly, fully human antibodies contain no murine sequences. These antibodies contain minimal murine sequences thus are less

15  Targeted Therapeutics in Cancer Treatment

405

Table 15.1  Intracellular targets and compounds in development Intracellular target Compound Generic name Trade name EGFR OSI774 Erlotinib Tarceva ZD1839 Gefitinib Iressa GW572016 Lapatinib Tykerb HKI-272 EKB-569 CI-1033 MEK CI-1040 PDO325901 ARRY_142886 (AZD6244) SRC BMS354825 Dasatinib Sprycel AZM475271 SKI-606 Apoptosis Apo2/TRAIL YM155 Aurora kinase AZD1152 MK-0457 Polo-like kinase ON01910 BI 2536 N-cadherin Exherin HSP-90 17-DMAG IPI504 KSP SB-743921 MK-0731 BCR-ABL, cKit, STI571 Imatinib mesylate Gleevec PDGFR AMN107 Nilotinib BMS-354825 Dasatinib Sprycel Multikinase BAY 43-9006 Sorafenib Nexavar SU11248 Sunitinib Sutent ZD6474 Zactima VEGFR PTK787/ZK222584 mTOR AY 22989 Rapamycin Sirolimus CCI-779 Temsirolimus Torisel RAD001 Everolimus Afinitor AP23573 FTI R115777 Tipifarnib Zarnestra CDK inhibitors L86-8275 Flavopiridol UCN-01 Bryostatin-1 CYC202 Roscovitine BMS387032 E7070 Indisulam PD0332991 Proteosome PS-341 Bortezomib Velcade KSP kinesis spindle protein; FTI farnesyl transferase inhibitor; CDK cyclin-dependent kinase inhibitor; HSP-90 heat shock protein-90; PDGFR platelet-derived growth factor receptor

406

C.D. Weekes and M. Hidalgo

immunogenic relative to murine antibodies, resulting in prolonged half-lives of these proteins [6]. Monoclonal antibodies have been developed to abrogate ligand–receptor interactions involved in oncogenesis. The physiologic consequence of receptor inhibition is the inability to initiate intracellular growth and survival signals. The target-dependent disruption of the ligand–receptor interaction occurs by direct antibody binding to the ligand or, in turn, binding to the ligand-binding domain of the receptor. Antibody-mediated receptor inhibition occurs primarily by steric hindrance and promotes receptor endocytosis and degradation. In rare instances, such as the case of the c-Met and death receptors, dual-arm monoclonal antibodies result in receptor activation. Antibody-based therapy has limited effect on receptors undergoing autophosphorylation independent of ligand–receptor binding. The US Food and Drug Administration (FDA) approval of Rituximab (Rituxan, Genentech Inc.) to treat patients with relapsed or refractory CD20-positive lowgrade follicular non-Hodgkin’s lymphoma in 1997 represented the first successful monoclonal antibody to obtain an indication for oncologic therapy. This success has been followed by the approval of a number of monoclonal antibodies targeting a variety of receptor in both solid tumors and hematologic malignancies (Table  15.2). Monoclonal antibodies also serve as carrier molecules to which either cellular toxins or radioactive isotopes are conjugated. This strategy allows for the delivery of highly toxic substances to the tumor cell while minimizing the exposure of adjacent normal cells. Gentuzumab ozogamicin (Mylotarg, Wyeth Pharmaceutical, Inc.) delivers the cellular toxin calichaemicin to CD33-expressing acute myelogenous leukemia cells. Similarly, both 90Y-Ytrium (irbitumomab tiuxetan, Zevalin, IDEC Pharmaceutical Corp.) and 131I-Iodine (tositumomab, Bexxar, GlaxoSmithKine) have been conjugated to anti-CD20 monoclonal antibodies for lymphoma cell-specific delivery of radiation. Antibody-based therapy possesses the potential to activate the tumor-suppressor function of the immune system through either antibody-dependent cell-mediated cytotoxicity (ADCC) or complement-dependent cytotoxicity (CDC). ADCC is activated upon the binding of the antibody Fc region to the Fcg-receptor (FcgR). ADCC is primarily mediated by the high affinity of IgG (IgG1 and IgG3) subclasses of antibodies to the FcgR (FcgRIIa and FcgRIIIa) in humans [7, 8]. Germline polymorphisms in the FcgRs affect their affinity for IgG: a histidine (H)/arginine (R) polymorphism at position 131 for FcgRIIa and a phenylalanine (P)/valine (V) polymorphism at position 158 for FcgRIIIa. The FcgRIIa-131H/H and FcgRIIIa-158V/V genotypes promote high-affinity binding to augment ADCC-mediated tumor growth inhibition [9, 10]. As predicted, patients harboring these genotypes achieved improved clinical outcomes when treated with rituximab therapy [11, 12]. Recently, both FcgRIIa-131H/H and FcgRIIIa-158V/V genotypes predicted the progressionfree survival of a cohort of colorectal cancer patients treated with cetuximab and irinotecan independent of Kras mutation status [13]. These findings support the clinical importance of ADCC in addition to target-dependent effects of antibodybased therapy.

15  Targeted Therapeutics in Cancer Treatment

407

Table  15.2  US Food and Drug Administration-approved monoclonal antibodies for cancer therapy Unconjugated Monoclonal Antibodies Antigenic Monoclonal Antibody target antibody construct

Isotype

Cancer indication

CD20 HER-2 CD52 VEGF EGFR

IgG1 IgG1 IgG1 IgG1 IgG1

Rituximab Trastuzumab Alemtuzumab Bevacizumab Cetuximab

Chimeric Humanized Humanized Humanized Chimeric

EGFR Panitumumab Human Conjugated Monoclonal Antibodies Antigenic Monoclonal target antibody Conjugate

IgG2

Non-Hodgkin’s lymphoma Breast cancer, gastric cancer Chronic lymphocytic leukemia Colorectal cancer Colorectal cancer, head and neck cancer Colorectal cancer

Isotype

Cancer indication

CD33

IgG4

Acute myelogenous leukemia

Y-Ytrium

IgG1

Non-Hodgkin’s lymphoma

I-Iodine

IgG2a

Non-Hodgkin’s lymphoma

CD20 CD20

Gentuzumab ozogamicin Irbitumomab tiuxetan Tositumomab

Calichaemicin 90

131

15.2.2 Tyrosine Kinase Inhibitors The human genome encodes for approximately 518 kinases. Almost all intracellular signals are conducted by a network of phosphorylation events. Kinase inhibition abrogates the physiologic signals promoting oncogenesis. As a result, a wide variety of tyrosine kinase inhibitors (TKIs) are undergoing clinical investigation. Protein kinases are defined by their ability to transfer the terminal phosphate of ATP to substrates containing serine, threonine, or tyrosine residues. Kinases typically possess a conserved secondary structure of 12 subdomains. These domains form a bi-lobed cleft structure to which ATP binds. The hinge portion of the kinase domain links the amino and carboxy ends of the domain; thereby, approximating the cleft walls. The central portion of the cleft houses the ATP-binding pocket. ATP binds to the cleft by hydrogen bond formation between adenosine and the kinase domain hinge region. The ribose and triphosphate groups of ATP bind in a hydrophilic channel extending to the substrate-binding site. All kinases possess a highly conserved flexible activation loop marked by the conserved aspartic acid/phenylalanine/glycine (DFG) or alanine/ proline/glutamic acid (APE) motifs [14, 15]. In the activated confirmation, it is phosphorylated and catalytically competent. The inactive confirmation results in the DFG motif blocking the substrate-binding site. The majority of current kinase inhibitors mimic ATP by presenting sites for hydrogen bonding with the hinge region. Rarely is the ribose domain of ATP utilized as is the case of the src inhibitor AZD0530. There are four distinct subclasses of kinase inhibitors. Type I inhibitors bind to the active conformation of the ATP kinase domain, the so-called DFG-in

408

C.D. Weekes and M. Hidalgo

conformation. These compounds represent the majority of ATP-competitive inhibitors. Type II inhibitors recognize the inactive conformation of the kinase, “DFG-out conformation.” Type II kinase inhibitors, such as imatinib and sorafenib share a common molecular framework that allows for the exploitation of the exposed conserved hydrogen bonds, resulting in the multikinase inhibition demonstrated by individual compounds within this class of kinase inhibitors. Type III kinase inhibitors bind regions adjacent to the kinase domain and allosteric modulators of kinase activity. These are highly selective kinase inhibitors due the use of unique binding sites and regulatory mechanisms distinct to specific kinases. CI-1040 occupies a pocket adjacent to the ATP-binding site to impart allosteric inhibition of MEK 1 and MEK 2. In contrast, the Akt inhibitor Akt-I-1 binds to the pleckstrain homology domain of Akt. Lastly, type IV kinase inhibitors form irreversible, covalent bonds to the kinase active domain. The EGFR inhibitor HKI-272 is representative of this class of compounds. In summary, current receptor tyrosine kinase (RTK)-targeted therapy consists primarily of antibodies and small molecule TKIs. Antibody-based therapy targets extracellular events to inhibit cancer cells from obtaining growth and survival cues. Additionally, antibody-based therapy possesses the potential for immune-mediated tumor suppression. Conversely, small molecule kinase inhibitors impede the propagation of intracellular oncogenic signals. The mechanism of kinase inhibition determines both the promiscuity and reversibility of kinase inhibition by these molecules. All of these factors have clinical implications for therapeutic benefit, toxicity, and induction of resistance to these compounds.

15.2.3 Nucleic Acids An antisense oligonucleotide (ASO) is a short single-stranded deoxyribonucleotide (typically 20 bp in length). Hybridization of ASO to messenger RNA by Watson– Crick base pairing leads to gene expression inhibition by RNase H activity stimulation resulting in target mRNA degradation. The design of an ASO to maximize gene silencing depends upon the following factors (a) secondary structure of the RNA prediction; (b) identification of RNA secondary local structures accessible for hybridization; (c) GC content and motif determination; (d) binding energy (DG37o) [16]. Approximately, only 2–5% of developed ASOs are successful in target-­ specific inhibition of gene expression. Naked, unmodified ASOs are highly susceptible to degradation by the abundant nucleases found in biologic fluids. The first-generation phosphorothioate-ASOs (PS-ASO) replace a phosphodiester bond nonbridging oxygen atom with sulfur moiety to form a phosphorothioate (PS) backbone [17]. Second-generation ASOs are characterized by a 10-base pair (bp) central gap of unmodified PS-ASO winged by either one of two alkyl-modified PS-ASOs of either 2¢-O-Methyl (2¢-MO) and 2¢-O-Methoxyethyl (2¢-MOE) on the 3¢ and 5¢ ends of molecule [18]. This modification augments nuclease resistance and binding affinity to target mRNAs, while

15  Targeted Therapeutics in Cancer Treatment

409

overcoming the inability of a completely alkyl-modified PS-ASO to promote RNase H activity. Third-generation ASOs are characterized by modification of the furanose ring of the nucleotide and were developed to improve target affinity, nuclease resistance, biostability and pharmacokinetics [19]. Peptide nucleic acid (PNA), locked nucleic acids (LNA) and phosphoroamidate morpholino oligomers (PMO) are the most studied third-generations ASOs. All of these compounds are nuclease and protease resistant. PNAs are highly biostable DNA mimics containing phosphoprotein backbones [20, 21]. Gene silencing occurs predominately via steric hindrance of translational machinery leading to protein knockdown. LNA are conformationally restricted nucleotides possessing a 2¢-O 4¢-C-methylene bridge in the b-D-ribofuranosyl configuration to enhance target affinity [22]. The use of chimeric LNA–DNA molecules to flank a 10 bp portion of PS-ASO similar to that of second-generation gapmers establishes RNase H activity [23]. PMO are ASOs in which the ribose sugar and phosphodiester bonds are replaced by a morpholino ring and a phosphoroamidate linkage, respectively [24]. Steric hindrance of ribosomal assembly results in translational arrest. Other thirdgeneration ASOs include N3¢-P5¢ phosphoroamidate, 2¢-fluoro-arabino nucleic acid, cyclohexene nucleic acid and tricycle-DNA. 15.2.3.1 RNA Interference Historically, nucleic acid-based oncologic therapy consisted of primarily ASO therapy. More recently, RNA interference (RNAi) has become a potential therapeutic option [25, 26]. This strategy is designed to selectively degrade messenger RNA (mRNA) of genes responsible for molecular process that promote oncogenesis. RNAi utilizes small-interfering RNA (siRNA) that are generated as a result of the cleavage of precursor double-stranded RNA by RNaseIII endonuclease dicer. The siRNA then enters the RNA-induced silencing complex (RISC), which is activated by thermodynamically governed RNA 5¢ strand selection. The selected strand serves as a template for the RISC complex to selectively degrade complementary mRNA [27]. The targeting of oncogenes makes RNAi a potential oncologic therapeutic intervention. This technology has been utilized both in vitro and in vivo to inhibit gene expression associated with tumor cell growth and survival. The limitation use of RNAi as an oncologic therapy is due to relatively large molecular weight and its polyanionic nature. As a result, siRNA molecules do not efficiently cross the cell membrane. The implementation of siRNA as an oncologic therapy will require more efficient delivery systems. Currently, these delivery systems can be segregated into viral and nonviral-based strategies. 15.2.3.2 Nucleic Acid Delivery Systems Delivery of nucleic acid compounds to the target lesions remains the primary impediment to implementation of nucleic acid technology as cancer therapy.

410

C.D. Weekes and M. Hidalgo

Nucleic acids undergo limited absorptive endocytosis due to their negative charge. Passive absorptive endocytosis in turn is inadequate for the delivery of large quantities of therapeutic nucleic acids. Viral vectors represent a highly efficient delivery system; however, their clinical utility is limited by the loading capacity and potential safety risks of some viruses. Nonviral vector delivery systems in development include hydrodynamic injection, cholesterol conjugation, cationic delivery systems, and microparticles [16, 28, 29]. Cationic delivery systems are characterized by their net positive charge, which facilitates complex interaction of polyanionic nucleic acids such as siRNA and ASOs and the interface with the negatively charged cell membrane. These systems are either lipid or polymer/peptide-based systems. Liposomes are chemically stable particles consisting of an aqueous core with the entrapped hydrophilic compound encased in a phospholipid bilayer. In contrast, lipoplexes spontaneously form between oppositely charged lipids and nucleic acids; however, are physiochemically unstable [30]. Liposomes have been PEGylated to improve their pharmacokinetic properties. Liposomes have been FDA approved to package doxorubicin and represents a viable delivery system for siRNA and ASOs alike. Polyethylimine (PEI) is the most widely investigated polymer for use as a siRNA delivery system. Recently, a PEGylated PEI containing VEGF-targeted siRNA conjugated with a RGD-targeting sequence inhibited tumor vasculature formation in  vivo [31, 32]. Cationic lipid carriers like N-[1-(2,3-dioleoyloxy)propyl]-N,N,N,-trimethylammonium chloride and N-[1-(2,3-dioleoyloxy)propyl]-N,N,N,-trimethylammonium methyl sulfate are the most widely used vectors for ASO internalization [33, 34]. Cell-penetrating peptides, in which the ASO is covalently conjugated to positively charged peptides to rapidly traverse the cell membrane by energy-dependent active transport mechanisms has been utilized [35]. Lastly, self-assembled nanoparticles consisting of a liposome, protamine, and DNA core (LPD) PEGylated with the nucleic acid particle possesses the capacity to deliver multiple siRNA-targeting multiple genes within a single nanoparticle. In addition, peptide sequences targeting a variety of receptors can be added to potentially add tumor cell specificity to the LPD siRNA delivery system [36, 37]. The ongoing improvement in delivery systems is required for the clinical implementation of therapeutic nucleic acid technology as a cancer therapy.

15.3 Cell Cycle The cell cycle forms the platform upon which all cellular processes are governed (Fig. 9.1). The cell cycle regulates the process of cellular proliferation, growth as well as cell division after DNA damage. The cell cycle can be divided into four phases, the transition of which is regulated by a series of checkpoints to insure genetic code fidelity. The quiescent phase (G0) represents the period of cellular dormancy during which the cell is not actively in the cell cycle. Preparation for DNA synthesis occurs during the initial gap 1 (G1) phase of the cell cycle. S phase

15  Targeted Therapeutics in Cancer Treatment

411

G0

M

Cyclin B CDC2

G1

G2

Cyclin D

Cyclin A

CDK4 CDK6

CDC2

Cyclin E CDK2

S Cyclin A CDK2

CDK Inhibitors

Fig. 15.1  Cell cycle

is characterized by DNA synthesis. Gap 2 (G2) is second preparatory phase necessary­for mitosis. Mitosis (M) culminates the cell cycle during which the newly synthesized DNA and nuclear contents migrates to the resultant daughter cells. Progression through the cycle is facilitated by complexes formed between cyclindependent kinases (CDK) and the regulatory protein cyclin. Inhibition of cell cycle progression is coordinated by a series of negative regulatory proteins known as CDK inhibitors. A detailed review of the cell cycle is beyond the context of this chapter and the reader is referred to more detailed review of the cell cycle by Sherr and Roberts [38]. The improved knowledge of cell cycle regulation combined with the understanding of its critical role in carcinogenesis has resulted in the targeting of cell cycle regulators as a potential therapeutic strategy. Targeting of the cell cycle and, in particular, CDKs stems from the prevalent perturbation of CDKs during carcinogenesis as well as the observation that CDK inhibition can induce apoptosis. Cancer cells are characterized by the loss of checkpoint control, which can occur by either cyclin overexpression or the inactivation of CDKs [39]. Cyclin D is commonly overexpressed in breast cancer and the translocation event between chromosomes 11 and 14 (t11:14) prominent in mantle cell lymphoma results in the overexpression of cyclin D by its juxtaposition to the immunoglobulin promoter region [40, 41]. p16 is an INK4 gene that frequently undergoes hypermethylation resulting in its epigenetic silencing. This event is

412

C.D. Weekes and M. Hidalgo

c­ ommonly observed in melanoma, colorectal, breast and lung cancers [42]. Pharmacologic CDK inhibitors (CDKI) function as cell cycle-specific agents by binding to the CDK catalytic domain as well as their indirect impact on regulatory pathways of CDKs [43]. Recently, the importance of transcription regulation by CDK-dependent phosphorylation of the C-terminal domain (CTD) of RNA ­polymerase II has been appreciated. CDK-dependent CTD phosphorylation results in RNA polymerase II inactivation; thereby, suppressing mRNA production and transcription during mitosis [43, 44]. CDKI-mediated inhibition of CTD phosphorylation may contribute to its proapoptotic effect [45, 46]. Therefore the goal of targeting the CDKs would be to reinstitute appropriate checkpoint regulation and resultant apoptosis. Currently, there are a number of compounds in various stages of development including Flavopiridol, UCN-01, Bryostatin-1, YC202, BMS 387032, E7070, and PD0332991 (Table 15.3). Flavopiridol is the most extensively studied molecule of this group of CDKIs and the results of its development will be used to illustrate the utility of this class of agents in cancer therapy. Flavopiridol is a pan-­­cyclindependent kinase inhibitor of CDK2, CDK4, and CDK6 at nanomolar concentrations. It effectively induces cell cycle arrest at both the G1/S and G2/M checkpoints. In addition to its effects as a cell cycle modulator, its administration has been shown to induce apoptosis by a variety of mechanisms [47, 48]. The antitumor properties of flavopiridol have been validated in vivo with xenograft models of prostate and squamous head and neck cancers [49, 50]. Clinical evaluation of flavopiridol indicates that its apoptotic effects are highly dose and schedule dependent. Flavopiridol administration as 30  min bolus followed by a 4-h infusion attains free drug levels of 250–300  nmol/L, which is necessary to induce apoptosis. This regimen produced nearly a 50% response rate in patients with fludarabine refractory chronic lymphocytic leukemia resulting in prolonged survival [51]. Flavopiridol, as well as bryostatin-1, have been ­combined

Table 15.3  Cyclin-dependent kinase inhibitors Drug

Cell cycle phase

Target

IC50 (nmol/L)

Phase of development

Flavopiridol

G1/S Arrest, G2

  100

Phase II

UCN-01 Bryostatin 1 Roscovitine

S, G2 G2 arrest S, G2

<260   100

Phase II Phase II Phase II

E7070 BMS387032 PD0332991

S, G2 S, G2 G1 arrest

CDK1, CDK2, CDK4/CDK6, CDK9-Cyclin T-RNA Pol II CDK2 CDK2 CDK2, CDK9CyclinT-RNA Pol II CDK2, CDK1 CDK2, CDK1 CDK4/6

  11   48   11

Phase II Phase II Phase II

15  Targeted Therapeutics in Cancer Treatment

413

with cytotoxic agents to overcome cell cycle-mediated drug resistance. Cell cycle-mediated drug resistance is characterized by the insensitivity to a chemotherapeutic due to the induction of checkpoint activation, thus resulting in cell cycle arrest. This was exemplified by the failure of irinotecan to demonstrate antitumor activity in a xenograft model after exposure to flavopiridol due to the induction of G2 cell cycle arrest. This strategy has been utilized in the clinical development of combinations of cytotoxic agents with flavopiridol and other CDKIs. The sequential administration of Irinotecan followed by flavopiridol divided by a 7 h interval given 4 of 6 weeks has been evaluated in a phase I study in 51 solid tumor patients. No DLTs were observed and peak flavopiridol concentrations were >2 mmol/L at dose of 50 mg/m2. Clinical activity was documented with 36% of patients attaining a partial response in a variety of tumors and an addition to those with prolonged stable disease [52]. A phase I study of the combination of flavopiridol with irinotecan and cisplatin is ongoing [53]. Similarly, a phase I trial is ongoing evaluating the combination of flavopiridol with gemcitabine or taxanes. The evaluation of flavopiridol demonstrates that the ability of preclinical models to direct strategy for rational combinations of small molecules and traditional cytotoxic agents. Targeting mitosis with small molecule inhibitors has recently become a practical option with the recent understanding of the function of Aurora kinase and polo-like kinase-1 (PLK-1) (Table  15.4). The proper segregation of DNA and activation of the anaphase-promoting complex (APC) requires the binding of the kinetochores to tubulin fibers. Aurora kinase B and survivin are integral components of the chromosome passenger complex (CPC) found within the kinetochore [54]. Aurora kinase B activation via phosphorylation allows the appropriate binding of tubulin to the kinetochore [55]. Aurora kinase B inhibition is associated with impaired chromosomal alignment, leading to abnormal chromosomal segregation, polyploidy, and eventually apoptosis. Several inhibitors of Aurora kinases are currently undergoing clinical investigation [56]. Thus far, these agents appear to be well tolerated with bone marrow toxicity being the principal side effect. Many of these compounds also inhibit VEGFR2, thus hypertension does occur. The polo-like kinase (PLK) protein family consists of at least four members in mammalian cells [57]. These proteins are characterized by a conserved c-terminal POLO box domain (PBD), in addition to a serine-threonine kinase domain. PLK-1 is overexpressed in a wide variety of cancers. PLK-1 appears to be pivotal for mitotic and meiotic division by facilitating centrosome maturation, mitotic spindle assembly, and APC activation [58]. PLK-1 activates CDK-1 early in mitosis to mediate the transition from G2 to mitosis. In addition to its well-described role in mitosis, PLK-1 has recently been demonstrated to promote cytokinesis by linking the spindle midzone to RhoA as well as to modulate the DNA damage checkpoint [59]. DNA damage results in the activation of checkpoint kinases ATM and ATR. These kinases subsequently prevent activation of effector kinases like Chk1, which ultimately prevents CDK-1 activation and the promotion of mitosis. In contrast, PLK-1 promotes CDK-1 activation and mitotic entry via its regulation cyclin B.

PHA-739358

CYC116 MK-0457

Aurora A Aurora B Aurora B/C Aurora A/B, JAK2, BCR-ABL T315I Aurora A/B, VEGFR2 Aurora A/B/C, JAK, FLT-3, BCR-ABL T315I Aurora A/B/C, FLT-3, BCR-ABL T315I

Kinesis spindle protein Kinesis spindle protein

Kinase spindle protein SB-743921 MK-0731

Aurora kinase MLN8054 R763 AZD1152 AT9283

N/A N/A

PLK-1 PLK-1

18 35

27

0.3 3

N/A

N/A N/A

N/A N/A

Aurora B IC50 (nM)

0.66

5 N/A N/A 3

N/A N/A

Aurora A IC50 (nM)

Target

Drug Polo-like kinase BI2536 ON01910

Table 15.4  Mitosis-specific inhibitors

120

N/A 4.6

N/A N/A 17 N/A

N/A N/A

N/A N/A

Aurora C IC50 (nM)

N/A

N/A N/A

N/A N/A N/A N/A

N/A N/A

0.8 10

PLK-1 IC50 (nM)

Phase II

Phase I Phase II

Phase I Phase I Phase II Phase I

Phase I Phase I

Phase I/II Phase IB

Phase of development

414 C.D. Weekes and M. Hidalgo

15  Targeted Therapeutics in Cancer Treatment

415

In addition, PLK-1 facilitates ubiquitin degradation of claspin, an adaptor protein necessary for Chk1 expression and G2 arrest. In turn PLK-1 function is controlled by the DNA damage checkpoint by two mechanisms. First, PLK-1 phosphorylation at T210 is inhibited in response to DNA damage. Secondly, DNA damage leads to the degradation of PLK-1 by the E3 ubiquitin ligase anaphase-promoting complex/ cyclosome (APC/C). Currently, there are two PLK-1-inhibiting compounds in clinical investigation. BI-2536 (Boehringher Ingelheim) is an ATP-competitive dihydropteridinone derivative. This compound inhibits PLK-1 enzymatic activity at an IC50 of 0.8  nM. BI-2536 is a highly selective PLK-1 inhibitor demonstrating a greater than 1,000-fold selectivity in comparison to a panel of 63 other kinases. In vitro and in vivo experiments of BI-2536 demonstrate its antiproliferative activity heralded by prometaphase arrest and subsequent apoptosis. The results of the phase I single-agent study have recently been reported. The MTD for a single infusion every 3 weeks was 200 mg [60]. Dose-limiting toxicity of grade 3 or 4 neutropenia occurred in 56% of patients. This compound is currently being evaluated in lung cancer, non-Hodgkin’s lymphoma, and acute myelogenous leukemia as single-agent therapy. ON01910 (Onconova Therapeutics, Inc.) is a non-ATP-competitive small molecule inhibitor of PLK-1 that likely interferes with the kinase domain by binding to the peptide-binding site. ON01910 is a selective inhibitor of PLK-1 enzymatic activity with IC50 of 9–10 nM. ON01910 treatment of cells induces mitotic arrest and apoptosis in a variety of tumor cells, with a IC50 ranging from 50 to 200 nM [61]. It also augments the antitumor effects of a variety of cytotoxic agents. ON01910 is now in early-phase clinical development [62]. Single-agent phase I analysis of this compound demonstrated the drug to be well tolerated at an MTD of 3,120  mg. Early activity was seen in a patient with ovarian carcinoma. ON01910 demonstrated growth inhibitory effect in preclinical evaluation of a pancreatic carcinoma model. The observed ON01910-induced growth inhibition was associated with ex vivo suppression of CDC25 phosphorylation and cyclin B1 mRNA and protein expression, thereby defining a potential biomarker of antitumor activity of ON01910 [63]. Currently, ON01910 is being evaluated in the phase I setting as a single agent in patients with refractory anemia with excess blasts. A second phase I study is also being conducted in combination with gemcitabine for solid tumors. Opportunities for combinatorial strategies targeting PLK-1 include conventional DNA-damaging agents, PARP inhibitors, histone deacetylase inhibitors (epigenetics), and heat shock protein 90 (HSP90) inhibitors. Mutations involving the c-terminus of PLK-1, which is the location of the PBD, inhibit the ability of PLK-1 to bind to HSP90. The alteration of PLK-1 protein binding to the chaperone protein HSP90 results in the instability of the PLK-1 protein. Cells harboring these mutations are exquisitely sensitive to the effects of the HSP90 inhibitor geldamycin. Similarly, exposure to hypericin results in loss of PLK-1 function and mitotic cell death. As a result, the combination of PLK-1 and HSP90 inhibitors warrants further investigation.

416

C.D. Weekes and M. Hidalgo

15.4 Signal Transduction and Protein Kinases Protein tyrosine kinases (PTKs) comprise a large fraction of the approximately 40 tumor-suppressor genes and over 100 dominant oncogenes described to date [64]. PTKs are also the largest group of dominant oncogenes with structural homology. PTKs evolved to mediate aspects of multicellular communication and development. Somatic mutations in this very small group of genes cause a significant fraction of human cancers, again emphasizing the inverse relationship between normal developmental regulation and oncogenesis [65]. Currently there are over 90 known PTK genes in the human genome; 58 encode transmembrane receptor PTK distributed into 20 subfamilies, and 32 encode cytoplasmic, nonreceptor PTKs in ten subfamilies [66, 67]. The RTKs have been organized into families based on sequence homology, structural characteristics, and distinct motifs in the extracellular domain. There are currently 20 known families in vertebrates. RTKs share several structural features. They are glycoproteins possessing an extracellular ligand-binding domain, which conveys ligand specificity, and a single hydrophobic transmembrane domain, anchoring the receptor to the membrane. Intracellular sequences typically contain regulatory regions in addition to the catalytic domain. Ligand binding induces activation of the intracellular tyrosine kinase domain leading to the initiation of signaling events specific for the receptor. RTK phosphorylation induces receptor dimerization with conformational changes that result in intermolecular phosphorylation at tyrosine residues at multiple sites. Receptor heterodimerization can also occur, as reported with transforming growth factor alpha interaction with receptor heterodimers comprising HER-2 and EGFR [68]. In malignant tumors, a number of these receptors are overexpressed or mutated, leading to abnormal cell proliferation. Nonreceptor tyrosine kinases are cytoplasmic proteins which transduce extracellular signals to downstream intermediates in pathways that regulate cell growth, activation, and differentiation. Many nonreceptor tyrosine kinases are linked to transmembrane receptors including those for peptide hormones and cytokines. Unlike RTKs, they lack transmembrane domains and ligand binding. They are activated by ligand binding to their associated receptors or events such as cell adhesion, calcium influx, or cell cycle progression. More than 30 members are classified into ten families. The serine-threonine kinases are almost all intracellular and include key mediators of carcinogenesis such as raf, Akt/protein kinase B, and MEK. Effector molecules recruited after ligand-induced phosphorylation of RTK’s include phospholipase C (PLC), phosphoinositide-3-kinase (PI3-kinase, and Ras. Conceptually, there are multiple potential key points of intervention to attack signaling pathways for cancer therapy. The first is the neutralization of ligands prior to their association with the receptors. This approach has been successfully validated with bevacizumab, a humanized monoclonal antibody targeting circulating vascular endothelial growth factor (VEGF) [69]. The second approach to abrogating signaling pathways is the direct inhibition of receptors. This can be achieved by preventing the

15  Targeted Therapeutics in Cancer Treatment

417

binding of growth factors to their receptors, exemplified by the success of cetuximab, a chimeric antibody directed against the epidermal growth factor receptor (EGFR) or by inhibiting the kinase activity of receptors with small molecule inhibitors of receptor phosphorylation as with erlotinib [70, 71]. The final approach relates to the inhibition of signaling by cytoplasmic secondary messengers, most of which are protein kinases. An example of this last approach is imatinib, an inhibitor of the kinase activity of bcr-abl, c-kit, and PDGFR [3, 4]. Bortezomib inhibition of IkB proteosomal degradation provides an excellent example of a modulator of protein degradation as a therapeutic target [72]. In this section, the development of small molecules inhibitors of the RTKs, nonreceptor tyrosine kinases and chaperone protein will be discussed to illustrate strategies for targeting intracellular signal transduction pathways.

15.4.1 Receptor Kinase Inhibition RTKs function as the cancer cell’s link to its microenvironment. As such, RTKs are positioned at the apex of cancer cell signal transduction. Currently, RTKs represent the most commonly utilized target in molecular therapeutics. Strategies for RTK inhibition applied to the EGFR, insulin-like growth factor-1 receptor (IGF-1R) and c-MET will be discussed as these receptors utilize common intracellular signal transduction pathways. Furthermore, cross talk between these receptors possesses clinical ramifications to the development of therapeutic strategies.

15.4.1.1 Epidermal Growth Factor Receptor Family The human epidermal growth factor receptor (HER) family consists of four receptors: HER1, also known as EGFR, HER2 (Erb2 or HER2/neu), HER3 (erb3), HER4. EGFR receptor family members share similar homology with the exception of HER 3, which lacks the kinase domain [73]. The tyrosine kinase domain of EGFR shares 80% homology with that of HER2 and HER4 [74]. Epidermal growth factor (EGF), transforming growth factor a (TGFa), and ampiregulin represent EGFR-exclusive ligands. In contrast, betacellulin and epiregulin bind to both EGFR and HER4. Receptor activation promotes receptor homodimerization and heterodimerization [75, 76]. HER2 is the preferred partner for heterodimerization with EGFR, although it lacks a known ligand [77]. Homodimers of EGFR are unstable whereas heterodimerization with HER2 results in EGFR stabilization and enhanced cell surface expression [78]. EGFR tyrosine kinase activation results in the autophosphorylation of up to five tyrosine residues in its carboxy terminus at positions Y992, Y1068, Y1086, Y1448, and Y1173 [79]. These phosphotyrosines serve as docking sites for adaptor proteins (Fig.  2). EGFR autophosphorylation result in the propagation of intracellular ­signals via at least five pathways: phosphatidylinositol 3-kinase PI3K/Akt,

418

C.D. Weekes and M. Hidalgo Growth Factor Receptor Tyrosine Kinase (EGFR, PDGFR/Kit, IGF-1R, cMet)

Ligand

Ligand Cell Membrane

Src

P

P

Ras

Activation

PTEN

P110

P

P85

PI3K

Adaptor Proteins

P Activation

GTP

Raf

Akt

P

P

MEK P

mTOR 4EBP1 eIF4E S6Kinase

ERK P

Proliferation Survival Angiogenesis Invasion

Nuclear Membrane

Fig. 15.2  Receptor tyrosine kinase intracellular signal cascade

­ itogen-activated protein kinase (MAPK), PLC, signal transducer and activator of m transcription (STAT) and SRC/focal adhesion kinase (FAK) pathways [80, 81]. These pathways exert EGFR-mediated control of cellular proliferation, transformation, metastasis, angiogenesis, and apoptosis [82, 83]. Tumor cells have developed alternative mechanisms for EGFR activation. First gene amplification results in receptor overexpression and ligand-independent receptor dimerization [84]. Secondly, tumor cells produce ligands (such as TGFa) that mediate autocrine EGFR activation [85]. Lastly, germline mutations in the EGFR gene are common. These can be classified into two main groups; tyrosine kinase domain mutation and truncating mutations in exons 2–7. Approximately, 90% of mutations occur in exons 18–21 that code for the tyrosine kinase domain. In-frame deletions of exon 19, codons 746–750 are responsible for upwards of 50% of these cases, while substitution of leucine 858 by arginine accounts for another 35–45%. The remaining mutations are either insertions in exon 20 or rare substitutions occurring in exons 18–21. Tyrosine kinase mutations possess oncogenic potential. The truncated isoforms of EGFR (EGFRvIII) described in glioblastoma are a consequence of deletion mutations in amino acids 6–273 involving the extracellular domain and are constitutively activated [86].

Anti-EGFR Strategies The primary strategies employed for the targeting of EGFR have been to prevent ligand–receptor binding by either monoclonal antibodies (MAb) directed against the extracellular domain of the receptor or the use of TKIs) (Table 15.5).

15  Targeted Therapeutics in Cancer Treatment

419

Table 15.5  HER receptor family-targeted therapy Monoclonal Antibodies Drugs

Antibody construct

Isotype Target

Developmental phase

Cetuximab

Chimeric

IgG1

EGFR

Panitumumab Trastuzumab

Human Humanized

IgG2 IgG1

EGFR HER2

FDA approved (colorectal and head and neck) FDA approved (colorectal) FDA approved (HER2amplified breast)

Tyrosine Kinase Inhibitors

Drugs

Mechanism of action

Target

EGFR HER-2 IC50 IC50 Developmental (nM) (nM) phase

Gefitinib

Reversible, ATP competitive Reversible, ATP competitive

EGFR

  3.1

EGFR

  0.56 512

EGFR, HER2

10.2

9.8

EGFR, HER2 EGFR, HER2

36   0.5

43 14

Approved – NSCLC Approved – NSCLC Approved – breast Phase I/II Phase II

EGFR, HER2

92

59

Phase I/II

EGFR, HER2, HER4 EGFR, HER2, HER3

22

32

Phase I Phase I/II

EGFR, HER2, HER4

 6

45

Phase I/II

Erlotinib Lapatinib ARRY-33543 BIBW-2992 HKI-272 BMS-599626 CI-1033 PF-00299804

ATP competitive Reversible, ATP competitive Irreversible, ATP mimetic Irreversible, ATP mimetic Irreversible, ATP mimetic

BMS-690514 XL-647

Reversible, ATP competitive

EGFR, HER2, HER4, VEGFR2 EGFR, HER2, HER4, VEGFR2, EPHB4

343

Phase I/II   0.3

16

Phase I

Three monoclonal antibodies targeting the EGFR family are currently utilized clinically. Both cetuximab (Erbitux) and Panitumumab (Vectabix) target EGFR, while trastuzumab (Herceptin) targets the HER2 receptor. The EGFR antibodies have been investigated in a variety of solid tumors. Cetuximab has demonstrated clinical efficacy in both metastatic colon cancer and squamous cell carcinoma of the head and neck [86]. Irinotecan-refractory colorectal cancer patients treated with cetuximab and irinotecan obtained a superior response rate of 22.9%, which translated into a prolonged time to progression (4.1 vs. 1.5 months, p < 0.001) then those receiving cetuximab monotherapy (11%) in a randomized phase III study [86]. Interestingly, response to cetuximab was not associated with immunohistochemical EGFR tumor cell expression [87]. Subsequently, a series of studies have demonstrated the efficacy of cetuximab in all lines of therapy for metastatic disease and also in combination with oxaliplatin [88, 89]. Current investigation is focused on the utility of combining EGFR therapy with bevacizumab.

420

C.D. Weekes and M. Hidalgo

Cetuximab has obtained FDA approval for treatment of both locally advanced and metastatic head and neck cancer. Treatment with the combination of cetuximab and external beam radiation therapy resulted in a significant improvement in time of local control of metastasis, progression-free survival, and overall survival (49.0 months vs. 29.3 months, HR 0.74, p = 0.03) in comparison to radiation therapy alone in a randomized phase III study [90]. Subsequently, the addition of cetuximab to 5-FU and platinum-based chemotherapy as first-line treatment for patients with metastatic head and neck cancer demonstrated superior median overall survival of 10.1 months in comparison to 7.4 months for chemotherapy alone (HR 0.80, p = 0.04) [91]. In contrast, panitumumab has currently demonstrated clinical benefit for colon cancer patients whose disease is refractory to all FDA-approved cytotoxic agents [92]. Single-agent trastuzumab was initially demonstrated to improve overall survival in HER2/Neu-positive breast cancer patients with metastatic disease, leading to its FDA approval in 1999 [2]. This study validated the use of pharmacodiagnostics to develop a clinical investigation strategy to demonstrate clinical efficacy of a mole­ cular agent. Current investigation is focused on the adjuvant therapy setting in which initial evidence of single-agent clinical benefit has been obtained [93, 94]. A similar pharmacodiagnostic approach is not being utilized for the clinical development of trastuzumab in HER2/Neu-positive gastric cancer patients [95]. Combined, these data demonstrate that monoclonal antibody therapy targeting the EGFR family is a viable strategy. The tyrosine kinase inhibitors in clinical investigation include gefitinib, erlotinib, lapatinib (GW572016), and HKI-272 (Table 4). Both gefitinib and erlotinib selectively and reversibly target EGFR, preventing EGFR autophosphorylation thereby blocking the signal cascade. Preclinical in  vivo testing demonstrated the antitumor effects of both agents in a variety of cancer models [96–98]. Phase I clinical investigation of both agents showed favorable toxicities primarily consisting of skin rash and diarrhea. DLTs associated with gefitinib dosing occurred at doses far beyond those required for antitumor effect [99–102]. Phase I studies combining gefitinib with cytotoxic agents were completed with a gefitinib dose of 250 or 500 mg combined with full-dose cytotoxic agents without incremental increase in toxicity. Gefitinb obtained early FDA approval based upon the principle of optimal biologic dose. Phase II evaluation of gefitinib at doses of either 250 or 500 mg in chemotherapy-resistant NSCLC patients demonstrated an improved response rate of 18.7% compared to 10.6% associated with fewer toxicities in patients treated with 250  mg [103, 104]. Unfortunately, phase III testing of gefitinib when combined with doublet chemotherapy or as a single agent compared with best supportive care failed to demonstrate a survival benefit [105, 106]. Recently, two noninferiority phase III studies in patients with advanced NSCLC have been completed in which gefitinib demonstrated similar median overall survival as either docetaxel or carboplatin and paclitaxel [107, 108]. In contrast to gefitinib, erlotinib was developed at the maximum-tolerated dose. Erlotinib is the first EGFR-targeted therapy to show significant survival benefit compared to placebo control, as demonstrated by a 6.7-month median survival compared to 4.7 months in patients with NSCLC in clinical progression or with chemotherapy refractory disease [109].

15  Targeted Therapeutics in Cancer Treatment

421

Additionally, erlotinib has received approval for metastatic pancreatic cancer therapy when combined with gemcitabine due to a modest improvement in overall survival [110]. These data would suggest that clinical development of targeted therapy at MTD may ultimately improve the ability to demonstrate clinical benefit in comparison to OBD strategies. Lapatinib represents the initial dual HER family tyrosine kinase inhibitor to attain FDA approval. Lapatinib targets EGFR and reversibly inhibits the phosphorylation of the ErbB2 [111, 112]. It has a typical TKI toxicity profile with diarrhea and rash [110]. Lapatinib has now been approved for use with capecitabine in trastumumab refractory breast cancer [113]. More recently, the dual EGFR–ErbB2 receptor inhibitor HKI-272 has shown to also inhibit the activation of the EGFR with the T790M mutations, which are known to induce erlotinib resistance. Other dual EGFR–ErbB2 TKIs are undergoing clinical investigation (Table 4). Other strategies such as EGFRtargeted strategies include dual targeting of EGFR with the simultaneous administration of monoclonal antibodies and TKI, resulting abrogation of both ligand-induced receptor activation and intracellular signal propagation. In vivo testing of this strategy demonstrated a synergistic antitumor effect of dual EGFR inhibition over either single agent [114]. This strategy is now under clinical investigation. Ultimately, the clinical investigation of EGFR receptor family inhibitors has been a successful endeavor. EGFR represents an integral receptor in oncogenesis. In toto, this avenue of investigation has demonstrated the role of primary and secondary receptor mutations as determinant of the clinical efficacy of TKIs. In addition, it has demonstrated the importance of cross talk between receptors as a result of overlapping signal transduction pathways as a mechanism of resistance to TKIs. Furthermore, cancer cells can recruit receptors that utilize similar communication pathways to overcome target receptor inactivation, such as that observed with c-MET phosphorylation upon NSCLC exposure to gefitinib. In addition, mutations of intracellular signaling proteins may overcome antibody-based external receptor inactivation to promote resistance to therapy. These factors amongst others are now being incorporated into the design of future early-phase clinical trials. 15.4.1.2 Insulin-Like Growth Factor-I Receptor Similar to EGFR, the Insulin-like growth factor-I receptor (IGF-1R) signal transduction pathway is comprised of multiple circulating ligands, such as IGF-I, IGF-II, and insulin, interacting with multiple membrane-bound receptors, such as IGF-1R, IGF-2R, insulin receptor-A (IR-A), insulin receptor-B (IR-B) to facilitate hormonal regulation of cancer cells. Additionally, circulating insulin-binding proteins (IGF-BPs) serve as extracellular docking proteins for ligands thus providing an additional level of pathway regulation. IGF-1R exists as a heterotetramer comprised of two extracellular ligand-binding a subunits and two b subunits with transmembrane and tyrosine kinase domains. Ligand binding to IGF-1R results in a conformational change and phosphorylation and the recruitment of adaptor ­proteins in the form of insulin-receptor substrates (IRS) and/or Src homology 2

422

C.D. Weekes and M. Hidalgo

domain-containing (SHC) proteins. As with EGFR, IGF-1R activation mediates oncogenesis by mitogenic, angiogenic, metastatic, and survival signals transmitted through the MAPK and PI3k/mTOR pathways [115]. Aberrant regulation of the IGF-1R in cancer cells occurs by overexpression of either the ligands (IGF-I or IGF-II) or the receptor. This can occur by gene amplification, loss of imprinting or overexpression of transcription factors. Ligand concentration can also be altered by the IGF-BP [116, 117]. Lastly, loss of IGF-2R, which functions as a decoy receptor, could promote IGF-1R cellular growth [118]. As with the EGFR pathway, several monoclonal antibodies and small molecule inhibitors of IGF-1R are currently in clinical development (Table 15.6). As a class, the monoclonal antibodies are the furthest in clinical development. CP-721,871 (Prizer) is a fully humanized IgG2 MoAb antagonist of IGF-1R. Phase I investigation Table 15.6  IGF1-R inhibitors Monoclonal antibodies Monoclonal Drug antibody class CP-751871 Human IgG2

Developmental phase Phase II/III

IMC-A12

Human IgG1

Phase II

R1507 AMG-479

Human IgG1 Human IgG2

Phase I/II Phase II

SCH-717454

mAb

Phase IB/II

MK-0646 BIIB022

Humanized Human IgG4

Phase II Phase I

Drug

TKI Class

Developmental phase

OSI-906

Reversible ATP competitive

Disease Breast, colorectal, NSCLC, prostate and Ewing sarcoma: NSCLC Breast, colorectal, head and neck, HCC, pancreatic, prostate and sarcoma Sarcoma Breast, lymphoma, ovarian, pancreatic and sarcoma Colorectal and pediatric cancers Colorectal

Tyrosine Kinase Inhibitors

XL-228 NVP-AEW541 NVP-ADW742 AG-1024 BMS-536924 BMS-554417 Nordihydroguareacetic acid (NDGA)

Reversible ATP competitive Reversible ATP competitive Non-ATP competitive Reversible ATP competitive

Disease

Phase I Phase I Phase I Phase I Phase I Phase I Phase I Phase I

CLL and ALL

15  Targeted Therapeutics in Cancer Treatment

423

of this agent identified 20 mg/kg as the maximally feasible dose without achieving MTD. The most common adverse events included hyperglycemia, elevation of liver transaminases, hyperuremia, anorexia, nausea, and fatigue [119]. Pharmacodynamic analysis identified increased serum levels of insulin and human growth hormone and transient decrease in IGF-1R-expression on circulating tumor cells (CTCs) with CP-721,871 administration [120]. It is now undergoing phase II and III evaluation in a variety of tumor types, including Ewing’s sarcoma and advanced NSCLC. The phase III study in NSCLC patients is designed to confirm the phase II data in which patients with untreated advanced NSCLC who received CP-721,871 in addition to carboplatin and paclitaxel (178 patients) achieved a superior response rate of 51% vs. 36%; p < 0.01 and progression-free survival compared to those treated with carboplatin and paclitaxel (143 patients) [121]. Other monoclonal antibodies targeting IGF-1R include AMG-479 (Amgen, Inc.), IMC-A12 (Imclone, Inc.), R1507 (Roche Pharmaceuticals), MK-606 (Merck), SCH-717454 (ScheringPlough), AVE-1642 (Sanofi-Aventis) and BIIB022 (Biogen Idec). All compounds have undergone initial phase I testing. As a class, early signs of activity have been observed in sarcoma patients and, in particular, patients with Ewing’s sarcoma. A complete response (AMG479) and two partial responses (R1057) were observed during phase I testing. Subsequent analysis has demonstrated cross talk between the IGF-1R pathway and EWS-FLI1 fusion protein in Ewing’s sarcomas. As such, sarcoma is disease under investigation in all compounds currently in phase II development. In addition, this finding demonstrates the ability of phase I testing to provide early signals of clinical activity as manifested by response or prolonged stable disease in rare tumor histologies, thus providing the opportunity for biology-driven clinical development of molecular compounds in unique niches. Small molecule inhibitors of the IGF-1R are also in early phases of clinical investigation. These include OSI-906 (OSI Oncology), XL-228 (Exelixis), NVP-AEW541 (Novartis), NVP-ADW742 (Novartis), AG-1024 (Merck), BMS-536924 (BristolMyers Squibb), BMS-554417 (Bristol-Myers Squibb), and Nordihydroguareacetic acid. As would be expected, hyperglycemia has been a common adverse event in phase I testing occurring in approximately 20% of patients. Hyperglycemia has been moderate at best and is reversible when managed with oral hypoglycemic agents. Initial combination therapy trials with other receptor tyrosine TKIs and nonreceptor kinase inhibitors and IGF-1R inhibitors are commencing. 15.4.1.3 Hepatocyte Growth Factor and c-MET The MET gene encodes a high-affinity receptor comprised of disulfide-linked a and b subunits for hepatocyte growth factor (HGF), also known as scatter factor [122]. HGF is a 90 kD multidomain glycoprotein that function as a heterodimer of a and b subunits [123]. The HGF a subunit contains the high-affinity c-MET-binding domain, while the b subunit is necessary for c-MET activation. HGF binding to c-MET induces receptor homodimerization and autophosphorylation of tyrosine residues Y1234 and Y1235. Receptor activation further results in the phosphorylation

424

C.D. Weekes and M. Hidalgo

of tyrosines Y1349 and Y1356 which can then serve as docking sites for intracellular adaptor proteins and propagation of secondary signals [124, 125]. c-MET activation results in the activation of a complex physiologic process termed “invasive growth,” comprised of cellular proliferation, invasion, and angiogenesis [123]. The aberrant activation of c-MET utilizes these properties of invasive growth to facilitate oncogenic transformation by promoting cellular proliferation, invasion, metastasis, angiogenesis, and inhibiting apoptosis. Similar to previously discussed RTKs, the MAPK and PI3K pathways transduce signals upon c-MET activation. MET is expressed in a number of epithelial and hematologic malignancies. In contrast, HGF is produced by both tumor cells and adjacent mesenchymal cells and thus can function both in an autocrine and paracrine manner. Activation of HGF/MET pathway in tumor cells occurs by overexpression of HGF and MET in addition to MET gene amplification and the acquisition of activating mutations [126]. The formation of heterodimers with co-receptors such as sematophorin results in c-MET activation and adds an additional layer complexity to c-MET regulation. HGF/c-MET Pathway Inhibition Strategies Initial strategies to inhibit the HGF/c-MET signal transduction pathway focused on preventing c-MET ligand binding. Antibodies targeting both HGF and c-MET have been developed and are in early phases of clinical investigation (Table 15.7). AMG102 (Amgen, Inc) is a fully human IgG2 antibody that selectively binds and neutralizes HGF [127]. Initial phase I testing defined 15 mg/kg as the MTD for AMG102. AMG102 is now undergoing phase II testing as a single agent in glioblastoma multiforme and renal cell carcinoma (RCC) and in combination with chemotherapy in patients with metastatic gastric cancer. The latter study is designed to test the hypothesis that HGF inhibition will be effective in patients with c-MET overexpression. A single armed c-MET antibody, OA5D5 (MetMAb, Genentech, Inc.) is currently in phase I clinical investigation. Initial analysis demonstrates OA5D5 safety to 30 mg/kg. A number of c-MET tyrosine kinase inhibitors are currently in early phases of clinical investigation. Four of these are highly selective for c-MET and include ARQ197 (ArQule), JNJ-38877605 (Johnson and Johnson), PF-04217903 (Pfizer) and SGX523 (SGX Pharmaceuticals). ARQ197 is a non-ATP-competitive inhibitor of the c-Met receptor. It potently inhibits both HGF-mediated and constitutively active c-Met phosphorylation [128]. ARQ197 is currently in phase I clinical investigation with only DLT of grade 3 fatigue being reported in a single patient [129]. The other compounds are all ATP-competitive inhibitors of c-MET catalytic activity and are in early phases of clinical investigation. The phase I studies of JNJ-38877605 and PF-04217903 were recently initiated and no clinical data is currently available. SGX523 demonstrated unexpected toxicities early in phase I investigation including renal insufficiency and has been discontinued for further investigation. The remainder of c-Met tyrosine kinase inhibitors under investigation are broadspectrum multitargeted kinase inhibitors. PF-02341066 is a multitargeted tyrosine kinase inhibitor with activity against c-Met and the anaplastic lymphoma kinase

15  Targeted Therapeutics in Cancer Treatment Table 15.7  HGF/c-Met receptor inhibitors Drug Class Monoclonal antibodies AMG-102 Antibody OA-5D5 – MetMAb Monovalent antibody Tyrosine kinase inhibitors PF-02417903 Selective kinase inhibitor, ATP competitive JNJ-38877605 Selective kinase inhibitor, ATP competitive SGX-523 Selective kinase inhibitor, ATP competitive PHA-665752 Selective kinase inhibitor, ATP competitive ARQ-197 Selective kinase inhibitor, non-ATP competitive PF-02341066 Multikinase inhibitor, ATP competitive XL-880 Broad-spectrum kinase inhibitor, ATP competitive XL-184 Broad-spectrum kinase inhibitor, ATP competitive MGCD-265 Broad-spectrum kinase inhibitor, ATP competitive MK-2461 Broad-spectrum kinase inhibitor MP-470 Broad-spectrum kinase inhibitor, ATP competitive

425

Target

Development phase

HGF c-Met

Phase I/II Phase I

c-Met

Phase I/II

c-Met

Phase I

c-Met

Phase I – discontinued

c-Met

Preclinical

c-Met

Phase I

c-Met, ALK

Phase I

c-Met, KDR

Phase I/II

c-Met, KDR, RET

Phase I/II

c-Met, Ron, Tie-2, VEGFR 1/2/3

Phase I

c-Met, KDR, FGFR 1/2/3, FLT 1/3/4 c-Met, c-Kit, FLT3, PDGFRa

Phase I/II Phase I

(ALK). PF-02341066 is in late phase of single-agent phase I clinical investigation. It has demonstrated initial early signs of clinical activity in patients harboring ALK translocations. GSK 1363089/XL880 (Exelixis) demonstrates high-binding affinities to both c-Met and VEGFR2. Observed toxicities in phase I investigations are consistent with classic VEGFR2 inhibition with hypertension being seen in 27% of patients. Interestingly, prolonged stable disease has been demonstrated in the phase I study. XL184 (Exelixis) is an orally bioavailable inhibitor of both c-Met and VEGFR2 that can resensitize cells to EGFR-targeted TKIs in preclinical models. XL184 is currently undergoing single-agent investigation, demonstrating early activity in patients with medullary thyroid cancer. MP470 (SuperGen) is another

426

C.D. Weekes and M. Hidalgo

orally bioavailable c-Met inhibitor, possessing additional kinase activity against c-Kit, PDGFRa, and Flt-3. It has demonstrated preclinical synergy with platinumbased chemotherapy agents and radiation therapy. Lastly, MK-2461 (Merck) is a strong inhibitor of c-Met, KDR, FGFR 1/2/3, and Flt 1/3/4. 15.4.1.4 Determinants of Response to RTK Therapy Molecular determinants of TKI-based therapy may be different than that of monoclonal antibodies. EGFR mutations appear to determine the clinical efficacy of tyrosine kinase inhibitors. EGFR tyrosine kinase-activating mutations resulting from in-frame deletions within exon 19 and point mutations in codon 858 (exon 21) have been associated with response to either gefitinib- or erlotinib-induced tyrosine kinase inhibition [130, 131]. Furthermore, these mutations preferentially segregate to adenocarcinomas in NSCLC patients with minimal smoking exposure as compared to smoker [71, 132]. EGFR overexpression has been associated with advanced disease states, chemotherapy resistance and poor prognosis. Recently, the interaction of EGFR and Kras was analyzed in a series of NSCLC patients treated with TKIs. EGFR mutation status was associated with sensitivity to EGFR TKIs irrespective of Kras mutation status and predicted clinical outcome [133]. This is an area of ongoing investigation. Similarly, mutations in c-Met receptor appear to impact efficacy of c-MET TKIs in preclinical models. Tumors with wild-type MET receptor expression are resistant to MET inhibitor PF-02341006, while cells harboring ATP-binding loop (V1094I and H1094R) and P-loop (M1250T) mutations are relatively sensitive to PF-02341066. Importantly, MK-2461 inhibits c-Met phosphorylation independent of receptor mutation status [134]. In contrast to TKIs, the acquisition of Kras-activating mutations appears to be the predominant determinant of anti-EGFR monoclonal antibodies clinical efficacy. This phenomenon has primarily been demonstrated in colorectal cancer patients. Kras is the predominate secondary signal transduction molecule involved in the transmission of EGFR signal transduction. Mutations occurring in codons 12, 13, and 61 account for 90% of mutations in the Kras gene and result in a constitutively active protein able to transduce putative EGFR-mediated signals independent of receptor function [130, 131]. Initial evidence of the predictive capacity of Kras mutation status to determine sensitivity to anti-EGFR monoclonal antibodies was obtained in a series of retrospective analyses of single-arm clinical trials. In these studies, patients with wild-type Kras had response rates ranging from 10 to 40%, whereas those with mutant Kras had substantially reduced response [135]. These data have now been confirmed with subset retrospective analyses of large phase III clinical trials in which EGFR antibodies have been combined with chemotherapy. Analyses of response rate and progression-free survival in terms of Kras mutation supports the hypothesis that Kras mutation is associated with lack of response and predicts decreased survival. The other ­important determinant of response to ­antibody-based­ EGFR-targeted therapy is the presence of FcgRIIIa mutations [13].

15  Targeted Therapeutics in Cancer Treatment

427

RTK crosstalk due to the use of common intracellular signal transduction pathways is another important determinant to the efficacy of RTK-targeted therapy [136, 137]. This has led to a variety of clinical studies in which RTK inhibitors are being combined with either mammalian target of rapamycin (mTOR) or MEK pathway-targeted compounds. This strategy would potentially inhibit compensatory pathway activation occurring through a secondary RTK upon effective inhibition of a primary RTK. An example is the relationship between EGFR and c-MET receptors. EGFR activation has been demonstrated to promote c-MET phosphorylation. Conversely, HGF-driven c-MET activation augments EGFR phosphorylation. Recently, MET amplification was observed in a gefitinib-resistant NSCLC line, resulting in MET- and EGFRindependent activation of ERBB3/PI3K/Akt signaling [138, 139]. Inhibition of both receptors was necessary for growth inhibition of these cells. Additionally, combination RTK clinical trials are also ongoing. Another strategy would be the development of clinical studies in which RTK inhibitors are combined with inhibitors of effectors of RTK pathway function. Enhanced incorporation systems, biology techniques, and bioinformatics will be necessary to develop clinically relevant biomarkers of clinical efficacy to fully appreciate the therapeutic benefit of this class of compounds.

15.4.2 Nonreceptor Kinase Inhibition Nonreceptor kinases are integral intracellular members of the signal transduction cascade. These proteins are responsible for the propagation of the signaling cascades from the RTKs to the nucleus. Both tyrosine and serine/threonine nonreceptor kinases can propagate signals arising from RTKs. The PI3Kinase/Akt/mTOR and MAPK pathways represent two key families of nonreceptor kinases that facilitate oncogenesis. Although other nonreceptor kinases promote oncogenesis, the PI3Kinase/Akt/mTOR and MAPK pathways represent the most commonly targeted nonreceptor kinase being developed for cancer therapy. The following section will discuss the development of compounds targeting these pathways. 15.4.2.1 PI3Kinase/Akt/Mammalian Target of Rapamycin Pathway Phosphatidylinositol 3-kinase (PI3K) represents the apical kinase in this signal transduction pathway. PI3K activation occurs due to ligand-dependent activation of RTKs, G-protein-coupled receptors, or integrins. Receptor-independent activation occurs resultant from RTK autophosphorylation and constitutively active Ras proteins. Activated PI3K catalyzes the conversion of phosphatidylinositol (4, 5)-­biphosphate (PIP2) to phosphatidylinositol-3, 4, 5-trisphospahte (PIP3). Phosphatase and tensin homolog deleted from chromosome 10 (PTEN) is a phosphatase that functions as a negative regulator of PI3K. Genetic mutation, LOH, methylation, aberrant expression of microRNA, and protein instability all lead to inhibition of PTEN gene or protein expression in a variety of tumor histologies. Acquisition of these abnormalities within

428

C.D. Weekes and M. Hidalgo

a cell can lead to aberrant PI3K/Akt/mTOR pathway activation and malignant transformation [140]. The importance of this pathway in cellular biology and malignant transformation is exemplified by the number of cancer-related syndromes that result from genomic inactivation of this pathway and include: Cowden’s syndrome (PTEN mutation), Peutz–Jegher’s syndrome (LKB1 mutation), tuberous sclerosis (TSC 1/2 mutation) and neurofibromatosis (NF1 mutation) [141–144]. The best characterized phosphorylation target of PI3K is the plecstrin homology domain of Akt. This phosphorylation stimulates the catalytic activity of Akt, resulting in the phosphorylation of a host of other proteins that affect cell growth, cell cycle entry, and cell survival. The precise mechanism of mTOR regulation by PI3K or Akt, however, is still not well understood. The tuberous sclerosis (TSC) complex functions as a modulator between PI3K/Akt and mTOR. Rheb (Ras homolog enriched in brain) is a small G protein that promotes mTOR signaling in an active GTP-bound conformation, whereas the tuberin–hamartin heterodimer inhibits Rheb by converting it to an inactive GDP-bound state. Upon activation, mTOR forms a scaffold complex with other proteins to form the mTORC1 and mTORC2 complexes. mTOR complex 1 (mTORC1) is comprised of mTOR, raptor (regulatory-associated protein of mTOR) and mLST8. Raptor, a 150 kDa evolutionarily conserved protein, may act as a scaffold protein linking mTOR to p70s6k and 4E-BP1. mTORC1 is specifically inhibited by the binding of rapamycin to FKBP12. mTOR complex 2 (mTORC2) is comprised of mTOR, LST8, rapamycin-independent companion of mTOR (rictor) and mSin1 (also known as MAPK-associated protein 1). Formation of the rapamycin–FKBP12 complex does not directly inhibit mTORC2 function but has been demonstrated to perturb mTORC2 complex formation in approximately 20% of cancer cell lines, resulting in inhibition of Akt signaling [140]. mTOR is a serine/threonine kinase which belongs to the family of phosphatidylinositol kinase-like kinases involved in the regulation of a wide range of growth-related cellular functions, including transcription, translation, membrane trafficking protein degradation, and reorganization of the actin cytoskeleton. mTOR functions as a nutrient and growth factor sensor, controlling cellular growth and proliferation. The principal downstream effects of mTOR is the control of cellular translation machinery, through two separate downstream pathways: the eukaryotic initiation factor 4E binding protein-1 (4E-BP1), The increase in translation of a subset of mRNAs brings about protein products that are required to traverse the G1/S checkpoint of the cell cycle. 15.4.2.2 mTOR-Targeting Agents Rapamycin is the prototypical first-generation mTOR inhibitor. These compounds form a complex with FK506-binding protein 12 (FKBP12) [145]. The resultant FKBP12–RAP complex abrogates mTORC1 function [146, 147]. First-generation mTOR inhibitors in clinical development include rapamycin and the structurally related compounds (rapalogs) Temsirolimus (CCI-779), Everolimus (RAD001), and Deferolimus (AP23573 or MK-8669) (Table  15.8). In addition to their well-­ characterized inhibitory effects on p70s6k and 4E-BP1, these compounds result in G1

ATP competitive

Akt Akt Akt Akt Akt Akt/JNK Akt, PKC

Akt API-2/triciribine PBI-05204 SR13668 GSK2110133 GSK2141795 KRX-0401 GSK690693

Tricyclic nucleoside

PI3K PI3K PI3K PI3K

PI3K BGT226 CAL101 PX-866 XL147

ATP competitive

mTOR-PI3K mTOR-PI3K mTOR-PI3K mTOR-PI3K

Phase I/II Phase I Phase I Phase I Phase I Phase I/II Phase I

Phase I/II Phase I/II Phase I Phase I

Phase I/II Phase I/II Phase I/II Phase I/II

mTOR mTOR mTOR

Everolimus Deferolimus AZD8055

mTOR-PI3K NVP-BEZ235 PF04691502 SF1126 XL765

Approved – RCC Phase III Approved – RCC Phase II Phase I/II

mTOR

Rapamycin derivative FKBP12 Rapamycin derivative FKBP12 mTORC-1:2 inhibitor

Development phase

Target

Table 15.8  PI3K/Akt/mTOR inhibitors Drug Mechanism of action mTOR Temsirolimus Rapamycin derivative FKBP12

Mantle cell lymphoma Lung, colon, breast, and gastric Bone sarcoma, leiomyosarcoma

Tumor

15  Targeted Therapeutics in Cancer Treatment 429

430

C.D. Weekes and M. Hidalgo

growth arrest, the induction of apoptosis in p53 null cells, autophagy and inhibit angiogenesis in  vitro [147–152]. Consistent with these effects, xenograft experiments demonstrate growth inhibition and not tumor regression [153]. All of the rapamycin analogs have undergone initial phase I clinical evaluation [153–158]. Despite the ubiquitous importance of the mTOR pathway in basic cellular function, these compounds were relatively well tolerated. Common toxicities among these compounds include hypercholesterolemia, hypertriglyceridemia, elevated transaminases, hyperglycemia, mucositis, nausea, diarrhea, rash, asthenia, and rarely pneumonitis. All compounds demonstrated predictable pharmacologic behavior and preliminary antitumor effects. A major partial response was seen in a temsirolimus-treated patient but the predominant effect among the compounds was prolonged disease stabilization [157, 158]. One patient with RCC obtained disease stabilization lasting beyond 14 months. The consistent observation of tumor response with mTOR inhibitors at nontoxic doses suggested that the optimal therapeutic dose may be lower than the MTD. As a result, doses of 25 mg intravenously weekly and oral administration of either 10 mg/day or 50 mg/week were recommended for phase II testing of temsirolimus or everolimus, respectively [153, 154]. Phase I testing of deferolimus defined weekly administration of 75  mg intravenously as the MTD [155, 156]. Initial pharmacodynamic analysis demonstrated inconsistent suppression of mTOR pathway activation in analyzed tissue specimens among the three compounds. The mTOR pathway has now been demonstrated to be valid therapeutic target for RCC therapy. Both temsirolimus and everolimus have demonstrated clinical efficacy in this patient population. Patients with untreated advanced RCC receiving single-agent temsirolimus achieved prolonged overall survival of 10.9 months vs. 7.3 and 8.4 months for those treated with interferon alfa or the combination, respectively, in a randomized phase III study [159]. Similarly, treatment of patients with advanced RCC, who failed VEGFR-targeted therapy, with everolimus in a doubleblinded placebo-controlled phase III study demonstrated statistically significant evidence prolonged median progression-free survival of 4 months vs. 1.8 months (HR = 0.3, 95% CI, 0.22–0.4; p < 0.0001) at the second planned interim analysis, leading to early study termination [160]. The single-agent effect of these compounds in RCC patients may be due to the inherent reliance of RCC on the VHL/ HIF-1 axis for oncogenesis, which is intimately regulated in part by mTOR [161]. Rapalogs, however, have shown activity in a phase III study of mantle lymphoma patients resulting in improved ORR of 22% and PFS of 4.8 months in patients receiving temsirolimus vs. investigator’s choice of therapy. These results reflects pathway addiction to cyclin D as the result of the t(11;14) in mantle cell lymphoma [162]. Recently, the interim results of a phase II study demonstrated the activity of temsirolimus in recurrent and metastatic endometrial cancer with 63% of patient attaining a partial response or disease stabilization, which likely reflect the high rate of PTEN deficiency [163]. The clinical evaluation of first-generation mTOR inhibitors in other solid tumors has met with less success. Temsirolimus has been studied in glioblastoma multiforme because of the overexpression of Akt is common in these tumors [164]. Preclinical

15  Targeted Therapeutics in Cancer Treatment

431

data has demonstrated that PTEN-defective tumors or those expressing high levels of Akt are exquisitely sensitive to mTOR-targeted therapy [165]. Unfortunately, this observation was not upheld in the clinical setting; however, high baseline p70s6k expression did correlate with radiographic response. Similar studies have been conducted either as single agent or in combination with cytotoxic chemotherapy with disappointing outcomes. This may be in part due to inappropriate patient selection or invalid pharmacodynamic assessment. The fact that the preclinical observation that either isogenic cell lines or PTEN-deficient tumors demonstrate exquisite sensitivity to rapalogs does not hold true in clinical situation may reflect that pathway activation due to knock-out models is not reflective of the genetics of clinical specimens. PD analysis tends to focus on demonstration of pathway inhibition, which also may have no true relevance to response to therapy. PD markers of downstream effects that are integral in mTOR cellular function may be more reflective of rapalog growth inhibition and need to be further developed. Additionally, current PD strategies have focused on PBMC, which do not demonstrate the same degree of pathway inhibition in comparison to tumor specimens and may not accurately approximate tumor effects of mTOR inhibitor exposure. The development of strategies to implement and improve the current mTOR pharmacodynamic markers is necessary to improve the clinical outcomes associated with these therapies. One potential mechanism of resistance to classic mTOR inhibitors is the compensatory activation of Akt or the MAPK pathways. As a result, second-generation mTOR inhibitors are entering clinical investigation that are dual mTOR/PI3K inhibitors. These compounds demonstrate preclinical activity in a wide variety of tumors possessing activating mutations proteins involved in both the Akt and MAPK pathways. It will be interesting to follow the development of these compounds and observe their ability to impart clinical benefit in tumors possessing any number of mutations resulting in mTOR activation where rapalogs have limited efficacy. These compounds are now entering phase I evaluation and limited data is available regarding toxicity and evidence of clinical antitumor activity and include NVP-BEZ235, PF-04691502, XL765, and SF1126 (Table  15.8). PI3K inhibitors are also in the initial phase of clinical investigation with limited clinical data available currently. This class of compounds includes PX-866, XL147, CAL-101, and BGT-226. Additional upstream activation from Akt represents an additional mode of resistance to mTOR inhibitors. Small molecule inhibitors are being evaluated in the phase I setting and include SR13668, GSK 2110183, GSK 2141795, GSK 690693, Triciribine, KRx-0401, and PBI-05204.

15.5 Mitogen-Activated Protein Kinase Family Numerous critical growth factors and cytokines transduce their signals from the cell membrane to the nucleus via protein kinase networks called signal transduction pathways, which have become major targets for anticancer drugs. An important example is the mitogen-activated protein kinase (MAPK) or extracellular

432

C.D. Weekes and M. Hidalgo

s­ ignal-regulated protein kinase (ERK) pathway. When extracellular growth factors such as the EGF bind to receptors (such as EGFR), conformational changes are induced in the receptor which lead to autophosphorylation, receptor dimerization, and recruitment of proteins such as RAS at the inner surface of the cell membrane. RAS stimulates RAF activation, which in turn phosphorylates MAPK kinase (MEK), which then phosphorylates and activates ERK. At each step in the pathway, phosphorylation of the next signaling member is required for activation and downstream phosphorylation of the next protein kinase. ERK coordinates responses to extracellular signals by regulating gene expression, cytoskeletal rearrangements, and metabolism as well as cell proliferation, differentiation, and apoptosis. This pathway has been shown to be constitutively activated in a number of human cancers. Activation of the pathway causes gene expression changes and changes in cell proliferation, survival, and differentiation.

15.5.1 Compounds in Development 15.5.1.1 RAS Inhibitors Ras has been demonstrated to be an integral component of initiation and maintenance of the malignant phenotype. Activating Ras mutations occur in greater than 20% of cancers [166]. There are three Ras homologues H-Ras, K-Ras, and N-Ras. Activating mutations in K-Ras occur most frequently. It can be found in a variety of cancers to include pancreas (90%), colon (50%), thyroid (50%), lung (30%), and leukemias (5–65%). The most common of these occurs in codons 12 and 13. Ras mutations involving H-Ras and N-Ras are substantially less common. Ras functions at the inner aspect of the cell membrane and requires posttranslational farnesylation for membrane localization. Fanesyltransferase inhibitors (FTIs) have been developed as Ras inhibitors. Compounds in clinical development include Tipifarnib (R115777), Lonafarnib (SCH-66,336), and BMS-214662. Tipifarnib is the most well-characterized compound of this class of oncologic agents (Table 15.9). Preclinical evaluation of tipifarnib demonstrated potent antiproliferative, antiapoptotic, and antiangiogenic effects. Clinical evaluation of the compound has occurred in a variety of tumors to include leukemias, breast, colorectal, non-small-cell lung cancer, pancreatic, and prostate cancer [167]. Unfortunately, tipfarnib failed to demonstrate evidence of clinical efficacy when evaluated in a number of phase II studies and a phase III study in patients with metastatic pancreatic cancer [168]. This may be in part a reflection of inability to inhibit K-Ras relative to H-Ras and N-Ras. In addition, mutant K-Ras may require greater level of FTI to inhibit membrane localization then either wild-type K-Ras or mutant H-Ras [169]. As a result of these disappointing findings, the clinical exploration of MAPK pathway inhibitors is now focused on the development of Raf and MEK inhibitors.

Raf-265 (CHIR-265) LBT613 AZ628

ATP-mimetic inhibitor

ATP-mimetic inhibitor

AAL881

Sorafenib

ATP-competitive inhibitor

XL-281

B-Raf V600E Raf B-Raf V600E Raf C-Raf B-Raf V600E Raf C-Raf B-Raf V600E Raf C-Raf VEGFR2, VEGFR3 Raf, VEGFR2 Raf, VEGFR2 B-Raf V600E Raf C-Raf DDR2, EphA2, EphB2, Lyn, Flt-2, FMS, VEGFR2

FTI FTI FTI

Ras Tipifarnib (R11577) Lonafarnib (SCH 66336) BMS-214662

Raf PLX-4032

17 <1 14 <25 7

MEK MEK MEK MEK MEK MEK MEK

105 34 29

100 31 4.5 6 2.5 940 220 430 22 38 6 3–60

199

IC50 (nM)

Target

Table 15.9  MEK pathway inhibitors Drug Mechanism of action MEK PD184352 (CI-1040) Noncompetitive inhibitor PD032590 Noncompetitive inhibitor ARRY-142886 (AZD6244) Noncompetitive inhibitor RDEA-119 Noncompetitive inhibitor AZD8330 Noncompetitive inhibitor GSK1120212 Noncompetitive inhibitor XL518 Noncompetitive inhibitor

Phase I Phase I Phase I

Approved (HCC:RCC)

Phase I

Phase I

Phase I

Phase II/III Phase II Phase II

Phase I/II Phase I Phase I/II Phase I Phase I Phase I Phase I

Development phase

15  Targeted Therapeutics in Cancer Treatment 433

434

C.D. Weekes and M. Hidalgo

15.5.1.2 Raf Inhibitors The Raf protein is an integral Ras effector protein. It is a member of structurally ­conserved serine-threonine kinases and includes; A-Raf, B-Raf, and C-Raf (Raf-1). The Raf pathway is activated in cancer cells by either upstream events such as growth factor receptor activation or oncogene mutations. Raf itself may contain activating mutations the most common of which is V600E. B-Raf mutations occur in nearly 70% of melanomas and are found in a number of other solid tumors to include colorectal and ovarian carcinomas [170]. Sorafenib (BAY43-9006, Nexavar) represents the first generation of Raf inhibitors in clinical investigation. It is a dual inhibitor of Raf and VEGF receptor (VEGFR). In vitro analysis demonstrates that it potently inhibits Raf-1, B-Raf, VEGR2, VEGFR3, and platelet-derived growth factor receptor (PDGFR). As such, it demonstrates both antiproliferative and antiangiogenic effects preclinical testing against a wide array of tumor histologies. RAF-265 (CHIR-265) is another compound in this class that has dual Raf and VEGFR inhibitory properties. Secondgeneration, highly selective, oral Raf inhibitors are now reaching early-phase clinical investigation and include XL-281 and PLX-4032. However, the most clinical experience with any of the compounds in the class is with that of Sorafenib. Sorafenib is currently FDA approved for the treatment of patients unresectable hepatocellular carcinoma and advanced RCC. The current recommended dose is 400 mg twice daily. Early clinical evaluation demonstrated dose-limiting toxicities of diarrhea, fatigue, and rash [171]. These toxicities were confirmed in phase II and III testing in which these toxicities, in addition to hand-and-foot syndrome, were common findings [172, 173]. Pharmacodynamic analysis of MAPK inactivation in peripheral blood mononuclear cells (PBMCs) demonstrated the loss of MAPK phosphorylation at a sorafenib dose of 200 mg, demonstrating this as a potential useful marker of pathway inhibition. The FDA approval of sorafenib for the treatment of patients with advanced RCC was based upon the results of the initial interim analysis of a phase III study and the findings of a randomized discontinuation phase II study [172, 173]. The phase III trial randomized low- to intermediate-risk patients with advanced RCC to be treated with either Sorafenib 400  mg twice daily (n = 451) vs. placebo (n = 452) in the second-line setting [174]. The median progression-free survival favored the sorafenib arm (5.5 months vs. 2.8 months, HR, 0.44; p < 0.01). The overall survival at the initial interim analysis was superior in the sorafenib cohort in comparison to placebo (HR, 0.72; p = 0.02). These results did not reach planned statistical significance; however, were supported by the results of a phase II randomized discontinuation study of sorafenib in this patient population [172]. This phase II study utilized a placebo-controlled randomized discontinuation design in which 202 patients were initially enrolled to be treated with Sorafenib for a 12-week run-in period. At the end of that period patients achieving ³25% tumor shrinkage were continued on the Sorafenib until evidence of disease progression. Those patients attaining disease stabilization were randomized to either Sorafenib or placebo for an additional 24 weeks, while patients with evidence of disease

15  Targeted Therapeutics in Cancer Treatment

435

progression or intolerable toxicity were taken off study at the end of the 12-week run-in period. This strategy allowed the investigators to separate slow tumor growth from resistance as demonstrated a favorable increase in median ­progression-free survival time from randomization of the sorafenib group relative to the placebo group of 24 weeks and 6 weeks, respectively. Patients treated with placebo were allowed to cross over and receive sorafenib at the time of progression. Combining these results led to the FDA approval of sorafenib for the treatment of patients with advanced RCC. The results of the SHARP trial led to the approval of sorafenib for the treatment of patients with unresectable hepatocellular carcinoma. This was a phase III trial in which a total of 602 patients with Childs–Pugh class A liver dysfunction, ECOG performance status of 2 or less and advanced hepatocellular carcinoma were randomized to receive either sorafenib 400  mg twice daily or placebo [173]. The median survival time for patients receiving sorafenib was 10.7 months in comparison to the placebo group of 7.9 months, p < 0.001. The time to radiographic progression was prolonged in the sorafenib group relative to the placebo group (5.5 months vs. 2.8 months; p < 0.001). These results led to the approval of sorafenib for the treatment of patients with advanced hepatocellular carcinoma and Childs–Pugh A liver dysfunction. Both RCC and hepatocellular carcinoma are highly vascular tumors in which VEGF signal transduction and angiogenesis are integral components of the maintenance of the malignant phenotype. The results of the clinical investigation of sorafenib in addition to the findings of the clinical trials with bevacizumab and sunitinib in RCC call into question the role of the B-Raf modulation by sorafenib in the ultimate clinical efficacy of this compound [175–177]. The clinical contribution of Raf inhibition will be addressed as the second-­generation, Raf-specific inhibitors progress through clinical development (Table 15.9). 15.5.1.3 MEK Inhibitors MEK (MAP/ERK kinase or MAPK kinase) occupies a central role in the MAPK pathway. Expression of constitutively active forms of MEK leads to transformation of cell lines [178, 179]. MEK kinases have dual kinase activity to promote phosphorylation of both serine/threonine and tyrosine residues. MEK1 and MEK2 are the two homologues that share 80% identity, resulting in very similar three-dimensional structures [180, 181]. Both kinases are highly specific, with no known substrates aside from ERK [182]. However, MEK2 is approximately seven times more catalytically active than MEK1, yet MEK2 knock-out mice are fully viable while MEK1 knock-out is embryonically lethal [183–185]. The general interpretation of these findings is that MEK1 is able to compensate for the absence of MEK2 [186]. There are several known MEK inhibitors (Table  15.9). The vast majority of MEK inhibitors in development are highly selective and demonstrate little cross reactivity with other protein kinases. Most of the known MEK inhibitors bind to a unique site adjacent to the ATP-binding pocket, resulting in noncompetitive MEK

436

C.D. Weekes and M. Hidalgo

inhibition [181]. This property is responsible for the selectivity of the MEK inhibitors­ despite MEK containing a highly conserved the ATP-binding site. PD98059 and U0126 represent the initial generation of specific MEK inhibitors [187, 188]. The compounds are not in clinical investigation due to pharmacokinetic properties and lack of in vivo efficacy. There are a number of compounds currently in various phases of clinical development. CI-1040 was an initial MEK inhibitor to undergo clinical evaluation [189]. CI-1040 is orally bioavailable compound demonstrating in  vitro target inhibition and growth inhibition with 1 mmol/L exposure. Xenograft experiments broadened the target histologies and demonstrated correlation of antitumor effects with baseline high p-ERK expression as well as low p-ERK levels in treated tumor tissue. Initial single-agent phase I testing of CI-1040 demonstrated no grade 3 or 4 toxicities [190]. Common toxicities were grade 1–2 diarrhea, fatigue, rash, and nausea/ vomiting. One patient with pancreatic cancer attained a PR lasting 12 months while stable disease was observed in 19 (25%) patients. High-fat foods were found to increase the AUC and Cmax by threefold to fivefold, respectively. Therefore, 800 mg BID with food was the recommended phase II dose. Pharmacodynamic analysis demonstrated a decrease in intratumor p-ERK levels upwards of 100%. Subsequently, a parallel arm phase II study of CI-1040 designed essentially as four simultaneous Simon two-staged phase II studies was performed in a variety of histologic subtypes [191]. However, the early stoppage criteria were not met and the study was closed prior to completion. The observed toxicities and pharmacokinetics were similar to the phase I study. Phosphorylated ERK was elevated in most tumor types and demonstrated a trend with disease stabilization rate (p < 0.055). PD0325901 is a second-generation MEK inhibitor demonstrating a 50-fold increase in potency compared to CI-1040 along with improved bioavailability and longer MED suppression (24 h compared to 6–8 h for CI-1040 [186]. Single-agent phase I study demonstrated the compound to be tolerable up to 20 mg BID. DLTs were acneiform rash, elevated liver enzymes and syncope. Initial evidence of target inhibition was obtained at lower doses. Ten patients achieved either partial response or disease stabilization. This compound has not been developed further. AZD-6244 (ARRY-142886) is another second-generation highly selective MEK inhibitor. Preclinical evaluation of this compound demonstrated that it inhibited ERK phosphorylation and exhibited growth inhibition in cell lines containing both B-Raf and Ras mutations that were associated with in vivo tumor regression. Single-agent phase I evaluation demonstrated DLTs of hypoxia, rash, and diarrhea, while common toxicities included peripheral edema, fatigue, blurred vision, nausea, and altered taste [192]. Extensive pharmacodynamic analysis of MEK inhibition demonstrated a sustained rapid dose-dependent inhibition of ERK phosphorylation in patient PBMCs. Additionally, patients with activating mutations in Ras and Raf demonstrated a prolonged median time on study of 3.5 months compared to 2 months for patients without mutations and were associated with intratumoral pathway inhibition. These observations supported the preclinical observation of the growth inhibitory effects of MEK inhibition in tumors with Ras and B-Raf activating mutations in Ras or B-Raf. AZD-6244 is now undergoing phase II testing in a variety of malignancies at the

15  Targeted Therapeutics in Cancer Treatment

437

recommended phase II dose of 200  mg twice daily to further evaluate the role of MEK inhibition in Ras- and B-Raf-mutated tumors. Additional MEK inhibitors in the initial phases of clinical investigation include RDEA-119, AZD8330 (ARRY424704), GSK 1120212, and XL-518. Defining the molecular determinants of resistance to these compounds is an area of active investigation.

15.6 SRC Kinase Inhibitors Src kinase is a multifunctional intracellular tyrosine kinase which has been implicated in the regulation of a variety of physiological and oncogenic processes such as proliferation, differentiation, survival, motility, and angiogenesis [193]. Src is a member of the Src kinase family including Lck, Fyn, Yes, Yrk, Blk, Fgr, Hck, Lyn, and Frk [194]. Src contains several functional domains; an N-terminal membraneassociation domain (SH4), a variable proline-rich sequence-binding domain (SH3), a phosphotyrosine-binding domain (SH2), a tyrosine kinase domain (SH1), and the Y530-containing C-terminal negative regulatory domain. c-Src is weakly oncogenic as compared with its viral homolog, v-src [195]. It appears that the fundamental mechanism by which src is altered in human tumors is by overexpression of its wild-type form [196]. Indeed, the current understanding of the role of src in cancer is as a facilitator of other signaling molecules rather than by being oncogenic on its own [197]. Src interacts with multiple cellular elements such as membrane receptors, steroid hormone receptors, G protein-regulated pathways, STATs, FAK, and adaptor proteins [197–199]. One of the better characterized functions of src kinase is its ability to activate the EGFR by phosphorylation of its residue Tyr 845. Src also influences receptor endocytosis and degradation [197]. Src’s regulation of FAK modulates focal adhesions to promote cancer metastasis [199]. Given its significant implications in cancer development and progression, Src has become a target for drug development [194]. Different strategies are being used including inhibition of protein–protein interaction, protein stability, and kinase activity. AP22408 inhibits the binding of proteins to the SH2 domain of Src [200]. Protein stability has been targeted as part of the more generalized approach of inhibiting the chaperone Hsp90 [201]. The majority of Src inhibitors in clinical investigation are type II tyrosine kinase inhibitors. As such, these compounds inhibit multiple intracellular tyrosine kinases. Dasatinib (BMS-354825) is the most clinically developed compound in this class. It is a dual src and Abl tyrosine kinase inhibitor. This compound possesses less stringent conformational requirement than imatinib to induce Abl kinase inhibition. As such, dasatinib can restore Abl kinase inhibition in imatinib-resistant cells. Dasatinib has received FDA approval for chronic myelogenous leukemia and Philadelphia chromosome-positive (Ph1+) acute lymphoblastic leukemia (ALL). Dasatinib is currently being investigated in a variety of solid tumors. Bosutinib (SKI-606) is a potent dual src and Abl kinase inhibitor. The agent inhibits src with an IC50 of 1.2 nM in enzyme assays. By virtue of its inhibitory activity against Abl

438

C.D. Weekes and M. Hidalgo

kinase, the drug exerts potent antitumor effects in CML models [202]. AZM475271 is an anilinoquinazoline demonstrating potent in vitro Src kinase inhibition [203]. A recent study in pancreatic cancer models shows that inhibition of Src kinase alone and, particularly, in combination with gemcitabine inhibited the growth and metastatic potential of pancreatic cancer cells [204]. Furthermore, gene signatures have recently been developed that are predictive of both bosutinib and AZD0530 antitumor effects in pancreas xenograft models [205, 206]. The clinical utilization of this strategy is currently being investigated in a phase II study. AZD0530 (AstraZenaca), XL-999 (Exelixis), and XL-228 (Exelixis) are additional multikinase targeting agents under phase I clinical investigation as src inhibitors.

15.7 Apoptosis Apoptosis is the programmed elimination of damaged, redundant, or unnecessary cells that is imperative to maintain cellular homeostasis. It is a complex multistep proteolytic pathway mediated primarily by cysteine proteases known as caspases. The organized activation of the caspases results in DNA fragmentation, chromatin condensation, cell shrinkage, and membrane blebbing. Apoptosis is triggered by irreparable genetic aberrations and cellular stress. Unlike their normal counterpart, cancer cells preferentially survive situations of high genetic aberration and cellular stress. Caspases exist as inactive zymogens. Initiator or “apical” caspases are responsible for the initiation of apoptosis upon appropriate stimulus and are activated by dimerization upon recruitment to oligomeric activation platforms. In contrast, “executioner or effector” caspases are activated by direct proteolytic events ­mediated by initiator caspases. Two distinct molecular pathways result in caspase activation and the initiation of apoptosis: the intrinsic and extrinsic pathways [207]. Intrinsic pathway activation occurs as a result of cytochrome C release from the mitochondria intermembrane space into the cytosol by the proapoptotic BCL-2 family members BAX and BAK [208]. Activation of the intrinsic pathway commonly occurs upon the recognition of cellular stress and gene toxicity by p53 but may also be activated by p53-independent mechanisms. Upon its release, cytochrome C forms the apoptosome by forming a complex with apoptotic proteaseactivating factor-1 (APAF-1) and caspase-9, resulting in caspase-9 activation. Conversely, the extrinsic apoptotic pathway activation is initiated by ligand binding to cell surface death receptors (tumor necrosis factor [TNF] receptor, FAS1, and TNF-related apoptosis inducing ligand [TRAIL] receptor, also known as Apo2). Activation of TNF receptor, Fas, and TRAIL by their respective ligands; TNF, fas-ligand, and TRAIL (Apo2L) result in the recruitment of the adaptor protein FAS-associated death domain (FADD) protein to the receptor. The dead effector domain (DED) of FADD recruits and activates caspase 8 and 10. Activated caspase-9 subsequently activates the effector caspases 3, 6, and 7 ultimately producing a positive feedback loop with the extrinsic pathway through cleavage of caspases 8 and 10 by caspase 6. Ultimately, both pathways merge by proteolytic

15  Targeted Therapeutics in Cancer Treatment

439

activation of the common pathway caspases 3, 6, and 7, due to extrinsic pathway activation of caspases 8 and 10 and the intrinsic pathway caspase-9 activation to execute the cleavage of apotosis target proteins integral for cellular integrity and result in the hallmark findings of apoptosis.

15.7.1 Prosurvival Signal Inhibition The ultimate regulation of the intrinsic pathway occurs by the formation of heterodimers of prosurvival and proapoptotic members of the BCL-2 protein family [209]. In addition to modulation of mitochondrial membrane permeability, BAX and BAK propagate apoptosis by stimulating the release of smac proteins which in turn inactivate inhibitors of apoptosis proteins (IAPs: XIAP, IAP1, IAP2, and survivin). XIAP functions as a direct caspase inhibitor, while other IAPs promote caspase proteosomal degradation by caspase ubiquitination. Cancer cells commonly overexpress prosurvival proteins such as BCL-2. In addition, the inhibitor of apoptosis protein, survivin is preferentially expressed in cancer cells when compared to their normal counterparts. Preclinical downregulation of BCL-2, XIAP, and survivin promotes cancer cell apoptosis. Based upon these findings, the inhibition of prosurvival protein function is an active therapeutic strategy currently under clinical investigation. 15.7.1.1 BCL-2 The predominant strategy employed to date to inhibit prosurvival protein function is the use of ASO. Oblimersen represents the most clinically advanced antisense molecule targeting BCL-2 (Table 15.10). It is a first-generation ASO and is currently in phase III clinical investigation in both solid and hematologic malignancies [210, 211]. Completed phase III studies have failed to demonstrate a survival benefit for the combining oblimersen with chemotherapy. Combination with either dacarbazine or fludaribine and cyclophosphamide did demonstrate improved overall response rates in comparison to chemotherapy alone; however, failed to translate into an overall survival benefit in melanoma and chronic lymphocytic leukemia patients, respectively [210, 211]. The combination of oblimersen with either dexamethasone or carboplatin and etoposide in myeloma and advance-stage small-cell lung cancer patients also failed to demonstrate clinical efficacy [212, 213]. Preclinical data suggest that short intermittent intravenous dosing increases intratumoral cellular loading and retention of the active oligonucleotide to enhance oblimersen antitumor activity. Dose escalation studies have demonstrated that weekly 2 h infusion attains maximal blood concentrations that are tenfold hire than continuous infusion dosing as was utilized in the aforementioned phase III clinical trials. Further investigation of the bolus dosing may demonstrate the clinical activity of oblimersen. SPC2996 (Santaris Pharma) is a high-affinity BLC-2 LNA RNA analogue. Preclinical data utilizing a human chronic lymphocytic leukemia (CLL) explant

Bcl-2

1st Generation ASO

Locked nucleic acid R. Enantiomer of Gossypol

Small molecule inhibitor

Small molecule inhibitor Small molecule inhibitor

SPC2966 AT-101

Obatoclax (GX 15-070)

ABT-737 ABT-263

Survivin XIAP

Small molecule inhibitor 2nd Generation ASO

Recombinant TRAIL ligand Recombinant Fas-ligand TRAIL-R1/DR4 agonist moAb

TRAIL-R2/DR5 agonist moAb

TRAIL-R2/DR5 agonist moAb TRAIL-R2/DR5 agonist moAb

EM-1421 AEG35156

Death Receptor Apo2/TRAIL APO-010 Mapatumumab (HGS-ETR1)

Conatumumab (AMG655)

Lexatumumab (HGS-ETR2) ApoMab

TRAIL-R2/DR5 TRAIL-R2/DR5

TRAIL-R2/DR5

TRAIL-R1/DR4 FAS TRAIL-R1/DR4

Survivin Survivin Survivin

Inhibitor of Apoptosis Protein LY2181308 (ISIS-237722) 2nd Generation ASO SPC3042 Locked nucleic acid YM-155 Small molecule inhibitor

BH-3 domain BH-3 domain

BH-3 domain

Bcl-2 BH-3 domain

Target

Mechanism of action

Name Prosurvival antagonist Oblimersen

Table 15.10  Apoptosis-targeted therapy

Phase I/II Phase I/II

Phase II

Phase I/II Phase I Phase II

Phase I/II Phase I/II

Phase II Phase I/II Phase II

Phase I Phase I

Phase I/II

Phase I/II Phase I/II

Phase II/III

Development phase

NSCLC, sarcoma, Non-Hodgkin’s lymphoma

HCC, NSCLC, Non-Hodgkin’s lymphoma, and myeloma NSCLC, sarcoma, colorectal, and pancreatic adenocarcinoma

Non-Hodgkin’s lymphoma, NSCLC

CLL, prostate HCC, CLL NSCLC, Non-Hodgkin’s lymphoma, melanoma, and prostate Glioblastoma multiforme CLL, NSCLC

CLL, lymphoma, SCLC

Breast, colorectal, gastric, GIST, CLL, NonHodgkin’s lymphoma, NSCLC, Merkel cell, prostate, renal: melanoma and myeloma CLL B-cell lymphoma/leukemia, prostate, NSCLC, glioblastoma multiforme, esophageal SCLC, Non-Hodgkin’s lymphoma, myelodysplastic syndrome, NSCLC and myeloma

Disease

440 C.D. Weekes and M. Hidalgo

15  Targeted Therapeutics in Cancer Treatment

441

xenograft model demonstrated the antileukemic effects of SPC2996 in  vivo. SPC2996 has now undergone initial phase I/II evaluation in patients with relapsed/ refractory CD5+CD20+CD23+ CLL [214]. The maximum-tolerated dose is 4 mg/kg. All six patients treated at this dose demonstrated a decrease in circulating CLL cells. Lymph node analysis in these patients demonstrated that 50% of patients analyzed demonstrated a decrease in total lymph nodes of at least 50%. Further clinical investigation is planned.

15.7.2 Inhibitor of Apoptosis Protein 15.7.2.1 Survivin and XIAP LY2181308 (also known as ISIS-23722) is a second-generation O-methoxymethylmodified ASO that specifically inhibits survivin mRNA (Table 15.9). A single-agent phase I dose escalation study of LY2181308 administered as a loading dose of a 3-h infusion on 3 consecutive days followed by weekly intravenous infusions defined 750  mg as the maximum-tolerated dose. The most frequent adverse events were transient aPTT prolongation without associated bleeding and grade 2 fatigue. No grade 3 or 4 drug-related toxicities were observed. Pre- and postdose tumor biopsies were obtained from patients treated at predicted biologically effective doses [214]. LY2181308 treatment resulted upwards of a 50% reduction in survivin mRNA expression in 11/15 evaluable paired samples. Fresh tumor analysis obtained from paired transendobronchial biopsy specimen demonstrated a near complete elimination of survivin positive cells and an increased fraction of ­apoptosis. LY2181308 is now undergoing phase II clinical investigation in ­leukemia, hepatocellular carcinoma, and prostate cancer patients. SPC3042 is another LNA RNA analogue demonstrating preclinical inhibition of survivin expression associated with apoptosis and sensitizes cancer cells to taxane chemotherapy both in vitro and in vivo [215]. Both YM-155 and EM-1421 are small molecule inhibitors of survivin. The MTD of YM-155was defined as 8 mg/m2/day when administered as a 168-h continuous infusion in a phase I study [216]. Reversible elevated serum creatinine was the doselimiting toxicity observed in two patients. The most common ­treatment-related toxicities were mucosal inflammation and elevated prothrombin time. Fourteen of 34 patients treated on study attained stable disease or minor response. This compound is currently in phase II investigation in both solid and hematologic malignancies. AEG35156 is a second-generation ASO targeting the mRNA of XIAP. Phase I evaluation of two continuous intravenous infusion schedules, either as a 3-day or 7-day infusion, has been completed [217]. The MTD for the 3-day and 7-day infusion schedules were £213 mg/m2/day and 125 mg/m2/day, respectively. The predominant toxicity with either schedule was elevated liver enzymes. Suppression of XIAP mRNA was observed for 72 h. The suppression of XIAP mRNA levels did correlate with objective response in a limited number of patients. AEG35156 is currently in phase II investigation in combination with a number of chemotherapy agents [218].

442

C.D. Weekes and M. Hidalgo

15.7.2.2 Peptidomimetics and Small Molecule Inhibitors of Prosurvival Proteins Disruption of the protein interactions between prosurvival and proapoptotic ­members of the BCL-2 protein family is another strategy that has entered clinical investigation. This molecular class of compounds targets the hydrophobic binding pocket of the BH-3 domain responsible for protein–protein interactions occurring between BCL-2 family members. Gossypol is enantiomeric polyphenol originally derived from cottonseed. Gossypol binds to the BH-3-binding pocket of BCL-2, BCL-XL, and MCL1. AT-101, the R-enantiomer has been demonstrated to be the most potent inhibitor of BCL-2 expression and apoptosis inducer. The single-agent phase I study defined 20 mg/ml administered daily for 21 of 28 days as the MTD [219]. Patients receiving a dose of 30 mg/ml experienced a DLT of small intestinal obstruction. Other common drug-related side effects were diarrhea and fatigue. A randomized placebo-controlled phase II analysis of AT-101 combined with docetaxel as second-line therapy for patients with advanced NSCLC demonstrated a median overall survival of 7.3 months compared to 5.6 months in patients receiving docetaxel and placebo, HR 0.6 p = 0.05. Pharmacodynamic evaluation demonstrated that MCL-1 interactions with Noxa may promote apoptosis. Additional results of phase II studies of AT-101 and chemotherapy are forthcoming. Obatoclax (GX15-070) binds to the BH-3-binding domain of several BCL-2 family members and demonstrates single-agent antitumor activity against a wide variety of tumor cells in  vitro. Three single-agent phase I studies of obatoclax administered as a 3-h infusion every 3 weeks in patients with refractory chronic lymphocytic leukemia, hematologic malignancies, and solid tumors, respectively, have been completed [220–222]. All three studies defined the maximal tolerated dose to be 28 mg/m2. Neurologic dose-limiting toxicity in the form of somnolence, euphoria, and ataxia was common amongst the studies. CLL patients achieved modest clinical efficacy demonstrated by improved anemia and thrombocytopenia and a reduction in circulating CLL cells. In a second study of patients with advanced leukemias and myelodysplastic syndrome, one patient with multilineage t(9:11) acute myelogenous leukemia attained a complete response for a period of 8  months. Obatoclax is currently in phase I/II investigation in combination with docetaxel and bortezomib. ABT-737 and the orally bioavailable variant ABT-263 bind with high affinity to the BH-3 domain of BCL-2, BCL-XL, and BCL-W but not MCL-1. Preclinical evaluation demonstrates that MCL-1 expression functions as an important mechanism of resistance. ABT-263 has now entered early-phase clinical investigation testing multiple dosing schedules in patients with refractory CLL and extensive-stage small-cell lung cancer. The schedules being evaluated are oral administration daily for 14 of a 21 days or daily administration for 21 consecutive days [223]. Thrombocytopenia is a uniform dose-limiting toxicity in all studies to date. Elevated liver enzymes and pulmonary toxicity have also been observed. The recommended phase II dose has not been defined. Early evidence of clinical activity has been manifested with objective responses as well as decrease in circulating CLL cells.

15  Targeted Therapeutics in Cancer Treatment

443

15.7.2.3 Direct Proapoptosis Activation Death receptor activation by either antibodies or recombinant ligands has ­developed as a clinical strategy to induce cancer cell apoptosis. To date, this strategy has focused on targeting the TRAIL receptors (TRAIL-R1, DR4 or TRAIL-R2, DR5). Death receptors are preferentially expressed on cancer cells. In addition, TRAIL has little activity in normal healthy cells. As a result, targeting the death receptors has the potential to be a relative cancer cell-specific strategy. Apo2/TRAIL is a recombinant human homotrimer of TRAIL that has undergone initial phase I testing (Table  15.9). Preclinical investigation demonstrates that it induces apoptosis in p53-independent manner in approximately 50% of cancer cell lines and has minimal effects on healthy noncancerous cells. Initial clinical investigation shows the compound to possess minimal toxicity when given as a 1-h infusion on a 3-week schedule. A confirmed partial response was observed in a patient with metastatic synovial chondrosarcoma. Combination chemotherapy trials are ongoing. APO010 is a recombinant Fas-ligand. It is a fusion protein of three fasligand extracellular domains combined with dimer forming collagen protein backbone. It is currently in initial phase I clinical investigation. The results of this study are not yet available. Death receptor agonist antibodies represent the most clinically advanced compounds utilizing the strategy of extrinsic apoptosis pathway activation. Mapatumumab (HGS-ETR1) represents the most clinically advanced of these agents. Mapatumumab is a fully humanized IgG1 TRAIL-R1 agonist antibody. It has demonstrated apoptosis in both cancer cell lines and xenografts in the presence of TRAIL. The MTD was defined as 10 mg/kg in single-agent phase I evaluation [224]. The most common toxicities were grade 1 or 2 fatigue, fever, or myalgia. Two patients developed grade 3 elevation of liver transaminases, which may reflect a direct effect of mapatumumab on TRAIL-expressing hepatocytes. Single-agent phase II testing of mapatumumab in patients with chemotherapy refractory NSCLC confirmed the safety and tolerability of the compound [225]. Mapatumumab was well tolerated up to doses of 20 mg/kg in combination with paclitaxel and carboplatin in a phase I study [226]. Five of 27 (19%) patients achieved at least a partial response with one patient attaining a pathologic complete response. Phase II studies are ongoing in a number of malignancies, including hepatocellular carcinoma. TRAIL-R2/DR5 agonist antibodies in clinical investigation include Apomab, Conatumumab (AMG 655), and lexatumumab (HGS-ERT2). Apomab is a fully human monoclonal IgG1 antibody. Single-agent phase I evaluation demonstrated it to be well tolerated when administered intravenously on day 1 of a 28-day cycle [227]. Reversible dose-limiting toxicity of grade 3 elevation of liver transaminase was observed in one patient. Initial combination chemotherapy trials have focused on colorectal cancer, where is undergoing evaluation in combination with FOLFOX and bevacizumab or in combination with irinotecan and cetuximab. It is also undergoing clinical investigation as a single agent for patients with chondrosarcoma and in combination with paclitaxel, carboplatin, and bevacizumab for NSCLC patients. Conatumumab (AMG 655) is also a fully human monoclonal antibody agonist of

444

C.D. Weekes and M. Hidalgo

TRAIL-R2/DR5. Conatumumab was well tolerated up to doses of 20 mg/kg when administered every 14 days in single-agent phase I evaluation [228]. Initial objective responses were observed in a patient with NSCLC and a PET response was seen in a CRC patient. Clinical investigation of this antibody is focused on NSCLC, colorectal, and pancreatic cancers as well as lymphoma. Colorectal studies are ongoing in combination with Panitumumab, mFOLFOX6 and bevacizumab, or FOLFIRI. Pancreatic cancer evaluation is focused on the combination of AMG-655 and gemcitabine. It is also being evaluated in combination with the IGF1-R antagonistic monoclonal antibody AMG-479 in a phase IB study. Lastly, AMG-655 is being investigated as a combination with either vorinostat or bortezomib in lymphoma patients. Consistent with other compounds in this class lexatumumab demonstrated hepatic toxicity at high doses in single-agent phase I testing; therefore, 10  mg/kg was selected as the recommended phase II dose [229, 230]. Single-agent evaluation has resulted in objective responses. A phase I/II study was conducted in which it was combined with cytotoxic chemotherapies. No pharmacokinetic interaction with the cytotoxic chemotherapies or increased toxicity was observed. Initial evidence of clinical benefit was observed in patients with colorectal or non-small-cell lung cancer. It is currently being evaluated in combination with gamma interferon.

15.8 Challenges in the Clinical Development of STI The development of small molecule inhibitors of biologic processes is quite complex. The first generation of these molecules currently in development has demonstrated that this avenue of therapy is a feasible approach to anticancer therapy. However, the development of these compounds bears an enormous financial burden thus making it imperative to define more efficacious ways of administering these medications. The further development of small molecule inhibitors will require novel phase I dose-finding study design, improved patient selection criteria, novel study endpoints as well as the development of rational combination therapy.

15.8.1 Clinical Trials Design Issues The goal of early-phase clinical studies is to determine the optimum dose and patient population best suited for definitive clinical investigation to determine efficacy. The classic endpoint of phase I trials is maximum-tolerated dose. A paradigm switch to optimal biologic dose will be needed for future clinical evaluation. For example, rapamycin and it analogues have been demonstrated to have antitumor effect in a dose-independent manner with these effects occurring well below the maximum-tolerated dose [157, 158]. This is likely the case for many other targeted agents. Pharmacodynamic assessment of efficacy needs to be incorporated into

15  Targeted Therapeutics in Cancer Treatment

445

early-phase clinical evaluation of new compounds. This will allow the determination of optimal biologic dose as part of the drug approval process. It will be imperative to provide the optimal biologic dose of agents and avoid cumulative toxicity that may incur as a result of inhibiting multiple cellular pathways simultaneously. Pharmacodynamic response can be measured in tumor biopsies obtained prior to and after the initiation of treatment or in surrogate markers. In addition, surrogate tissues such as PBMC or skin biopsies have been used for this purpose with varying degrees of success. Current criticisms of a pharmacodynamic approach to the development of biologic agents would include the poor reliability of the results obtained with these studies. Additionally, results obtained in normal surrogate tissue may not reflect drug effects observed in malignant tissue. Lastly, target inhibition does not necessarily correlate with drug efficacy as there may be a myriad of intracellular mechanism resulting in drug resistance. Modification of phase II and phase III study design in addition to pharmacodynamic approaches will be necessary to efficiently evaluate biologic agents in the future. Novel ways of optimizing trial design include the multinomial method and the randomized discontinuation design. The clinical development of Sorafenib in patients with metastatic RCC demonstrated the importance of novel study design in the development of biologic agents [172]. This phase II study utilized a placebocontrolled randomized discontinuation design in which 202 patients were initially enrolled to be treated with Sorafenib for a 12-week run-in period. At the end of that period, patients achieving ³25% tumor shrinkage were continued on the Sorafenib until evidence of disease progression. Those patients attaining disease stabilization were randomized to either Sorafenib or placebo for an additional 24 weeks, while patients with evidence of disease progression or intolerable toxicity were taken off study at the end of the 12-week run-in period. This strategy allowed the investigators to separate slow tumor growth from resistance as demonstrated by a 32% increase in progression-free survival (50% vs. 18%) for those patients treated with Sorafenib amongst those patient initially achieving disease stabilization. A criticism of the randomized discontinuation design is that the information lost on the nonrandomized patients may be of a significant magnitude to underpower these studies compared to the classic design of randomizing all patients.

15.8.2 Patient Selection The development of biologic therapy has forced the introduction of novel clinical trial design. Classic clinical evaluation of a therapeutic compound consisted of treating an unselected group of patients with a given cancer with the compound and the assessment of objective response by traditional imaging techniques. Appropriate patient selection will be imperative for the development of future biologic agents. The importance of patient selection is exemplified by the clinical development of trastuzumab, gefitinib, and cetuximab. The registration trial for trastuzumab was performed in patients with metastatic breast cancer with documented overexpression

446

C.D. Weekes and M. Hidalgo

of HER2 [2]. The addition of trastuzumab to traditional chemotherapy was associated with an 18% (50% vs. 32%) increase in objective response, which translated to a 5-month improvement in median survival (25.1 vs. 20.3). HER2 overexpression occurs in 25–30% of metastatic breast cancer patients. Performing this study in a group of unselected metastatic breast cancer patients would be expected to produce objective responses in less than 5% of patients and would likely be associated with an increased rate of cardiotoxicity over the 27% observed in the published study. It is likely that trastuzumab would have been deemed too toxic for a minimal survival benefit to recommend FDA approval for the treatment of metastatic breast cancer patients. Patient selection issues regarding the clinical development of cetuximab and gefitinib relate to the receptor activation status. Cetuximab received FDA approval for the treatment of irinotecan-refractory metastatic colorectal cancer independent of EGFR expression status [86]. Similarly, gefitinib was approved by the FDA for chemotherapy refractory advanced stage NSCLC in the second-line setting. Subsequently, it was determined that the tumor sensitivity to gefitinib and later erlotinib is due to the presence of activating mutations in the kinase domain of EGFR [130, 131]. Furthermore, these mutations segregate to patients of Asian descent and nonsmokers. Clinical studies are ongoing to test the hypothesis of that the presence of activating mutations in certain patient populations may provide a survival benefit. Similarly, it has now been determined that activating mutations in Kras mediate resistance to EGFR-targeted antibodies [62]. Similarly, pp70s6 kinase activity has been associated with radiologic response in patients with glioblastoma multiforme treated with temsirolimus [164]. Future clinical development of biologic agents targeting receptor kinases or intracellular kinases will need to select patients based on evidence of target activation not based on histology. Histology may become important in the development of therapeutic combinations with cytotoxic agents as well as combinatorial therapy including hormonal therapy as there is clear evidence that perturbation of hormonal signals can impact RTK-mediated signal transduction.

15.8.3 Study Endpoints Novel endpoints need to be explored in addition to novel study design and patient selection for early-phase clinical studies. As stated previously, compounds targeting signaling events are not likely to result in tumor response in the classic sense. Therefore, alternative endpoints to evaluate efficacy must be validated and accepted to enhance our understanding of how to use these drugs and to avoid mislabeling compounds as ineffective. Possible alternative endpoints that have proposed include time to progression, the proportion of patients without progressive disease as their best response, progression rate, symptomatic benefit, measures of target inhibition, positron emission tomography scanning, and reduction in tumor markers. The combination of novel study design and novel therapeutic endpoints will help to efficiently investigate the use of compounds targeting intracellular signal events.

15  Targeted Therapeutics in Cancer Treatment

447

The traditional model of tumor size reduction as assessed by classic radiologic test may not be the proper endpoint for the development of biologic agents. Many of these molecules are cytostatic not cytocidal when evaluated in vitro and may be expected to result in disease stabilization rather than tumor response in human subjects. Additionally, the results of the extensive early-phase trials of various biologic agents have demonstrated evidence of disease stabilization associated with significant palliative effect, suggesting the presence of biologic effect. Furthermore, there are rare incidences where a given tumor is dependent on a single biologic pathway for oncogenesis. Additionally, the new biologic agents may preferentially impact tumor cell metabolism. These factors would suggest that functional imaging may better represent tumor response to biologic therapy rather than traditional radiologic methods of assessing response. This was demonstrated by the treatment of VHL null RCC cell with temsirolimus in a xenograft model in which stability of HIF-1a was correlated with temsirolimus sensitivity and tumor response as measured by [18F] fluoro-2-deoxy-D-glucose (FDG) positron emission tomography (PET) [231]. The development of biologic agents targeting molecules regulating angiogenesis further exemplifies the need for the validation of nontraditional measurements of tumor response. Dynamic contrastenhanced (DCE) magnetic resonance imaging (MRI) and diffusion-weighted MRI are two examples of novel imaging techniques that allow the quantification of various aspects of angiogenesis including perfusion, vascular permeability, and necrosis to be assessed [232]. These are particularly important as it is postulated that a component of the antitumor effect of bevacizumab is due to the modulation of vascular permeability [233].

15.8.4 Combination Therapy The evolution of biologic therapy will require the use of rationale combination therapy. This is in part due to the complex mechanisms of resistance to biologic therapies. First, cross talk between membrane receptors is a known mechanism. This is evidenced by the emerging role of HER2 in EGFR signaling and vice versa. Overexpression of HER2 can result in enhanced EGFR signaling [75, 77, 78]. Therefore, one could envisage a strategy targeting both HER2 and EGFR either as a dual kinase inhibitor or by combining antibody therapy with TKI-based therapy. Nullification of antioncogenic downstream effectors may also result in resistance to biologic agents. EGFR inhibitors augment the protein levels of the CDKI p27 resulting in cell cycle arrest. Proteosomal degradation of p27 in response to EGFR inhibition may result in a resistant phenotype. Conversely, the upregulation of oncogenic downstream effectors either independent or in response to biologic agents is a common mechanism to induce resistance [234]. In the case of EGFR inhibition, loss of PTEN, overexpression of Akt as well as the functional activation of the PI3K and ERK pathways may occur de novo or in response to EGFR inhibition resulting in abrogation of the therapeutic effect of EGFR inhibition. The importance of this

448

C.D. Weekes and M. Hidalgo

mode of resistance is demonstrated by the association between the reduction of Akt activity in response to EMD7200 therapy and therapeutic benefit [235]. Lastly, the augmentation of complementary pathways may also provide a critical mechanism to either bypass the growth inhibition effects or provide a secondary signal to propagate oncogenesis. This would be particularly important for epithelial tumors, which are less likely to be dependent upon a single oncogenic pathway to mediate oncogenesis. A prime example of this phenomenon is the acquired overexpression of vascular endothelial cell growth factor receptor (VEGFR) in response to antibody therapy targeting EGFR. The use of next-generation agent ZD6474 targeting both EGFR and VEGFR is an example of a strategy that may allow compensation for such events [236]. Additionally, as the oncology community gains a better understanding of the predicted biologic responses to a given biologic agent and the timing of such events, sequential rotation of small molecule inhibitors combined with conventional cytotoxic agents may become a practical strategy to overcome the inevitable resistance to therapy that will eventually occur. The rationale for the development of combination molecular therapies is ultimately going to require a systems biology approach. Currently, there are compounds in development that target almost every major cellular process. The effective use of these compounds will also require the incorporation of biostastician, whose focus is a systems approach to combinatorial drug development. Understanding the complex mechanisms of resistance to biologicbased therapy will be imperative to designing effective combinations of biologic, cytotoxic, and immunologic therapy in the future.

References 1. Maloney DG, Grillo-Lopex AJ, White CA, Bodkin D, Schilder RJ, Neidhart JA, Janakiraman N, Foon KA, Liles TM, Dallaire BK, Wey K, Royston I, Davis T, Lefy R. IDEC-C2B8 (Rituximab) anti-CD20 monoclonal antibody therapy in patients with relapsed low-grade non-Hodgkin’s lymphoma. Blood. 90(6):2188–2195, 1997 2. Slamon DJ, Leyland-Jones B, Shak S, Fuchs H, Paton V, Bajamonde A, Fleming T, Eiermann W, Wolter J, Pegram M, Baselga J, Norton L. Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2. N Engl J Med. 344:783–792, 2001 3. Druker BJ, Talpaz M, Resta DJ, Peng B, Buchdunger E, Ford JM, Lydon NB, Kantarjian H, Capdeville R, Ohno-Jones S, Sawyers CL. Efficacy and safety of a specific inhibitor of the BCR-ABL tyrosine kinase in chronic myeloid leukemia. N Engl J Med. 344:1031–1037, 2001 4. Demetri GD, von Mehren M, Blanke CD, Van den Abbeele AD, Eisenberg B, Roberts PJ, Heinrich MC, Tuvenson DA, Singer S, Janicek M, et  al. Efficacy and safety of imatinib mesylate in advanced gastrointestinal stromal tumors. N Engl J Med. 347:472–480, 2002 5. Kohler G, Milstein C. Continuous cultures of fused cells secreting antibody of predefined specificity. Nature. 256:495–497, 1975 6. Clynes R. Antitumor antibodies in the treatment of cancer: Fc receptors link opsonic antibody with cellular immunity. Hematol Oncol Clin North Am. 20:585–612, 2006 7. Iannello A, Ahmad A. Role of antibody-dependent cell-mediated cytotoxicity in the efficacy of therapeutic anti-cancer monoclonal antibodies. Cancer Metastasis Rev. 24:487–499, 2005

15  Targeted Therapeutics in Cancer Treatment

449

8. Goldspy RA, Kindt TJ, Osborne BA, et  al. Immunology, Fifth Edition. New York: WH Freeman and Company, 1–551, 2003 9. Koene HR, Kleijer M, Algra J, et al. FcgRIIIa-158V/F polymorphism influences the binding of IgG by natural killer cell FcgRIIIa, independently of the FcgRIIIa-48L/R/H phenotype. Blood. 90:1109–1114, 1997 10. Van Sorge NM, van der Pol WL, van de Winkel JGJ. FcgR polymorphisms: implications for function, disease susceptibility and immunotherapy. Tissue Antigens. 61:189–202, 2003 11. Cartron C, Dacheux L, Salles G, et al. Therapeutic activity of humanized anti-CD20 monoclonal antibody and polymorphism in IgG Fc receptor FcgRIIIa gene. Blood. 99:754–758, 2002 12. Weng W-K, Levy R. Two immunoglobulin G fragment C receptor polymorphisms independently predict response to rituximab in patients with follicular lymphoma. J Clin Oncol. 21:3940–3947, 2003 13. Bibeau F, Lopez-Crapez E, Di Fiore F, Thezenas S, Ychou M, Blanchard F, Lamy A, Penault-Llorca F, Frebourg T, Michel P, Sabourin J-C, Boissiere-Michot F. Impact of FcgRIIa-FcgRIIIa polymorphisms and KRAS mutations on the clinical outcome of patients with metastatic colorectal cancer treated with cetuximab plus irinotecan. J Clin Oncol. 27:1122–1129, 2009 14. Traxler P, Furet P. Strategies toward the design of novel and selective protein tyrosine kinase inhibitors. Pharmacol Ther. 82:195–206, 1999 15. Liu Y, Gray NS. Rational design of inhibitors that bind to inactivate kinase conformations. Nat Chem Biol. 2:358–364, 2006 16. Chan JHP, Lim S, Wong WSF. Antisense oligonucleotides: from design to therapeutic application. Clin Exp Pharm Physiol. 33:533–540, 2006 17. Eckstein F. Phosphorothioate oligonucleotides: what is their origin and what is unique about them? Antisense Nucleic Acid Drug Dev. 10:117–121, 2000 18. Altmann KH, Fabbro D, Dean NM, et  al. Second-generation antisense oligonucleotides: structure-activity relationships and the design of improved signal-transduction inhibitors. Biochem Soc Trans. 24:630–637, 1996 19. Gleave ME, Monia BP. Antisense therapy for cancer. Nat Rev Cancer. 5:468–479, 2005 20. Nielsen PE, Egholm M, Berg RH, Buchardt O. Sequence-selective recognition of DNA by strand displacement with a thymine-substituted polyamide. Science. 254:1497–1500, 1991 21. Nielsen PE. PNA technology. Mol Biotechnol. 26:233–248, 2004. 22. Vester B, Wengel J. LNA (locked nucleotide acid): high-affinity targeting of complementary RNA and DNA. Biochemistry. 43:13233–13241, 2004 23. Kurreck J, Wyszko E, Gillen C, Erdmann VA. Design of antisense oligonucleotides stabilized by locked nucleic acid. Nucleic Acids Res. 30:1911–1918, 2002 24. Amantana A, Iversen PL. Pharmacokinetics and biodistribution of phosphorodiamidate morpholino antisense oligomers. Curr Opin Pharmacol. 5:550–555, 2005 25. Fire A, Xu S, Montgomery MK, Kostas SA, Driver SE, Mello CC. Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature. 391:806– 811, 1998 26. Jana S, Chakraborty C, Nandi S, Deb JK. RNA interference: potential therapeutic targets. Appl Microbiol Biotechnol. 65:649–657, 2004 27. Martinez J, Patkaniowska A, Urlaub H, Luhrmann R, Tuschl T. Single-stranded antisense siRNAs guide target RNA cleavage in RNAi. Cell. 110:563–574, 2002 28. Lewis DL, Wolff JA. Systemic siRNA delivery via hydrodynamic intravascular injection. Adv Drug Deliv Rev. 59:115–123, 2007 29. Gao K, Huang L. Nonviral methods for siRNA delivery. Mol Pharm. 6:651–658, 2009 30. de Fougerolles AR. Delivery vehicles for small interfering RNA in vivo. Hum Gene Ther. 19:125–132, 2008 31. Minakuchi Y, Takeshita F, Kosaka N, Sasaki H, Yamamoto Y, Kouno M, Honma K, Nagahara S, Hanai K, Sano A, Kato T, Terada M, Ochiya T. Atelocollagen-mediated synthetic small interfering RNA delivery for effective gene silencing in vitro and in vivo. Nucleic Acids Res. 32:e109, 2004

450

C.D. Weekes and M. Hidalgo

32. Takei Y, Kadomatsu K, Yuzawa Y, Matsuo S, Muramatsu T. A small interfering RNA targeting vascular endothelial growth factor as cancer therapeutics. Cancer Res. 64:3365–3370, 2004. 33. Lysik MA, Wu-Pong S. Innovations in oligonucleotide drug delivery. J Pharm Sci. 92:1559– 1573, 2003 34. Fattal E, Couvreur P, Dubernet C. ‘Smart’ delivery of antisense oligonucleotides by anionic pH-sensitive liposomes. Adv Drug Deliv Rev. 56:931–946, 2004 35. Jarver P, Langel U. The use of cell-penetrating peptides as a tool for gene regulation. Drug Discov Today. 9:395–402, 2004 36. Xu Y, Szoka FC, Jr. Mechanism of DNA release from cationic liposome/DNA complexes used in cell transfection. Biochemistry. 35:5616–5623, 1996 37. Li SD, Chono S, Huang L. Efficient oncogene silencing and metastasis inhibition via systemic delivery of siRNA. Mol Ther. 16:942–946, 2008 38. Sherr CJ, Roberts JM. CDK Inhibitors: positive and negative regulators of G-1 phase progression. Genes Dev. 13:1125–1126, 1999 39. Malumbres M, Barbacid M. To cycle or not to cycle: a critical decision in cancer. Nat Rev Cancer. 1:222–231, 2001 40. Buckley M, Sweeney KJ, Hamilton JA, Sini RL, Manning DL, Nicholson RI, DeFazio A, Watts CK, Musgrove EA, Sutherland RL. Expression and amplification of cyclin genes in human breast cancer. Oncogene. 8:2127–2133, 1993 41. Yatabe Y, Suzuki R, Tobinai K, Matsuno Y, Ichinohasama R, Okamoto M, Yamaguchi M, Tamaru J, Uike N, Hashimoto Y, et  al. Significance of cyclin D1 overexpression for the diagnosis of mantle cell lymphoma: a clinicopathologic comparison of cyclin D1-positive ML and cyclin D1-negative MCL-like-B-cell lymphoma. Blood. 95:2253–2261, 2000 42. Shahjehan WA, Laird P, DeMeester T. DNA methylation: an alternative pathway to cancer. Ann Surg. 234:10–20, 2001 43. Senderowicz AM. Cyclin-dependent kinases as targets for cancer therapy. In: Cancer Chemotherapy and Biological Response Modifiers. (Giaccone G, Schilsky R, Sondel P, eds), New York, NY, Elsevier Science, pp 169–188, 2002 44. Oelgeschlager T. Regulation of RNA polymerase II activity by CTD phosphorylation and cell cycle control. J Cell Physiol. 190:160–169, 2002 45. Kobor M, Greenblatt J. Regulation of transcription elongation by phosphorylation. Biochim Biophys Acta. 13:261–275, 2002 46. Koumenis C, Giaccia A. Transformed cell require continuous activity of RNA polymerase II to resist oncogene-induced apoptosis. Mol Cell Biol. 17:7306–7316, 1997 47. Te Poele R, Okorokov A, Joel S. RNA synthesis block by 5,6-dichloro-1-beta-D-ribofuranosylbenzimidazole (DRB) triggers p53-dependent apoptosis in human colon carcinoma cells. Oncogene. 18:5765–5772, 1999 48. Losiewicz MD, Carlson BA, Kaur G, Sausville EA, Worland PJ. Potent inhibition of cdc2 kinase activity by the flavonoid L86-8275. Biochem Biophys Res Commun. 201:589–595, 1994 49. Drees M, Dengler WA, Roth T, Labonte H, Mayo J, Malspeis L, Grever M, Sausville EA, Fiebig HH. Flavopiridol (L86-8275): selective antitumor activity in vitro and activity in vivo for prostate carcinoma cells. Clin Cancer Res. 3:271–279, 1997 50. Patel V, Senderowicz AM, Pinto D, Igishi T, Raffeld M, Quintanilla-Martinez L, Ensley JF, Sausville EA, Gutkind JS. Flavopiridol, a novel cyclin-dependent kinase inhibitor, suppresses the growth of head and neck squamous cell carcinomas by inducing apoptosis. J Clin Invest. 102:1674–1681, 1998 51. Byrd JC, Lin T, Dalton JT, Wu B, Phelps MA, Fischer B, Moran M, Blum KA, Rovin B, Brooker-McEldowney M, Broering S, Schaaf LF, Johnson AJ, Lucas DM, Heerema NA, Lozanski G, Young DC, Suarez IR, Colevas AD, Grever MR. Flavopiridol administered using a pharmacologically derived schedule is associated with marked clinical efficacy in refractory, genetically high-risk chronic lymphocytic leukemia. Blood. 109:399–404, 2007

15  Targeted Therapeutics in Cancer Treatment

451

52. Shah MA, Kortmansky J, Motwani M, Drobnjak M, Gonen M, Yi S, Weyerbacher A, Cordon-Cardo C, Lefkowitz R, Brenner B, et al. A phase I/pharmacologic study of weekly sequential Irinotecan (CPT) and flavopiridol. Clin Cancer Res. 11:3836–3845, 2005 53. Shah MA, Kortmansky J, Gonen M, Tse A, Lefkowitz R, Kelsen D, Colevas D, Winkelman J, Yi S, Schwartz G. A phase I study of weekly sequential Irinotecan (CPT), cisplatin (CIS) and flavopiridol (F). J Clin Oncol Suppl. 22:14S, 2004 54. Lampson MA, Renduchitala K, Khodjakov A, Kapoor TM. Correcting improper chromosome-spindle attachments during cell division. Nat Cell Biol. 6:232–237, 2004 55. Lampson MA, Kapoor TM. The human mitotic checkpoint protein BubR1 regulates chromosome-spindle attachments. Nat Cell Biol. 7:93–98, 2005 56. Gautschi O, Heighway J, Mack PC, Purnell PR, Lara PM, Gandara DR. Aurora kinases as anticancer drug targets. Clin Cancer Res. 14:1639–1648, 2008 57. Duncan P, Pollet N, Niehrs C, Nigg EA. Cloning and characterization of Plx2 and Plx3, two additional Polo-like kinases from Xenopus laevis. Exp Cell Res. 270:78–87, 2001 58. Kotani S, Tugendreich S, Fujii M, Jorgensen PM, Watanabe N, Hoog C, Hieter P, Todokoro K, et  al. PKA and MPF-activated polo-like kinase regulate anaphase-promoting complex activity and mitosis progression. Mol Cell. 1:371–380, 1998 59. Takaki T, Trenz K, Costanzo V, Petronczki M. Polo-like kinase 1 reaches beyond mitosis-cytokinesis, DNA damage response and development. Curr Opin Cell Biol. 20:650–660, 2008 60. Mross K, Frost A, Steinbild S, Hedbom S, Rentschler J, Kaiser R, Rouyrre N, Trommeshauser D, Hoesl CE, Munzert G. Phase I dose escalation and pharmacokinetic study of BI 2536, a novel polo-like kinase 1 inhibitor, in patients with advanced solid tumors. J Clin Oncol. 26:5511–5517, 2008 61. Gumireddy K, Reddy M, Cosenza SC, Boominathan R, Baker SJ, Papathi N, Jiang J, Holland J, Reddy EP. ON01910, a non-ATP-competitive small molecule inhibitor of Plk1, is a potent anticancer agent. Cancer Cell. 7:275–286, 2005 62. Jimeno A, Li J, Messersmith WA, Laheru D, Rudek MA, Maniar M, Hidalgo M, Baker SD, Donehower RC. Phase I study of ON 01910.Na, a novel modulator of the Polo-like kinase 1 pathway, in adult patients with solid tumors. J Clin Oncol. 26:5504–5510, 2008 63. Jimeno A, Chan A, Cusatis G, Zhang X, Wheelhouse J, Solomon A, Chan F, Zhao M, Cosenza SC, Ramana Reddy MV, Rudek MA, Kulesza P, Donehower RC, Reddy EP, Hidalgo M. Evaluation of the novel mitotic modulator ON 01910.Na in pancreatic cancer and preclinical development of an ex vivo predictive assay. Oncogene. 28:610–618, 2009 64. Futreal P, Kasprzyk A, Birney E, Mullikin JC, Wooster R, Stratton MR. Cancer and genomics. Nature. 409:850–852, 2001 65. Blume-Jensen P, Hunter T. Oncogenic kinase signaling. Nature. 411:355–365, 2001. 66. Plowman GD, Sudarsanam S, Bingham J, Whyte D, Hunter T. The protein kinases of Caenorhabditis elegans: a model for signal transduction in multicellular organisms. Proc Natl Acad Sci U S A. 96:13603–13610, 1999 67. Robinson DR, Wu YM, Lin SF. The protein tyrosine kinase family of the human genome. Oncogene. 19:5548–5557, 2000 68. Kolibaba KS, Druker BJ. Protein tyrosine kinases and cancer. Biochim Biophys Acta. 1333:F217–F248, 1997 69. Susman E. Bevacizumab adds survival benefit in colorectal cancer. Lancet Oncol. 6:136, 2005 70. Meyerhardt JA, Mayer RJ. Systemic therapy for colorectal cancer. N Engl J Med. 352:476– 487, 2005 71. Woodburn JR. The epidermal growth factor receptor and its inhibition in cancer therapy. Pharmacol Ther. 82:241–250, 1999 72. Mitsiades CS, Mitsiades N, Hideshima T, Richardson PG, Anderson KC. Proteasome inhibitors as therapeutics. Essays Biochem. 41:205–218, 2005 73. Wells A. EGF receptor. Int J Biochem Cell Biol. 31:637–643, 1999 74. Arteaga CL. The epidermal growth factor receptor: from mutant oncogene in nonhuman cancers to therapeutic target in human neoplasia. J Clin Oncol Suppl. 18:32S–40S, 2001

452

C.D. Weekes and M. Hidalgo

75. Pinkas-Kramarski R, Soussan L, Waterman H, Levkowitz G, Alroy I, Klapper L, Lavi S, Seger R, Ratzkin BJ, Sela M, Yarden Y. Diversification of Neu differentiation factor and epidermal growth factor signaling by combinatorial receptor interactions. EMBO J. 15:2452–2467, 1996 76. Yarden Y, Sliwkowski MS. Untangling the ErbB signaling network. Nat Rev Mol Cell Biol. 2:127–137, 2001 77. Graus-Porta D, Beerli RR, Daly JM, Hynes NE. ErbB-2, the preferred heterodimerization partner of all ErbB receptors, is a mediator of lateral signaling. EMBO J. 16:1647–1655, 1997 78. Worthylake R, Opresko LK, Wiley HS. ErbB-2 amplification inhibits down-regulation and induces constitutive activation of both ErbB-2 and epidermal growth factor receptors. J Biol Chem. 247:8865–8874, 1999 79. Schlessinger J. Cell signaling by receptor tyrosine kinases. Cell. 103:211–225, 2000 80. Blenis J. Signal transduction via the MAP kinases: proceed at your RSK. Proc Natl Acad Sci U S A. 90:5889–5892, 1993 81. Burgering BM, Coffer PJ. Protein kinase B (c-Akt) in phosphatidylinositol-3-OH kinase signal transduction. Nature. 376:599–602, 1995 82. Lewis TS, Shapiro PS, Ahn NG. Signal transduction through MAP kinase cascades. Adv Cancer Res. 74:49–139, 1998 83. Cantley LC. The phosphoinositide 3-kinase pathway. Science. 296:1655–1657, 2002 84. Grandis JR, Melhem MF, Gooding WE, Day R, Holst VA, Wagener MM, Drenning SD, Tweardy DJ. Levels of TGF-alpha and EGFR protein in head and neck squamous cell carcinoma and patient survival. J Natl Cancer Inst. 90:824–832, 1998 85. Nishikawa R, Ji XD, Harmon RC, Lazar CS, Gill GN, Cavenee WK, Huang HJ. A mutant epidermal growth factor receptor common in human glioma confers enhanced tumorigenicity. Proc Natl Acad Sci U S A. 91:7727–7731, 1994 86. Cunningham D, Humblet Y, Siena S, Khayat D, Bleiberg H, Santoro A, Bets D, Mueser M, Harstrick A, Verslype C, Chau I, Van Cutsem E. Cetuximab monotherapy and cetuximab plus irinotecan in irinotecan-refractory metastatic colorectal cancer. N Engl J Med. 22:337– 345, 2004 87. Chung KY, Shia J, Kemeny NE, Shah M, Schwartz GK, Tse A, Hamilton A, Pan D, Schrag D, Schwartz L, Klimstra DS, Fridman D, Kelsen DP, Saltz LB. Cetuximab shows activity in colorectal cancer patients with tumors that do not express the epidermal growth factor receptor by immunohistochemistry. J Clin Oncol. 9:1803–1810, 2005 88. Van Cutsem E, Köhne C-H, Hitre E, Zaluski J, Chien C-R, Makhson A, D’Haens G, Pintér T, Lim R, Bodoky G, Roh J, Folprecht G, Ruff P, Stroh C, Tejpar S, Schlichting M, Nippgen J, Rougier P. Cetuximab and chemotherapy as initial treatment for metastatic colorectal cancer. N Engl J Med. 360:1408–1417, 2009 89. Bokemeyer C, Bondarenko I, Makhson A, Hartmann JT, Aparicio J, de Braud F, Donea S, Ludwig H, Schuch G, Stroh C, Loos AH, Zubel A, Koralewski P. Fluorouracil, leucovorin, and oxaliplatin with and without cetuximab in the first-line treatment of metastatic colorectal cancer. J Clin Oncol. 27:663–671, 2009 90. Bonner JA, Harari PM, Giralt J, Azarnia N, Shin DM, Cohen RB, Jones CU, Sur R, Raben D, Jassem J, Ove R, Kies MS, Baselga J, Youssoufian H, Amellal N, Rowinsky EK, Ang KK. Radiotherapy plus cetuximab for squamous-cell carcinoma of the head and neck. N Engl J Med. 354:567–578, 2006 91. Vermorken JB, Mesia R, Rivera F, Remenar E, Kawecki A, Rottey S, Erfan J, Zabolotnyy D, Kienzer HR, Cupissol D, Peyrade F, Benasso M, Vynnychenko I, De Raucourt D, Bokemeyer C, Schueler A, Amellal N, Hitt R. Platinum-based chemotherapy plus cetuximab in head and neck cancer. N Engl J Med. 359:1116–1127, 2008 92. Gibson TB, Ranganathan A, Grothey A. Randomized phase III trial results of panitumumab, a fully human anti-epidermal growth factor receptor monoclonal antibody, in metastatic colorectal cancer. Clin Colorectal Cancer. 6:29–31, 2006 93. Piccart-Gebhart MJ, Procter M, Leyland-Jones B, Goldhirsch A, Untch M, Smith I, Gianni L, Baselga J, Bell R, Jackisch C, Cameron D, Dowsett M, Barrios CH, Steger G, Huang CS,

15  Targeted Therapeutics in Cancer Treatment

453

Andersson M, Inbar M, Lichinitser M, Lang I, Nitz U, Iwata H, Thomssen C, Lohrisch C, Suter TM, Ruschoff J, Suto T, Greatorex V, Ward C, Straehle C, McFadden E, Dolci MS, Gelber RD. Trastuzumab after adjuvant chemotherapy in HER2-positive breast cancer. N Engl J Med. 353:1659–1672, 2005 94. Romond EH, Perez EA, Bryant J, Suman VJ, Geyer CE Jr, Davidson NE, Tan-Chiu E, Martino S, Paik S, Kaufman PA, Swain SM, Pisansky TM, Fehrenbacher L, Kutteh LA, Vogel VG, Visscher DW, Yothers G, Jenkins RB, Brown AM, Dakhil SR, Mamounas EP, Lingle WL, Klein PM, Ingle JN, Wolmark N. Trastuzumab plus adjuvant chemotherapy for operable HER2-positive breast cancer. N Engl J Med. 353:1673–1684, 2005 95. Van Cutsem E, Kang Y, Chung H, Shen L, Sawaki A, Lordick F, Hill J, Lehle M, Feyereislova A, Bang Y. Efficacy results from the ToGA trial: a phase III study of trastuzumab added to standard chemotherapy (CT) in first-line human epidermal growth factor receptor 2 (HER2)positive advanced gastric cancer (GC). J Clin Oncol. 27:18s, 2009 96. Sirotnak FM, Zakowski MF, Miller VA, Scher HI, Kris MG. Efficacy of cytotoxic agents against human tumor xenografts is markedly enhanced by coadministration of ZD1839 (Iressa), an inhibitor of EGFR tyrosine kinase. Clin Cancer Res. 6:4885–4892, 2000 97. Pollack VA, Savage DM, Baker DA, Tsaparikos KE, Sloan DE, Moyer JD, Barbacci EG, Pustilnik LR, Smolarek TA, Davis JA, et al. Inhibition of epidermal growth factor receptorassociated tyrosine phosphorylation in human carcinomas with CP-358,774: dynamics of receptor inhibition in situ and antitumor effects in athymic mice. J Pharmacol Exp Ther. 291:739–748, 1999 98. Moyer JD, Barbacci EG, Iwata KK, Arnold L, Boman B, Cunningham A, DiOrio C, Doty J, Morin AJ, Moyer MP, et al. Induction of apoptosis and cell cycle arrest by CP-358,774, an inhibitor of epidermal growth factor receptor tyrosine kinase. Cancer Res. 57: 4838–4848, 1997 99. Baselga J, Rischin D, Ranson M, Calvert H, Raymond E, Keiback DG, Kaye SB, Gianni L, Harris A, Bjork T, et  al. Phase I safety, pharmacokinetic, and pharmacodynamic trial of ZD1839, a selective oral epidermal growth factor receptor tyrosine kinase inhibitor, in patients with five selected solid tumor types. J Clin Oncol. 20:4292–4302, 2002 100. Herbst RS, Maddox AM, Rothenberg ML, Small EJ, Rubin EH, Baselga J, Rojo F, Hong WK, Swaisland H, Averbuch SD, et al. Selective oral epidermal growth factor receptor tyrosine kinase inhibitor ZD1839 is generally well-tolerated and has activity in non-­ small-cell lung cancer and other solid tumors: results of a phase I trial. J Clin Oncol. 20:3815–3825, 2002 101. Ranson M, Hammond LA, Ferry D, Kris M, Tullo A, Murray PI, Miller V, Averbuch S, Ochs J, Morris C, et al. ZD1839, a selective oral epidermal growth factor receptor-tyrosine kinase inhibitor, is well tolerated and active in patients with solid, malignant tumors: results of a phase I trial. J Clin Oncol. 20:2240–2250, 2002. 102. Hidalgo M, Siu LL, Nemunaitis J, Rizzo J, Hammond LA, Takimoto C, Eckhardt SG, Tolcher A, Britten CD, Denis L, et  al. Phase I and pharmacologic study of OSI-774, an epidermal growth factor receptor tyrosine kinase inhibitor, in patients with solid malignancies. J Clin Oncol. 19:3267–3279, 2001 103. Fukuoka M, Yano S, Giaccone G, Tamura T, Nakagawa K, Douillard JY, Nishiwaki Y, Vansteenkiste J, Kudoh S, Rischin D, Eek R, et al. Multi-institutional randomized phase II trial of gefitinib for previously treated patients with advanced non-small-cell lung cancer. J Clin Oncol. 21:2237–2246, 2003 104. Kris MG, Natale RB, Herbst RS, Lynch TJ, Jr, Prager D, Belani CP, Schiller JH, Kelly K, Spiridonidis H, Sandler A, et al. Efficacy of gefitinib, an inhibitor of the epidermal growth factor receptor tyrosine kinase, in symptomatic patients with non-small cell lung cancer: a randomized trial. JAMA. 290:2149–2158, 2004. 105. Giaccone G, Herbst RS, Manegold C, Scagliotti G, Rosell R, Miller V, Natale RB, Schiller JH, Von Pawel J, Pluzanska A, et al. Gefitinib in combination with gemcitabine and cisplatin in advanced non-small-cell lung cancer: a phase III trial – INTACT 1. J Clin Oncol. 22:777–784, 2004

454

C.D. Weekes and M. Hidalgo

106. Herbst R, Prager D, Hermann R, Fehrenbacher L, Johnson BE, Sandler A, Kris MG, Tran HT, Klein P, Li X. et al. TRIBUTE – a phase III trial of erlotinib HCl (OSI-774) combined with carboplatin and paclitaxel (CP) chemotherapy in advanced non-small cell lung cancer (NSCLC). J Clin Oncol. 23:5892–5899, 2005 107. Kim ES, Hirsh V, Mok T, Socinski MA, Gervais R, Wu YL, Li LY, Watkins CL, Sellers MV, Lowe ES, Sun Y, Liao ML, Osterlind K, Reck M, Armour AA, Shepherd FA, Lippman SM, Douillard JY. Gefitinib versus docetaxel in previously treated non-small-cell lung cancer (INTEREST): a randomised phase III trial. Lancet. 372:1809–1818, 2008 108. Mok T, et al. Phase III, randomized, open-label, first line study of gefitinib vs. carboplatin/ paclitaxel in clinically selected patients with advanced non-small-cell lung cancer (NSCLC) (iPASS). Ann Oncol. 10(Suppl 8), 2008; abstr LBA2 109. Shepherd F, Pereira J, Ciuleanu T, Tan EH, Hirsh V, Thongprasert S, Bezjak A, Tu D, Santabarbara P, Seymour L. A randomized placebo-controlled trial of erlotinib in patients with advanced non-small cell lung cancer (NSCLC) following failure of 1st line or 2nd line chemotherapy. A National Cancer Institute of Canada Clinical Trials Group (NCIC CTG) trial. J Clin Oncol Suppl. 22:14S, 2004 110. Moore M, Goldstein D, Hamm J, Figer A, Hecht JR, Gallinger S, Au HJ, Murawa P, Walde D, Wolff RA, Campos D, Lim R, Ding K, Clark G, Voskoglou-Nomikos G, Ptasynski M and Parulekar W. Erlotinib plus gemcitabine compared with gemcitabine alone in patients with advanced pancreatic cancer: a phase III trial of the National Cancer Institute of Canada Clinical Trials Group. J Clin Oncol. 25:1960–1966, 2007 111. Burris H, Taylor C, Jones S, Pandite L, Smith DA, Versola M, Stead A, Whitehead B, Spector N, Wilding G, et al. A phase I study of GW572016 in patients with solid tumors. J Clin Oncol Suppl. 22:248, 2003 112. Erlichman C, Hidalgo M, Boni JP, Martins P, Quinn SE, Zacharchuk C, Amorusi P, Adjei AA, Rowinsky EK. Phase I study of EKB-569, an irreversible inhibitor of the epidermal growth factor receptor, in patients with advanced solid tumors. J Clin Oncol. 24:2252–2260, 2006 113. Geyer CE, Forster J, Lindquist D, Chan S, Romieu CG, Pienkowski T, Jagiello-Gruszfeld A, Crown J, Chan A, Kaufman B, Skarlos D, Campone M, Davidson N, Berger M, Oliva C, Rubin SD, Stein S, Cameron D. Lapatinib plus capecitabine for HER2-positive advanced breast cancer. N Engl J Med. 355:2733–2743, 2006 114. Jimeno A, Rubio-Viquiera B, Amador ML, Oppenheimer D, Bouraroud N, Kuleza P, Sebastiani V, Maitra A, Hidalgo M. Epidermal growth factor receptor dynamics influences response to epidermal growth factor targeted agents. Cancer Res. 65:3003–3010, 2005 115. Yee D. Targeting insulin-like growth factor pathways. Br J Cancer. 94:465–468, 2006 116. Samani AA, Yakar S, LeRoith D, Brodt P. The role of the IGF system in cancer growth and metastasis: overview and recent insights. Endocr Rev. 28:20–47, 2007 117. Boulle N, Logie A, Gicquel C, Perin L, Le Bouc Y. Increased levels of insulin-like growth factor II (IGF-II) and IGF-binding protein-2 are associated with malignancy in sporadic adrenocortical tumors. J Clin Endocrinol Metab. 83:1713–1720, 1998 118. Butler AA, Blakesley VA, Poulaki V, et  al. Stimulation of tumor growth by recombinant human insulin-like growth factor-I (IGF-I) is dependent on the dose and the level of IGF-I receptor expression. Cancer Res. 58:3021–3027, 1998 119. Haluska P, Shaw HM, Batzel GN, et al. Phase I dose escalation study of the antiinsulin-like growth factor-I receptor monoclonal antibody CP751,871 in patients with refractory solid tumors. Clin Cancer Res. 13:5834–5840, 2007 120. de Bono JS, Attard G, Adjei A, et  al. Potential applications for circulating tumor cells expressing the insulin-like growth factor-I receptor. Clin Cancer Res, 13:3611–3616, 2007 121. Karp DD, Paz-Ares LG, Novello S, et  al. CP751,871 in combination with paclitaxel and carboplatin in squamous NSCLC. J Clin Oncol. 26:8015, 2008 122. Bottaro DP, Rubin JS, Faletto DL, et al. Identification of the hepatocyte growth factor receptor as c-met proto-oncogene product. Science. 251:802–804, 1991 123. Birchmeier C, Birchmeier W, Gherardi E, VandeWoude GF. Met, metastasis, motility and more. Nat Rev Mol Cell Biol. 4:915–925, 2003

15  Targeted Therapeutics in Cancer Treatment

455

124. Ponzetto C, Bardelli A, Maina F, et al. A novel recognition motif for phosphatidylinositol 3-kinase binding mediates its association with the hepatocyte growth factor. Mol Cell Biol. 13:4600–4608, 1993 125. Ponzetto C, Bardelli A, Zhen Z, et al. A multifunctional docking site mediates signaling and transformation by the hepatocyte growth factor/scatter factor receptor family. Cell. 77:261–271, 1994 126. Schmidt L, Duh FM, Chen F, et al. Germline and somatic mutations in the tyrosine kinase domain of the MET proto-oncogene in papillary renal carcinomas. Nat Genet. 16:68–73, 1997 127. Gordon MS, Mendelson D, Sweeney C, et  al. Interim results from a first-in-human study with AMG102, a fully human monoclonal antibody that neutralizes hepatocyte growth factor (HGF), the ligand to c-Met receptor, in patients with advanced solid tumors. J Clin Oncol. 18S:3551, 2007 128. Jeay S, Munshi N, Hill J, et  al. ARQ197, a highly selective small molecule inhibitor of c-Met, with selective antitumor properties in a broad spectrum of human cancer cells. Presented at 98th AACR Annual Meeting, 2007; abstr 3525 129. Garcia A, Rosen L, Cunningham CC, et al. Phase I study of ARQ197, a selective inhibitor of the c-Met RTK in patients with metastatic solid tumors reaches recommended phase 2 dose. J Clin Oncol. 25, 2007; abstr 3525 130. Paez JG, Janne PA, Lee JC, Tracy S, Greulich H, Gabriel S, Herman P, Kaye FJ, Lindeman N, Boggon TJ, et al. EGFR mutations in lung cancer: correlation with clinical response to gefitinib therapy. Science. 304:1497–1500, 2004 131. Pao W, Miller V, Zakowski M, Doherty J, Politi K, Sarkaria I, Singh B, Heelan R, Rusch V, Fulton L, et al. EGF receptor gene mutations are common in lung cancers from never smokers and are associated with sensitivity of tumors to gefitinib and erlotinib. Proc Natl Acad Sci U S A. 101:13306–13311, 2004 132. Nicholson RI, Gee JM, Harper ME. EGFR and cancer prognosis. Eur J Cancer Suppl. 37:S9–S15, 2001 133. Jackman DM, Miller VA, Cioffredi LA, Yeap BY, Jänne PA, Riely GJ, Ruiz MG, Giaccone G, Sequist LV, Johnson BE. Impact of epidermal growth factor receptor and KRAS mutations on clinical outcomes in previously untreated non-small cell lung cancer patients: results of an online tumor registry of clinical trials. Clin Cancer Res. 15(16):5267–5273, 2009 134. Zhou HY, Lee JH, et al. An orally available small molecule inhibitor of c-Met, PF2341066, exhibits cytoreductive antitumor efficacy through antiproliferative and antiangiogenic mechanisms. Cancer Res. 67:4408–4417, 2007 135. Jimeno A, Messersmith WA, Hirsch FR, Franklin WA, Eckhardt SG. KRAS mutations and sensitivity to epidermal growth factor receptor inhibitors in colorectal cancer: practical application of patient selection. J Clin Oncol. 27:1130–1136, 2009 136. Zeng Q, Chen S, You Z, et al. Hepatocyte growth factor inhibits anoikis in head and neck squamous cell carcinoma cells by activation of ERK and Akt signaling independent of NFkB. J Biol Chem. 277:25203–25208, 2002 137. Tulasne D, Foveau B. The shadow of death on the MET tyrosine kinase receptor. Cell Death Differ. 15:427–434, 2008 138. Bean J, Brennan C, Shih JY, Riely G, Viale A, Wang L, Chitale D, Motoi N, Szoke J, Broderick S, Balak M, Chang WC, Yu CJ, Gazdar A, Pass H, Rusch V, Gerald W, Huang SF, Yang PC, Miller V, Ladanyi M, Yang CH, Pao W. MET amplification occurs with or without T790M mutations in EGFR mutant lung tumors with acquired resistance to gefitinib or erlotinib. Proc Natl Acad Sci U S A. 104:20932–20937, 2007 139. Engelman JA, Zejnullahu K, Mitsudomi T, Song Y, Hyland C, Park JO, Lindeman N, Gale CM, Zhao X, Christensen J, Kosaka T, Holmes AJ, Rogers AM, Cappuzzo F, Mok T, Lee C, Johnson BE, Cantley LC, Jänne PA. MET amplification leads to gefitinib resistance in lung cancer by activating ERBB3 signaling. Science. 316:1039–1043, 2007 140. Sabatini DM. mTOR and cancer: insights into a complex relationship. Nat Rev Cancer. 6:729–734, 2006

456

C.D. Weekes and M. Hidalgo

141. Liaw D, Marsh DJ, Li J, et al. Germline mutations of the PTEN gene in Cowden disease, an inherited breast and thyroid cancer syndrome. Nat Genet. 16:64–67, 1997 142. Shaw RJ, Bardeesy N, Manning BD, et al. The LKB1 tumor suppressor negatively regulates mTOR signaling. Cancer Cell. 6:91–99, 2004 143. Inoki K, Zhu T, Guan KL. TSC2 mediates cellular energy response to control cell growth and survival. Cell. 115:577–590, 2003 144. Johannessen CM, Reczek EE, James MF, et al. The NF1 tumor suppressor critically ­regulates TSC2 and mTOR. Proc Natl Acad Sci U S A. 102:8573–8578, 2005 145. Koltin Y, Faucette L, Bergsma DJ, Levy MA, Cafferkey R, Koser PL, Johnson RK, Livi GP. Rapamycin sensitivity in Saccharomyces cerevisiae is mediated by a peptidyl-prolyl cis-trans isomerase related to human FK506-binding protein. Mol Cell Biol. 11:1718–1723, 1991 146. Fruman DA, Wood MA, Gjertson CK, Katz HR, Burakoff SJ, Bierer BE. FK506 binding protein 12 mediates sensitivity to both FK506 and rapamycin in murine mast cells. Eur J Immunol. 25:563–571, 1995 147. Seufferlein T, Rozengurt E. Rapamycin inhibits constitutive p70s6k phosphorylation, cell proliferation, and colony formation in small cell lung cancer cells. Cancer Res. 56:3895– 3897, 1996 148. Grewe M, Gansauge F, Schmid RM, Adler G, Seufferlein T. Regulation of cell growth and cyclin D1 expression by the constitutively active FRAP-p70s6K pathway in human pancreatic cancer cells. Cancer Res. 59:3581–3587, 1999 149. Hashemolhosseini S, Nagamine Y, Morley SJ, Desrivieres S, Mercep L, Ferrari S. Rapamycin inhibition of the G1 to S transition is mediated by effects on cyclin D1 mRNA and protein stability. J Biol Chem. 273:14424–14429, 1998 150. Huang S, Liu LN, Hosoi H, Dilling MB, Shikata T, Houghton PJ. p53/p21CIP1 cooperate in enforcing rapamycin-induced G1 arrest and determine the cellular response to rapamycin. Cancer Res. 61:3373–3381, 2001 151. Guba M, von Breitenbuch P, Steinbauer M, Koehl G, Flegel S, Hornung M, Burns CJ, Zuelke C, Farkas S, Anthuber M, et al. Rapamycin inhibits primary and metastatic tumor growth by antiangiogenesis: involvement of vascular endothelial growth factor. Nat Med. 8:128–135, 2002 152. Yu Y, Sato JD. MAP kinases, phosphatidylinositol 3-kinase, and p70 S6 kinase mediate the mitogenic response of human endothelial cells to vascular endothelial growth factor. J Cell Physiol. 178:235–246, 1999. 153. O’Donnell A, Faivre S, Burris HA, 3rd, Rea D, Papadimitrakopoulou V, Shand N, Lane HA, Hazell K, Zoellner U, Kovarik JM, Brock C, Jones S, Raymond E, Judson I. Phase I pharmacokinetic and pharmacodynamic study of the oral mammalian target of rapamycin inhibitor everolimus in patients with advanced solid tumors. J Clin Oncol. 26:1588–1595, 2008 154. Tabernero J, Rojo F, Calvo E, Burris H, Judson I, Hazell K, Martinelli E, Ramon y Cajal S, Jones S, Vidal L, Shand N, Macarulla T, Ramos FJ, Dimitrijevic S, Zoellner U, Tang P, Stumm M, Lane HA, Lebwohl D, Baselga J. Dose- and schedule-dependent inhibition of the mammalian target of rapamycin pathway with everolimus: a phase I tumor pharmacodynamic study in patients with advanced solid tumors. J Clin Oncol. 26:1603–1610, 2008 155. Hartford CM, Desai AA, Janisch L, Karrison T, Rivera VM, Berk L, Loewy JW, Kindler H, Stadler WM, Knowles HL, Bedrosian C, Ratain MJ. A phase I trial to determine the safety, tolerability, and maximum tolerated dose of deforolimus in patients with advanced malignancies. Clin Cancer Res. 15:1428–1434, 2009 156. Mita MM, Mita AC, Chu QS, Rowinsky EK, Fetterly GJ, Goldston M, Patnaik A, Mathews L, Ricart AD, Mays T, Knowles H, Rivera VM, Kreisberg J, Bedrosian CL, Tolcher AW. Phase I trial of the novel mammalian target of rapamycin inhibitor deforolimus (AP23573; MK-8669) administered intravenously daily for 5 days every 2 weeks to patients with advanced malignancies. J Clin Oncol. 26:361–367, 2008 157. Hidalgo M, Buckner JC, Erlichman C, Pollack MS, Boni JP, Dukart G, Marshall B, Speicher L, Moore L, Rowinsky EK. A phase I and pharmacokinetic study of temsirolimus (CCI-779)

15  Targeted Therapeutics in Cancer Treatment

457

administered intravenously daily for 5 days every 2 weeks to patients with advanced cancer. Clin Cancer Res. 12:5755–5763, 2006 158. Raymond E, Alexander J, Faivre S, Vera K, Materman E, Boni J, Leister C, Korth-Bradley J, Hanauske A, Jean-Pierre A. Safety and pharmacokinetics of escalated doses of weekly intravenous infusion of CCI-779, a novel mTOR inhibitor, in patients with cancer. J Clin Oncol. 22:2336–2347, 2004 159. Hudes G, Carducci M, Tomczak P, Dutcher J, Figlin R, Kapoor A, Staroslawska E, Sosman J, McDermott D, Bodrogi I, Kovacevic Z, Lesovoy V, Schmidt-Wolf IG, Barbarash O, Gokmen E, O’Toole T, Lustgarten S, Moore L, Motzer RJ; Global ARCC Trial. Temsirolimus, interferon alfa, or both for advanced renal-cell carcinoma. N Engl J Med. 356:2271–2281, 2007 160. Motzer RJ, Escudier B, Oudard S, Hutson TE, Porta C, Bracarda S, Grünwald V, Thompson JA, Figlin RA, Hollaender N, Urbanowitz G, Berg WJ, Kay A, Lebwohl D, Ravaud A; RECORD-1 Study Group. Efficacy of everolimus in advanced renal cell carcinoma: a doubleblind, randomised, placebo-controlled phase III trial. Lancet. 372:449–456, 2008 161. Kim WY, Kaelin WG. Role of VHL gene mutation in human cancer. J Clin Oncol. 22:4991– 5004, 2004 162. Hess G, Herbrecht R, Romaguera J, Verhoef G, Crump M, Gisselbrecht C, Laurell A, Offner F, Strahs A, Berkenblit A, Hanushevsky O, Clancy J, Hewes B, Moore L, Coiffier B. Phase III study to evaluate temsirolimus compared with investigator’s choice therapy for the treatment of relapsed or refractory mantle cell lymphoma. J Clin Oncol. 27:3822–3829, 2009 163. Oza AM, Elit L, Provencher D, Biagi JJ, Panasci L, Sederias J, Dancey JE, Tsao MS, Eisenhauer EA. A phase II study of temsirolimus (CCI-779) in patients with metastatic and/ or locally advanced recurrent endometrial cancer previously treated with chemotherapy: NCIC CTG IND 160b. J Clin Oncol. 26:18S, 2008; abstr 5516 164. Chang SM, Kuhn J, Wen P, Greenberg H, Schiff D, Conrad C, Fink K, Robins HI, Cloughesy T, De Angelis L, et al. Phase I/Pharmacokinetic study of CCI-779 in patients with recurrent malignant glioma on enzyme-induced antiepileptic drugs. Invest New Drugs. 22:427–435, 2004 165. Neshat MS, Mellinghoff IK, Tran C, Stiles B, Thomas G, Petersen R, Frost P, Gibbons JJ, Wu H, Sawyers CL. Enhanced sensitivity of PTEN-deficient tumors to inhibition of FRAP/ mTOR. Proc Natl Acad Sci U S A. 98:10314–10319, 2001 166. Rajalingam K, Schreck R, Rapp UR, Albert S. Ras oncogenes and their downstream targets. Biochim Biophys Acta. 1773:1177–1195, 2007 167. Zhu K, Hamilton AD, Sebti SM. Farnesyl transferase inhibitors as anticancer agents: current status. Curr Opin Investig Drugs. 4:1428–1435, 2004 168. Van Cutsem E, van de Velde H, Karasek P, Oettle H, Vervenne WL, Szawlowski A, Schoffski P, Post S, Verslype C, Neumann H, Safran H, Humblet Y, Perez Ruixo J, Ma Y, Von Hoff D. Phase III trial of gemcitabine plus tipifarnib compared with gemcitabine plus placebo in advanced pancreatic cancer. J Clin Oncol. 22:1430–1438, 2004 169. End DW, Smets G, Todd AV, Applegate TL, Fuery CJ, Angibaud P, Venet M, Sanz G, Poignet H, Skrzat S, Devine A, Wouters W, Bowden C. Characterization of the antitumor effects of the selective farnesyl protein transferase inhibitor R115777 in vivo and in vitro. Cancer Res. 61:131–137, 2001 170. Davies H, Bignell GR, Cox C, Stephens P, Edkins S, Clegg S, Teague J, Woffendin H, Garnett MJ, Bottomley W, Davis N, Dicks E, Ewing R, Floyd Y, Gray K, Hall S, Hawes R, Hughes J, Kosmidou V, Menzies A, Mould C, Parker A, Stevens C, Watt S, Hooper S, Wilson R, Jayatilake H, Gusterson BA, Cooper C, Shipley J, Hargrave D, Pritchard-Jones K, Maitland N, Chenevix-Trench G, Riggins GJ, Bigner DD, Palmieri G, Cossu A, Flanagan A, Nicholson A, Ho JW, Leung SY, Yuen ST, Weber BL, Seigler HF, Darrow TL, Paterson H, Marais R, Marshall CJ, Wooster R, Stratton MR, Futreal PA. Mutations of BRAF gene in human cancers. Nature. 417:949–954, 2002 171. Strumberg D, Richly H, Hilger RA. Phase I clinical and pharmacokinetic study of the Novel Raf kinase and vascular endothelial growth factor receptor inhibitor BAY 43-9006 in patients with advanced refractory solid tumors. J Clin Oncol. 23:965–972, 2005

458

C.D. Weekes and M. Hidalgo

172. Ratain MJ, Eisen T, Stadler WM, et al. Phase II placebo-controlled randomized discontinuation trial of sorafenib in patients with metastatic renal cell carcinoma. J Clin Oncol. 24:2505–2512, 2006 173. Llovet JM, Ricci S, Mazzaferro V, et  al. Sorefenib in advanced hepatocellular carcinoma. N Engl J Med. 359:378–390, 2008 174. Escudier B, Eisen T, Stadler WM, Szczylik C, Oudard S, Siebels M, Negrier S, Chevreau C, Solska E, Desai AA, Rolland F, Demkow T, Hutson TE, Gore M, Freeman S, Schwartz B, Shan M, Simantov R, Bukowski RM; TARGET Study Group. Sorafenib in advanced clearcell renal-cell carcinoma. N Engl J Med. 356:125–134, 2007 175. Motzer RJ, Hutson TE, Tomczak P, Michaelson MD, Bukowski RM, Oudard S, Negrier S, Szczylik C, Pili R, Bjarnason GA, Garcia-del-Muro X, Sosman JA, Solska E, Wilding G, Thompson JA, Kim ST, Chen I, Huang X, Figlin RA. Overall survival and updated results for sunitinib compared with interferon alfa in patients with metastatic renal cell carcinoma. J Clin Oncol. 27:3584–3590, 2009 176. Escudier B, Pluzanska A, Koralewski P, Ravaud A, Bracarda S, Szczylik C, Chevreau C, Filipek M, Melichar B, Bajetta E, Gorbunova V, Bay JO, Bodrogi I, Jagiello-Gruszfeld A, Moore N; AVOREN Trial investigators. Bevacizumab plus interferon alfa-2a for treatment of metastatic renal cell carcinoma: a randomised, double-blind phase III trial. Lancet. 370:2103–2111, 2007 177. Rini BI, Halabi S, Rosenberg JE, Stadler WM, Vaena DA, Ou SS, Archer L, Atkins JN, Picus J, Czaykowski P, Dutcher J, Small EJ. Bevacizumab plus interferon alfa compared with interferon alfa monotherapy in patients with metastatic renal cell carcinoma: CALGB 90206. J Clin Oncol. 26:5422–5428, 2008 178. Cowley S, Paterson H, Kemp P, Marshall CJ. Activation of MAP kinase kinase is necessary and sufficient for PC12 differentiation and for transformation of NIH 3T3 cells. Cell. 77:841–852, 1994 179. Mansour SJ, Matten WT, Hermann AS, Candia JM, Rong S, Fukasawa K, Vande Woude GF, Ahn NG. Transformation of mammalian cells by constitutively active MAP kinase kinase. Science. 265:966–970, 1994 180. Brott BK, Alessandrini A, Largaespada DA, Copeland NG, Jenkins NA, Crews CM, Erikson RL. MEK2 is a kinase related to MEK1 and is differentially expressed in murine tissues. Cell Growth Differ. 4:921–929, 1993 181. Ohren JF, Chen H, Pavlovsky A, Whitehead C, Zhang E, Kuffa P, Yan C, McConnell P, Spessard C, Banotai C, et al. Structures of human MAP kinase kinase 1 (MEK1) and MEK2 describe novel noncompetitive kinase inhibition. Nat Struct Mol Biol. 11:1192–1197, 2004 182. Seger R, Ahn NG, Posada J, Munar ES, Jensen AM, Cooper JA, Cobb MH, Krebs EG. Purification and characterization of mitogen-activated protein kinase activator(s) from epidermal growth factor-stimulated A431 cells. J Biol Chem. 267:14373–14381, 2003 183. Zheng CF, Guan KL. Cloning and characterization of two distinct human extracellular ­signal-regulated kinase activator kinases, MEK1 and MEK2. J Biol Chem. 268: 11435–11439, 1993 184. Giroux S, Tremblay M, Bernard D, Cardin-Girard JF, Aubry S, Larouche L, Rousseau S, Huot J, Landry J, Jeannotte L, Charron J. Embryonic death of Mek1-deficient mice reveals a role for the kinase in angiogenesis in the labyrinthine region of the placenta. Curr Biol. 9:369–372, 2002 185. Belanger LF, Roy S, Tremblay M, Brott B, Steff AM, Mourad W, Hugo P, Erikson R, Charron J. Mek2 is dispensible for mouse growth and development. Mol Cell Biol. 23: 4778–4787, 2003 186. Sebolt-Leopold JS, Herrera R. Targeting the mitogen-activated protein kinase cascade to treat cancer. Nat Rev Cancer. 4:937–947, 2004 187. Alessi DR, Cuenda A, Cohen P, Dudley DT, Saltiel AR. PD 098059 is a specific inhibitor of the activation of mitogen-activated protein kinase kinase in vitro and in vivo. J Biol Chem. 270:27489–27494, 1995

15  Targeted Therapeutics in Cancer Treatment

459

188. Favata MF, Horiuchi KY, Manos EJ, Daulerio AJ, Stradley DA, Feeser WS, Van Dyk DE, Pitts WJ, Earl RA, Hobbs F, et  al. Identification of a novel inhibitor of mitogen-activated protein kinase kinase. J Biol Chem. 273:18623–18632, 1998 189. Allen LF, Sebolt-Leopold J, Meyer MB. CI-1040 (PD184352), a targeted signal transduction inhibitor of MEK (MAPKK). Semin Oncol Suppl. 16:105–116, 2003 190. Lorusso PM, Krishnamurthi S, Rinehart JR, Nabell L, Croghan G, Varterasian M, Sadis SS, Menon SS, Leopold J, Meyer MB, et al. A phase 1-2 clinical study of a second generation oral MEK inhibitor, PD 0325901 in patients with advanced cancer. J Clin Oncol Suppl. 23:16S, 2005 191. Rinehart J, Adjei AA, Lorusso PM, Waterhouse D, Hecht JR, Natale RB, Hamid O, Varterasian M, Asbury P, Kaldjian P, et al. Multicenter phase II study of the oral MEK inhibitor, CI-1040, in patients with advanced non-small-cell lung, breast, colon, and pancreatic cancer. J Clin Oncol. 22:4456–4462, 2004 192. Adjei AA, Cohen RB, Franklin W, Morris C, Wilson D, Molina JR, Hanson LJ, Gore L, Chow L, Leong S, Maloney L, Gordon G, Simmons H, Marlow A, Litwiler K, Brown S, Poch G, Kane K, Haney J, Eckhardt SG. Phase I pharmacokinetic and pharmacodynamic study of the oral, small-molecule mitogen-activated protein kinase kinase 1/2 inhibitor AZD6244 (ARRY142886) in patients with advanced cancers. J Clin Oncol. 26:2139–2146, 2008 193. Ishizawar R, Parsons SJ. c-Src and cooperating partners in human cancer. Cancer Cell. 6:209–214, 2004. 194. Sawyer T, Boyce B, Dalgarno D, Iuliucci J. Src inhibitors: genomics to therapeutics. Expert Opin Investig Drugs. 10:1327–1344, 2001 195. Biscardi JS, Ishizawar RC, Silva CM, Parsons SJ. Tyrosine kinase signaling in breast cancer: epidermal growth factor receptor and c-Src interactions in breast cancer. Breast Cancer Res. 2:203–210, 2000 196. Biscardi JS, Tice DA, Parsons SJ. c-Src, receptor tyrosine kinases, and human cancer. Adv Cancer Res. 76:61–119, 1999 197. Biscardi JS, Maa MC, Tice DA, Cox ME, Leu TH, Parsons SJ. c-Src-mediated phosphorylation of the epidermal growth factor receptor on Tyr845 and Tyr1101 is associated with modulation of receptor function. J Biol Chem. 274:8335–8343, 1999 198. Duxbury MS. Inhibition of SRC tyrosine kinase impairs inherent and acquired gemcitabine resistance in human pancreatic adenocarcinoma cells. Clin Cancer Res. 10:2307–2318, 2004 199. Zhang Q, Thomas SM, Xi S, Smithgall TE, Siegfried JM, Kamens J, Gooding WE, Grandis JR. SRC family kinases mediate epidermal growth factor receptor ligand cleavage, proliferation, and invasion of head and neck cancer cells. Cancer Res. 64:6166–6173, 2004 200. Shakespeare W, Yang M, Bohacek R, Cerasoli F, Stebbins K, Sundaramoorthi R, Azimioara M, Vu C, Pradeepan S, Metcalf C, et al. Structure-based design of an osteoclast-selective, nonpeptide src homology 2 inhibitor with in vivo antiresorptive activity. Proc Natl Acad Sci U S A. 97:9373–9378, 2000 201. Workman P. Overview: translating Hsp90 biology into Hsp90 drugs. Curr Cancer Drug Targets. 3:297–300, 2003 202. Golas JM, Arndt K, Etienne C, Lucas J, Nardin D, Gibbons J, Frost P, Ye F, Boschelli DH, Boschelli F. SKI-606, a 4-anilino-3-quinolinecarbonitrile dual inhibitor of Src and Abl kinases, is a potent antiproliferative agent against chronic myelogenous leukemia cells in culture and causes regression of K562 xenografts in nude mice. Cancer Res. 63:375–381, 2002 203. Ple PA, Green TPHennequin LF, Curwen J, Fennell M, Allen J, Lambert-Van Der Brempt C, Costello G. Discovery of a new class of anilinoquinazoline inhibitors with high affinity and specificity for the tyrosine kinase domain of c-Src. J Med Chem. 47:871–887, 2004 204. Yezhelyev MV, Koehl G, Guba M, Brabletz T, Jauch KW, Ryan A, Barge A, Green T, Fennell M, Bruns CJ, et al. Inhibition of SRC tyrosine kinase as treatment for human pancreatic cancer growing orthotopically in nude mice. Clin Cancer Res. 10:8028–8036, 2004

460

C.D. Weekes and M. Hidalgo

205. Messersmith WA, Rajeshkumar NV, Tan AC, Wang XF, Diesl V, Choe SE, Follettie M, Coughlin C, Boschelli F, Garcia-Garcia E, Lopez-Rios F, Jimeno A, Hidalgo M. Efficacy and pharmacodynamic effects of bosutinib (SKI-606), a Src/Abl inhibitor, in freshly generated human pancreas cancer xenografts. Mol Cancer Ther. 8:1484–1493, 2009 206. Rajeshkumar NV, Tan AC, De Oliveira E, Womack C, Wombwell H, Morgan S, Warren MV, Walker J, Green TP, Jimeno A, Messersmith WA, Hidalgo M. Antitumor effects and biomarkers of activity of AZD0530, a Src inhibitor, in pancreatic cancer. Clin Cancer Res. 15:4138–4146, 2009 207. Ashkenazi A. Targeting death and decoy receptors in the tumor necrosis factor superfamily. Nat Rev Cancer. 2:420–430, 2002 208. Kim I, Xu W, Reed JC. Cell death and endoplasmic reticulum stress: disease relevance and therapeutic opportunities. Nat Rev Drug Discov. 7:1013–1030, 2008 209. Lessene G, Czabotar PE, Colman PM. BCL-2 family antagonists for cancer therapy. Nat Rev Drug Discov. 7:989–1000, 2008 210. O’brien S, Moore JO, Boyd TE, et al. Randomized phase III trial of fludaribine plus cyclophosphamide with or without oblimersen sodium (Bcl-2 antisense) in patients with relapsed or refractory chronic lymphocytic leukemia. J Clin Oncol. 25:1114–1120, 2007 211. Bedikian AY, Millward M, Pehamberger H, et al. Bcl-2 antisense (oblimersen sodium) plus dacarbazine in patients with advanced melanoma: the Oblimersen Melanoma Study Group. J Clin Oncol. 24:4738–4745, 2006 212. Chanan-Khan AA, Niesvizky R, Hohl RJ, Zimmerman TM, Christiansen NP, Schiller GJ, Callander N, Lister J, Oken M, Jagannath S. Phase III randomised study of dexamethasone with or without oblimersen sodium for patients with advanced multiple myeloma. Leuk Lymphoma. 50:559–565, 2009 213. Rudin CM, Salgia R, Wang X, et al. Randomized phase II study of carboplatin and etoposide with or without the bcl-2 antisense oligonucleotide oblimersen for extensive-small cell lung cancer: CALGB 30103. J Clin Oncol. 26:870–876, 2008 214. Frieden M, Orum H. The application of locked nucleic acids in the treatment of cancer. IDrugs. 9:706–711, 2006 215. Hansen JB, Fisker N, Westergaard M, Kjaerulff LS, Hansen HF, Thrue CA, Rosenbohm C, Wissenbach M, Orum H, Koch T. SPC3042: a proapoptotic survivin inhibitor. Mol Cancer Ther. 7:2736–2745, 2008 216. Satoh T, Okamoto I, Miyazaki M, Morinaga R, Tsuya A, Hasegawa Y, Terashima M, Ueda S, Fukuoka M, Ariyoshi Y, Saito T, Masuda N, Watanabe H, Taguchi T, Kakihara T, Aoyama Y, Hashimoto Y, Nakagawa K. Phase I study of YM155, a novel survivin suppressant, in patients with advanced solid tumors. Clin Cancer Res. 15:3872–3880, 2009 217. Dean E, Jodrell D, Connolly K, Danson S, Jolivet J, Durkin J, Morris S, Jowle D, Ward T, Cummings J, Dickinson G, Aarons L, Lacasse E, Robson L, Dive C, Ranson M. Phase I trial of AEG35156 administered as a 7-day and 3-day continuous intravenous infusion in patients with advanced refractory cancer. J Clin Oncol. 27:1660–1666, 2009 218. Schimmer AD, Estey EH, Borthakur G, Carter BZ, Schiller GJ, Tallman MS, Altman JK, Karp JE, Kassis J, Hedley DW, Brandwein J, Xu W, Mak DH, Lacasse E, Jacob C, Morris SJ, Jolivet J, Andreeff M. Phase I/II trial of AEG35156 X-linked inhibitor of apoptosis protein antisense oligonucleotide combined with idarubicin and cytarabine in patients with relapsed or primary refractory acute myeloid leukemia. J Clin Oncol. 27(28):4741–4746, 2009 219. Liu G, Kelly WK, Wilding G, Leopold L, Brill K, Somer B. An open-label, multicenter, phase I/II study of single-agent AT-101 in men with castrate-resistant prostate cancer. Clin Cancer Res. 15:3172–3176, 2009 220. Schimmer AD, O’Brien S, Kantarjian H, Brandwein J, Cheson BD, Minden MD, Yee K, Ravandi F, Giles F, Schuh A, Gupta V, Andreeff M, Koller C, Chang H, Kamel-Reid S, Berger M, Viallet J, Borthakur G. A phase I study of the pan bcl-2 family inhibitor obatoclax mesylate in patients with advanced hematologic malignancies Clin Cancer Res. 14:8295–8301, 2008 221. O’Brien SM, Claxton DF, Crump M, Faderl S, Kipps T, Keating MJ, Viallet J, Cheson BD. Phase I study of obatoclax mesylate (GX15-070), a small molecule pan-Bcl-2 family antagonist, in patients with advanced chronic lymphocytic leukemia. Blood. 113:299–305, 2009

15  Targeted Therapeutics in Cancer Treatment

461

222. Firozvi K, Hwang J, Hansen N, et al. A phase I study of the panCL2 family inhibitor GX15070, administered as a 3-hour weekly infusion in patients with refractory solid tumors or lymphomas. Proc Am Soc Clin Oncol. 24:141s, 2006; abstr 3081 223. Wilson WH, Tulpule A, Levine AM, et al. A phase I/2a study evaluating the safety, pharmacokinetics, and efficacy of ABT-263 in subjects with refractory or relapsed lymphoid malignancies. Blood. 110, 2007; abstr 1371 224. Tolcher AW, Mita M, Meropol NJ, von Mehren M, Patnaik A, Padavic K, Hill M, Mays T, McCoy T, Fox NL, Halpern W, Corey A, Cohen RB. Phase I pharmacokinetic and biologic correlative study of mapatumumab, a fully human monoclonal antibody with agonist activity to tumor necrosis factor-related apoptosis-inducing ligand receptor-1. J Clin Oncol. 25:1390–1395, 2007 225. Greco FA, Bonomi P, Crawford J, Kelly K, Oh Y, Halpern W, Lo L, Gallant G, Klein J. Phase 2 study of mapatumumab, a fully human agonistic monoclonal antibody which targets and activates the TRAIL receptor-1, in patients with advanced non-small cell lung cancer. Lung Cancer. 61:82–90, 2008 226. Leong S, Cohen RB, Gustafson DL, Langer CJ, Camidge DR, Padavic K, Gore L, Smith M, Chow LQ, von Mehren M, O’Bryant C, Hariharan S, Diab S, Fox NL, Miceli R, Eckhardt SG. Mapatumumab, an antibody targeting TRAIL-R1, in combination with paclitaxel and carboplatin in patients with advanced solid malignancies: results of a phase i and pharmacokinetic study. J Clin Oncol. 27(26):4413–4421, 2009 227. Camidge DR, Herbst RS, Gordon M, et al. A phase I safety and pharmacokinetic study of Apomab, a human DR5 agonist antibody in patients with advanced cancer. Proc Am Soc Clin Oncol. 25(18S), 2007; abstr 3582 228. LoRusso P, Hong D, Heath E, et  al. First-in-human study of AMG 655, a pro-apoptotic TRAIL receptor-2 agonist, in adult patients with advanced solid tumors. Proc Am Soc Clin Oncol. 25, 2007; abstr 3534 229. Plummer R, Attard G, Pacey S, Li L, Razak A, Perrett R, Barrett M, Judson I, Kaye S, Fox NL, Halpern W, Corey A, Calvert H, de Bono J. Phase 1 and pharmacokinetic study of lexatumumab in patients with advanced cancers. Clin Cancer Res. 13(20):6187–6194, 2007 230. Wakelee HA, Patnaik A, Sikic BI, Mita M, Fox NL, Miceli R, Ullrich SJ, Fisher GA, Tolcher AW. Phase I and pharmacokinetic study of lexatumumab (HGS-ETR2) given every 2 weeks in patients with advanced solid tumors. Ann Oncol. 21(2):376–381, 2010 231. Thomas GV, Tran C, Mellinghoff IK, Welsbie DS, Chan E, Fueger B, Czernin J, Sawyers CL. Hypoxia-inducible factor determines sensitivity to inhibitors of mTOR in kidney cancer. Nat Med. 12:122–127, 2006 232. Hylton N. Dynamic contrast enhanced – magnetic resonance imaging as an imaging ­biomarker. J Clin Oncol. 24:3293–3298, 2004 233. Willett CG, Boucher Y, di Tomaso E, Duda DG, Munn LL, Tong RT, Chung DC, Sahani DV, Kalva SP, Kozin SV, et al. Direct evidence that the VEGF-specific antibody bevacizumab has antivascular effects in human rectal cancer. Nat Med. 10:45–147, 2004 234. Wu X, Rubin M, Fan Z, DeBlasio T, Soos T, Koff A, Mendelsohn J. Involvement of p27KIP1 in GI arrest mediated by an anti-epidermal growth factor receptor monoclonal antibody. Oncogene. 12:1397–1403, 1996 235. Salazar R, Tabernero J, Rojo F, Jimenez E, Montaner I, Casado E, Sala G, Tillner J, Malik R, Baselaga J, et al. Dose-dependent inhibition of the EGFR and signaling pathways with the anti-EGFR monoclonal antibody (MAb) EMD 7200 administered every three weeks (q3w). A phase I pharmacokinetic/pharmacodynamic (PK/PD) study to define the optimal biological dose (OBD). J Clin Oncol. 22:14S, 2004 236. Ciardiello F, Bianco R, Caputo R, Caputo R, Damiano V, Troiani T, Melisi D, De Vita F, De Placido S, Bianco AR, et al. Antitumor activity of ZD6474, a vascular endothelial growth factor receptor tyrosine kinase inhibitor, in human cancer cells with acquired resistance to antiepidermal growth factor receptor therapy. Clin Cancer Res. 10:784–793, 2004

Chapter 16

Cancer Chemoprevention Christopher H. Lieu, William N. William Jr, and Scott M. Lippman

16.1 Introduction Current multidisciplinary strategies for the treatment of various stages of invasive disease have not substantially improved the morbidity or mortality of major ­epithelial cancers. Developing chemopreventive agents that can divert the carcinogenic process and inhibit the development of tumors may greatly reduce the serious public health consequences of epithelial cancers. Cancer chemoprevention was first described by Sporn in 1976 as the use of natural, synthetic, or biologic chemical agents to reverse, suppress, or prevent carcinogenic progression [1]. This definition includes reducing the risk of and treating intraepithelial neoplasia (IEN) [2, 3]. Major clinical chemoprevention successes include vaccines targeting hepatitis B virus to prevent hepatocellular carcinoma and targeting human papillomavirus to prevent cervical cancer. Molecular-targeted agents such as celecoxib (as an adjunct in treating familial adenomatous polyposis and in reducing sporadic ­adenoma risk), tamoxifen and raloxifene (in reducing breast cancer risk), and finasteride (in reducing prostate cancer risk) have also shown activity in major cancer prevention trials [4–7]. Although providing proof of principle for cancer chemoprevention, ­landmark studies of these agents also required up to 12 years and 32,400 subjects to show statistically significant results. These massive logistics have become increasingly difficult to repeat, and newer chemoprevention ­trials will need streamlined designs comprising, for example, high-risk patients, intermediate endpoints, and innovative clinical/statistical designs in order to reduce their sizes, durations, and expense. Implementing such design strategies potentially will lead to widespread, personalized and effective cancer prevention that can reduce the far-reaching burden of cancer.

S.M. Lippman (*) Department of Thoracic/Head and Neck Medical Oncology, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Blvd., Unit 432, Houston, TX 77030-4009, USA e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_16, © Springer Science+Business Media, LLC 2011

463

464

C.H. Lieu et al.

This chapter discusses fundamental biologic concepts of cancer chemoprevention and important aspects of chemoprevention trial design, including the considerations for selecting populations, drugs, and endpoints for optimizing risk–benefit. Effective designs for these aspects can improve chemoprevention trials and hasten their ability to effect positive changes in the standard of care.

16.2 Biologic Concepts The three key concepts behind chemoprevention are that carcinogenesis is a multistep, multipath, and multifocal process [8]. Multistep carcinogenesis is the gradual accumulation of genetic and epigenetic alterations that progress toward neoplastic transformation. Theoretically, halting this progression may delay or inhibit the development of cancer. Multipath carcinogenesis is the disruption of multiple functional pathways leading to the development of cancer, including the inhibition of apoptosis, self-sufficiency in growth signals, and sustained angiogenesis. Multifocal carcinogenesis includes field carcinogenesis (the consequence of exposing an epithelial surface to carcinogens with the resultant development of genetic damage in normal-appearing mucosa) [9] and clonal expansion. Diffuse epithelial injury from carcinogen exposure can translate into an increased risk of cancer development over an entire field. Both injuries to tissue and host–environment interactions play integral roles in cancer risk. The role of host susceptibility has recently been highlighted by three independent studies demonstrating that single nucleotide polymorphisms (SNPs) at 15q24/15q25 were associated with increased risk of lung cancer [10–12]. Interestingly, in addition to lung cancer, these SNPs have also been associated with nicotine dependence and peripheral arterial disease [12, 13]. The 15q24/15q25 chromosome regions map to genes that encode nicotinic acetylcholine receptor subunits, which are expressed in neurons, alveolar epithelial cells, pulmonary neuroendocrine cells, and lung cancer cell lines. These subunits bind to N¢nitrosonornicotine and other potential lung carcinogens and could represent important targets for cancer chemoprevention [11]. Persistent tissue injury (from exposure to environmental carcinogens, such as tobacco, ultraviolet light, and even infections, e.g., human papillomavirus and H. pylori) can lead to cellular dysfunction and aberrant pathway activation which can result in premalignant epithelial changes [14, 15]. This includes genetic and epigenetic changes such as loss of heterozygosity (LOH) and methylation and global transcriptome changes (inflammation pathways). These changes can occur in patches of clonally and subclonally related cells. The clonal patches at high cancer risk can be defined as IEN. IEN is a premalignant lesion that occurs in epithelial tissues. Accumulating mutations and loss of cellular control functions can cause progressive phenotypic changes from normal histology to IEN to superficial cancer and finally to

16  Cancer Chemoprevention

465

invasive disease. As a result, IEN is frequently found in the late stages in the pathway from normal tissue to invasive cancer. In 2002, the AACR IEN Task Force recommended focusing chemopreventive drug development on IEN because of its close association between IEN and invasive cancer [2]. Reducing IEN would also decrease overall cancer risk and the need for invasive interventions. IENs are sometimes recognized clinically as diseases that require treatment, and not merely a cancer risk factor. Examples of this include colorectal adenomas in familial adenomatous polyposis, breast ductal carcinoma in situ, and grades 2 and 3 cervical IEN. Progression of dysplasia to cancer generally occurs over a long period of time. For example, colorectal adenomas may form over a period of time as long as 5–15 years, and progression from an adenoma to a colorectal carcinoma requires another 5–15 years [16–19]. In breast cancer, progression from atypical hyperplasia through DCIS to adenocarcinoma usually requires 10–20 years or more [20, 21]. The invasive cancer risk associated with IEN can be illustrated by the adenoma– carcinoma sequence in colon cancer. The ability of adenomas to progress to cancer varies by its histological growth pattern, size, and severity of dysplasia. For example, 2–5% of tubular adenomas progress to invasive disease versus 20–55% of villous adenomas [17]. Risk of malignancy is lower for adenomas <1 cm in diameter, but increases dramatically as the diameter increases above 1  cm. Of the 70% of adenomatous polyps that are mildly dysplastic, <5% will progress to cancer. However, up to 55% of adenomas of severely dysplastic adenomas will eventually progress to cancer [16]. The use of IEN to establish cancer risk is not without its challenges. Not all IEN can be easily detected or measured. Currently, most IENs are detected by direct visualization, such as in the colon, esophagus, cervix, and skin. In these tissues, the precancerous lesion can be identified, stained, and in some cancers, defined and imaged (e.g., cervix). This is exceedingly more difficult in organs that are not easily visualized, such as the prostate, breast, pancreas, and liver. The American Association for Cancer Research recommendations regarding the development of chemoprevention strategies updated in 2006 stressed the importance of the use of advances in imaging to provide the tools for further refining IEN detection and measurements [3]. This included the use of improved diagnostic methodologies such as the confocal microscope, breast ductal lavage and ductoscopy, the LIFE scope for visualizing bronchial tissue, and the magnifying endoscope for colorectal monitoring in order to assure adequate visualization and monitoring of IEN. Despite a higher cancer risk when compared to the general population, only a relatively small percentage of patients harboring an IEN will eventually present progression to cancer. As a result, it will be essential in future drug development trials to determine more refined ways to predict invasive cancer risk, both in patients with and even without histologic evidence of IEN. These methods include advances in proteomics and genomics, such as gene chip analysis, to identify patients with molecular alterations (i.e., “molecular IEN”) likely to result

466

C.H. Lieu et al.

in, phenotypically, full malignant transformation. As an example, Galipeau et al. have demonstrated that LOH and DNA content profiles could identify cases of Barrett’s esophagus likely to progress to cancer (0 vs. 79% 6-year cancer risk in the favorable and unfavorable groups, respectively) [22]. LOH has also been demonstrated to be a molecular risk factor for oral cancer (in patients with oral leukoplakia) [23, 24].

16.3 Population Selection Chemoprevention studies will target three major patient populations. Chemo­ prevention will target relatively healthy patients who are at a risk of a new ­malignancy in hopes of reducing this risk. These patients will have high-risk ­features, such as strong family histories or exposures, but have yet to develop a primary malignancy. The second group of patients targeted by chemoprevention will be patients with premalignant lesions, such as colon adenomas or oral leukoplakia. The focus will be on halting the transition from premalignant to malignant cells. Finally, chemoprevention will preclude the development of second primary tumors in patients with definitively treated primary cancer with no evidence of disease. The sample size calculation in a randomized clinical trial takes into account three major factors: the power of the study, the event rate in the control arm, and the expected magnitude of effect of the intervention. In chemoprevention studies, sample size calculation is particularly important, not only to maximize the chances of success of the trial, but also to minimize exposure of a “relatively healthy” population to the possible adverse events of the intervention. This is illustrated by the ATBC (Alpha-Tocopherol Beta-Carotene Cancer Prevention Study Group) trial, in which 29,133 male smokers were randomized to placebo or active treatment with alpha-tocopherol and/or beta-carotene; the primary endpoint of the trial was lung cancer development. The overall incidence of lung cancer at the time of the first study analysis in this unselected population was only 3%, and there was a statistically significant increase in the number of lung cancer cases and deaths due to lung cancer for individuals receiving beta-carotene [25]. Therefore, in most scenarios of chemoprevention, selection of a population at highest cancer risk (which increases the event rate in the control arm) will be key to reduce the sample size of a study, minimize exposure of a number of patients at lower cancer risk to potential toxic interventions, while maintaining the study’s power to detect a benefit from the intervention in the most relevant population. Chemoprevention in an unselected population may only be feasible when the agent employed carries minimal toxicity, is highly effective, and the endpoint is relatively common. An example is the use of HPV vaccines for the prevention of cervical cancer – a quadrivalent vaccine against HPV 6, 11, 16, and 18 tested in 5,455 individuals had 100% efficacy in preventing genital warts, vulvar and cervical IENs or cancer at 7 months in women

16  Cancer Chemoprevention

467

with no virologic evidence of HPV, with <0.1% vaccine-related serious adverse events [26].

16.3.1 Cancer Risk Modeling Cancer risk assessments are currently based on clinical and demographic criteria. These can be bundled in models, such as the Gail model, which has the ability to predict a 5-year invasive breast cancer risk of 1.66% or greater by taking into account five clinical factors. This model has been used for inclusion in large-scale breast cancer prevention trials [27]. Typically, the risk of cancer calculated by these models is significantly lower than the risk of cancer associated with a diagnosis of an advanced premalignant lesion. This is evidenced by the high 5-year risk of breast cancer of 1.67% in the Gail model versus the diagnosis of a breast DCIS which is associated with a 5-year risk of 13% (following resection and radiation) [5]. Prediction models based on clinical and demographic criteria have also been developed for lung cancer and, for the first time, bladder cancer (which includes exposure to tobacco, aromatic amines, and other carcinogens as covariates) [28–31]. The lung cancer model has been expanded to include genomic features, with increased accuracy [32]. Similarly, adding mutagen sensitivity phenotype data to the bladder cancer model has also led to improved discriminatory ability [31]. Hence, ­promising cancer risk models that incorporate clinical, histopathological, genetic, ­metabolic, and environmental parameters (the pharmaco-ecogenetic profile) will be pivotal to improve the prediction accuracy at the individual level and identify ideal candidates for chemoprevention. Molecular data may also improve the prediction ability of already established risk factors. Methylation markers in the sputum of chronic smokers, for example, precede development of lung cancer [33]. In oral leukoplakia, LOH at specific loci has been shown to be associated with subsequent invasive oral cancer in three ­different studies [23, 24, 34]. In patients with a resected oral cancer, LOH in oral premalignant lesions developed at the surgical site is also predictive of recurrence/ second primary tumors [35]. Specific cyclin D1 genotypes have also been linked to the progression of laryngeal dysplastic lesions [36]. In patients with DCIS, worse outcome was associated with the presence of biomarkers indicative of an abrogated response to cellular stress [37]. These and other molecular markers of cancer risk may be useful in selecting patients for participation in clinical chemoprevention trials. The ongoing Erlotinib Prevention of Oral Cancer (EPOC) trial is a pioneering prevention trial with a high-risk entry criterion based on molecular assessment [38]. In EPOC, patients with an oral premalignant lesion or an oral cancer treated with curative intent are first evaluated for LOH at 3p, 9p, and other key chromosomal sites. Patients who do not fit the prespecified molecular high-risk criteria (based on retrospective studies) [23, 35] are deemed to be at low cancer risk (i.e., <5% in 3 years)

468

C.H. Lieu et al.

and are not offered participation in the active treatment portion of the trial. On the other hand, patients with a high-risk LOH profile are randomized to treatment with placebo or erlotinib for 1 year. The estimated event rate (i.e., 3-year oral cancer incidence) in the control arm of the trial is 35% (for patients without a prior history of oral cancer) to 65% (for patients with a history of prior oral cancer) [23, 35]. With such high figures, the design of the trial calls for randomizing only 150 patients with an 85% power (and 5% type I error rate) to detect a 40% reduction in oral cancer (the primary endpoint of the trial) in the experimental arm. Hence, EPOC is an example of how selection of a study population based on molecular markers can reduce the scale of prevention trials with definitive endpoints and limit the intervention to truly high-risk individuals.

16.3.2 Convergent Trial Design Convergent trial designs also focus on selecting a population at high cancer risk for chemoprevention interventions, either in the setting of early phase clinical testing or late phase studies (e.g., trials including a mixture of patients with advanced premalignant lesions – “late prevention” – and curatively treated earlystage invasive lesions – “early treatment”). As advances continue to be made in the molecular biology of carcinogenesis, the distinction between preinvasive and invasive neoplasia is decreasing. As drug development continues to expand, the development of targeted therapy and prevention drugs continues to converge. Advances in screening and early detection are also contributing to this convergence as radiographic studies may allow the detection of IEN or early cancers, and advances in proteomics are bridging the gap between microscopic disease and clinical cancer. IEN and invasive cancer cells share similar cellular alterations that include increased proliferation and loss of apoptosis. This also includes the initial acquisition of a vascular supply that is a necessary step for a lesion to grow, known as the angiogenic switch [39]. Due to the similar qualities of both of these cells, moleculartargeted agents have the potential to be used in both chemoprevention and cancer therapy. This is probably best described by the Breast Cancer Prevention Trial of the selective estrogen receptor modulator tamoxifen in women at an increased risk of breast cancer [5]. The decrease in breast cancer in the tamoxifen arm was most likely due to the treatment of subclinical, microscopic, undetected breast cancer in high-risk women without any clinical evidence of disease. Regardless of the tumor-specific site, any agent that could delay the transformation of preinvasive cells or target subclinically invasive cells would be of tremendous clinical benefit in the cancer prevention. It is impossible to delineate the distinct line between IEN and cancer as there is a substantial overlap between the molecular characteristics of IEN and invasive cells. As a result, there is a middle ground that represents an opportunity for convergent expertise that includes prevention and therapy. Any agent that has a molecular target relevant to advanced cancer

16  Cancer Chemoprevention

469

may also be relevant to precancer, supporting the notion that phase I trials involving novel drugs will be relevant to prevention as well. Convergent trial design has been described by Lippman and Heymach [40]. Early-phase testing would be the most likely, but not the only, scenario for convergent trials. Three early-phase trial designs for convergent prevention and therapy would include: 1. Dose-finding trial in one type of cancer in order to elucidate the maximally tolerated, active dose in order to pursue further testing in phase II trials for either prevention or therapy with preliminary biomarker responses as a correlative opportunity. This trial may suggest doses of the study drug that may have activity in prevention that may differ from doses that have activity in cancer therapy. 2. A therapy design in advanced cancer patients with imbedded prevention endpoints (e.g., IEN, tumor-response) for agents with preventative potential. 3. Trials involving treating patients with IEN or early-stage cancer with a molecularly targeted agent prior to any surgical resection to assess IEN and tumor response As advances continue in molecular-targeted drug development, convergent trial design will aid greatly in controlling the increasing costs of challenges of drug development and potentially promote collaboration between chemoprevention and cancer therapy within the area of opportunity between preclinical and clinical diseases. The convergence of cancer therapy with cancer prevention should eventually streamline the development of targeted drugs and will improve the overall control of major cancers.

16.3.3 Hereditary Cancer Syndromes Hereditary cancer syndromes are also an excellent platform for the development of chemopreventive strategies. Some of the advantages of selecting this population for chemoprevention drug development include the fact that the risk of a specific malignancy is typically higher when compared with the general population; the potential hazards associated with the intervention are more easily accepted, given the perceived increased cancer risk; the molecular defects driving carcinogenesis are usually known, thus facilitating the understanding of pathophysiologic mechanisms of chemoprevention; the knowledge gained can potentially be translated to a broader population, since many of the pathways disrupted in hereditary cancer syndromes are also altered in sporadic cancers. Familial adeomatous polyposis (FAP) is an excellent example of the development of cyclooxygenase (COX) inhibitors for adenoma prevention, first within the context of the hereditary syndrome, and then having the concept exported to sporadic forms of the disease. FAP is a rare inherited disease caused by a mutation in the APC gene. FAP is clinically characterized by numerous polyps in the

470

C.H. Lieu et al.

colon, which, if untreated, virtually always transform into colon cancer. The first placebo-controlled randomized trial of a nonsteroidal anti-inflammatory agent in the prevention of colorectal polyps was conducted by Labayle et al. The authors observed a significant decrease in the number of polyps with the use of sulindac when compared with placebo in ten patients with FAP (P < 0.01) [41]. Subsequently, three other placebo-controlled trials (with a total number of 66 patients) confirmed the ability of sulindac to reduce the size and number of established colorectal polyps in the same population with the genetic syndrome [42–44]. However, in patients genotypically affected by familial adenomatous polyposis but still phenotypically normal (i.e., primary prevention), there was no effect of sulindac in the development of new adenomas and the authors concluded that colectomy remains the treatment of choice for primary prophylaxis of these patients [45]. In 2000, Steinbach et al. reported the results of the randomized trial of celecoxib (a selective COX-2 inhibitor) versus placebo in 77 patients with familial adenomatous polyposis. The study drug significantly decreased the burden of colorectal polyps and became US Food and Drug Administration (FDA)-approved for the reduction of polyp number in patients with this genetic disorder, despite such a small-scale trial [46]. The results of COX inhibition in adenoma prevention in FAP, associated with preclinical and epidemiologic observational data, formed the basis for the study of COX inhibition in sporadic colorectal adenomas. Indeed, aspirin and the COX-2 inhibitors celecoxib and rofecoxib were shown to be effective in this setting, but the trials had to enroll a much larger number of patients (i.e., up to 2,587) to demonstrate such benefit [4, 47–52].

16.4 Selection of Agents Characteristics of the optimal chemopreventive agent include a high therapeutic index (i.e., an agent that would efficiently prevent the emergence of or eliminate established premalignant clones in the majority of patients, with minimal or no toxicity), ease of administration (thus increasing the likelihood of compliance to the intervention), and cost-effectiveness. So far, chemoprevention trials have used a number of agents to disrupt the carcinogenic process, ranging from natural compounds (such as green tea) to vitamins and micronutrients, to anti-inflammatory and immunomodulatory agents, to vaccines, and to molecular-targeted drugs [4, 53–57]. Although not always identifying chemopreventive agents with acceptable risk–benefit profiles, these trials are related to important basic and translational research involving molecular biomarkers and contributing to a better understanding of risk assessment, the carcinogenic process, mechanisms of drug actions, and factors of resistance and sensitivity to a preventive agent. The chemoprevention field is now moving toward refining molecular biomarkers to be used in drug development and intelligent trial design.

16  Cancer Chemoprevention

471

16.4.1 Mechanism-Based Selection of Chemopreventive Agents A key component of rational preventive drug development is clinically relevant preclinical models for prevention that allow mechanistic studies of agents with potential application in clinical prevention trials. The importance of developing solid preclinical data before implementing a phase III trial was highlighted in a recent editorial by Gann on the results of the prostate cancer prevention trials SELECT (Selenium and Vitamin E Cancer Prevention Trial) and PHS II (Physician’s Health Study) [58–60]. SELECT evaluated selenium and/or vitamin E in 35,533 men of the general population ³50 years old, and PHS II evaluated vitamin C and/ or vitamin E in 14,641 male physicians ³50 years old. Both trials failed to demonstrate a prostate cancer preventive effect of any of the experimental treatments. The enthusiasm for studying these agents largely came from dietary observational ­studies and from secondary endpoint analyses of the ATBC Study and Nutritional Prevention of Cancer trial [61, 62]. Gann argues that first-generation phase III nutritional chemoprevention trials (ATBC and Nutritional Prevention of Cancer) might have been too reliant on observational studies subject to bias and that SELECT and PHS II might have been too reliant on secondary findings of the ATBC and Nutritional Prevention of Cancer studies that could have occurred by chance. Therefore, future phase III prevention trials should be based on causal effects validated in preclinical work and early-phase clinical trials. Although many preventive strategies are developed on the basis of therapeutic activity in advanced disease, as was the case with tamoxifen, lack of activity in invasive cancer should not halt evaluation of a candidate agent for prevention, if it is supported by preclinical evidence. On the other hand, activity in advanced malignancy does not necessarily mean that the agent will have preventive effects as well. Two studies demonstrate the possible lack of an association between preventive and therapeutic effects. Bergers et al. used an elegant mouse model in demonstrating that antiangiogenic agents have different ability in inhibiting the angiogenic switch (a relevant step in the transformation of preinvasive to invasive lesions) versus inhibiting persistent angiogenesis (a relevant step for invasive tumor growth); some agents were more effective in one versus the other setting [63]. William et al. demonstrated that the lack of phase II clinical activity of high-dose fenretinide in oral leukoplakia potentially could be explained by in  vitro studies showing a lesser effect of high-dose fenretinide in premalignant cells (versus in malignant cells) [64]. These two studies underscore the importance of evaluating an agent in preclinical models specifically reflecting the transformation of premalignancy to malignancy before clinical testing. An elegant example of successful translation of preclinical findings to the clinical arena is the development of COX-2 inhibitors for colorectal adenoma prevention. The landmark study of Oshima et al. demonstrated the contribution of COX-2 to colorectal carcinogenesis: knockout COX-2 mutations were induced in the Apcdelta716 knockout mice (an animal model for human familial adenomatous

472

C.H. Lieu et al.

polyposis). The mice with the COX-2 mutation exhibited a dramatic decrease in the number of polyps when compared with the Apcdelta716 knockout littermates with a wild-type COX-2 gene [65]. These findings were corroborated in humans with the study of Steinbach et al. of celecoxib in FAP patients described earlier and then translated to the sporadic adenoma setting in three randomized-placebo controlled studies: the APPROVe (Adenomatous Polyp Prevention on Vioxx) trial (using rofecoxib) and the APC (Adenoma Prevention with Celecoxib) and PreSAP (Prevention of Colorectal Sporadic Adenomatous Polyps) trials (using various doses of celecoxib) [4, 46, 51, 52]. All three studies demonstrated a statistically significant reduction in the cumulative 3-year incidence of adenomas, with hazard rations raging from 0.55 to 0.76. Nonetheless, celecoxib-related cardiovascular adverse events identifies in these studies precluded the widespread use of selective COX-2 inhibitors, as discussed in greater detail elsewhere in this chapter [66, 67].

16.4.2 Molecular Biomarkers Great progress is being made in genomic, proteomic, and novel imaging approaches for assessing molecular biomarkers of carcinogenesis, cancer risk, and preventive drug effects. Molecular biomarkers can serve as targets for chemoprevention. Additionally, they may also be used to improve the accuracy of established risk factors, to identify a target for modulation as well as factors of resistance or sensitivity to an intervention, and to monitor response to therapy. Molecular biomarkers may also be studied in animal models and early clinical trials to determine effectiveness and safety. There are several characteristics of ideal molecular biomarkers that can be targeted for successful chemoprevention studies. One characteristic is the relevance of the biomarker to the carcinogenic process that it models, e.g., if a single gene or protein is overexpressed or mutated in precancers or cancers, it may be a target for modulation as long as it is associated with neoplastic progression. A promising chemopreventive approach with potentially higher likelihood of success would be to identify biomarkers that, at the same time, confer increased cancer risk, participate in carcinogenesis, and are druggable targets (either directly or indirectly). An example of such approach is cyclin D1 in head and neck premalignant lesions. Recent studies suggest that a specific cyclin D1 genotype is associated with a high cancer risk in patients with head and neck premalignant lesions [36, 68]. Though no direct cyclin D1 inhibitors are yet in clinical use, there are agents that target EGFR and mTOR, which are both upstream of cyclin D1. Currently, one agent, erlotinib, is being tested in a phase III oral cancer prevention trial (EPOC). Another example of indirect inhibition of a key target for carcinogenesis is AKT, an upstream regulator of mTOR, in prostate preneoplasia. Activation of AKT1 leads to the development of high-grade prostatic IEN (HGPIN). A clinical trial recently reported that mTOR inhibition with everolimus induced apoptosis of epithelial cells and the complete reversal of the neoplastic phenotype [69].

16  Cancer Chemoprevention

473

Modulation of an ideal biomarker would also be linked to clinical benefit, commensurate with the biologic rationale and preclinical activity and should also be able to be quantified directly (e.g., tyrosine kinase activity) [70]. Quantification can also be completed by measuring an upstream or downstream target, such as S6 kinase activity or phosphorylation of 4-EBP or pS6 to measure mammalian target of rapamycin (mTOR) inhibition [71]. Finally, if the measurement of these biomarkers can be obtained in a cost–effective, noninvasive method, the potential for efficient identification and measurement of targeted biomarkers will aid future studies in chemoprevention greatly.

16.4.3 Combination Strategies One of the biologic concepts behind chemoprevention is that the development of cancer is a multipath process. As a result, another strategy behind chemoprevention trials will be to prevent carcinogenesis by targeting multiple pathways with a combination of agents. Preclinical work has shown that a combination of two drugs could afford protection from neoplasia in animal studies [72–74]. This concept was recently validated for the first time in the clinical setting in a randomized, placebocontrolled, double-blind trial using the combination of difluoromethylornithine and sulindac to prevent sporadic colorectal adenomas [75]. Neither drug had been shown to be clinically active at low doses, and the combination of both drugs has been shown to be the most active regimen to date. This illustrates one theoretical advantage of combination regimens, which have been carefully evaluated in the preclinical setting – the potential for use of lower doses of drugs with additive/synergistic properties, thus sparing patients from the toxic effects of high, therapeutic doses while maintaining efficacy. Interestingly, the difluoromethylornithine and sulindac study only needed to accrue 375 patients to show a large statistically significant benefit, whereas preceding single-agent studies required 1,000 to over 2,000 patients in order to show a lesser benefit [4, 48, 51, 52]. The clinically meaningful findings along with the small sample size of this combination trial reflect the need for further clinical investigation of the use of multiple drugs for chemoprevention of cancer. Other combination strategies that show promise in preclinical trials include EGFR and COX-2 signaling pathways, IGF-1 and mTOR, and the combination of a histone deacetylase inhibitor with a DNA methyltransferase inhibitor [76–78].

16.4.4 Chemopreventive Agents in Infection-Related Cancers and Vaccines Inflammation and infection promote neoplastic development through the stimulation of signal transduction pathways, tumor suppressor inhibition, and the production of growth-promoting cytokines [79]. Extraordinarily successful

474

C.H. Lieu et al.

chemoprevention have involved vaccines against human papillomavirus (cancers of the cervix, vulva, and vagina) [26, 55] and hepatitis B (hepatocellular carcinoma) [80]. The Females United to Unilaterally Reduce Endo/Ectocervical Disease (FUTURE) II trial, for example, found that a quadrivalent vaccine against HPV 6, 11, 16, and 18 reduced the risk of grades 2 and 3 cervical IEN, adenocarcinoma in situ, and cancer related to HPV 16 or 18 by 98% [55]. These vaccines demonstrate the success or molecular targeting through immunization against infections associated with neoplasia. A developing area of chemoprevention research involves targeting infections related to other tumor types, such as targeting H. pylori to prevent gastric cancer. Individuals with chronic H. pylori infection can develop gastric atrophy and intestinal metaplasia and are at an increased risk of developing precancerous gastric lesions [81]. A recent randomized controlled trial of 544 patients with early gastric cancer found that prophylactic eradication of H. pylori after endoscopic resection of early gastric cancer significantly reduced the development of metachronous gastric carcinoma [82]. Over the past year, the link between cancer and infection has been better characterized. As such knowledge becomes available, promising areas of future research include targeting HPV to prevent oropharyngeal cancer, herpes virus 8 to prevent Kaposi sarcoma, Epstein-Barr virus to prevent nasopharyngeal cancer, and perhaps even targeting poliomavirus, which has been recently associated with Merkel cell carcinoma [83]. Another area of promising molecular-based immunization strategy is the production of antibodies with new effector functions against known tumor targets or to identify new targets for therapeutic antibodies [84]. One animal study found that mice vaccinated with a prostate stem cell antigen-based vaccine induced MHC class I expression and cytokine production within tumors leading to low Gleason scores and increased survival [85]. The potential importance of the immune system in chemoprevention is also supported by the ability of patients to mount a vigorous T-cell response to autologous premalignant cells known as monoclonal gammopathy of undetermined significance (the precursor to multiple myeloma). T cells from myeloma marrow lacked this tumor-specific rapid effector function [57, 86]. This suggested a possible role for the immune system in influencing the early growth of transformed cells prior to the development of clinical cancer, which could be used for the development of novel immune-based chemopreventive strategies.

16.5 Endpoint Selection and Optimizing Risk–Benefit Besides the selection of a suitable population and agent, several other aspects of study design and conduct are key to the success of a clinical trial and have been discussed elsewhere in this book. Two features, however, are particularly relevant to chemoprevention studies: choice of definitive and intermediate endpoints and optimization of an intervention’s risk–benefit ratio.

16  Cancer Chemoprevention

475

16.5.1 Definitive and Intermediate Endpoints The design of a cancer chemoprevention trial will affect the willingness of patients to enter the trial. A questionnaire administered to 1,463 adults in a periodic health examination center revealed an overall lack of enthusiasm among healthy individuals for enrolling in possible chemopreventive trials; [87] reasons included randomization [88]. However, study participants were more likely to agree to randomization in shorter (1 year or less) rather than longer trials. This research points to the importance of intermediate endpoints that can be evaluated sooner than the cancer endpoint or short-term interventions for enhancing recruitment to clinical prevention trials. Overall survival is the most definitive endpoint for clinical chemoprevention; the other definitive endpoint, cancer incidence, may not relate to prolonged survival. The follow-up for overall survival, however, is prohibitively longer than that of cancer incidence, which has been used in successful chemoprevention trials but still generally requires many years follow-up and thousands of trial participants. Intermediate or surrogate endpoint biomarkers (SEBs) certainly have value for early-phase clinical trials and may, in future, be an alternative to the cancer endpoint that can reduce the duration and sample size of definitive phase III clinical trials. In theory, these biomarkers would correlate with a cancer endpoint and thus provide evidence of an agent’s efficacy prior to its definitive testing. Unfortunately, no SEBs developed to date have been confirmed to correlate with a definitive cancer endpoint. For example, a prespecified substudy of the APC trial evaluated the SEB potential of aberrant crypt foci (ACF), which are tiny lesions at the earliest stage of colorectal carcinogenesis. There was no significant modulation of ACF by celecoxib in this study, and the presence and number of nondysplastic ACF did not correlate with a higher risk of synchronous advanced or recurrent adenomas [89]. Another interesting potential SEB is prostate-specific antigen (PSA) in lieu of prostate cancer, but PSA also has yet to be validated as an SEB for cancer prevention [90]. A valid SEB must be as reliable for assessing intervention efficacy as is a definitive endpoint. Although SEBs theoretically could reduce the scale of new definitive chemoprevention trials, validated correlation with the cancer endpoint will be needed before SEBs can be substituted for definitive cancer endpoints. The potential SEB oral leukoplakia illustrates the importance of this point. In a recently reported analysis of the longest-term trial of retinoids in oral leukoplakia, leukoplakia response to the intervention was only moderately associated with long-term oral cancer-free survival after a median follow-up of ³5.9 years [91]. The limited association of leukoplakia response with oral cancer development might be due to underlying molecular aberrations. Mao et al. have demonstrated the persistence of LOH in patients with head and neck premalignancy, even after a complete clinical and histologic response to a chemopreventive combination of 13-cis retinoic acid, interferon-alpha, and alpha-tocopherol [14]. This finding underscores the need for molecular surrogate endpoints with independent value and the potential to enhance the predictive value of potential clinical SEBs such as oral leukoplakia.

476

C.H. Lieu et al.

Ideal SEBs should correlate with long-term cancer development, be modulated by a preventive agent, show variable expression between normal and tumor tissue, and correlate with clinical response. The evolution of potential SEBs will allow chemoprevention drug development to become more efficient and cost-effective.

16.5.2 Optimizing the Risk–Benefit Ratio An important concept in chemoprevention trials is the possibility that risks associated with the intervention may actually outweigh its potential benefits. This concept was illustrated by the COX-2 inhibitor trials, where the cardiovascular toxicity associated with rofecoxib and celecoxib has curtailed their use in colorectal cancer prevention. This is despite their established ability to reduce colorectal adenomas in the APPROVe, PreSAP, and APC trials. Modern chemoprevention trial designs should embed strategies to identify the population that will have an increased ­benefit from or a reduced risk with the intervention. This is independent of the selection of a population at high cancer risk. While markers of high cancer risk are prognostic factors, they may not necessarily (and frequently will not) predict which individuals will profit from an intervention. Predictive markers of efficacy may be based on simple clinical, demographic, or histopathologic criteria or may be identified with molecular-based research. In the APC trial, for example, individuals with ³3 adenomas at baseline, age ³60 years, and at least one parent with a history of colorectal cancer seemed to benefit more (i.e., 73% adenoma risk reduction) from celecoxib 400 mg, than the general population enrolled in the study (i.e., 45% adenoma risk reduction) [51]. In the molecular biomarker arena, the ornithine decarboxylase G316A genotype, for example, has been identified as a predictive marker of benefit from aspirin for adenoma prevention [92, 93]. Planning for baseline and posttreatment biospecimens collection in the context of chemoprevention studies will streamline identification of predictive, prognostic, and possible surrogate markers, as well as potential novel therapeutic targets to be explored in future studies. Identification of predictive markers of toxicity also represents an important aspect of optimizing the risk/benefit ratio. The studies of colorectal adenomas have also pioneered this aspect of clinical research in the setting of chemoprevention. As mentioned earlier, despite the efficacy of COX-2 inhibitors in preventing adenomas, both rofecoxib and celecoxib have been associated with increased incidence of cardiovascular events. In the APC trial, for example, serious cardiovascular adverse events occurred in 1% of the individuals receiving placebo versus 2.6 and 3.4% of the individuals receiving celecoxib 200 and 400 mg, respectively, twice daily [67]. To better characterize the risk for adverse events, Solomon et al. analyzed six different placebo-controlled trials of celecoxib for nonarthritic conditions. While this analysis confirmed the cardiovascular toxicity of the drug, it also demonstrated that a scoring system computing baseline clinical characteristics (including age, blood pressure, lipid profile, smoking status, aspirin use, diabetes, and prior history of cardiovascular disease) could stratify patients into groups with no, moderate, or

16  Cancer Chemoprevention

477

marked risk for celecoxib-induced cardiovascular adverse events [94]. This information could be potentially useful in selecting individuals for adenoma prevention with celecoxib. Molecular-based toxicity risk assessment has not been widely studied in chemoprevention settings. Nonetheless, experience from advanced cancer and nonneoplastic conditions provide evidence for its potential utility. For example, individuals homozygous for the UGT1A1*28 polymorphism have been shown to be at higher risk for irinotecan-induced hematologic toxicities; individuals with certain alleles of CYP2C9 and VKORC1 are at increased risk of warfarininduced over-anticoagulation at the initiation of treatment [95, 96].

16.6 Conclusions Cancer chemoprevention integrating molecular-based drug development continues to evolve rapidly. Identifying mechanisms involved in carcinogenesis has led to ­significant progress in the prevention of colorectal, breast, and prostate cancer. Striking results of the recent randomized trials and FDA approval of the HPV vaccine are a major step forward not only for the prevention of cervical neoplasia but also for the entire field of cancer prevention. Future research in chemoprevention will continue to utilize molecular-targeted approaches for drug selection, patient selection, and outcome evaluation. This research will parallel, and at times be combined with, molecular-targeted advances in cancer therapy, ultimately streamlining the development of safe and effective interventions that prevent cancer and/ or IEN.

References 1. Sporn MB, Dunlop NM, Newton DL, Smith JM. Prevention of chemical carcinogenesis by vitamin A and its synthetic analogs (retinoids). Fed Proc 1976;35(6):1332–8. 2. O’Shaughnessy JA, Kelloff GJ, Gordon GB, et al. Treatment and prevention of intraepithelial neoplasia: an important target for accelerated new agent development. Clin Cancer Res 2002;8(2):314–46. 3. Kelloff GJ, Lippman SM, Dannenberg AJ, et al. Progress in chemoprevention drug development: the promise of molecular biomarkers for prevention of intraepithelial neoplasia and cancer – a plan to move forward. Clin Cancer Res 2006;12(12):3661–97. 4. Arber N, Eagle CJ, Spicak J, et al. Celecoxib for the prevention of colorectal adenomatous polyps. N Engl J Med 2006;355(9):885–95. 5. Fisher B, Costantino JP, Wickerham DL, et  al. Tamoxifen for prevention of breast cancer: report of the National Surgical Adjuvant Breast and Bowel Project P-1 Study. J Natl Cancer Inst 1998;90(18):1371–88. 6. Cummings SR, Eckert S, Krueger KA, et al. The effect of raloxifene on risk of breast cancer in postmenopausal women: results from the MORE randomized trial. Multiple Outcomes of Raloxifene Evaluation. JAMA 1999;281(23):2189–97. 7. Thompson IM, Lucia MS, Redman MW, et  al. Finasteride decreases the risk of prostatic intraepithelial neoplasia. J Urol 2007;178(1):107–9; discussion 10.

478

C.H. Lieu et al.

8. Lippman SM, Hong WK. Cancer prevention science and practice. Cancer Res 2002;62(18):5119–25. 9. Slaughter DP, Southwick HW, Smejkal W. “Field cancerization” in oral stratified squamous epithelium. Clinical implications of multicentric origin. Cancer 1953;6:963–8. 10. Amos CI, Wu X, Broderick P, et al. Genome-wide association scan of tag SNPs identifies a susceptibility locus for lung cancer at 15q25.1. Nat Genet 2008;40(5):616–22. 11. Hung RJ, McKay JD, Gaborieau V, et al. A susceptibility locus for lung cancer maps to nicotinic acetylcholine receptor subunit genes on 15q25. Nature 2008;452(7187):633–7. 12. Thorgeirsson TE, Geller F, Sulem P, et al. A variant associated with nicotine dependence, lung cancer and peripheral arterial disease. Nature 2008;452(7187):638–42. 13. Spitz MR, Amos CI, Dong Q, Lin J, Wu X. The CHRNA5-A3 region on chromosome 15q2425.1 is a risk factor both for nicotine dependence and for lung cancer. J Natl Cancer Inst 2008;100(21):1552–6. 14. Mao L, El-Naggar AK, Papadimitrakopoulou V, et al. Phenotype and genotype of advanced premalignant head and neck lesions after chemopreventive therapy. J Natl Cancer Inst 1998;90(20):1545–51. 15. Spira A, Beane J, Shah V, et al. Effects of cigarette smoke on the human airway epithelial cell transcriptome. Proc Natl Acad Sci U S A 2004;101(27):10143–8. 16. Hamilton SR. Pathology and biology of colorectal neoplasia. In: Young GP, Levin B, Rozen P, eds. Prevention and Early Detection of Colorectal Cancer. London: W.B. Saunders; 1996. 17. Hamilton SR. The adenoma-adenocarcinoma sequence in the large bowel: variations on a theme. J Cell Biochem Suppl 1992;16G:41–6. 18. Day DW, Morson BC. The adenoma-carcinoma sequence. In: Morson BC, ed. The Pathogenesis of Colorectal Cancer. Philadelphia: W.B. Saunders; 1978. 19. A multicenter study of colorectal adenomas. Rationale, objectives, methods and characteristics of the study cohort. The Multicentric Study of Colorectal Adenomas (SMAC) Workgroup. Tumori 1995;81:157–63 20. Frykberg ER, Bland K. In situ breast carcinoma. Adv Surg 1993;26:29–72. 21. Page DL, Dupont WD, Rogers LW, Rados MS. Atypical hyperplastic lesions in the female breast. A long-term follow-up study. Cancer 1985;55:2698–708. 22. Galipeau PC, Li X, Blount PL, et al. NSAIDs modulate CDKN2A, TP53, and DNA content risk for progression to esophageal adenocarcinoma. PLoS Med 2007;4(2):e67. 23. Rosin MP, Cheng X, Poh C, et al. Use of allelic loss to predict malignant risk for low-grade oral epithelial dysplasia. Clin Cancer Res 2000;6(2):357–62. 24. Mao L, Lee JS, Fan YH, et al. Frequent microsatellite alterations at chromosomes 9p21 and 3p14 in oral premalignant lesions and their value in cancer risk assessment. Nat Med 1996;2(6):682–5. 25. Omenn GS, Goodman GE, Thornquist MD, et al. Effects of a combination of beta carotene and vitamin A on lung cancer and cardiovascular disease. N Engl J Med 1996;334(18):1150–5. 26. Garland SM, Hernandez-Avila M, Wheeler CM, et al. Quadrivalent vaccine against human papillomavirus to prevent anogenital diseases. N Engl J Med 2007;356(19):1928–43. 27. Fisher B, Dignam J, Wolmark N, et al. Tamoxifen in treatment of intraductal breast cancer: National Surgical Adjuvant Breast and Bowel Project B-24 randomised controlled trial. Lancet 1999;353(9169):1993–2000. 28. Spitz MR, Etzel CJ, Dong Q, et al. An expanded risk prediction model of lung cancer. Cancer Prev Res 2008;1(4):250–4. 29. Spitz MR, Hong WK, Amos CI, et  al. A risk model for prediction of lung cancer. J Natl Cancer Inst 2007;99(9):715–26. 30. Bach PB, Kattan MW, Thornquist MD, et al. Variations in lung cancer risk among smokers. J Natl Cancer Inst 2003;95(6):470–8. 31. Wu X, Ros M, Gu J, Kiemeney L. Epidemiology and genetic susceptibility to bladder cancer. BJU Int 2008;102(9 Pt B):1207–15. 32. Beane J, Sebastiani P, Whitfield TH, et al. A prediction model for lung cancer diagnosis that integrates genomic and clinical features. Cancer Prev Res 2008;1(1):56–64.

16  Cancer Chemoprevention

479

33. Belinsky SA, Liechty KC, Gentry FD, et al. Promoter hypermethylation of multiple genes in sputum precedes lung cancer incidence in a high-risk cohort. Cancer Res 2006;66(6): 3338–44. 34. Partridge M, Emilion G, Pateromichelakis S, A’Hern R, Phillips E, Langdon J. Allelic imbalance at chromosomal loci implicated in the pathogenesis of oral precancer, cumulative loss and its relationship with progression to cancer. Oral Oncol 1998;34(2):77–83. 35. Rosin MP, Lam WL, Poh C, et al. 3p14 and 9p21 loss is a simple tool for predicting second oral malignancy at previously treated oral cancer sites. Cancer Res 2002;62(22):6447–50. 36. Izzo JG, Papadimitrakopoulou VA, Liu DD, et al. Cyclin D1 genotype, response to biochemoprevention, and progression rate to upper aerodigestive tract cancer. J Natl Cancer Inst 2003;95(3):198–205. 37. Gauthier ML, Berman HK, Miller C, et  al. Abrogated response to cellular stress identifies DCIS associated with subsequent tumor events and defines basal-like breast tumors. Cancer Cell 2007;12(5):479–91. 38. William WN, Heymach JV, Kim ES, Lippman SM. Molecular targets for cancer chemoprevention. Nat Rev Drug Discov 2009;8(3):213–25. 39. Hanahan D, Folkman J. Patterns and emerging mechanisms of the angiogenic switch during tumorigenesis. Cell 1996;86(3):353–64. 40. Lippman SM, Heymach JV. The convergent development of molecular-targeted drugs for cancer treatment and prevention. Clin Cancer Res 2007;13(14):4035–41. 41. Labayle D, Fischer D, Vielh P, et al. Sulindac causes regression of rectal polyps in familial adenomatous polyposis. Gastroenterology 1991;101(3):635–9. 42. Nugent KP, Farmer KC, Spigelman AD, Williams CB, Phillips RK. Randomized controlled trial of the effect of sulindac on duodenal and rectal polyposis and cell proliferation in patients with familial adenomatous polyposis. Br J Surg 1993;80(12):1618–9. 43. Giardiello FM, Hamilton SR, Krush AJ, et al. Treatment of colonic and rectal adenomas with sulindac in familial adenomatous polyposis. N Engl J Med 1993;328(18):1313–6. 44. Keller JJ, Offerhaus GJ, Polak M, et al. Rectal epithelial apoptosis in familial adenomatous polyposis patients treated with sulindac. Gut 1999;45(6):822–8. 45. Giardiello FM, Yang VW, Hylind LM, et al. Primary chemoprevention of familial adenomatous polyposis with sulindac. N Engl J Med 2002;346(14):1054–9. 46. Steinbach G, Lynch PM, Phillips RK, et  al. The effect of celecoxib, a cyclooxygenase-2 inhibitor, in familial adenomatous polyposis. N Engl J Med 2000;342(26):1946–52. 47. Benamouzig R, Deyra J, Martin A, et al. Daily soluble aspirin and prevention of colorectal adenoma recurrence: one-year results of the APACC trial. Gastroenterology 2003;125(2):328–36. 48. Baron JA, Cole BF, Sandler RS, et  al. A randomized trial of aspirin to prevent colorectal adenomas. N Engl J Med 2003;348(10):891–9. 49. Logan RF, Grainge MJ, Shepherd VC, Armitage NC, Muir KR. Aspirin and folic acid for the prevention of recurrent colorectal adenomas. Gastroenterology 2008;134(1):29–38. 50. Sandler RS, Halabi S, Baron JA, et  al. A randomized trial of aspirin to prevent colorectal adenomas in patients with previous colorectal cancer. N Engl J Med 2003;348(10):883–90. 51. Bertagnolli MM, Eagle CJ, Zauber AG, et al. Celecoxib for the prevention of sporadic colorectal adenomas. N Engl J Med 2006;355(9):873–84. 52. Baron JA, Sandler RS, Bresalier RS, et al. A randomized trial of rofecoxib for the chemoprevention of colorectal adenomas. Gastroenterology 2006;131(6):1674–82. 53. Shimizu M, Fukutomi Y, Ninomiya M, et  al. Green tea extracts for the prevention of metachronous colorectal adenomas: a pilot study. Cancer Epidemiol Biomarkers Prev 2008;17(11):3020–5. 54. Klein EA, Thompson IM, Lippman SM, et al. SELECT: the selenium and vitamin E cancer prevention trial. Urol Oncol 2003;21(1):59–65. 55. FUTURE II Study Group. Quadrivalent vaccine against human papillomavirus to prevent high-grade cervical lesions. N Engl J Med 2007;356(19):1915–27.

480

C.H. Lieu et al.

56. Lipkin SM, Rhee J, Zell J, Meyskens F, Iwata K. Molecular cancer prevention: Phase IIa trial of Erlotinib for IPMNs and the prevention of pancreatic cancer. AACR Meeting Abstracts 2006;2006(3):CS04–03. 57. Spisek R, Dhodapkar MV. Immunoprevention of cancer. Hematol Oncol Clin North Am 2006;20(3):735–50. 58. Gann PH. Randomized trials of antioxidant supplementation for cancer prevention: first bias, now chance –next, cause. JAMA 2009;301(1):102–3. 59. Lippman SM, Klein EA, Goodman PJ, et al. Effect of selenium and vitamin e on risk of prostate cancer and other cancers: The Selenium and Vitamin E Cancer Prevention Trial (SELECT). JAMA 2009:301(1):39–51. 60. Gaziano JM, Glynn RJ, Christen WG, et al. Vitamins E and C in the prevention of prostate and total cancer in men. The Physicians’ Health Study II Randomized Controlled Trial. JAMA 2009;301(1):52–62. 61. The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers. The Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group. N Engl J Med 1994;330(15):1029–35. 62. Clark LC, Combs GF, Jr, Turnbull BW, et al. Effects of selenium supplementation for cancer prevention in patients with carcinoma of the skin. A randomized controlled trial. Nutritional Prevention of Cancer Study Group. JAMA 1996;276(24):1957–63. 63. Bergers G, Javaherian K, Lo K-M, Folkman J, Hanahan D. Effects of angiogenesis inhibitors on multistage carcinogenesis in mice. Science 1999;284(5415):808–12. 64. William WN, Lee JJ, Lippman SM, et al. High-dose fenretinide in oral leukoplakia. Cancer Prev Res 2009;2:22–6. 65. Oshima M, Dinchuk JE, Kargman SL, et  al. Suppression of intestinal polyposis in Apc delta716 knockout mice by inhibition of cyclooxygenase 2 (COX-2). Cell 1996;87(5):803–9. 66. Bresalier RS, Sandler RS, Quan H, et al. Cardiovascular events associated with rofecoxib in a colorectal adenoma chemoprevention trial. N Engl J Med 2005;352(11):1092–102. 67. Solomon SD, McMurray JJ, Pfeffer MA, et al. Cardiovascular risk associated with celecoxib in a clinical trial for colorectal adenoma prevention. N Engl J Med 2005;352(11):1071–80. 68. Papadimitrakopoulou V, Izzo JG, Liu DD, et  al. Cyclin D1 and Cancer Development in Laryngeal Premalignancy Patients. Cancer Prev Res 2009;2:14–21. 69. Majumder PK, Febbo PG, Bikoff R, et al. mTOR inhibition reverses Akt-dependent prostate intraepithelial neoplasia through regulation of apoptotic and HIF-1-dependent pathways. Nat Med 2004;10(6):594–601. 70. Kelloff GJ, Fay JR, Steele VE, et al. Epidermal growth factor receptor tyrosine kinase inhibitors as potential cancer chemopreventives. Cancer Epidemiol Biomarkers Prev 1996;5(8):657–66. 71. Sansal I, Sellers WR. The biology and clinical relevance of the PTEN tumor suppressor pathway. J Clin Oncol 2004:22(14)2954–63. 72. Gupta RA, DuBois RN. Combinations for cancer prevention. Nat Med 2000;6(9):974–5. 73. Sporn MB. Combination chemoprevention of cancer. Nature 1980;287:107–8. 74. Torrance CJ, Jackson PE, Montgomery E, et al. Combinatorial chemoprevention of intestinal neoplasia. Nat Med 2000;6(9):1024–8. 75. Meyskens FL, Jr, McLaren CE, Pelot D, et al. difluoromethylornithine plus sulindac for the prevention of sporadic colorectal adenomas: a randomized placebo-controlled, double-blind trial. Cancer Prev Res 2008;1(1):32–8. 76. Zhang X, Chen Z, Choe MS, et al. Tumor growth inhibition by simultaneously blocking epidermal growth factor receptor and cyclooxygenase-2 in a xenograft model. Clin Cancer Res 2005;11:6261–9. 77. O’Reilly KE, Rojo F, She QB, et  al. mTOR inhibition induces upstream receptor tyrosine kinase signaling and activates Akt. Cancer Res 2006;66(3):1500–8. 78. Belinsky SA, Klinge DM, Stidley CA, et  al. Inhibition of DNA methylation and histone deacetylation prevents murine lung cancer. Cancer Res 2003;63(21):7089–93.

16  Cancer Chemoprevention

481

79. Philip M, Rowley DA, Schreiber H. Inflammation as a tumor promoter in cancer induction. Semin Cancer Biol 2004;14(6):433–9. 80. Chang MH, Chen CJ, Lai MS, et  al. Universal hepatitis B vaccination in Taiwan and the incidence of hepatocellular carcinoma in children. Taiwan Childhood Hepatoma Study Group. N Engl J Med 1997;336(26):1855–9. 81. Wolfgang Fischbach, Annie On-On Chan, Benjamin Chun-Yu Wong. Helicobacter pylori and gastric malignancy. Helicobacter 2005;10:34–9. 82. Fukase K, Kato M, Kikuchi S, et al. Effect of eradication of Helicobacter pylori on incidence of metachronous gastric carcinoma after endoscopic resection of early gastric cancer: an openlabel, randomised controlled trial. Lancet 2008;372:392–7. 83. Feng H, Shuda M, Chang Y, Moore PS. Clonal integration of a polyomavirus in human Merkel cell carcinoma. Science 2008;319(5866):1096–100. 84. Finn OJ. Cancer immunology. N Engl J Med 2008;358(25):2704–15. 85. Garcia-Hernandez Mde L, Gray A, Hubby B, Klinger OJ, Kast WM. Prostate stem cell antigen vaccination induces a long-term protective immune response against prostate cancer in the absence of autoimmunity. Cancer Res 2008;68(3):861–9. 86. Dhodapkar MV, Krasovsky J, Osman K, Geller MD. Vigorous premalignancy-specific effector T cell response in the bone marrow of patients with monoclonal gammopathy. J Exp Med 2003;198(11):1753–7. 87. Maisonneuve AS, Huiart L, Rabayrol L, et al. Acceptability of cancer chemoprevention trials: impact of the design. Int J Med Sci 2008;5:244–7. 88. Evans DGR, Lalloo F, Shenton A, Boggis C, Howell A. Uptake of screening and prevention in women at very high risk of breast cancer. The Lancet 2001;358(9285):889–90. 89. Cho NL, Redston M, Zauber AG, et al. Aberrant crypt foci in the adenoma prevention with celecoxib trial. Cancer Prev Res 2008;1(1):21–31. 90. Horwitz EM, Vicini FA, Ziaja EL, Dmuchowski CF, Stromberg JS, Martinez AA. The correlation between the astro consensus panel definition of biochemical failure and clinical outcome for patients with prostate cancer treated with external beam irradiation. Int J Radiat Oncol Biol Phys 1998;41(2):267–72. 91. Papadimitrakopoulou VA, Lee JJ, William WN, Jr, et al. Randomized trial of 13-cis retinoic acid compared with retinyl palmitate with or without beta-carotene in oral premalignancy. J Clin Oncol 2009;27(4):599–604. 92. Hubner RA, Muir KR, Liu JF, Logan RF, Grainge MJ, Houlston RS. Ornithine decarboxylase G316A genotype is prognostic for colorectal adenoma recurrence and predicts efficacy of aspirin chemoprevention. Clin Cancer Res 2008;14(8):2303–9. 93. Martinez ME, O’Brien TG, Fultz KE, et  al. Pronounced reduction in adenoma recurrence associated with aspirin use and a polymorphism in the ornithine decarboxylase gene. Proc Natl Acad Sci U S A 2003;100(13):7859–64. 94. Solomon SD, Wittes J, Finn PV, et  al. Cardiovascular risk of celecoxib in 6 randomized ­placebo-controlled trials: the cross trial safety analysis. Circulation 2008;117(16):2104–13. 95. Hoskins JM, Goldberg RM, Qu P, Ibrahim JG, McLeod HL. UGT1A1*28 genotype and ­irinotecan-induced ne-utropenia: dose matters. J Natl Cancer Inst 2007;99(17):1290–5. 96. Dumas TE, Hawke RL, Lee CR. Warfarin dosing and the promise of pharmacogenomics. Curr Clin Pharmacol 2007;2(1):11–21.

Chapter 17

Combined Modality Therapy in Cancer Management David Raben and Kyle Rusthoven

17.1 Introduction The last 30 years have ushered in a convincing argument that combining chemotherapeutic drugs, with actions against specific aspects of the cancer cell cycle or DNA, with ionizing radiation results in clinically meaningful improvements in locoregional control and survival in many epithelial cancers and gliomas. Many patients with locally advanced cancers will receive radiation with concurrent chemotherapy or biologic agents. We continue to optimize drug combinations and sequencing with radiation, and newer agents have emerged targeting specific cancer pathways, such as the epidermal growth factor receptor (EGFR) and vascular endothelial growth factor receptor (VEGFR) pathways. In radiation oncology, continuous improvements during the last century in radiotherapy delivery, using both radioactive isotopes and linear accelerators, have increased the utilization of radiation therapy in the multidisciplinary management of cancer. Recent advances in the field of radiobiology have furthered our understanding of critical molecular pathways. Specific tumor cell characteristics associated with sensitivity to radiation and relative radioresistance have been identified. Modern systemic therapies used concurrently with radiation are designed to manipulate these tumor characteristics in order to maximize the sensitivity of tumor cells to the lethal effects of ionizing radiation. Current studies are integrating highly conformal methods of radiotherapy delivery, such as intensity-modulated radiation therapy (IMRT), image-guided radiation therapy (IGRT), stereotactic body radiation (SBRT), and proton therapy, with systemic therapies aimed at specific biologic pathways in an attempt to further improve the therapeutic ratio. In this chapter, we discuss combined modality therapy (CMT) and its pertinence to a variety of disease sites, highlighting specific examples of the benefits and drawbacks of CMT. Furthermore, we discuss current research areas and future directions with CMT. D. Raben (*) Department of Radiation Oncology, University of Colorado Denver, Anschutz Cancer Pavilion, MS F-706, 1665 N. Ursula St., Suite 1032, Aurora, CO 80045-0510, USA e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_17, © Springer Science+Business Media, LLC 2011

483

484

D. Raben and K. Rusthoven

17.2 A Brief Explanation of Radiation Therapy Techniques and Modalities From the discovery of X-rays by Wilhelm Rongten in 1895 to the development of the cathode ray tube by William Coolidge in 1913 to the development of linear accelerators after 1945, many meaningful advances have allowed for the delivery of focused and highly accurate radiation therapy to any site in the human body. The first linear accelerator was based in Hammersmith Hospital in London. These machines generated electron energies through the propagation of electromagnetic waves in a waveguide hitting against a tungsten target. Through Bremstrahlung interactions deeply penetrating packets of energy, referred to as photons, were generated which could be directed and used for therapeutic purposes.

17.2.1 Radiation Treatment Planning Multiple critical steps are necessary in order to deliver focused radiation to the target, the specifics of which vary for different sites of the body. The first step is the radiotherapy simulation. Modern treatment planning most commonly utilizes sophisticated helical computed tomography (CT) units; however, MRI and combined CT and positron emission tomography (PET)-based simulators have also been described [1, 2]. These simulation technologies provide the radiation oncologist with improved visual representation of soft tissue anatomy and allow for improved delineation of tumor extent and critical structure location. In this process, patients are typically positioned on a CT table and are immobilized using a variety of techniques. Immobilization reduces both interfraction and intrafraction movement allowing for a reduction in the required margins around a tumor. Helical or axial CT slices are obtained at 3–5 mm intervals. Reference marks or tattoos facilitate three-point setup of the treatment isocenter. In some instances, internal fiducial markers, such as gold seeds, or motion tracking devices are placed within a tumor or organ, such as in the setting of prostate or liver tumors, prior to CT simulation and are used for daily positioning and motion feedback [3–5]. Images from the CT simulation are then transferred to a planning system that is used for target volume delineation and treatment planning. Often, the radiation oncologist will also acquire a diagnostic quality MRI or PET scan with the patient in the treatment position and fuse these diagnostic studies to the planning CT to assist in the delineation of target volumes [6–10]. The radiation oncologist then contours the gross tumor volumes (GTVs), and adds margins to account for ­microscopic tumor extension (clinical target volume or CTV) and margins for setup error and intrafraction motion (planning target volume or PTV) [11]. In cooperation with dosimetrists and physicists, radiation beams are positioned to optimally encompass the PTV while minimizing the dose to adjacent organs at risk. Dose constraints are placed on critical normal surrounding tissue in order to not exceed the radiation tolerance doses. Recent advances in radiation planning techniques and hardware have facilitated improvements in normal-tissue sparing without compromising target volume coverage. Newer linear accelerators encompass CT or MRI built-in to employ daily soft

17  Combined Modality Therapy in Cancer Management

485

tissue localization and enable the radiation oncologists to “dose paint” different doses of radiation to low- and high-risk areas (Fig. 17.1). From a radiation software and delivery perspective, we have begun to utilize on a routine basis IMRT. With IMRT, the dose delivered to organs at risk is reduced compared with older planning techniques such as 3D conformal or 2D-based treatment planning, particularly in the setting of irregular or concave target volumes [12–15]. Improved organ at risk delineation with CT planning combined with more tightly focused dose delivery using IMRT has translated into improved toxicity profiles when compared with historical controls. For example, in squamous cell carcinoma (SCC) of the head and neck mucosa (HNSCC), IMRT allows for excellent target coverage while simultaneously decreasing the mean dose delivered to the parotid glands. This has resulted in a reduction in the rates of late xerostomia without compromising tumor control rates [14, 16, 17]. IMRT can be delivered in two ways: (1) using an initial IMRT plan for the treatment of microscopic disease followed by a cone-down IMRT plan to boost gross disease or (2) single-phase treatment using simultaneous integrated boost (SIB) technique. SIB technique allows for simultaneous delivery of multiple different fractional doses to target volumes, also known as “dose painting” [18, 19]. This technique ­precludes the need to modify fields during radiation therapy (Fig. 17.2). The use of daily image guidance has allowed for incremental improvements in setup accuracy allowing for further reductions in margins around a tumor. Historically, setup accuracy has been verified using weekly portal images, with poor soft tissue resolution, and primarily using references to bony anatomy. More recently, IGRT utilizes ultrasound, kV X-rays, in-room CT (CT-on-rails), or a CT

Fig. 17.1  Examples of new linear accelerators that contain CT and MRI; mobile accelerators and incorporation of mini-micro leaf collimation offer greater sophistication and allow for “dose painting” as demonstrated in Fig. 17.2

486

D. Raben and K. Rusthoven

Fig. 17.2  Example of differing dose clouds of radiation for a patient with a base of tongue cancer receiving combined chemoradiation representing different total doses prescribed to areas of gross tumor volume (7,000 cGy as shown in red) and lower risk areas (4,500–5,000 cGy as shown in pink and blue)

scanner integrated into the linear accelerator (cone-beam CT) to localize the target volume and reposition the patient before each fraction using bony or soft tissue anatomy [20–24]. IGRT allows for PTV margins as small as 3 mm in the setting of prostate cancer [25], 5 mm for head and neck cancers [26], and 3–7 mm for lung and liver tumors simulated using four-dimensional simulation techniques [27–29], thereby further reducing the dose delivered to adjacent organs at risk. IGRT can be used in conjunction with IMRT to further optimize the accuracy and precision of radiotherapy delivery [25].

17.2.2 Biological Basis of Radiation Therapy as It Pertains to Combined Modality Therapy Radiation biology is the study of the effects of radiation on the inner workings of the cellular machinery. Cancer cell killing by ionizing radiation occurs through both direct DNA double-strand break damage and indirectly through the generation of hydroxyl radicals that, in turn, results in lethal or sublethal DNA damage [30]. The

17  Combined Modality Therapy in Cancer Management

487

process is incredibly complex and the death of a cancer cell due to irreparable damage to DNA from ionizing radiation relies on variables such as the amount and type of radiation energy deposited, the extent of DNA damage, and the repair capacity of a cancer cell after it senses the damage. Tissue and environmental factors also play a role in determining if the reproductive integrity of a cancer cell will be destroyed. The effect of ionizing radiation on cancer cells in the context of the four “R”s of radiobiology is: repair, reassortment, repopulation, and reoxygenation [30]. DNA damage is divided into lethal, sublethal, and potentially lethal damage. Repair refers to the capacity of irradiated cells to repair this DNA damage. The rationale for fractionating radiation over an extended period of time is based in part on exploiting potential differences between the repair capacities of cancer cells and the surrounding normal tissue. Lethal damage refers to irreversible damage to DNA that leads to certain cell death (ex double-stranded DNA breaks); sublethal damage is equated with radiation-based damage that can be repaired at given time and without further insults from additional fractionated radiation. Sublethal damage repair is responsible for the shoulder region of the cell survival curve. The repair of sublethal damage is associated with a variety of molecular checkpoint pathways that are immediately activated to begin homologous and nonhomologous repair of DNA single- and double-strand breaks. These include ATM, Rad 51, p53, DNA-PK, Chk1-2, and PARP, and inhibitors of the latter (Chk1-2 and PARP) are under investigation as radiosensitizers [31–33]. Potentially lethal damage can be repaired under certain conditions; however, the potential for repair is dependent upon the surrounding tumor and tissue microenvironment. The reader is referred to additional sources for a more detailed discussion of this topic [30]. Regarding reassortment, it should be mentioned that G2/M is considered the most radiosensitive part of the cell cycle. Cancer cells of different origins vary in their sensitivity to radiation, and a specific cancer will present with an asynchronous population in various phases of the cell cycle. After each fraction of radiation, the cancer cells in the sensitive aspects of the cell cycle may be lethally damaged, allowing resistant cells to redistribute themselves through the cell cycle, with new cells entering the radiosensitive G2/M phase. One of the most important contributors to radiation sensitivity relates to the presence of oxygen. Oxygen is required for the generation of hydroxyl radicals and it need only be present a short time during radiation delivery to be effective. Simplistic experiments in the 1960s involving pinching off skin (and blood supply) resulting in reduced skin reactions to ionizing radiation [34], leading investigators to suspect that lack of oxygen might also contribute to radioresistance. Thomlinson and Gray furthered the premise that alterations in tumor hypoxia could modify responses to radiation [35]. The presence of hypoxic cancer cells was noted to be primarily within the center of tumors, at least 150  mm away from the blood vessels and beyond the diffusion distance of oxygen. Animal experiments comparing the effects of radiation on xenografts based on breathing pure oxygen room air or under hypoxic conditions confirmed different survival curves [36, 37]. The increase in relative biologic effectiveness of photon radiotherapy in the oxygenated vs. hypoxic setting can be determined using the oxygen enhancement ratio. In general, the ­percentage of hypoxic cells within a tumor averages around 15%, which could potentially prevent radiation from effectively controlling tumors. During a

488

D. Raben and K. Rusthoven

fractionated course of radiation, however, tumor cells that are killed no longer require oxygen leading to reoxygenation of previously hypoxic cells as they moved closer to the surrounding vasculature. Animal experiments have confirmed these hypotheses, [38, 39] and several theories have evolved to possibly explain the hypoxia phenomenon including “diffusion-limited hypoxia,” briefly discussed above, and hypoxia based on changes in vascular permeability, mediated by changes in interstitial pressure within and around tumor vasculature [40]. Repopulation refers to the ability of cancer cells to regenerate through a variety of mechanisms, including decreased cell loss through resistance to radiation damage and upregulation of growth factor elements to drive proliferation. Repopulation can occur at any time, but appears to primarily begin after a lag time from the initial insults caused by ionizing radiation. In head and neck cancer, delays in radiation delivery can lead to significantly reduced locoregional control with classic studies by Withers et al. suggesting a loss of 0.6 Gy/day with increasing treatment times [41, 42]. Clinical trials have proven that a shorter time course when delivering radiation alone improves locoregional control [43–46].

17.3 Rationale for Combined Chemotherapy and Radiation A litany of clinical evidence has been attained demonstrating enhanced tumor control with the addition of concurrent chemotherapy to radiation (Tables 17.1–17.5). A clear rationale exists for this marriage of modalities in that chemotherapy and radiotherapy can combine in an additive, or even supra-additive, fashion to enhance tumor kill by multiple mechanisms. Prior to clinical trials of concurrent chemoradiotherapy, preclinical data demonstrated a synergistic effect. Systemic therapies sensitize tumors to the effects of radiotherapy using a variety of mechanisms: for example, mitomycin- C and tirapazamine alter the radiosensitivity of hypoxic cells; 5-fluorouracil and cisplatin interfere with DNA damage repair during radiation; and cetuximab (C225) prevents the cell proliferation and radioresistance mediated by the EGFR signaling pathway. With concurrent chemotherapy, the dose–response curve is shifted to the left (increased biologic effect per gray) for both tumor and normal tissue. However, because chemotherapy has a differential effect on tumor and normal tissue, the magnitude of this increase in biologic effect is greater for the tumor, thereby improving the therapeutic ratio. Thus, at a given level of normaltissue injury, a greater likelihood of tumor control is achievable. Various techniques are available to measure the effects of chemotherapy and radiotherapy either in vivo or in vitro. In vitro methods entail subjecting cells in culture to various treatments to assess response. Examples of in  vitro methods include simple tumor growth measurements, tumor cure (TCT50) assays, dilution assays, and lung colony assays [30]. Although cell survival curves generated by laboratory assays demonstrate the effect of combined modality treatment, they say nothing regarding its mechanism. The possible interactions of combined chemotherapy and radiation were promulgated by Steel [47]. The conceptual cornerstones

(continued)

Table 17.1  Summary of phase III clinical trials of concurrent chemoradiation for selected gastrointestinal tumors and glioblastoma Comments Study Trial design N Results Pancreas Adjuvant Only trial to show statistically significant Kalser et al. Patients with R0 resection randomized to CRT    43 Improved DFS, OS with survival advantage; limited to patients [141] (40 Gy/20 fx split course with two cycles CRT; Med S: 20 months with margin-negative resection; of 5-FU 500 mg/m2 days 1–3 followed by vs. 11 months (p = 0.03) maintenance 5-FU in CRT arm 2 years weekly bolus 5-FU) vs. observation   114 Trend toward improvement Trial included tumors of pancreatic head Klinkenbijl Patients with resected pT1-2N0-1M0 dz and periampullary region; limiting in OS with CRT; Med et al. [142] randomized to CRT (40 Gy/20 fx split power to detect survival difference in S: 17.1 months vs. course with two cycles 5-FU 25 mg/kg/d) vs. pancreatic tumors; included patients 12.6 months (p = 0.099) observation with positive margins Neoptolemos Patients with resected pancreatic cancer   289 Worse survival in CRT arm; 2 × 2 factorial design clouds interpretation; et al. [143] randomized in 2 × 2 factorial design to CRT poor description of RT technique; Med S: 15.9 months vs. (40 Gy/20 fx split course with two cycles of allowed range of doses up to 60 Gy 17.9 months (p = 0.05) 5-FU 500 mg/m2 days 1–3) vs. no CRT (30% of patients received nonuniform dose of RT or no RT); no RT QA described [144] Unresectable Older trial conducted in 1960s; low-RT    64 Improved survival with Moertel et al. Patients with unresectable PCa randomized doses used CRT; Med S: 6.4 months [145] to RT (35–40 Gy in 3–4 weeks) vs. same vs. 10.3 months RT plus 5-FU RT alone arm closed early (25 patients   194 Improved survival with Moertel et al. Patients with unresectable PCa randomized to enrolled) due to poor results; confirms CRT and dose-escalation; [146] three arms: (1) 60 Gy, split-course RT benefit of CRT; suggestion of dose– Med S: (1) 5.3 months alone, (2) 40 Gy, split course plus 5-FU, or response between 40 and 60 Gy with vs. (2) 8.4 months vs. (3) (3) 60 Gy split course plus 5-FU chemoRT 11.4 months

17  Combined Modality Therapy in Cancer Management 489

Patients with unresectable PCa randomized to CRT (59.4 Gy/33 fx with 5-FU 1 g/m2 days 2–5, 28–31 and MMC 11,410 mg/m2 on day 2) vs. RT alone

Rectal Adjuvant Gastrointestinal Patients who had underwent “curative” resection were randomized to four arms: Tumor Study (1) observation, (2) chemo alone (5-FU, Group [151] MeCCNU), (3) RT alone (40–48 Gy to pelvis), (4) CRT Krook et al. Patients with resected pT3-4 or pN + rectal [152] cancer randomized to adjuvant RT (50.4 Gy/28 fx) vs. adjuvant CRT (MeCCNU/5-FU × 1   50.4 Gy/28 fx with 5-FU   MeCCNU/5-FU × 1)

Gastrointestinal Patients with unresectable PCa randomized to CRT (54 Gy/30 fx with two cycles of Tumor Study 5-FU followed by adjuvant SMF) vs. SMF Group [148] chemotherapy alone Klaassen et al. Patients with unresectable PCa or postoperative [149] with gross residual disease were randomized to CRT (40 Gy/20 fx with 5-FU) vs. 5-FU alone Loehrer et al. Patients with unresectable PCa randomized to [150] CRT (RT 50.4 Gy/28 fx plus Gem 600 mg/m2 weekly followed by five cycles of Gem alone at 1,000 mg/m2 weekly 3× every 4 weeks) vs. Gem alone 7×

Cohen et al. [147]

Table 17.1  (continued) Study Trial design

Improved LC and DFS with CRT (arm 4) vs. observation (1); no improvement in OS CRT associated with improved LC, DM, DSS, and OS; OS improved by 29%, p = 0.025

  204

Improved survival with CRT; Med S: 11.0  months vs. 9.2 months (p = 0.044)

   74

  227

No difference in survival; Med S: 8.3 months vs. 8.2 months

No significant difference in DFS or OS; Med S: 7.1 months (CRT) vs. 8.4 months (RT) (p = 0.16) Improved survival with CRT; Med S: 9.7  months vs. 7.4 months

Results

   91

   43

  114

N

Only trial of adjuvant chemoRT demonstrating significant survival benefit over RT alone in rectal cancer; trial performed prior to era of TME surgery

Results of this trial served as basis for recommendation of adjuvant therapy in rectal cancer

First trial showing survival benefit with Gem-based CRT over Gem alone; grade 4 toxicity significantly increased in CRT arm (41% vs. 6%)

Low-RT dose for definitive treatment

First trial to demonstrate benefit of CRT vs. chemotherapy alone; small study

Only trial showing no benefit with the addition of chemo to RT in this setting; heme toxicity significantly increased with chemoRT

Comments

490 D. Raben and K. Rusthoven

Bujko et al. [157, 158]

Braendengen et al. [156]

Gérard et al. [155]

Neoadjuvant Bosset et al. [154]

Wolmark et al. [153]

  694

  207 Patients with unresectable T4 rectal cancer randomized to neoadjuvant CRT (50 Gy/25 fx with concurrent 5-FU/LV followed by surgery and 16 weeks of adjuvant 5-FU/LV) vs. RT alone Patients with resectable T3-T4 rectal cancer   312 randomized to neoadjuvant RT (25 Gy/5 fx followed by surgery within 7 days) vs. chemoRT (50.4 Gy/28 fx with bolus 350 mg/ m2 days 1–5 and LV 20 mg/m2 q4 weeks 2×) followed by surgery 4–6 weeks later

Patients with T3-T4 rectal cancer treated with 1,011 neoadjuvant RT (45 Gy/25 fx to posterior pelvis) and randomized to four arms: (1) neo-RT alone, (2) neo-CRT, (3) neo-RT and adjuvant chemo, (4) neo-CRT and adjuvant chemo. Chemo: 350 mg/m2 days 1–5 and LV 20 mg/m2 q4 weeks 2× Patients with T3-T4 rectal cancer randomized to   733 neoadjuvant CRT (45 Gy/25 fx to posterior pelvis plus 350 mg/m2 days 1–5 and LV 20 mg/m2 q4 weeks 2×) vs RT alone. All patients received adjuvant 5-FU/LV

Patients with resected stage II–III rectal cancer randomized to adjuvant chemotherapy alone (5-FU/LV or MOF) for five cycles vs. CRT (same chemo plus 50.4 Gy/28 fx)

(continued)

The addition of chemotherapy to neoadjuvant RT improves local control, but not OS in this study; the addition of postoperative chemotherapy was associated with trend toward improvement in 3 year OS (67.2% vs. 63.2%, p = 0.12) CRT associated with increased pCR rate Decreased LF with CRT: (11.4% vs. 3.6%) and increased grade 8.1% vs. 16.5% 3–4 acute toxicity (14.6% vs. 2.7%); (p < 0.05). No difference Unlike the EORTC study (Bosset et al. in OS [154]), adjuvant chemotherapy in RT alone arm did not abrogate the LC benefit of neoadjuvant CRT CRT associated with improved Significant benefits of combined chemoRT in patients with locally advanced T4 rate of R0 resection, pCR, LC, and DFS; trend toward primary tumors; CRT improves the rate of R0 resection pCR improvement in OS (66% sv 53%, p = 0.09) No difference in LC, DFS, Trial difficult to interpret because of different RT fractionation schemes and OS, or late toxicity; time intervals between neoadjuvant acute toxicity increased treatment and surgery; CRT associated in CRT arm; increased with increased acute toxicity, but also pCR (16% vs. 1%) and associated with decreased residual decreased rate of positive tumor size and higher rates of pCR and radial margins (4% vs. radial margin-negative surgery 13%) with CRT Decreased LF with the addition of chemo: 17.1% (RT alone) vs. 7.6–9.6% (chemo arms); no difference in OS (p = 0.84)

No difference in RFS, DFS, All female patients treated with 5-FU/LV; male patients randomized to 5-FU/LV and OS; improved LC vs. MOF; increased skin toxicity and with CRT (13% vs. 8%, leukopenia in CRT arm p = 0.02)

17  Combined Modality Therapy in Cancer Management 491

N

Patients with GBM (age <70; stable steroid requirements) randomized to adjuvant RT alone (60 Gy/30 fx) vs. CRT (60 Gy/30 fx with 75 mg/m2 TMZ followed by adjuvant TMZ 200 mg/m2 5/28 days × 6 cycles)   573

Comments

Improved PFS and OS with CRT; Med S 14.6 months vs. 12.1 months; 2 year OS: 26.5% vs. 10.4% (p < 0.001)

Established TMZ-based chemo RT as a standard of care in patients <70 years; 84% of patients had debulking surgery; grade 3–4 heme toxicity in only 7% in CMT arm

CRT improves local control and cancer CRT improved LC and specific mortality (CSM); no overall CSM, but not OS; survival benefit observed; acute 3 year LC: 61% vs. 39% toxicity increased with CMT (p < 0.001); 3 year CSM: 28% vs. 39% (p = 0.02); 3 year OS: 65% vs. 58% (p = 0.25) CRT improves local control and organ CRT improved LC (18% preservation compared with RT alone; increase; p = 0.02) and no difference in overall acute or late colostomy-free survival toxicity; increased risk of late anal (32% increase; p = 0.002), ulcers in CMT arm but not OS (p = 0.17)

Results

DFS disease-free survival, RFS relapse-free survival, PFS progression-free survival, DSS disease-specific survival, OS overall survival, SMF streptozocin, mitomycin, fluorouracil, 5-FU fluourouracil, Gem Gemicitabine, MeCCNU semustine, MOF semustine, vincristine, 5-FU, LV leukovorin, MMC mitomycin-C, TMZ temozolomide, EBRT external beam radiation therapy, TME total mesorectal excision

Glioblastoma Stupp et al. [161]

Anal canal   585 UKCCCR Anal Patients with ³ T2N0 SCC of the anal canal randomized to RT alone (45 Gy/25 fx Cancer Trial followed by boost for responders 6 weeks Working later to 15–25 Gy with EBRT or Ir192 brachy) Party [159 ] vs. CRT (same RT plus 5-FU 1 g/m2 days 1–4 or 750 mg/m2 days 1–5 in weeks 1 and 5 and MMC 12 mg/m2 day 1, week 1 only) Bartelink et al. Patients with T3-T4 or N + anal canal cancer   110 [160] randomized to RT alone (45 Gy/25 fx followed by boost of 15 Gy (CR) or 20 Gy (PR) 6 weeks later) vs. CRT (same RT plus 750 mg/m2 days 1–5 in weeks 1 and 5 and MMC 15 mg/m2 day 1, week 1 only)

Table 17.1  (continued) Study Trial design

492 D. Raben and K. Rusthoven

547

226

263

Larynx Forastiere et al. [83] c

Oropharynx Calais et al. [78]

Oropharynx/hypopharynx Semrau et al. [82] c AFxCB

Daily

Daily

Daily Daily Daily Daily Daily

Carbo/5-FU

Carbo/5-FU

CDDP

CDDP b CDDP b CDDP/5-FU CDDP CDDP b

16

42

56

47 65 54 59 78

26

51

54

78 80 72 70 78

0.016

0.02

ns

0.005 0.006 0.002 0.049 0.97

Postoperative Cooper et al. [88] 459 Daily CDDP 47 56 0.09 Bernier et al. [89] c 370 Daily CDDP 40 53 0.02   88 Daily CDDP 13 36 <0.01 Bachaud et al. [166] c HFx hyperfractionated radiation, AFxCB accelerated fractionation with concomitant boost, CDDP cisplatin, Carbo carboplatin, 5-FU 5-flourouracil, C225 cetuximab a  Two-year survival rates reported b  Three cycles of adjuvant CDDP/5-FU also given c  Five-year survival rates reported

147 221 284 350 348

Nasopharynx Al-Sarraf et al. [77] Wee et al. [162] Lin et al. [163] c Chan et al. [164] c Lee et al. [165]

Table  17.2  Summary of 3-year overall survival rates from phase III trials of concurrent chemoradiotherapy vs. radiation alone in patients with locally advanced head and neck cancer ChemoRT Study N RT Concurrent chemo RT alone (%) (%) p Multiple sites Adelstein et al. [76] 295 Daily CDDP 23 37 <0.001 Brizel et al. [80] 122 HFx; split course CDDP; 5-FU 34 55 0.07 Wendt et al. [79] 298 HFx; split course CDDP; 5-FU 24 48 <0.001 Jeremic et al. [81]a 130 HFx CDDP 49 68 0.008 Bonner et al. [84] 424 Daily; HFx or AFxCB C225 45 55 0.05

17  Combined Modality Therapy in Cancer Management 493

Table 17.3  Summary of median survival from phase III trials of comparing sequential vs. concurrent chemoradiation in stage III NSCLC Concurrent Sequential chemoRT chemoRT (months) (months) Study N RT Concurrent chemo Curran et al. [103] 610 63 Gy/34 fx CDDP/vinblastine 14.6 17.0 Furuse et al. [104] 320 56 Gy/28 fx CDDP/vindesine/MMC 13.3 16.5 Spilt course Fournel et al. [105] 205 66 Gy/33 fx CDDP/VP-16 14.5 16.3 Zatloukal et al. [106] 102 60 Gy/30 fx CDDP/vinorelbine 12.9 16.6 CDDP cisplatin, VP-16 etoposide, MMC mitomycin-C

0.24 0.029

p 0.046 0.039

494 D. Raben and K. Rusthoven

Table 17.4  Summary of 5-year survival rates from phase III trials comparing concurrent chemoradiation vs. radiation alone in cervical cancer Concurrent Study N FIGO stage PA nodal staging RT Concurrent chemo RT alone (%) chemoRT GOG 85a [121] 388 IIB-IVA Surgical 81 Gy to point A; EBRT; CDDP/5-FU 50b 64b and LDR brachy 526 IIB-IVA Surgical 81 Gy to point A, EBRT, CDDP/5-FU/ 40 61 GOG 120a [120] and LDR brachy hydroxyurea 526 IIB-IVA Surgical 81 Gy to point A, EBRT, CDDP 40 60 GOG 120a [120] and LDR brachy CDDP/5-FU 52 73 RTOG 9001 403 IB-IIA ³ 5 cm; Surgical or 85 Gy to point A, EBRT, [119] IIB-IVA clinical and LDR brachy; EFRT in control arm CDDP 58 62 80 Gy to point A, EBRT NCIC [122] 259 IB-IVA ³ 5 cm Clinical plus LDR/MDR/HDR or node brachy positive GOG 123 [118] 374 IB2 CDDP 74 83 Surgical staging 75 Gy to point A, EBRT, optional and LDR brachy followed by type I hysterectomy 81 Clinical 49.3 Gy/29 fx EBRT to CDDP/5-FU (two 71 GOG 109c [117] 268 IA2-IIA with: pelvis concurrent and positive two adjuvant) margins, positive nodes, or parametrial involvement at surgery LDR low dose rate, MDR moderate dose rate, HDR high dose rate, EBRT external beam radiation therapy, CDDP cisplatin a Control arm was RT and concurrent hydroxyurea b Estimated c Four-year survival 0.003

0.008

0.42

<0.001

<0.001

<0.001

p 0.033

17  Combined Modality Therapy in Cancer Management 495

14.8

  9.2

0.005

36

14.2

0.002

53.8

  9.3

0.15

17.6

CDDP; 5-FU weeks 1, 5, 8, 50 Gy/25 fx in CRT and 11 arm; 64 Gy/32 fx in RT alone 5-FU; MMC Smith et al. [135] 221 60 Gy/30 fx with optional surgical evaluation at 40 Gy SC split course, CDDP cisplatin, 5-FU 5-fluorouracil, LV leukovorin

Herskovic et al. [133, 134]

123

0.57

22.2

0.03

<0.001

p

0.78

18.6

ChemoRT (months)

<0.01

16

RT alone (months)

p

Adjuvant chemoRT (months)

Definitive

Table 17.5  Summary of median survival rates from phase III trials of concurrent chemoradiation in esophageal cancer Surgery alone Study N RT Concurrent chemo (months) Adjuvant Walsh et al. [167] 113 Neoadjuvant CDDP; 5-FU 11 40 Gy/15 fx Bosset et al. [138] 122 Neoadjuvant CDDP 18.6 37 Gy/10 fx SC Burmeister et al. [168] 128 Neoadjuvant CDDP; 5-FU 19.3 35 Gy/15 fx CDDP; 5-FU; vinblastine 16.9 Urba et al. [169] 100 Neoadjuvant 45 Gy/30 fx BID Tepper et al. [170]   56 Neoadjuvant CDDP; 5-FU 21.5 50.4 Gy/28 fx MacDonald et al.a [140] 556 Adjuvant 45 Gy/ 5-FU/LV × 5; two cycles 27 25 fx concurrent

496 D. Raben and K. Rusthoven

17  Combined Modality Therapy in Cancer Management

497

of the interaction were spatial cooperation and toxicity independence, referring to the targeting of different anatomic sites by the respective modalities without overlapping toxicity. While providing a rationale for combined treatment, the original framework outlined by Steel did not propose a direct interaction of the modalities on common tissues, nor provide an explanation for the supra-additive effect on tumors when delivered in combination. The proposed mechanisms by which chemotherapy increases the sensitivity of tumors to cell kill by ionizing radiation are numerous and, in this section, we discuss the mechanisms of action for the more commonly utilized agents in CMT. Cisplatin (CDDP) has been administered more frequently with radiotherapy than any other agent. The water-soluble drug is converted intracellularly to its active agent, which subsequently reacts with nuclear DNA to form interstrand and intrastrand crosslinks. Crosslink formation then triggers a cascade involving signaling pathways, checkpoint activation, DNA repair activity, and apoptosis [48]. Several mechanisms have been proposed to account for the synergy of CDDP and radiation [49–51], but the leading hypothesis is that CDDP inhibits repair of radiation-induced DNA damage, specifically, sublethal radiation damage repair [52] and has inhibitory effects on nonhomologous DNA repair through the DNA-dependent protein kinase (DNA-PK) pathway [53–55]. Other mechanisms have been suggested, including: enhanced formation of toxic platinum intermediates via ­radiation-induced free radicals, increased permanence of DNA damage by the way of CDDP-mediated free electron scavenging, increased cellular uptake of CDDP in the presence of radiation, and cell-cycle disruption. Understanding mechanisms of intrinsic or acquired resistance to CDDP are ongoing. Expression of ERCC1 and RRM1 genes has been linked to CDDP resistance in nonsmall cell lung cancer (NSCLC) [56–58]. Newer agents such as oxaliplatin, a third-generation platinum compound, appear to have similar radiosensitizing properties to CDDP; however, may interact differently in regards to activating mismatch repair gene complexes and have shown activity in cell systems resistant to CDDP [59–61]. Satraplatin is an oral platinum analog and may be active in CDDPresistant cancers. Owing to its ease of administration, satraplatin is an ideal candidate agent to investigate in combination with radiation in malignancies where CDDP and radiation are commonly used [62]. 5-Flourouracil (5-FU) is another agent incorporated in the chemotherapy regimens administered with radiation. The drug is converted to its cytotoxic form by multiple pathways resulting in depletion of thymidine 5¢-monophosphate and thymidine 5¢-triphosphate with subsequent derangement of DNA synthesis and repair. The combination of 5-FU and radiation can clearly be synergistic, although the exact mechanism is incompletely understood. It has been proposed that the mechanism of action is related to rapid progression of cells through S-phase (when they are relatively resistant to radiation) due to the presence of drug [63]. Other studies have suggested that 5-FU synergizes with radiation by eliminating the radiation-induced G2 cycle arrest, thereby reducing the likelihood of sublethal damage repair, or directly inhibiting the repair of DNA double-strand breaks from radiation. Byfield et al. studied the radiosensitizing effect of 5-FU on two adenocarcinoma cell lines. In this study, only postradiation incubation with 5-FU had any supra-additive effects on cell kill, and the

498

D. Raben and K. Rusthoven

effect of preradiation 5-FU exposure was only additive [64]. The enhanced cell killing was maximized if the tumor cells were continuously exposed to 5-FU for 48 h following the X-ray exposure. These findings support the theory that the primary radiosensitizing effects of 5-FU are mediated by inhibiting sublethal damage repair and have served as the rationale for protracted venous infusion administration of 5-FU over a course of radiation therapy. Capecitabine (Xeloda®) is an oral fluoropyrimidine carbamate that is metabolized in vivo to fluorouracil by thymidylate phosphorylase (TP). Thymidylate phosphorylase expression is higher in colorectal tumors compared with normal tissues. Schüller et al. compared 5-FU concentrations in tumor vs. normal tissues in 19 patients treated with capecitabine. In this study, the concentration of 5-FU was 3.2 times higher in colorectal tumors compared with healthy tissues [65]. These findings demonstrate that capecitabine is tumor-selective in the treatment of gastrointestinal primaries and has the potential to improve the therapeutic ratio. Moreover, the specificity of capecitabine may be maximized in the setting of concurrent chemoradiation, as radiation has been shown to increase intratumoral TP expression enhance the efficacy of capecitabine [66]. Taxanes have proven to be potent radiosensitizers with promising results in the treatment of tumors of multiple sites. The success of taxanes in combination with radiotherapy highlights the influence of the cell cycle on radiation sensitivity, as first described by Terasima and Tolmach over 40 years ago [67]. Taxanes bind to b-tubulin and increase polymerization to promote stable microtubule generation thereby arresting cells in the G2/M phase. As a result, cells accumulate in the G2/M phase, the most sensitive phase of the cell cycle, and the efficacy of cell kill by ionizing radiation is maximized. One of the newer drugs commonly used in combined modality regimens is temozolomide. Temozolomide (TMZ) is an oral alkylating agent, used in high-grade gliomas, that is rapidly absorbed and spontaneously converted to its active metabolite, MTIC. Methylation at the sixth position of guanine by TMZ triggers the activation of the mismatch repair process [68]. This results in G2 checkpoint activation leading to G2/M cell-cycle arrest and eventually to induction of apoptosis [69–71]. Cells then exposed to radiation in the G2/M phase are most susceptible to lethal DNA damage. Synergistic effects have been described with temozolomide and RT in preclinical glioma models using combined treatment in vitro and in vivo. The efficiency of TMZ and RT is dependent on the absence of methyl guanine methyltransferase (MGMT) expression. These concepts were elucidated by Wild-Bode et al., who evaluated the effect of adding TMZ to radiation in four GBM cell lines, two with and two without the expression of MGMT [72]. TMZ enhanced the radiation response in glioma cell lines without MGMT expression, but not those with MGMT expression. In MGMT-negative cells, there was significantly increased double-strand DNA damage with the addition of TMZ to RT compared with RT alone. In MGMT-positive GBM cells, TMZ did not enhance the effect of RT; however, the addition O6BG, an inhibitor of MGMT, appeared to enhance the sensitivity of these MGMT-positive cells to combined TMZ and RT. The importance of MGMT expression, as measured by MGMT promoter methylation, in predicting the outcome in patients with glioblastoma treated with temozolomide-based

17  Combined Modality Therapy in Cancer Management

499

chemoradiation has been validated in patients treated on a large clinical trial. A reanalysis of the landmark EORTC/NCIC trial, which established adjuvant TMZ-based chemoradiotherapy as the standard of care for GBM, demonstrated that MGMT promoter methylation (inactivating MGMT gene expression) was associated with improved prognosis and was predictive of response to TMZ [73].

17.4 Examples of Combined Modality Success The combination of radiosensitizing chemotherapy with ionizing radiation has proven benefits on clinical outcomes in numerous solid organ malignancies. A comprehensive discussion of all clinical trials and disease sites is beyond the scope of this chapter. As such, we will limit our discussion to specific sites with substantial clinical evidence supporting chemoradiation. In this section, we review the relevant literature regarding chemoradiation for tumors of the head and neck, lung, esophagus, and cervix. A summary of trials of concurrent chemoradiation for other sites is shown in Table 17.1.

17.4.1 Head and Neck Cancer Since the early 1990s, CMT for locally advanced head and neck cancer has become the preferred approach for inoperable patients. Merlano established the early foundations for this concept with a randomized trial comparing radiotherapy alone to alternating chemoradiation for patients with locally advanced head and neck cancer [74]. In this phase III trial, 157 patients with unresectable SCC of the head and neck were randomized to receive either alternating chemotherapy [four courses of cisplatin (20 mg/m2) and fluorouracil (200 mg/m2), given daily for five consecutive days during weeks 1, 4, 7, and 10] and radiotherapy (three courses of 20 Gy in ten daily fractions during weeks 2–3, 5–6, and 8–9) or radiotherapy alone (70 Gy in 35 daily fractions). As reported in the follow-up manuscript in 1996 [75], a statistically improved survival rate was observed with the combined modality approach (24% 5-year survival) vs. radiotherapy alone (10% 5-year survival). Statistical improvements in progression-free survival and locoregional control were also observed with CMT. The local control of 64% with combined chemoradiation compared favorably with historical controls. Concurrent platinum-based chemotherapy added fractionated radiotherapy is associated with the improved rates of locoregional control and survival. For locally advanced or unresectable tumors of most pharyngeal sites, concurrent ­platinum-based chemoradiation is associated with improved survival compared with radiation alone (Table  17.2). The benefit of chemoradiation is independent of radiation fractionation; concurrent chemotherapy has been shown to improve survival in the setting of once-daily radiation [76–78], hyperfractionated radiation [79–81], and accelerated radiation [82]. For tumors of the larynx, locoregional control and organ

500

D. Raben and K. Rusthoven

p­ reservation are improved with chemoradiation, but survival is not increased compared with radiation alone or sequential chemotherapy and radiation [83]. Similarly, little data exists suggesting a survival benefit for tumors of the hypopharynx. While several trials have accrued patients with hypopharyngeal primary tumors, only the trials by Semrau and Bonner analyzed the survival outcomes by head and neck site [82, 84]. In these trials, the survival benefit of CMT was limited to tumors of the oropharynx and did not apply to patients with hypopharyngeal primaries. Pignon et al. performed an individual patient data meta-analysis to estimate the influence of chemotherapy on survival in HNSCC [85]. For patients treated with any chemotherapy, overall survival was improved by 4% at 2 and 5  years. Among patients receiving concurrent chemotherapy, ­survival was improved by 8% at 5 years. CMT has also been used as an alternative to radical surgery as part of an organ preservation strategy. The landmark Veterans Affairs Larynx Cancer trial compared a larynx preservation approach using three cycles of induction chemotherapy (CDDP/5-fluorouracil) followed by definitive radiation therapy (66–76  Gy) with laryngectomy and postoperative radiotherapy in 332 patients with resectable stage III–IV larynx cancer [86]. Patients in the larynx preservation arm had response evaluation after two cycles, with responders receiving another cycle of chemotherapy followed by radiation and nonresponders proceeding to laryngectomy. Overall survival at 2 years was 68% in both study arms. In the larynx preservation arm, 85% of patients responded to two cycles of chemotherapy and 64% of patients were disease-free with an intact larynx at 2 years. A similar study performed by the European Organization for the Research and Treatment of Cancer (EORTC) established the feasibility of organ preservation for locally advanced tumors of the pyriform sinus or aryepiglotic folds [87]. A subsequent study, RTOG 91-11, evaluated the role of concurrent chemoradiation in a larynx preservation approach [83]. In this phase III trial, 547 patients with stage III–IV larynx cancer, without significant thyroid cartilage erosion, were randomized to radiotherapy alone or induction cisplatin/5-FU followed by radiotherapy (same as experimental arm of VA Larynx trial) or concurrent cisplatin (100 mg/m2 days 1, 22, and 43) and radiotherapy. All patients were treated to 70 Gy in 35 fractions. Statistical improvements in locoregional control (78% with concurrent cisplatin and radiation vs. 61% for induction chemotherapy and 56% for radiotherapy alone) and laryngectomy-free survival were observed with concurrent chemoradiation; however, no benefit was observed for overall survival (74–76% in all arms) with mature follow-up. Higher rates of acute mucosal toxicity were seen with concurrent chemoradiation in RTOG 91-11. Overall, severe (grade ³ 3) toxicity was 77% concurrent arm compared with only 51% in the sequential arm and 47% in the radiotherapy alone arm. The absence of a survival benefit with concurrent chemoradiation may be related to increased acute and chronic toxicity, particularly dysphagia and aspiration, and necessitates consideration in the design of future organ preservation protocols using chemoradiation. In the postoperative setting, survival benefits have been observed with the use of adjuvant concurrent chemoradiation for patients with high-risk pathologic features. In the landmark EORTC 22931 and RTOG 95-01 trials [88, 89], patients with highrisk features [EORTC: any T3-T4 primary (except T3N0 larynx), T1-2N2-3, or

17  Combined Modality Therapy in Cancer Management

501

T1-2N0-1 tumor with any of the following high-risk features: positive margins, extracapsular extension (ECE), perineural invasion (PNI), vascular tumor embolism or level IV or level V lymph node metastases from an oral cavity or oropharyngeal primary; RTOG: two or more positive lymph nodes, ECE or positive margins] were randomized postoperatively to adjuvant RT (60–66  Gy) with or without three cycles of concurrent CDDP 100 mg/m2. In both trials, locoregional control and disease-free survival were improved in the CMT arm. Overall survival was improved in the EORTC trial (p = 0.02), but not the RTOG trial (p = 0.19). A recent reanalysis of EORTC 22931 and RTOG 95-01 revealed that the patients most likely to benefit from postoperative chemoradiation were those with positive margins and ECE [90]. Finding a combined modality regimen that improves outcomes compared with radiation alone without increasing severe toxicity is challenging. Consistent across all trials of concurrent chemoradiation therapy in head and neck cancer is an increase in the incidence of high-grade acute mucositis and dysphagia. Because of this increased toxicity, recent efforts have been directed at identifying effective radiosensitizing systemic therapy which does not heighten the toxicity profile of radiation alone. EGFR overexpression has shown to be an independent poor prognostic factor in HNSCC [91]. Ang et al. reviewed tumor specimens from the RTOG 90-03 trial for EGFR expression via immunohistochemistry (IHC) [92]. This study revealed a significant correlation between high-EGFR expression and poor disease-free and overall survival, independent of T- and N-stage. Locoregional control was markedly diminished in patients with high-EGFR expression although no correlation with distant metastases was observed. It is important to note that the examined specimens were limited to those patients randomized to the control (conventionally fractionated) arm of the trial. Preclinical data confirms that transfection of EGFR expression confers radioresistance in human HNSCC tumors [93]. Chung et al. looked at this phenomenon in patients treated with chemoradiation and solidified the importance of EGFR expression or EGFR gene amplification by fluorescence in situ hybridization (FISH) as a prognostic factor [94, 95]. Hence, it appears that EGFR inhibition has the potential to decrease radioresistance and improve outcomes in patients with HNSCC. In a phase III trial, Bonner et al. [84] randomized 413 patients with stage III– IVB SCCA of the oropharynx, larynx, or hypopharynx to receive radiotherapy alone or radiotherapy plus weekly C225 (loading dose 400 mg/m2 one week prior to start of RT, then 250 mg/m2 weekly). Sixty percent of patients had oropharyngeal primaries, 75% had stage IV disease, and 30% had T4 disease. The RT was conventional once-daily treatment, hyperfractionation, or accelerated radiotherapy with concomitant boost. The majority (59%) of patients were treated with the concomitant boost approach. The results revealed a significant improvement in locoregional control, progression-free survival, and overall survival in the radiotherapy plus cetuximab arm. At a median follow-up of 54 months, overall survival was statistically superior with the combination arm (median 49.0  months vs. 29.3  months; p = 0.03) (Fig. 17.3). In a recent update of this trial, the survival benefit persisted at 5 years for patients treated with combined cetuximab and radiation [96]. Patients with prominent rash fared much better. The median survival benefits from C225

502

D. Raben and K. Rusthoven

Proportion with Locoregional Control

1.0

RT (n=213) 134 Events Median duration (mo) 14.9 2-year rate (%) 40.7

0.9 0.8 0.7

RT + C (n=211) 110 24.4 50.3

0.6 0.5 0.4 0.3 0.2

HR = 0.68 (0.52– 0.89) p = 0.005

0.1 0.0 0

6

12

18

24

30

36

42

48

54

60

Locoregional Control (months) Fig.  17.3  Kaplan–Meir plots for the primary endpoint for locoregional control (LRC). The ­outcomes with the RT alone arm does at least as well or better than RT in prior large studies. The median duration of LRC was 14.9 months vs. 24.4 months; 2 year LRC rates of 40.7% (RT) vs. 50.3% (RT/C) were observed. A significant decrease was observed in the risk of locoregional recurrence (32%) with a hazard ratio of 0.68, and a p value of 0.005. Updated analysis at 5 years continues to demonstrate a statistical LRC in favor of RT/C

and radiation appear quite favorable to traditional chemoradiation approaches [97]. At least one randomized clinical trial, RTOG 0522, is currently underway investigating the role of C225 in addition to concurrent CDDP-based chemoradiation in locally advanced head and neck cancer. The results of this trial are eagerly anticipated.

17.4.2 Nonsmall Cell Lung Cancer Similar to head and neck cancer, nonsmall cell lung cancer (NSCLC) is a locally aggressive malignancy historically associated with poor local tumor control rates with radiation alone. Radiation Therapy Oncology Group 73-01 demonstrated two critical concepts in the management of nonmetastatic NSCLC: First, that intensification of local therapy with radiation dose escalation is associated with improved local tumor control and, second, that improved local control is associated with improved survival [98]. Based on these findings, there was sound rationale for intensification of local therapy using a combination of chemotherapy and radiation. CALGB 8433 was a critical study demonstrating a survival advantage with the use of sequential chemotherapy and radiation therapy compared with radiation alone [99].

17  Combined Modality Therapy in Cancer Management

503

In this trial, 155 patients with stage III NSCLC were randomized to CDDP (100 mg/m2 on days 1 and 29) and vinblastine (5 mg/m2 on days 1, 8, 15, 22, and 29) followed by RT (60 Gy starting on day 50) vs. RT alone (starting day 1). The sequential chemoradiation arm was associated with improved median survival (13.8 months vs. 9.7 months, p = 0.007) and 5-year survival (13% vs. 6%, p = 0.012) [100]. These results were later confirmed in a trial by the RTOG [101]. During this same era, European investigators were testing the feasibility and efficacy of concurrent CDDP chemotherapy with radiation. In a three-arm phase II trial conducted by the EORTC, 331 patients with locally advanced, unresectable NSCLC were randomized to split-course radiotherapy alone (30  Gy in ten fractions   3-week break   25  Gy in ten fractions), split-course radiotherapy with weekly CDDP (30 mg/m2), or split-course radiotherapy with daily CDDP (6 mg/m2) [102]. Splitcourse RT with daily CDDP was associated with improved 2-year local control (31% vs. 19%, p = 0.003) and 3-year survival (16% vs. 2%, p = 0.009) compared with RT alone. Split-course RT with weekly CDDP was associated with improved local control, but not survival, compared with RT alone. More recently, trials comparing sequential and concurrent chemotherapy have demonstrated a survival advantage with the concurrent approach (Table  17.3). RTOG 94-10 was a three-arm phase III trial comparing the sequential chemoradiation (CDDP 100 mg/m2 days 1 and 29; Vinblastine 5 mg/m2 days 1, 8, 15, 22, and 29 with RT 63 Gy in 34 fractions) with two experimental arms: one arm combining once-daily RT (63 Gy in 34 fractions) with the same chemotherapy given concurrently and a second arm combining hyperfractionated radiation (69.6 Gy in 1.2 Gy/ fraction given twice daily) with CDDP (50 mg/m2 days 1, 8, 29, and 36) and oral VP-16 (50  mg BID for five consecutive days in weeks 1, 2, 5, and 6) [103]. Concurrent chemotherapy with once-daily RT was associated with improved survival compared with sequential chemoradiation (median 17.0  months vs. 14.6 months, p = 0.046), at the cost of increased acute grade 3–4 esophagitis (25% vs. 4%). No increase in late toxicity was observed. By contrast, there was no survival benefit observed with concurrent chemotherapy and twice-daily radiation. The benefit of concurrent over sequential chemotherapy has been confirmed by other investigators [104–106]. In each of these trials, acute toxicity, most commonly esophagitis, was increased in the concurrent arm. Is there a role for consolidation chemotherapy as part of the overall combined modality approach in lung cancer? The Southwest Oncology Group (SWOG 95-04) began to build upon the concurrent chemoradiation story with a singlearm phase II trial. Eighty-three patients with stage IIIB NSCLC with adequate pulmonary function (defined as FEV1 ³ 2  L or FEV1 in the contralateral lung ³800 mL) were treated with chemotherapy (CDDP 50 mg/m2 days 1, 8, 29, and 36 and VP-16 50 mg/m2 days 1–5, 29–33) and radiation (61 Gy in 33 fractions) followed by three cycles of adjuvant docetaxel chemotherapy (75  mg/m2 every 21 days) [107]. Median progression-free and overall survival rates with this regimen were 16 and 26  months, respectively. These outcomes are excellent compared with historical controls, and now serve as a benchmark for patients with stage IIIB NSCLC. A recently completed trial by the Hoosier Oncology Group

504

D. Raben and K. Rusthoven

(HOG 0124) confirmed the favorable outcomes with concurrent CDDP and VP-16 chemotherapy. In this trial, however, there was added toxicity but no survival benefit, observed with the addition of consolidation docetaxel to the backbone of thoracic RT with concurrent CDDP/VP-16 [108], thereby questioning the role of consolidation chemotherapy. EGFR is expressed in 80–90% of NSCLC and is overexpressed in 45–70% overall and in 57–92% of squamous cell tumors [109, 110]. Early clinical results using C225 in addition to chemoradiation appear promising. A recently reported phase II study, RTOG 0324, evaluated chemoradiation [63 Gy in 35 fractions with weekly carboplatin (AUC2) and paclitaxel (45 mg/m2)] with concurrent Cetuximab in 84 patients with unresectable stage III NSCLC [111]. Cetuximab was given as a loading dose (400 mg/m2) 1 week before the start of RT and weekly (250 mg/m2) during chemoradiation. All patients received two cycles of consolidation: carboplatin (AUC 6) and paclitaxel (200  mg/m2). Median survival with this regimen was 22.7 months which is the best reported in any previous RTOG lung cancer trial. The RTOG has incorporated cetuximab into amended RTOG protocol 0617 comparing 60–74 Gy with concurrent chemoradiation [112].

17.4.3 Cervical Cancer Cervical cancer represents another excellent example of a disease site that has demonstrated success in combining chemotherapy with radiation. Locally advanced cervical cancer (FIGO stage IIB-IVA) has historically been managed with primary radiation therapy using a combination of pelvic radiotherapy and intracavitary brachytherapy. Despite high cumulative doses of radiation (80–85 Gy), local failure occurs in approximately 35% of patients managed with radiation alone [113, 114]. Furthermore, after surgery for early stage disease (IB-IIA), many patients have high-risk pathologic factors, including lymphovascular invasion, deep stromal invasion, positive margins, positive lymph nodes or occult parametrial invasion, and require postoperative pelvic radiation therapy to achieve a high rate of locoregional control [115, 116]. Because of the challenges posed by the high rate of local failure with surgery or radiation alone, there was heightened interest in chemoradiation as a means of intensifying locoregional therapy. Six large randomized clinical trials have compared radiation therapy alone with radiation therapy plus chemotherapy in a variety of settings (Table 17.4). Five of these six trials demonstrated a significant overall survival advantage with concurrent chemoradiation, leading the National Cancer Institute to issue a clinical announcement recommending the addition of cisplatin-based chemotherapy for high-risk cervical cancer patients requiring treatment with radiation therapy [117]. One trial evaluated chemoradiation in the postoperative setting. Gynecologic Oncology Group (GOG) 109 compared adjuvant pelvic radiation alone (49.3 Gy in 29 fractions) with adjuvant chemoradiation with four cycles of concurrent and adjuvant cisplatin/5-FU [118]. Eligible patients had FIGO stage IA2-IIA with

17  Combined Modality Therapy in Cancer Management

505

positive margins, positive lymph nodes, or microscopic parametrial invasion. In this study, 4-year progression-free survival (80% vs. 63%, p = 0.003) and 4-year overall survival (81% vs. 71%, p = 0.007) were improved with chemoradiation. GOG 123 compared radiation (75  Gy total dose to point A) with concurrent chemoradiation (cisplatin 40 mg/m2 weekly) followed by consolidation extrafascial (type I) hysterectomy for patients with FIGO stage IB2 disease [119]. With a median follow-up of 36  months, progression-free survival and overall survival were improved by 49 and 46%, respectively. Finally, four trials compared definitive radiation therapy with or without concurrent cisplatin-based chemotherapy in patients with locally advance (FIGO IIB-IVA) or bulky IB-IIA disease [120–123]. Three of these trials, GOG 85, GOG 120, and RTOG 90-01, demonstrated significant improvements in progression-free and overall survival with chemoradiation [120–122]. In these trials, the relative improvement in progression-free survival ranged from 21 to 51% and the relative improvements in overall survival ranged from 24 to 52%. Notably, GOG 120, compared two CDDP-based chemotherapy arms: one arm using concurrent cisplatin (50 mg/m2 on days 1 and 29), 5-FU (4 g/m2 in 96 h in weeks 1 and 5), and hydroxyurea (2 g/m2 twice weekly for 6 weeks) and another arm using weekly cisplatin (40  mg/m2) alone. Both cisplatin-containing arms improved progression-free and overall survival compared with hydroxyurea and radiation (control arm), and there was no difference in outcomes between the cisplatin-containing arms, suggesting that cisplatin is the most important component of concurrent chemotherapy. Moreover, the three drug arm was associated with significantly more grade 3–4 hematologic toxicity (p < 0.001) compared with the cisplatin-only arm, leading the GOG to adopt weekly cisplatin (40 mg/m2) as the standard arm for future clinical trials. The importance of concurrent cisplatin was confirmed in GOG 165, which compared concurrent weekly cisplatin (40 mg/ m2) with protracted venous infusion 5-FU alone (225  mg/m2 daily) in a similar patient population. This trial demonstrated an excess risk of progression and death in the 5-FU alone arm [124]. Of note, one trial performed by the National Cancer Institute of Canada failed to demonstrate an improvement in progression-free or overall survival with the addition of weekly cisplatin chemotherapy (40 mg/m2) to radiotherapy [123]. The authors offer several potential explanations for the absence of benefit with chemoradiation in their trial; however, they conclude that the balance of evidence supports the use of concurrent chemoradiation in patients with locally advanced disease. In most of the trials discussed above, not unexpectedly, acute toxicity was increased with the addition of chemotherapy to radiation. The high-grade acute toxicities increased with cisplatin-based chemotherapy were predominantly hematologic and gastrointestinal. One exception was the RTOG 90-01 trial, in which the radiotherapy in the radiation alone arm was delivered to an extended field (with upper border at L1–L2 interspace) and the radiotherapy in the combined modality arm was to the pelvis only. In this trial, acute toxicity was similar in both arms. Current trials are underway testing new agents in addition to the cisplatin-based chemoradiotherapy backbone in an attempt to further improve locoregional control

506

D. Raben and K. Rusthoven

and survival. The Gynecologic Oncology Group trial 219 is a randomized phase III trial comparing chemoradiation with weekly cisplatin with or without the addition of tirpazamine, a hypoxic radiosensitizer [125]. Furthermore, RTOG 0417 is a phase II trial prospectively testing the safety and efficacy of the addition of bevacizumab (a monocolonal antibody against vascular endothelial growth factor receptor) to cisplatin-based chemoradiation [126].

17.5 Esophageal Cancer Both surgery and radiation therapy have been used as the principle local treatment modality in the management of esophageal cancer. The dose of radiation that can be safely delivered is limited by the tolerance of adjacent normal tissues, specifically the lung, heart, and spinal cord [127], and by the large superior– inferior margins required to encompass microscopic extension of disease [128]. As a result, radiation alone is rarely able to control large primary tumors and the predominant pattern of progression after radiation alone is disease persistence [129]. Moreover, in large surgical series, locoregional relapse is a component of failure in approximately one-third of patients [130]. Concurrent chemoradiation has been shown to improve outcomes in both the definitive and neoadjuvant settings (Table  17.5). Definitive chemoradiation and neoadjuvant chemoradiation followed by esophagectomy has yielded similar outcomes in patients with SCCs [131, 132], whereas patients with esophageal adenocarcinomas are typically managed with trimodality therapy, using neoadjuvant chemoradiation followed by esophagectomy. For patients managed with a nonoperative approach, the addition of cisplatin and 5-flourouracil chemotherapy has been shown to significantly improve progression-free and overall survival. RTOG 85-01 established important foundations for esophageal cancer management. This phase III clinical trial randomized 129 patients with biopsy-proven thoracic esophageal cancer without tracheal invasion to treatment with radiation therapy alone (64 Gy in 32 fractions) vs. chemoradiation (50 Gy in 25 fractions with CDDP 75 mg/m2 and 5-FU 1 g/m2 days 1–4 in weeks 1, 5, 8, and 11) [133]. Eighty-two percent of the patients enrolled had SCC. Radiation was delivered to the entire esophagus and supraclavicular fossa for the first 30 Gy followed by cone-down boost to the primary tumor plus 5 cm craniocaudal margins. Median (14.2 months vs. 9.3 months) and 5-year survival (27% vs. 0%) were significantly improved in the combined modality arm (p < 0.001) [134] despite the fact that the radiation dose was lower by ~25% compared with the radiation alone arm The addition of chemotherapy reduced the rate of both disease persistence (37% vs. 25%) and distant failure (30% vs. 16%) [129]. The survival benefit with chemoradiation was later confirmed in an Eastern Cooperative Oncology Group (EST-1282) using concurrent 5-FU and Mitomycin-C [135]. In both trials, acute gastrointestinal and hematologic toxicity was significantly increased in the combined modality arm, but late toxicities were not significantly different.

17  Combined Modality Therapy in Cancer Management

507

Only a small number of patients developed tracheoesophaeal fistulas. A current phase III study by the RTOG (RTOG 0436) is testing the addition of weekly cetuximab to chemoradiation (CDDP/paclitaxel) in patients treated without surgery [136]. The use of chemoradiation as neoadjuvant therapy has been shown to improve disease control and survival for patients with resectable esophageal cancer. A summary of the relevant trials is shown in Table 17.5. A recent individual patient data meta-analysis, including 1,298 patients enrolled in ten randomized clinical trials, confirmed a significant improvement in 5-year overall survival (7%, p = 0.002) and disease-free survival (4%, p = 0.036) with neoadjuvant chemoradiation over surgery alone [137]. In the largest of these neoadjuvant trials, conducted by the EORTC, perioperative mortality was significantly increased in the trimodality arm compared with the surgery alone arm [138]. This trial compared neoadjuvant split-course chemoradiation (37  Gy per ten fractions with two cycles of concurrent CDDP 80 mg/m2) followed by transthoracic esophagectomy with transthoracic esophagectomy alone in 297 patients with SCC. Disease-free survival (p = 0.003) and local control (p = 0.02) were significantly improved with neaodjuvant therapy, but survival was not different due to the excess postoperative mortality in the trimodality arm. Postoperative mortality was 12.3% in the trimodality arm compared with 3.6% in the surgery alone arm (p = 0.012). The hypofractionated radiation schedule (3.7  Gy/fraction) in this trial may have also contributed to the excess mortality observed. Notably, postoperative mortality was not increased with trimodality therapy in the meta-analysis [137]. In addition to the trials specific to patients with primary tumors of the esophagus, trials of CMT in gastric cancer have also allowed the enrollment of patients with adenocarcinoma of the gastroesophageal junction. MacDonald et al. reported a statistically significant disease-free and overall survival advantage with adjuvant chemoradiation compared with surgery alone for patients with locally advanced gastric cancer in a phase III Intergroup study [139]. After gastrectomy, 556 patients with stage IB-IV(M0) disease were randomized to observation vs. adjuvant chemoradiation (5-FU 425 mg/m2 and leukovorin 20 mg/m2 × five cycles with 45  Gy/25  fx of RT given with the second and third cycles of chemotherapy). Radiation was delivered to the anastomotic site, gastric tumor bed and regional lymph nodes. Relapse-free (median: 30  months vs. 19  months; p < 0.001) and overall survival (median: 36 months vs. 27 months; p = 0.005) were improved in the chemoradiation arm. Outcomes were not analyzed separately for tumors of the stomach vs. GE junction. The most common grade ³3 toxicities attributable to chemoradiation were hematologic (54%) and gastrointestinal (33%). The MAGIC trial, a phase III study conducted in patients with stage II–IV(M0) lesions of the stomach or lower third of the esophagus demonstrated a similar survival advantage to that observed in the Intergroup study using perioperative epirubicin, cisplatin, and 5-FU chemotherapy (three cycles before and three cycles after surgery) compared with gastrectomy alone [140]. Both adjuvant chemoradiation and perioperative ECF chemotherapy represent standard approaches in the management of resectable gastric cancer.

508

D. Raben and K. Rusthoven

17.6 Conclusions CMT in the definitive and adjuvant treatment of various solid organ malignancies has resulted in meaningful improvements in locoregional control, organ preservation, and overall survival. The majority of clinical evidence supporting concurrent chemoradiation has been with cisplatin and 5-fluorouracil-based regimens; however, newer cytotoxic drugs, such as the taxanes and oxiliplatin, oral 5-FU analogs, such as capecitibine, as well as molecularly targeted agents, such as cetuximab, have shown great promise in laboratory studies and/or clinical trials. There are no free lunches with CMT. In most sites, concurrent chemoradiation is associated with a significant increase in acute toxicity. As a result, future efforts will focus on the incorporation of radiosensitizing systemic therapies with nonadditive toxicity profiles. Radioprotectors such as palifermin, a KGF stimulator, may provide reduced mucosal and intestinal toxicity when CMT is utilized. Molecular targeted therapies directed against EGFR and VEGFR are currently being tested in clinical trials, and novel agents directed against new molecular targets or simultaneously against multiple targets are currently being investigated.

References 1. Mizowaki T, Araki N, Nagata Y, et al: The use of a permanent magnetic resonance imaging system for radiotherapy treatment planning of bone metastases. Int J Radiat Oncol Biol Phys. 2001; 49: 605–11. 2. De Ruysscher D, Wanders S, Minken A, et al: Effects of radiotherapy planning with a dedicated combined PET-CT-simulator of patients with non-small cell lung cancer on dose limiting normal tissues and radiation dose-escalation: a planning study. Radiother Oncol. 2005; 77: 5–10. 3. Moseley DJ, White EA, Wiltshire KL, et al: Comparison of localization performance with implanted fiducial markers and cone-beam computed tomography for on-line image-guided radiotherapy of the prostate. Int J Radiat Oncol Biol Phys. 2007; 67: 942–53. 4. Schiffner DC, Gottschalk AR, Lometti M, et al: Daily electronic portal imaging of implanted gold seed fiducials in patients undergoing radiotherapy after radical prostatectomy. Int J Radiat Oncol Biol Phys. 2007; 67: 610–9. 5. Balter JM, Dawson LA, Kazanjian S, et al: Determination of ventilatory liver movement via radiographic evaluation of diaphragm position. Int J Radiat Oncol Biol Phys. 2001; 51: 267–70. 6. MacManus M, Nestle U, Rosenzweig KE, et  al: Use of PET and PET/CT for radiation therapy planning: IAEA expert report 2006–2007. Radiother Oncol. 2009; 91(1):85–94. 7. Cantwell CP, Setty BN, Holalkere N, et  al: Liver lesion detection and characterization in patients with colorectal cancer: a comparison of low radiation dose non-enhanced PET/CT, contrast-enhanced PET/CT, and liver MRI. J Comput Assist Tomogr. 2008; 32: 738–44. 8. Wang D, Schultz CJ, Jursinic PA, et al: Initial experience of FDG-PET/CT guided IMRT of head-and-neck carcinoma. Int J Radiat Oncol Biol Phys. 2006; 65: 143–51. 9. Deniaud-Alexandre E, Touboul E, Lerouge D, et al: Impact of computed tomography and 18F-deoxyglucose coincidence detection emission tomography image fusion for optimization of conformal radiotherapy in non-small-cell lung cancer. Int J Radiat Oncol Biol Phys. 2005; 63: 1432–41.

17  Combined Modality Therapy in Cancer Management

509

10. Pech M, Mohnike K, Wieners G, et  al: Radiotherapy of liver metastases. Comparison of target volumes and dose-volume histograms employing CT- or MRI-based treatment planning. Strahlenther Onkol. 2008; 184: 256–61. 11. International Commission on Radiation Units and Measurement (ICRU). Prescribing, recording and reporting photon beam therapy. ICRU Report No. 50. Washington, DC: ICRU; 1993. 12. Nutting CM, Convery DJ, Cosgrove VP, et al: Reduction of small and large bowel irradiation using an optimized intensity-modulated pelvic radiotherapy technique in patients with prostate cancer. Int J Radiat Oncol Biol Phys. 2000; 48: 649–56. 13. De Meerleer GO, Vakaet LA, De Gersem WR, et al: Radiotherapy of prostate cancer with or without intensity modulated beams: a planning comparison. Int J Radiat Oncol Biol Phys. 2000; 47: 639–48. 14. Chao KS, Deasy JO, Markman J, et al: A prospective study of salivary function sparing in patients with head-and-neck cancers receiving intensity-modulated or three-dimensional radiation therapy: initial results. Int J Radiat Oncol Biol Phys. 2001; 49: 907–16. 15. Eisbruch A, Dawson LA, Kim HM, et al: Conformal and intensity modulated irradiation of head and neck cancer: the potential for improved target irradiation, salivary gland function, and quality of life. Acta Otorhinolaryngol Belg. 1999; 53: 271–5. 16. Eisbruch A, Kim HM, Terrell JE, et al: Xerostomia and its predictors following parotid-sparing irradiation of head-and-neck cancer. Int J Radiat Oncol Biol Phys. 2001; 50: 695–704. 17. Dawson LA, Anzai Y, Marsh L, et al: Patterns of local-regional recurrence following parotidsparing conformal and segmental intensity-modulated radiotherapy for head and neck cancer. Int J Radiat Oncol Biol Phys. 2000; 46: 1117–26. 18. Wu Q, Mohan R, Morris M, et al: Simultaneous integrated boost intensity-modulated radiotherapy for locally advanced head-and-neck squamous cell carcinomas. I: dosimetric results. Int J Radiat Oncol Biol Phys. 2003; 56: 573–85. 19. McCammon R, Rusthoven KE, Kavanagh B, et al: Toxicity assessment of pelvic intensitymodulated radiotherapy with hypofractionated simultaneous integrated boost to prostate for intermediate- and high-risk prostate cancer. Int J Radiat Oncol Biol Phys. 2009; 75(2):413–20. 20. Mageras GS, Mechalakos J: Planning in the IGRT context: closing the loop. Semin Radiat Oncol. 2007; 17: 268–77. 21. Grills IS, Hugo G, Kestin LL, et al: Image-guided radiotherapy via daily online cone-beam CT substantially reduces margin requirements for stereotactic lung radiotherapy. Int J Radiat Oncol Biol Phys. 2008; 70: 1045–56. 22. Fuss M, Salter BJ, Cavanaugh SX, et al: Daily ultrasound-based image-guided targeting for radiotherapy of upper abdominal malignancies. Int J Radiat Oncol Biol Phys. 2004; 59: 1245–56. 23. Hawkins MA, Brock KK, Eccles C, et al: Assessment of residual error in liver position using kV cone-beam computed tomography for liver cancer high-precision radiation therapy. Int J Radiat Oncol Biol Phys. 2006; 66: 610–9. 24. Ryu S, Yin FF, Rock J, et al: Image-guided and intensity-modulated radiosurgery for patients with spinal metastasis. Cancer. 2003; 97: 2013–8. 25. Chung HT, Xia P, Chan LW, et al: Does image-guided radiotherapy improve toxicity profile in whole pelvic-treated high-risk prostate cancer? Comparison between IG-IMRT and IMRT. Int J Radiat Oncol Biol Phys. 2009; 73: 53–60. 26. Zhang L, Garden AS, Lo J, et al: Multiple regions-of-interest analysis of setup uncertainties for head-and-neck cancer radiotherapy. Int J Radiat Oncol Biol Phys 2006; 64: 1559–69. 27. Slotman BJ, Lagerwaard FJ, Senan S: 4D imaging for target definition in stereotactic radiotherapy for lung cancer. Acta Oncol. 2006; 45: 966–72. 28. D’Souza WD, Nazareth DP, Zhang B, et al: The use of gated and 4D CT imaging in planning for stereotactic body radiation therapy. Med Dosim. 2007; 32: 92–101. 29. Xi M, Liu MZ, Deng XW, et  al: Defining internal target volume (ITV) for hepatocellular carcinoma using four-dimensional CT. Radiother Oncol. 2007; 84: 272–8.

510

D. Raben and K. Rusthoven

30. Hall EJ, Giaccia AJ: Radiobiology for the Radiologist, 6th Edition. Lippincott Williams & Wilkins, Philadelphia, PA, 2006. 31. Shinohara ET, Geng L, Tan J, et al: DNA-dependent protein kinase is a molecular target for the development of noncytotoxic radiation-sensitizing drugs. Cancer Res. 2005; 65: 4987–92. 32. Veuger SJ, Hunter JE, Durkacz BW: Ionizing radiation-induced NF-kappa B activation requires PARP-1 function to confer radioresistance. Oncogene. 2009; 28(6):832–42. Epub 2008 Dec 8. 33. Russo AL, Kwon HC, Burgan WE, et al: In vitro and in vivo radiosensitization of glioblastoma cells by the poly (ADP-ribose) polymerase inhibitor E7016. Clin Cancer Res. 2009; 15: 607–12. 34. Berry RJ, Schwarz G, Ellis RE, et al: The effect of hypoxia on the skin response of mice to divided doses of 15 MeV electrons. Int J Radiat Biol Relat Stud Phys Chem Med. 1967; 12: 293–6. 35. Thomlinson R, Gray L: The histological structure of some human lung cancers and the possible implications for radiotherapy. Br J Cancer. 1955; 9: 539–49. 36. Rockwell S, Moulder JE, Martin D: Tumor-to-tumor variability in the hypoxic fractions of experimental rodent tumors. Radiother Oncol. 1984; 2: 57–64. 37. Moulder JE, Rockwell S: Hypoxic fractions of solid tumors: experimental techniques, methods of analysis, and a survey of existing data. Int J Radiat Oncol Biol Phys. 1984; 10: 695–712. 38. van Putten L, Kallman R: Oxygenation status of a transplantable tumor during fractionated radiation therapy. J Natl Cancer Inst. 1968; 40: 441–51. 39. Kallman RF, Jardine LJ, Johnson CW: Effects of different schedules of dose fractionation on the oxygenation status of a transplantable mouse sarcoma. J Natl Cancer Inst. 1970; 44: 369–77. 40. Brown JM: Evidence for acutely hypoxic cells in mouse tumours, and a possible mechanism of reoxygenation. Br J Radiol. 1979; 52: 650–6. 41. Withers H, Taylor J, Maciejewski B: The hazard of accelerated tumor clonogen repopulation during radiotherapy. Acta Oncol. 1988; 27: 131–46. 42. Maciejewski B, Withers HR, Taylor JM, et al: Dose fractionation and regeneration in radiotherapy for cancer of the oral cavity and oropharynx: tumor dose-response and repopulation. Int J Radiat Oncol Biol Phys. 1989; 16: 831–43. 43. Fu KK, Pajak TF, Trotti A, et al: A Radiation Therapy Oncology Group (RTOG) phase III randomized study to compare hyperfractionation and two variants of accelerated fractionation to standard fractionation radiotherapy for head and neck squamous cell carcinomas: first report of RTOG 9003. Int J Radiat Oncol Biol Phys. 2000; 48: 7–16. 44. Overgaard J, Hansen HS, Specht L, et  al: Five compared with six fractions per week of conventional radiotherapy of squamous-cell carcinoma of head and neck: DAHANCA 6 and 7 randomised controlled trial. Lancet. 2003; 362: 933–40. 45. Bourhis J, Overgaard J, Audry H, et  al: Hyperfractionated or accelerated radiotherapy in head and neck cancer: a meta-analysis. Lancet. 2006; 368: 843–54. 46. Turrisi AT 3rd, Kim K, Blum R, et al: Twice-daily compared with once-daily thoracic radiotherapy in limited small-cell lung cancer treated concurrently with cisplatin and etoposide. N Engl J Med. 1999; 340: 265–71. 47. Steel GG: Exploitable mechanisms in combined radiotherapy-chemotherapy: the concept of additivity. Int J Radiat Oncol Biol Phys. 1979; 5: 85–91. 48. Nias AH: Radiation and platinum drug interaction. Int J Radiat Biol Relat Stud Phys Chem Med. 1985; 48: 297–314. 49. Murthy AK, Rossof AH, Anderson KM, Hendrickson FR: Cytotoxicity and influence on radiation dose response curve of cis-diamminedichloroplatinum II (cis-DDP). Int J Radiat Oncol Biol Phys. 1979; 5: 1411–5. 50. Luk KH, Ross GY, Phillips TL, Goldstein LS: The interaction of radiation and cis-diamminedichloroplatinum (II) in intestinal crypt cells. Int J Radiat Oncol Biol Phys. 1979; 5: 1417–20.

17  Combined Modality Therapy in Cancer Management

511

51. Overgaard J, Khan AR: Selective enhancement of radiation response in a C3H mammary carcinoma by cisplatin. Cancer Treat Rep. 1981; 65: 501–3. 52. Dewitt L: Combined treatment of radiation and cis-diamminedichloroplatinum (II): a review of experimental and clinical data. Int J Radiat Oncol Biol Phys. 1987; 13: 403–26. 53. Frit P, Canitrot Y, Muller C, et al: Cross-resistance to ionizing radiation in a murine leukemic cell line resistant to cis-dichlorodiammineplatinum(II): role of Ku autoantigen. Mol Pharmacol. 1999; 56: 141–6. 54. Turchi JJ, Henkels KM, Zhou Y: Cisplatin-DNA adducts inhibit translocation of the Ku subunits of DNA-PK. Nucleic Acids Res. 2000; 28: 4634–41. 55. Boeckman HJ, Trego KS, Turchi JJ: Cisplatin sensitizes cancer cells to ionizing radiation via inhibition of nonhomologous end joining. Mol Cancer Res. 2005; 3: 277–85. 56. Bepler G, Kusmartseva I, Sharma S, et al: RRM1 modulated in vitro and in vivo efficacy of gemcitabine and platinum in non-small-cell lung cancer. J Clin Oncol. 2006; 24: 4731–7. 57. Olaussen KA, Dunant A, Fouret P, et al: DNA repair by ERCC1 in non-small-cell lung cancer and cisplatin-based adjuvant chemotherapy. N Engl J Med. 2006; 355: 983–91. 58. Simon GR, Ismail-Khan R, Bepler G: Nuclear excision repair-based personalized therapy for non-small cell lung cancer: from hypothesis to reality. Int J Biochem Cell Biol. 2007; 39: 1318–28. 59. Fink D, Aebi S, Howell SB: The role of DNA mismatch repair in drug resistance. Clin Cancer Res. 1998; 4: 1–6. 60. Aebi S, Fink D, Gordon R, et al: Resistance to cytotoxic drugs in DNA mismatch repairdeficient cells. Clin Cancer Res. 1997; 3: 1763–7. 61. Rave-Fränk M, Schmidberger H, Christiansen H, et al: Comparison of the combined action of oxaliplatin or cisplatin and radiation in cervical and lung cancer cells. Int J Radiat Biol. 2007; 83: 41–7. 62. Choy H, Park C, Yao M: Current status and future prospects for satraplatin, an oral platinum analogue. Clin Cancer Res. 2008; 14: 1633–8. 63. Davis MA, Tang HY, Maybaum J, et al: Dependence of fluorodeoxyuridine-mediated radiosensitization on S phase progression. Int J Radiat Oncol Biol Phys. 1995; 67: 509–17. 64. Byfield JE, Calabro-Jones P, Klisak I, Kulhanian F: Pharmacologic requirements for obtaining sensitization of human tumor cells in vitro to combined 5-Fluorouracil or ftorafur and X rays. Int J Radiat Oncol Biol Phys. 1982; 8: 1923–33. 65. Schüller J, Cassidy J, Dumont E, et al: Preferential activation of capecitabine in tumor following oral administration to colorectal cancer patients. Cancer Chemother Pharmacol. 2000; 45: 291–7. 66. Sawada N, Ishikawa T, Sekiguchi F, et al: X-ray irradiation induces thymidine phosphorylase and enhances the efficacy of capecitabine (Xeloda) in human cancer xenografts. Clin Cancer Res. 1999; 5: 2948–53. 67. Terasima T, Tolmach LJ: Changes in x-ray sensitivity of HeLa cells during the division cycle. Nature 1961; 190: 1210–1. 68. Kaina B, Ziouta A, Ochs K, et  al: Chromosomal instability, reproductive cell death and apoptosis induced by O6-methylguanine in Mex–, Mex_ and methylation-tolerant mismatch repair compromised cells: facts and models. Mutat Res. 1997; 381: 227–41. 69. Hirose Y, Berger MS, Pieper RO: p53 effects both the duration of G2/M arrest and the fate of temozolomide-treated human glioblastoma cells. Cancer Res. 2001; 61: 1957–63. 70. Hirose Y, Berger MS, Pieper RO: Abrogation of the Chk1-mediated G(2) checkpoint pathway potentiates temozolomide-induced toxicity in a p53-independent manner in human glioblastoma cells. Cancer Res. 2001; 61: 5843–9. 71. Hirose Y, Kreklau EL, Erickson LC, et  al: Delayed repletion of O6-methylguanine-DNA methyltransferase resulting in failure to protect the human glioblastoma cell line SF767 from temozolomide induced cytotoxicity. J Neurosurg. 2003; 98: 591–8. 72. Wild-Bode C, Weller M, Rimner A, et  al: Sublethal irradiation promotes migration and invasiveness of glioma cells: implications for radiotherapy of human glioblastoma. Cancer Res. 2001; 61: 2744–50.

512

D. Raben and K. Rusthoven

73. Hegi ME, Diserens AC, Gorlia T, et al: MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med. 2005; 352: 997–1003. 74. Merlano M, Vitale V, Rosso R, et al. Treatment of advanced squamous-cell carcinoma of the head and neck with alternating chemotherapy and radiotherapy. N Engl J Med. 1992; 327: 1115–21. 75. Merlano M, Benasso M, Corvo R, et al: Five-year update of a randomized trial of alternating radiotherapy and chemotherapy compared with radiotherapy alone in treatment of unresectable squamous cell carcinoma of the head and neck. J Natl Cancer Inst. 1996; 88: 583–9. 76. Adelstein DJ, Li Y, Adams GL, et al: An intergroup phase III comparison of standard radiation therapy and two schedules of concurrent chemoradiotherapy in patients with unresectable squamous cell head and neck cancer. J Clin Oncol. 2003; 21: 92–8. 77. Al-Sarraf M, LeBlanc M, Shanker Giri PG, et al: Chemoradiotherapy versus radiotherapy in patients with advanced nasopharyngeal cancer: phase III randomized intergroup study 0099. J Clin Oncol. 1998; 16: 1310–7. 78. Calais G, Alfonsi M, Bardet E, et al: Randomized trial of radiation therapy versus concomitant chemotherapy and radiation therapy for advanced-stage oropharynx carcinoma. J Natl Cancer Inst. 1999; 91: 2081–6. 79. Wendt TG, Grabenbauer GG, Rodel CM, et  al: Simultaneous radiochemotherapy versus radiotherapy alone in advanced head and neck cancer: a randomized multicenter study. J Clin Oncol. 1998; 16: 1318–24. 80. Brizel DM, Albers ME, Fisher SR, et al: Hyperfractionated irradiation with or without concurrent chemotherapy for locally advanced head and neck cancer. N Engl J Med. 1998; 338: 1798–804. 81. Jeremic B, Shibamoto Y, Milicic B, et al: Hyperfractionated radiation therapy with or without concurrent low-dose daily cisplatin in locally advanced squamous cell carcinoma of the head and neck: a prospective randomized trial. J Clin Oncol. 2000; 18: 1458–64. 82. Semrau R, Mueller RP, Stuetzer H, et al: Efficacy of intensified hyperfractionated and accelerated radiotherapy and concurrent chemotherapy with carboplatin and 5-fluorouracil: updated results of a randomized multicentric trial in advanced head-and-neck cancer. Int J Radiat Oncol Biol Phys. 2006; 64: 1308–16. 83. Forastiere AA, Goepfert H, Maor M, et al: Concurrent chemotherapy and radiotherapy for organ preservation in advanced laryngeal cancer. N Engl J Med. 2003; 349: 2091–8. 84. Bonner JA, Harari PM, Giralt J, et al: Radiotherapy plus cetuximab for squamous-cell carcinoma of the head and neck. N Engl J Med. 2006; 354: 567–78. 85. Pignon JP, Bourhis J, Domenge C, et al: Chemotherapy added to locoregional treatment for head and neck squamous-cell carcinoma: three meta-analyses of updated individual data. Lancet. 2000; 355: 949–55. 86. The Department of Veterans Affairs Laryngeal Cancer Study Group: Induction chemotherapy plus radiation compared with surgery plus radiation in patients with advanced laryngeal cancer. N Engl J Med. 1991; 324: 1685–90. 87. Lefebvre JL, Chevalier D, Luboinski B, et al: Larynx preservation in pyriform sinus cancer: preliminary results of a European Organization for Research and Treatment of Cancer phase III trial. EORTC Head and Neck Cancer Cooperative Group. J Natl Cancer Inst. 1996; 88: 890–9. 88. Cooper JS, Pajak TF, Forastierre AA, et al: Postoperative concurrent radiotherapy and chemotherapy for high-risk squamous-cell carcinoma of the head and neck. N Engl J Med. 2004; 350: 1937–44. 89. Bernier J, Domenge C, Ozsahin M, et al: Postoperative irradiation with or without concomitant chemotherapy for locally advanced head and neck cancer. N Engl J Med. 2004; 350: 1945–52. 90. Bernier J, Cooper JS, Pajak TF, et al: Defining risk levels in locally advanced head and neck cancer: a comparative analysis of concurrent postoperative radiation plus chemotherapy trials of the EORTC (#22931) and RTOG (#9501). Head Neck. 2005; 27: 843–50. 91. Rubin Grandis J, Melhem MF, Gooding WE, et al: Levels of TGF-alpha and EGFR protein in head and neck squamous cell carcinoma and patient survival. J Natl Cancer Inst. 1998; 90: 824–32.

17  Combined Modality Therapy in Cancer Management

513

92. Ang KK, Berkey BA, Tu X, et al: Impact of epidermal growth factor receptor expression on survival and pattern of relapse in patients with advanced head and neck carcinoma. Cancer Res. 2002; 62: 7350–6. 93. Liang K, Ang KK, Milas L, et al: The epidermal growth factor receptor mediates radioresistance. Int J Radiat Oncol Biol Phys. 2003; 57: 246–54. 94. Chung CH, Ely K, McGavran L, Varella-Garcia M, et al: Increased epidermal growth factor receptor gene copy number is associated with poor prognosis in head and neck squamous cell carcinoma. J Clin Oncol. 2006; 24: 4170–6. 95. Chung CH, Parker J, Levy S, et  al: Gene expression profiles as markers of aggressive ­disease-EGFR as a factor. Int J Radiat Oncol Biol Phys. 2007; 69: S1. 96. Bonner JA, Giralt J, Harari PM, et al: Prolongation of survival with the addition of cetuximab to radiation in patients with locoregionally advanced head and neck cancer (SCCHN): fiveyear results from a randomized trial. American Society for Therapeutic Radiology and Oncology 2008 Annual Meeting Late breaking Abstracts. Int J Radiat Oncol Biol Phys. 2008; 72(1): e1–4, DOI: 10.1016/S0360-3016(08)03646-8 (LB3). 97. Bernier J, Schneider D: Cetuximab combined with radiotherapy: an alternative to chemoradiotherapy for patients with locally advanced squamous cell carcinomas of the head and neck? Eur J Cancer. 2007; 43: 35–45. 98. Perez CA, Pajak TF, Rubin P, et  al: Long-term observations of the patterns of failure in patients with unresectable non-oat cell carcinoma of the lung treated with definitive radiotherapy. Cancer. 1987; 59: 1874–81. 99. Dillman RO, Seagren SL, Propert KJ, et al: A randomized trial of induction chemotherapy plus high-dose radiation versus radiation alone in stage III non-small-cell lung cancer. N Engl J Med. 1990; 323: 940–5. 100. Dillman RO, Herndon J, Seagren SL, et al: Improved survival in stage III non-small-cell lung cancer: seven-year follow-up of cancer and leukemia group B (CALGB) 8433 trial. J Natl Cancer Inst. 1996; 88: 1210–5. 101. Sause WT, Scott C, Taylor S, et al: Radiation Therapy Oncology Group (RTOG) 88-08 and Eastern Cooperative Oncology Group (ECOG) 4588: preliminary results of a phase III trial in regionally advanced, unresectable non-small-cell lung cancer. J Natl Cancer Inst. 1995; 87: 198–205. 102. Schaake-Koning C, van den Bogaert W, Dalesio O, et al: Effects of concomitant cisplatin and radiotherapy on inoperable non-small-cell lung cancer. N Engl J Med. 1992; 326: 524–30. 103. Curran WJ, Scott CB, Langer CJ, et al: Long-term benefit is observed in a phase III comparison of sequential vs concurrent chemo-radiation for patients with unresected stage III NSCLC: RTOG 9410. Proc Am Soc Clin Oncol. 22: 2003; S2499. 104. Furuse K, Fukuoka M, Kawahara M, et al: Phase III study of concurrent versus sequential thoracic radiotherapy in combination with mitomycin, vindesine, and cisplatin in unresectable stage III non-small-cell lung cancer. J Clin Oncol. 1999; 17: 2692–9. 105. Fournel P, Robinet G, Thomas P, et al: Randomized phase III trial of sequential chemoradiotherapy compared with concurrent chemoradiotherapy in locally advanced non-small-cell lung cancer: Groupe Lyon-Saint-Etienne d’Oncologie Thoracique-Groupe Français de Pneumo-Cancérologie NPC 95-01 Study. J Clin Oncol. 2005; 23: 5910–7. 106. Zatloukal P, Petruzelka L, Zemanova M, et  al: Concurrent versus sequential chemoradiotherapy with cisplatin and vinorelbine in locally advanced non-small cell lung cancer: a randomized study. Lung Cancer. 2004; 46: 87–98. 107. Gandara DR, Chansky K, Albain KS, et al: Consolidation docetaxel after concurrent chemoradiotherapy in stage IIIB non-small-cell lung cancer: phase II Southwest Oncology Group Study S9504. J Clin Oncol. 2003; 21: 2004–10. 108. Hanna N, Neubauer M, Yiannoutsos C, et  al: Phase III study of cisplatin, etoposide, and concurrent chest radiation with or without consolidation docetaxel in patients with inoperable stage III non-small-cell lung cancer: the Hoosier Oncology Group and U.S. Oncology. J Clin Oncol. 2008; 26: 5755–60.

514

D. Raben and K. Rusthoven

109. Rusch V, Baselga J, Cordon-Cardo C, et al. Differential expression of the epidermal growth factor receptor and its ligands in primary non-small cell lung cancers and adjacent benign lung. Cancer Res. 1993; 53: 2379–85. 110. Rusch V, Klimstra D, Venkatraman E, et al. Overexpression of the epidermal growth factor receptor and its ligand transforming growth factor alpha is frequent in resectable non-small cell lung cancer but does not predict tumor progression. Clin Cancer Res. 1997; 3: 515–22. 111. Blumenschein GR, Paulus R, Curran WJ, et  al: A phase II study of cetuximab (C225) in combination with chemoradiation (CRT) in patients (PTS) with stage IIIA/B non-small cell lung cancer (NSCLC): a report of the 2 year and median survival (MS) for the RTOG 0324 trial. Proc Am Soc Clin Oncol. 2008; 26: S7516. 112. Radiation Therapy Oncology website: Protocol RTOG 0617. Available at: www.rtog.org (accessed 25 February 2009). 113. Rose PG, Ali S, Watkins E, Thigpen JT, et al: Long-term follow-up of a randomized trial comparing concurrent single agent cisplatin, cisplatin-based combination chemotherapy, or hydroxyurea during pelvic irradiation for locally advanced cervical cancer: a Gynecologic Oncology Group Study. J Clin Oncol. 2007; 25: 2804–10. 114. Eifel PJ, Winter K, Morris M, et al: Pelvic irradiation with concurrent chemotherapy versus pelvic and para-aortic irradiation for high-risk cervical cancer: an update of radiation therapy oncology group trial (RTOG) 90-01. J Clin Oncol. 2004; 22: 872–80. 115. Sedlis A, Bundy BN, Rotman MZ, et al: A randomized trial of pelvic radiation therapy versus no further therapy in selected patients with stage IB carcinoma of the cervix after radical hysterectomy and pelvic lymphadenectomy: A Gynecologic Oncology Group Study. Gynecol Oncol. 1999; 73: 177–83. 116. Landoni F, Maneo A, Colombo A, et al: Randomised study of radical surgery versus radiotherapy for stage Ib-IIa cervical cancer. Lancet. 1997; 350: 535–40. 117. National Cancer Institute website: Available at: www.cancer.gov (accessed 25 February 2009). 118. Peters WA 3rd, Liu PY, Barrett RJ 2nd, et al: Concurrent chemotherapy and pelvic radiation therapy compared with pelvic radiation therapy alone as adjuvant therapy after radical surgery in high-risk early-stage cancer of the cervix. J Clin Oncol. 2000; 18: 1606–13. 119. Keys HM, Bundy BN, Stehman FB, et al: Cisplatin, radiation, and adjuvant hysterectomy compared with radiation and adjuvant hysterectomy for bulky stage IB cervical carcinoma. N Engl J Med. 1999; 340: 1154–61. 120. Morris M, Eifel PJ, Lu J, et  al: Pelvic radiation with concurrent chemotherapy compared with pelvic and para-aortic radiation for high-risk cervical cancer. N Engl J Med. 1999; 340: 1137–43. 121. Rose PG, Bundy BN, Watkins EB, et al: Concurrent cisplatin-based radiotherapy and chemotherapy for locally advanced cervical cancer. N Engl J Med. 1999; 340: 1144–53. 122. Whitney CW, Sause W, Bundy BN, et al: Randomized comparison of fluorouracil plus cisplatin versus hydroxyurea as an adjunct to radiation therapy in stage IIB-IVA carcinoma of the cervix with negative para-aortic lymph nodes: a Gynecologic Oncology Group and Southwest Oncology Group study. J Clin Oncol. 1999; 17: 1339–48. 123. Pearcey R, Brundage M, Drouin P, et al: Phase III trial comparing radical radiotherapy with and without cisplatin chemotherapy in patients with advanced squamous cell cancer of the cervix. J Clin Oncol. 2002; 20: 966–72. 124. Lanciano R, Calkins A, Bundy BN, et al: Randomized comparison of weekly cisplatin or protracted venous infusion of fluorouracil in combination with pelvic radiation in advanced cervix cancer: a gynecologic oncology group study. J Clin Oncol. 2005; 23: 8289–95. 125. Gynecology Oncology Group website: Protocol GOG 219. Available at: www.gog.org (accessed 25 February 2009). 126. Radiation Therapy Oncology Group website: Protocol RTOG 0417. Available at: www.rtog. org (accessed 25 February 2009). 127. Emami B, Lyman J, Brown A, et al: Tolerance of normal tissue to therapeutic irradiation. Int J Radiat Oncol Biol Phys. 1991; 21: 109–22.

17  Combined Modality Therapy in Cancer Management

515

128. Komaki R, Liao Z, Forster K, et al: Target definition and contouring in carcinoma of the lung and esophagus. Rays. 2003; 28: 225–36. 129. Cooper JS, Guo MD, Herskovic A, et al: Chemoradiotherapy of locally advanced esophageal cancer: long-term follow-up of a prospective randomized trial (RTOG 85-01). Radiation Therapy Oncology Group. JAMA. 1999; 281: 1623–7. 130. Hulscher JB, van Sandick JW, de Boer AG, et al: Extended transthoracic resection compared with limited transhiatal resection for adenocarcinoma of the esophagus. N Engl J Med. 2002; 347: 1662–9. 131. Stahl M, Stuschke M, Lehmann N, et  al: Chemoradiation with and without surgery in patients with locally advanced squamous cell carcinoma of the esophagus. J Clin Oncol. 2005; 23: 2310–7. 132. Bedenne L, Michel P, Bouché O, et al: Chemoradiation followed by surgery compared with chemoradiation alone in squamous cancer of the esophagus: FFCD 9102. J Clin Oncol. 2007; 25: 1160–8. 133. Herskovic A, Martz K, Al-Sarraf M, et al: Combined chemotherapy and radiotherapy compared with radiotherapy alone in patients with cancer of the esophagus. N Engl J Med. 1992; 326: 1593–8. 134. Al-Sarraf M, Martz K, Herskovic A, et al: Progress report of combined chemoradiotherapy versus radiotherapy alone in patients with esophageal cancer: an intergroup study. J Clin Oncol. 1997; 15: 277–84. 135. Smith TJ, Ryan LM, Douglass HO Jr, et al: Combined chemoradiotherapy vs. radiotherapy alone for early stage squamous cell carcinoma of the esophagus: a study of the Eastern Cooperative Oncology Group. Int J Radiat Oncol Biol Phys. 1998; 42: 269–76. 136. Radiation Therapy Oncology Group website: Protocol RTOG 0436. Available at: www.rtog. org (accessed 25 February 2009). 137. Thirion P, Maillard E, Pignon J, et al: Individual patient data-based meta-analysis assessing the effect of preoperative chemo-radiotherapy in respectable oesophagel cancer. Proc Am Soc Ther Radiol Oncol. 2008; 71: S158. 138. Bosset JF, Gignoux M, Triboulet JP, et al: Chemoradiotherapy followed by surgery compared with surgery alone in squamous-cell cancer of the esophagus. N Engl J Med. 1997; 337: 161–7. 139. Macdonald JS, Smalley SR, Benedetti J, et al: Chemoradiotherapy after surgery compared with surgery alone for adenocarcinoma of the stomach or gastroesophageal junction. N Engl J Med. 2001; 345: 725–30. 140. Cunningham D, Allum WH, Stenning SP, et al: Perioperative chemotherapy versus surgery alone for resectable gastroesophageal cancer. N Engl J Med. 2006; 355: 11–20. 141. Kalser MH, Ellenberg SS: Pancreatic cancer. Adjuvant combined radiation and chemotherapy following curative resection. Arch Surg. 1985; 120: 899–903. 142. Klinkenbijl JH, Jeekel J, Sahmoud T, et al: Adjuvant radiotherapy and 5-fluorouracil after curative resection of cancer of the pancreas and periampullary region: phase III trial of the EORTC gastrointestinal tract cancer cooperative group. Ann Surg. 1999; 230: 776–82. 143. Neoptolemos JP, Stocken DD, Friess H, et al: A randomized trial of chemoradiotherapy and chemotherapy after resection of pancreatic cancer. N Engl J Med. 2004; 350: 1200–10. 144. Koshy MC, Landry JC, Cavanaugh SX, et  al: A challenge to the therapeutic nihilism of ESPAC-1. Int J Radiat Oncol Biol Phys. 2005; 61: 965–6. 145. Moertel CG, Frytak S, Hahn RG, et  al: Therapy of locally unresectable pancreatic carcinoma: a randomized comparison of high dose (6000 rads) radiation alone, moderate dose radiation (4000 rads + 5-fluorouracil), and high dose radiation + 5-fluorouracil: The Gastrointestinal Tumor Study Group. Cancer. 1981; 48: 1705–10. 146. Moertel CG, Childs DS Jr, Reitemeier RJ, et al: Combined 5-fluorouracil and supervoltage radiation therapy of locally unresectable gastrointestinal cancer. Lancet. 1969; 2: 865–7. 147. Cohen SJ, Dobelbower R Jr, Lipsitz S, et al: A randomized phase III study of radiotherapy alone or with 5-fluorouracil and mitomycin-C in patients with locally advanced adenocarcinoma of the pancreas: Eastern Cooperative Oncology Group study E8282. Int J Radiat Oncol Biol Phys. 2005; 62: 1345–50.

516

D. Raben and K. Rusthoven

148. Gastrointestinal Tumor Study Group: Treatment of locally unresectable carcinoma of the pancreas: comparison of combined-modality therapy (chemotherapy plus radiotherapy) to chemotherapy alone. J Natl Cancer Inst. 1988; 80: 751–5. 149. Klaassen DJ, MacIntyre JM, Catton GE, et al: Treatment of locally unresectable cancer of the stomach and pancreas: a randomized comparison of 5-flourouracil alone with radiation plus concurrent and maintenance 5-flourouracil – An Eastern Cooperative Oncology Group study. J Clin Oncol. 1985; 3: 373–8. 150. Loehrer PJ, Powell ME, Cardenes HR, et al: A randomized phase III study of gemcitabine in combination with radiation therapy versus gemcitabine alone in patients with localized, unresectable pancreatic cancer: E4201. J Clin Oncol. 2008; 26: S4506. 151. Gastrointestinal Tumor Study Group: Prolongation of the disease-free interval in surgically treated rectal carcinoma. N Engl J Med. 1985; 312: 1465–72. 152. Krook JE, Moertel CG, Gunderson LL, et al: Effective surgical adjuvant therapy for highrisk rectal carcinoma. N Engl J Med. 1991; 324: 709–15. 153. Wolmark N, Wieand HS, Hyams DM, et  al: Randomized trial of postoperative adjuvant chemotherapy with or without radiotherapy for carcinoma of the rectum: National Surgical Adjuvant Breast and Bowel Project Protocol R-02. J Natl Cancer Inst. 2000; 92: 388–96. 154. Bosset JF, Collette L, Calais G, et al: Chemotherapy with preoperative radiotherapy in rectal cancer. N Engl J Med. 2006; 355: 1114–23. 155. Gérard JP, Conroy T, Bonnetain F, et al: Preoperative radiotherapy with or without concurrent fluorouracil and leucovorin in T3-4 rectal cancers: results of FFCD 9203. J Clin Oncol. 2006; 24: 4620–5. 156. Braendengen M, Tveit KM, Berglund A, et al: Randomized phase III study comparing preoperative radiotherapy with chemoradiotherapy in nonresectable rectal cancer. J Clin Oncol. 2008; 26: 3687–94. 157. Bujko K, Nowacki MP, Nasierowska-Guttmejer A, et al: Sphincter preservation following preoperative radiotherapy for rectal cancer: report of a randomised trial comparing shortterm radiotherapy vs. conventionally fractionated radiochemotherapy. Radiother Oncol. 2004; 72: 15–24. 158. Bujko K, Nowacki MP, Nasierowska-Guttmejer A, et al: Long-term results of a randomized trial comparing preoperative short-course radiotherapy with preoperative conventionally fractionated chemoradiation for rectal cancer. Br J Surg. 2006; 93: 1215–23. 159. UKCCCR Anal Cancer Trial Working Party: Epidermoid anal cancer: results from the UKCCCR randomised trial of radiotherapy alone versus radiotherapy, 5-fluorouracil, and mitomycin. Lancet. 1996; 348: 1049–54. 160. Bartelink H, Roelofsen F, Eschwege F, et al: Concomitant radiotherapy and chemotherapy is superior to radiotherapy alone in the treatment of locally advanced anal cancer: results of a phase III randomized trial of the European Organization for Research and Treatment of Cancer Radiotherapy and Gastrointestinal Cooperative Groups. J Clin Oncol. 1997; 15: 2040–9. 161. Stupp R, Mason WP, van den Bent MJ, et al: Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med. 2005; 352: 987–96. 162. Wee J, Tan EH, Tai BC, et al: Randomized trial of radiotherapy versus concurrent chemoradiotherapy followed by adjuvant chemotherapy in patients with American Joint Committee on Cancer/International Union against cancer stage III and IV nasopharyngeal cancer of the endemic variety. J Clin Oncol. 2005; 23: 6730–8. 163. Lin JC, Jan JS, Hsu CY, et  al: Phase III study of concurrent chemoradiotherapy versus radiotherapy alone for advanced nasopharyngeal carcinoma: positive effect on overall and progression-free survival. J Clin Oncol. 2003; 21: 631–7. 164. Chan AT, Leung SF, Ngan RK, et al: Overall survival after concurrent cisplatin-radiotherapy compared with radiotherapy alone in locoregionally advanced nasopharyngeal carcinoma. J Natl Cancer Inst. 2005; 97: 536–9.

17  Combined Modality Therapy in Cancer Management

517

165. Lee AW, Lau WH, Tung SY, et al: Preliminary results of a randomized study on therapeutic gain by concurrent chemotherapy for regionally-advanced nasopharyngeal carcinoma: ­NPC-9901 Trial by the Hong Kong Nasopharyngeal Cancer Study Group. J Clin Oncol. 2005; 23: 6966–75. 166. Bachaud JM, Cohen-Jonathan E, Alzieu C, et al: Combined postoperative radiotherapy and weekly cisplatin infusion for locally advanced head and neck carcinoma: final report of a randomized trial. Int J Radiat Oncol Biol Phys. 1996; 36: 999–1004. 167. Walsh TN, Noonan N, Hollywood D, et al: A comparison of multimodal therapy and surgery for esophageal adenocarcinoma. N Engl J Med. 1996; 335: 462–7. 168. Burmeister BH, Smithers BM, Gebski V, et al: Surgery alone versus chemoradiotherapy followed by surgery for resectable cancer of the oesophagus: a randomised controlled phase III trial. Lancet Oncol. 2005; 6: 659–68. 169. Urba SG, Orringer MB, Turrisi A, et al: Randomized trial of preoperative chemoradiation versus surgery alone in patients with locoregional esophageal carcinoma. J Clin Oncol. 2001; 19: 305–13. 170. Tepper J, Krasna MJ, Niedzwiecki D, et al: Phase III trial of trimodality therapy with cisplatin, fluorouracil, radiotherapy, and surgery compared with surgery alone for esophageal cancer: CALGB 9781.J Clin Oncol. 2008; 26: 1086–92.

Chapter 18

Cancer Vaccines Daniel Laheru

18.1 Introduction Cancer vaccines offer the unique opportunity to provide specific and direct antitumor recognition and killing by recruiting both T- and B-cell arms of the immune system while avoiding nonspecific toxicities. Because of this exquisite sensitivity and specificity, cancer vaccines in theory could also be safely integrated with surgery, radiation, and chemotherapy. Thus, the major advantage of immune-based therapies lies in their ability to specifically target the transformed tumor cell relative to the normal cell of origin. While a number of tumor-specific antigens have been reported most notably in melanoma and renal cell cancer [1–3], the clinical translation into the development of effective immunotherapy has been to date limited [4–7]. These observations have revealed that the immunologic interaction between tumor and host is complex and involves a delicate balance of tumor antigen recognition vs. tumor escape through immune regulatory pathways [8, 9]. As we begin to understand more about these mechanisms of immune modulation, new opportunities for immunotherapy have emerged. A number of novel immunotherapeutic approaches have been developed. They range from antigen-targeted immunotoxins to vaccines that enhance tumor-specific antibody and cellular responses. A few pancreatic cancer-associated antigens have now been identified as candidate targets of both antibody and cellular responses, particularly T-cell responses. This section will review the important features of an effective antitumor immune response, discuss the results of some of the more promising strategies that are currently under clinical development, and foreshadow what can be expected in the near future. D. Laheru (*) Skip Viragh Center for Pancreas Cancer Clinical Research and Patient Care, The Sol Goldman Pancreatic Cancer Research Center, The Sidney Kimmel Comprehensive Cancer Center Bunting-Blaustein, The Johns Hopkins University School of Medicine, CRB Room G89, 1650 Orleans Street, Baltimore, MD 21231, USA e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_18, © Springer Science+Business Media, LLC 2011

519

520

D. Laheru

18.2 Features of the Immune System That Are Required for Successful Cancer Immunotherapy The immune system comprises a number of cell types which, when activated, are extremely efficient at recognizing and killing their target [10]. In particular, B cells and T cells each possess vast arrays of clonally distributed antigen receptors that enable them to recognize foreign antigens and to discriminate self from nonself. In fact, it has been estimated that both T and B cells can express more than a million different antigen-specific receptors through recombination of the genes encoding for their receptor at the time of maturation in the thymus and bone marrow, respectively. Therefore, the immune system should have an unlimited capability to recognize antigenic differences between normal and malignant cells, whether they are in the form of the product of a new genetic alteration, a reactivated embryonic gene, or an overexpressed gene. Other cell types that are likely involved in immune recognition of cancer include professional antigen-presenting cells (APCs), particularly dendritic cells (DCs) and macrophages, and natural killer cells (NK cells).

18.2.1 Tumors Use Multiple Mechanisms To Evade Immune Recognition Despite these unique features of the immune system that make it possible to recognize self from nonself, it has long been known that human cancers are, in general, poorly immunogenic. In order for cancers to proliferate, tumor cells must develop local and systemic mechanisms that allow them to escape immune recognition [11]. Dissection of the complex mechanisms involved in immune tolerance to tumors is ongoing in preclinical models and in patients. For immune-based therapies to be successful at treating cancer, they must incorporate strategies that can bypass these mechanisms of immune evasion. Pertinent features of these local and systemic mechanisms by which tumors evade immune recognition are described below and in Table 18.1. It is also now clear that the context in which the antigen is presented to the T cell determines whether or not the T cell subsequently becomes activated or suppressed. In the absence of the appropriate co-stimulatory signals, engagement of the T-cell receptor (TCR) can lead to ignorance, anergy, or even apoptotic T-cell death. There are a number of families of regulatory molecules that have been identified and that play a role in T-cell activation/ regulation. The B7 family is the best characterized. What is now known is that these molecules have both stimulating and downregulatory ligands (see Fig. 18.1).

18.2.1.1 Local Processes It is clear that the tumor and the surrounding microenvironment play a critical role in immune evasion. In the local tumor environment, the loss of appropriate

18  Cancer Vaccines

521

Table 18.1  Mechanisms of tumor immune evasion Alteration to immune response Local factors Immune cell molecules HLA class I; TAP; Beta-2 downregulated microglobulin Immune cell inhibitors IL-10; TGF-beta; COX-2; upregulated VEGF; B7-H1; B7-H4 Immune checkpoints B7-H1 signaling disrupted suppressed by tumors

Defects in immune cell localization

Loss of co-stimulation

T-regulatory cell accumulation in tumors; peripheral deletion of activated T cells via T-regulatory cells N/A

Cellular effects

T-cell apoptosis increased

Systemic factors

IL-1; IL-6; IL-10; TGF-beta B7.1/B7.2/CTLA-4 signaling by dendritic cells; B7-DC/PD-1 signaling by dendritic cells; B7-H1/PD-1 signaling by dendritic cells Peripheral deletion of activated T cells via T-regulatory cells

B7 family of molecules; OX-40; CD-40 Inhibition of dendritic cell maturation by production of: VEGF; COX-2

Local-direct cell–cell interaction; systemic-cytokine mediated IL interleukin, TGF tumor growth factor, COX cyclooxygenase, VEGF vascular endothelial growth factor, TAP transport-associated protein

s­ ignaling may occur by downregulation of human leukocyte antigen (HLA) class I molecules through a variety of processes such as loss of expression of HLA class I alleles and b-2 microglobulin, and downregulation of transporter associated with antigen processing (TAP) which transports antigenic peptides to the site in the cell where they bind to newly synthesized HLA molecules for presentation to T cells. Such alterations within the tumor cell are not uncommon since tumors have unstable genomes. The instability of the genome is also responsible for the frequently described loss of tumor antigen expression. In addition to the loss of appropriate antigen recognition molecules on the tumor cell, tumor cells naturally lack expression of co-stimulation molecules, which are required for effective T-cell activation. Additional local mechanisms of immune escape are known to occur including tumor cell secretion of immunosuppressive cytokines such as tumor growth factor (TGF)-b1, TGF-b2, and IL-10 which can further limit host defenses by interfering with the production of inflammatory cytokines and/or the recruitment of APCs to the tumor microenvironment [12, 13].

18.2.1.2 Systemic Processes Efficient immune priming against tumor cells is dependent on TCR recognition of specific peptide fragments derived from the tumor cell in the context of the appropriate

522

D. Laheru

B7-1 (CD80) B7-2 (CD86)

CD28

++

CD152 (CTLA) B7-H1 (PD-L1) B7-DC (PD-L2)

B7-H4

Antigen Presenting Cell/Tumor Cell

--

PD-1

--

? receptor

++

? receptor

--

T cell

Fig.  18.1  T-cell activation vs. suppression. Efficient immune priming against tumor cells is dependent on T-cell receptor recognition of specific peptide fragments derived from the tumor cell and processed by the antigen-presenting cell in the context of the appropriate HLA class I molecule and co-stimulatory molecule. The context in which the antigen is presented to the T cell determines whether or not the T cell subsequently becomes activated or suppressed. In the absence of the appropriate co-stimulatory signals, engagement of the T-cell receptor can lead to ignorance, anergy, or even apoptotic T-cell death. There are a number of families of regulatory molecules that have been identified and that play a role in T-cell activation/regulation. The B7 family is the best characterized. What is now known is that these molecules have both stimulating and downregulatory ligands. Some of these B7 molecule family members preside predominately on professional APCs (B7-1, 2, B7-DC, B7-H1) while others preside predominately on peripheral organs or on the tumor (B7-H4)

HLA class I molecule and co-stimulatory molecules. The context in which the antigen is presented to the T cell appears to determine whether or not the T cell subsequently becomes activated. In the absence of the appropriate co-stimulatory signals, engagement of the TCR can lead to ignorance, anergy, or even apoptotic T-cell death. Even in the presence of appropriate co-stimulatory signals, systemic mechanisms are in place to actively regulate all types of antigen-specific immune responses including tumor-specific immune responses. In fact, many years ago it had been observed that some chemotherapeutic agents (cyclophosphamide and some of the alkylating agents) enhance immune-based therapies when given in sequence with them. It wasn’t until recently that studies have been able to provide a mechanism for this observation. It now appears that the tumors induce regulatory T cells (Treg) (previously called suppressor T cells) that secrete cytokines including IL-10, turn off activated T cells, and also induce T-cell apoptosis. The induction of

18  Cancer Vaccines

523

these Treg both in peripheral blood as well as in the tumor proper or in associated tumor stroma appears to be tumor antigen specific, although these T cells function in an antigen-unrestricted manner. The induction of these regulatory T cells is likely a natural mechanism by which the host immune system provides checks and balances to control all types of immune responses. This response is likely only one of many natural mechanisms of regulation that need to be temporarily overcome to effectively induce systemic immune responses to GI malignancies such as pancreas, gastric, or colon cancer [14–18]. There are other mechanisms of systemic immune suppression that are less well defined. Alterations in expression of the TCR on the T cell’s surface results in downregulation of downstream signal transduction pathways. This in turn leads to decreased T-cell proliferation and function. It has already been demonstrated that expression of the TCR-z chain, which is the large intracytoplasmic homodimer associated with the CD3 complex, is selectively reduced in patients with pancreatic adenocarcinoma and other solid tumors [11]. The syndrome of cancer cachexia is thought to play a significant role in systemic immune suppression associated with pancreatic adenocarcinoma. The majority of pancreatic cancer patients develop the characteristics of this syndrome including weight loss through depletion of adipose tissue and skeletal muscle mass as well as the development of secondary states of immune deficiency as manifested by suppressed DTH responses to common antigens [12, 13]. A number of cytokines that have been implicated in the production of cachexia are known immunosuppressive cytokines including IL-1, IL-6, IL-10, and TGF-b.

18.3 Immunotherapy Clinical Trials Currently, immunotherapy strategies under clinical development can be broadly divided into passive and active therapeutic approaches. Passive immunotherapy mainly involves the use of unlabeled or labeled monoclonal antibodies that are specifically raised against tumor antigens. Advantages include specific targeting of tumor cells while sparing normal tissue, relative ease of administration, and low toxicity profile. The major disadvantages include the absence of T-cell activation which therefore precludes T cell-mediated cytotoxic killing and the generation of memory immune responses. In addition, a potential limiting factor in its use involves tumor heterogeneity. Specifically, all tumor cells within a proliferating mass may not express the antigen being targeted by the antibody. Furthermore, successful delivery of the antibody to the tumor’s microenvironment depends on adequate vascularization, which can be inhibited by areas of necrosis. Antibodies have so far been the most successful form of immunotherapy clinically. To date, bevacizumab, a fully humanized monoclonal antibody to vascular endothelial growth factor-A (VEGF-A) has been demonstrated to improve overall survival when combined with chemotherapy in first- or second-line treatment for metastatic colorectal cancer [19–21]. In addition, cetuximab and panitumumab are

524

D. Laheru

chimeric and humanized monoclonal antibodies, respectively, to the epidermal growth factor receptor (EGFR). Cetuximab and panitumumab have been approved for Irinotecan intolerant or chemotherapy refractory colorectal cancer [22, 23]. There is also strong evidence that inhibiting both VEGF and EGFR pathways are interrelated and when both pathways are inhibited there is an increase in antitumor activity [24, 25]. Moreover, the combination of bevacizumab and cetuximab either with or without irinotecan demonstrated activity in patients with irinotecan refractory disease [26]. Recently, it has also been demonstrated that cetuximab can be safely combined with irinotecan as well as with oxaliplatin-based chemotherapy in the first-line setting with improved efficacy over chemotherapy alone [27, 28]. Given the benefit of bevacizumab in the first and second line and of cetuximab and panitumumab in the more chemotherapy refractory population, it was a logical extension to consider a combination of VEGF and EGF inhibition integrated with first-line chemotherapy. However, a study examining the combination of first-line chemotherapy integrated with bevacizumab and panitumumab vs. chemotherapy and bevacizumab alone was discontinued after a planned interim analysis as patients on the chemotherapy and bevacizumab and panitumumab arm had increased serious adverse events including deaths with a decrease in progressionfree survival compared to the standard arm [29]. An explanation for these findings is still unclear. Several hypotheses have been raised including the notion that combined VEGF–EGFR pathway inhibition was accentuated when combined with chemotherapy. For example, VEGF inhibition could have enhanced cetuximab’s known side effects including diarrhea and skin rash by inhibiting tissue repair. Furthermore, VEGF inhibition could have increased the incidence of pulmonary emboli. Perhaps one of the most significant advances in the management of advanced colorectal cancer has been the recognition that patients with tumors who have KRAS mutations at codons 12 and 13 do not benefit from EGFR inhibition. Several studies have provided key evidence that support KRAS mutation status as predictive of response to EGFR inhibitors [27, 28, 30]. The data is of sufficient strength that is now considered the standard of care to have a patient’s colorectal cancer tumor tested for KRAS mutations in an appropriately accredited laboratory. If a KRAS mutation in either codon 12 or 13 is detected, then such a patient should not be treated with an anti-EGFR antibody as part of their care [31]. There are a number of ongoing studies that are determining the benefit of integrating either cetuximab (NCCTG-N0147, PETACC-8) or bevacizumab (ECOG 5202) in patients with resected stage II or III disease. However, to date a number of other monoclonal antibodies targeting other colorectal cancer cell surface proteins such as glycoprotein 17-1A have been tested in the adjuvant setting with no appreciable additional benefit over surgery and chemotherapy alone [32]. There are no approved monoclonal antibodies demonstrated have been approved for treatment of pancreatic or gastric cancer as of yet. Active immunotherapy is typically divided into nonspecific and specific processes. Nonspecific therapy attempts to augment an immune response without actually targeting a specific tumor antigen. In contrast, active specific or vaccine therapy, targets specific tumor antigens as a result of the induction of antigen-specific

18  Cancer Vaccines

525

B cell- or T cell-mediated immune responses. The major advantage of active specific therapy include the ability to generate antigen-specific memory T-cell responses that are capable of being reactivated if tumor cells expressing the same antigen profile recur. The induction of cellular immune responses has the added benefit of allowing natural access to the microenvironment of the tumor. Preclinical studies have already shown that T cell-mediated vaccine therapy can induce antitumor immune responses that are potent enough to eradicate colorectal tumors [13]. Translation of these vaccine approaches into therapies for patients with pancreatic and colorectal adenocarcinoma are ongoing in phase I and II clinical development.

18.3.1 Antigen-Based Vaccines A few candidate colorectal and pancreatic antigens recognized by B and T cells have already been identified and fall into several categories including reactivated embryonic genes [carcinoembryonic antigen (CEA)], mutated oncogenes/suppressor genes (k-ras and p53), altered mucins (MUC-1), and overexpressed tissuespecific genes (HER-2/neu and Gastrin-17). Viral vector, protein, and peptide vaccines employing some of these antigens have been tested in phase I and II clinical trials. Although T-cell responses have been observed, these responses have not yet been correlated with clinical regressions [1, 4]. Mutated k-ras vaccines have been the most extensively studied peptide/proteinbased vaccine approach in patients with pancreatic adenocarcinoma. In the largest study, patients with either resected or advanced pancreatic adenocarcinoma were intradermally administered a 17 amino acid peptide containing either the specific k-ras codon 12 mutation (resected disease) or a mixture of four k-ras peptides containing the four most common mutations (advanced disease) [33]. Human granulocyte/macrophage colony-stimulating factor (GM-CSF) (40 mg) was administered intradermally 15 min prior to peptide vaccination. Patients were vaccinated weekly for 4 weeks and were given booster injections at weeks 6 and 10. Peptide vaccination was well tolerated in all 48 patients. Of the 48 vaccinated patients, 43 were evaluable for induction of immune response. A positive DTH (measured as >5 mm induration 48 h post vaccination) was observed in 21/43 evaluable patients. In addition, the peptide vaccine elicited a positive mutated k-ras-specific proliferative T-cell response in the peripheral blood of 17/43 evaluable patients. Mean survival of patients following resection was 25.6 months. In the group with advanced disease, stable disease was seen in 11/34 evaluable patients. An immune response (defined as either a positive DTH or a proliferative T-cell response) was observed in 20 of the 34 treated patients including all 11 patients demonstrating stable disease. The median survival in the group who demonstrated an immune response was 148 vs. 61 days in the group that did not demonstrate an immune response (P = 0.0002). Heat shock proteins (HSP) are ubiquitous and highly conserved cellular proteins that are upregulated during cell stress. They are thought to have multiple functions

526

D. Laheru

including helping newly synthesized polypeptides fold, assisting in protein transport, and associating with peptides generated during protein degradation. They are also thought to stimulate macrophage and DC activation and assist in re-presentation of peptides. Preclinical studies have shown that HSPs isolated from tumor cells can serve as potent vaccines by taking advantage of their role as a peptide transporter and as a stimulator of APCs. This approach has recently been tested in patients with resected pancreatic adenocarcinoma from whom HSP could be obtained and purified [34]. A follow-up clinical trial using this approach is currently under investigation. Mucin-1 (MUC-1) is a glycosylated transmembrane protein that is uniquely characterized by an extracellular domain that consists of a variable number of tandem repeats of 20 amino acids rich in proline, serine, and threonine residues. While normally present lining ductal epithelial surfaces including the gastrointestinal tract, altered MUC-1 is overexpressed on the cell surface of many cancers including pancreatic adenocarcinoma. Data from animal and phase I clinical studies have demonstrated that HLA-unrestricted T cells isolated from patients with MUC-1 expressing tumors can recognize these exposed epitopes and can induce MUC-1specific responses [35, 36]. CEA is another glycoprotein that is overexpressed in a number of gastrointestinal malignancies including colorectal, gastric, and pancreatic cancers. A CEA vaccine approach has been tested in 58 patients with CEA expressing advanced tumors, including 35 patients with colorectal cancer and seven patients with other gastrointestinal malignancies [37]. A recombinant vaccinia virus containing the CEA gene (rV-CEA) was generated as the vaccinia virus is capable of infecting APCs and could therefore potentially present CEA to both CD4+ and CD8+ T cells. In addition, a second recombinant anti-CEA vaccine, avipox-CEA (rF-CEA) was generated. The avipox virus is similar to the vaccinia virus but is not capable of infecting mammalian cells and would therefore pose a decreased risk for a systemic infection. Patients also received a triad of co-stimulatory molecules that included human B7-1, ICAM-1, and LFA-3. For this study, patients were not restricted based on HLA typing. Patients were vaccinated once every 28 days treated in a staged design beginning first with dose escalation of rF-CEA administered subcutaneously and intradermally starting from 4 × 106 pfu to 4 × 108 pfu. The second-stage integrated rV-CEA administered intradermally starting at 1 × 106 pfu–1 × 108 pfu with rF-CEA. The third stage combined both rF-CEA and rV-CEA + GM-CSF at a fixed dose of 100 mg subcutaneously into vaccine injection sites days 1–4. No significant toxicities were identified. Twenty-three patients had stable disease for at least 4 months with 14 of these patients having prolonged stable disease of greater than 6 months. Eleven patients had decreasing or stable CEA values and one patient had a complete response. Enhanced CEA-specific T-cell responses as measured by ELISPOT was identified in the majority of patients tested. The cadherin family is divided into several subfamilies, including CDH1/ E-cadherin, CDH2/N-cadherin, and CDH3/P-cadherin, designated by their tissue distribution. CDH1 is the predominant cadherin family member expressed in all epithelial tissues. It is postulated that CDH1 functions as a tumor suppressor,

18  Cancer Vaccines

527

negatively regulating the invasion and metastasis of tumor cells, in several malignancies. Targets against cadherin proteins have been identified in pancreas, colon, and gastric cancers [38].

18.3.2 Whole Tumor Cell Vaccines Whole tumor cell vaccine approaches involve the use of autologous or allogeneic tumor cells to stimulate an immune response. However, studies aimed at dissecting antitumor immune responses have confirmed that most tumors are not naturally immunogenic, and preclinical models suggests that the failure of the immune system to reject spontaneously arising tumors is unrelated to the absence of sufficiently immunogenic tumor antigens. Instead, the problem is derived from the immune system’s inability to appropriately respond to these antigens. These findings have lead to the concept that a tumor cell can become more immunogenic if engineered to secrete immune-activating cytokines. Tumor cells genetically modified to secrete immune-activating cytokines have been extensively studied for their ability to induce systemic antitumor immune responses. Preclinical studies have shown that these vaccines can induce immune responses potent enough to cure mice of preestablished tumor. In one comparison study of ten cytokines, GM-CSF was most potent, generating systemic immunity dependent on both CD4+ and CD8+ T cells. GM-CSF is known to be involved in the recruitment and differentiation of bone marrow-derived DCs and DCs are known to be the most efficient APCs at activating T cells. Studies aimed at optimizing this cytokine-secreting tumor vaccine approach confirmed that GM-CSF secretion must be at the site of relevant tumor antigen. Simple injection of soluble GM-CSF along with the appropriate tumor cells does not provide sustained local levels required to provide a sufficient immunologic boost. This information suggested that the mere presence of GM-CSF was not sufficient. Rather, the sustained release and duration of GM-CSF secretion appeared to be critical for priming the immune response. Furthermore, high levels must be sustained for several days. In the preclinical data, it appeared that a minimum of 35 ng/106 cells/24 h is necessary to generate effective antitumor immunity [39]. The results of a phase I study testing irradiated allogeneic pancreatic tumor cell lines transfected with GM-CSF as adjuvant treatment administered in sequence with adjuvant chemoradiation in patients with resected pancreatic adenocarcinoma was reported. Fourteen patients with stage 2 or 3 disease received an initial vaccination 8 weeks following pancreaticoduodenectomy. This was a dose escalation study in which three patients each received 1 × 107, 5 × 107, and 1 × 108, and five patients received 5 × 108 vaccine cells. Study patients were jointly enrolled in an adjuvant chemoradiation protocol for 6 months. Following the completion of adjuvant chemoradiation, patients were reassessed and those who were still in remission were treated with three additional vaccinations given 1 month apart at the same original dose that they received for the first vaccination. Few toxicities were observed. Systemic GM-CSF levels were measured to assess the longevity of vaccine cells at

528

D. Laheru

the immunizing site. Serum GM-CSF levels could be detected for up to 96  h following vaccination. Postvaccination DTH responses to autologous tumor cells were observed in one of three patients receiving 1 × 108 and in two of four patients receiving 5 × 108 vaccine cells [40]. A larger follow-up study in this same patient population are ongoing. Of interest, overexpressed pancreas cancer tumor antigens have now been identified via a number of different strategies including serial analysis of gene expression (SAGE) [41, 42]. The most relevant candidate proteins can be ranked in order of importance for screening based on the following additional criteria (1) proteins that are nonmutated and would be generalizable to most patients with that cancer; (2) proteins thought to be of biological importance to tumor growth and disease progression; and (3) proteins that are not expressed or minimally expressed in normal tissue. Mesothelin is a candidate pancreatic tumor antigen that was recently identified using this approach [43]. Mesothelin is a transmembrane glycoprotein and derives from a larger protein, mesothelin/­megakaryocytepotentiating factor (MPF) [44] Mesothelin is overexpressed by most pancreatic tumors [41, 45]. This antigen was recently identified as a T-cell target using lymphocytes that were isolated from three pancreatic cancer patients who had been immunized with an allogeneic, GM-CSF-secreting pancreatic tumor vaccine and who demonstrated other evidence of immune and clinical responses [45].

18.4 New Immunotherapy Targets As additional immune-relevant pancreatic, colorectal, or gastric tumor antigens are identified, the next significant challenge lies in developing strategies to improve the in vivo delivery of these antigens to APCs and thereby allow effective antigen processing and presentation and subsequent activation of a potent antitumor immune response. DCs are now accepted as the most efficient APCs in B- and T-cell activation. Several clinical trials have tested ex vivo expanded and primed DCs as a vaccine approach [46]. However, these studies have revealed the difficulty in reliably producing phenotypically mature DCs for clinical testing as only mature DC’s are capable of efficiently presenting antigens to T cells. If antigen is not presented in the proper context by mature DCs, immune downregulation or tolerance can occur. It has been shown in animal models that immature DCs induce T-cell tolerance [47–50]. As an alternative to DC-based delivery, recombinant viral- and bacterialvector delivery systems are currently under development or are already undergoing clinical testing. The use of modified viral particles or targeted bacteria to deliver tumor antigens to the immune system is based on the innate ability of the agent to efficiently infect APCs in  vivo. Early approaches have included viruses such as vaccinia. However, the use of immunogenic vectors in cancer patients who have been previously exposed to a similar vector often induces vigorous antivector immune responses before effective priming against the tumor antigen can occur. As such, other viral particles and bacterial delivery systems are currently nearing or are already undergoing clinical development for the treatment of pancreatic cancer, including fowlpox viruses and Listeria monocytogenes [51, 52].

18  Cancer Vaccines

529

18.4.1 Targeting Immune Checkpoints In spite of encouraging results from immunotherapy-based clinical trials, it is clear that tumors still grow despite the detection of tumor-specific immune responses. To explain this observation, it has been postulated that patients with cancer develop peripheral tolerance to their tumor. Insights into the mechanisms that underlie immune tolerance have provided opportunities for designing combinatorial immunebased interventions that enhance the antitumor immune response. For example, preclinical studies and early clinical trials in patients with prostate cancer and melanoma have demonstrated that downregulation of signaling through CTLA-4, using an antagonist monoclonal antibody, increases antitumor immunity in some patients, even when administered as a single agent [53–58]. Phase I clinical trials that analyze the effects of combining antibodies that block CTLA-4 signaling with antigen-targeted vaccination in patients with pancreatic cancer are about to begin. Treg cells are now accepted as another immune checkpoint for the systemic regulation of the antigen-specific T-cell responses at the tumor site. A number of preclinical studies have demonstrated that the administration of Treg-inhibiting agents – either immune-modulating doses of chemotherapy or an IL-2-receptortargeted antibody that depletes Treg cells – to naïve hosts increases the antitumor effects of immune-based therapies [59–61]. A phase II study compared a whole-cell pancreatic cancer vaccine given either alone, or in combination with immunemodulating doses of the Treg-inhibiting chemotherapeutic agent cyclophosphamide, in patients with metastatic pancreatic cancer who were previously treated with two or more chemotherapies [62]. The study reported an increased number of patients experiencing progression-free survival in the cohort that received cyclophosphamide plus the vaccine (40% of patients at 16 weeks), compared to the cohort that received the vaccine alone (16% of patients at 16 weeks). More importantly, postimmunization increases in mesothelin-specific T cells were observed almost exclusively in patients with prolonged survival. The side effects associated with this vaccine approach are limited to local, transient, vaccine skin-site reactions. These side effects are usually tolerable and self-limiting, lasting no more than 2 weeks and requiring minimal, if any, intervention. The fact that the side effects are minimal and tolerable allows such a vaccine approach to be easily integrated with other treatment modalities. The results of these studies will provide direction for the future development of vaccines in pancreatic cancer. For example, immune-based therapies are currently being combined with targeted therapies that are believed to have multiple mechanisms (immune and nonimmune mediated) of antitumor activity such as inhibitors to EGFR and VEGF receptor.

18.4.2 Future Expectations There are challenges that must still be overcome if immune-based therapies are to play an important role in the treatment of advanced cancer. First, immune-based

530

D. Laheru

strategies must be able to circumvent the genetic alterations within a tumor cell that result in their ability to evade immunologic recognition and eradication. Typically, genetic alterations result in the loss of antigen expression or the ability to adequately present antigen for adequate immune system activation. One possible solution to this problem is to design polyvalent vaccines and antibodies that target several tumor rejection antigens. Until recently, the number of known pancreatic tumor-associated proteins have been few in number. With the recent sequencing of the human genome as well as the recent sequencing of some cancer including pancreatic cancer [63], and the availability of rapid gene profiling techniques that are being employed to identified genes involved in pancreatic tumor formation, it is now possible to identify candidate tumor antigens that are also the targets of the immune system. It is therefore likely that new targets will be tested in immunebased therapeutic strategies in the near future. It is also unlikely that immunotherapy alone will be able to overcome mechanisms that functionally inactivate tumor-specific T cells. Consequently, it might be possible to enhance the effects of immune-based approaches by combining the cytoreductive and/or immune-modulating elements of chemotherapy and radiotherapy with the tumor cell cytotoxic specificity of immunotherapy. Furthermore, as the mechanisms of tumor tolerance become better understood, it should also be possible to directly modulate specific molecular targets that are found to be involved in regulating T-cell activation and suppression. Ultimately, the success of immune-based therapies against pancreatic cancer will depend on the development of multiple strategies that can be applied in synergy with immunotherapy.

References 1. Vieweg J, Jackson A. Antigenic targets for renal cell carcinoma immunotherapy. Expert Opin Biol Ther 2004;4(11):1791–1801. 2. Mulders P, Bleumer I, Oosterwijk E. Tumor antigens and markers in renal cell carcinoma. Urol Clin North Am 2003;30:455–465. 3. Wang RF, Rosenberg SA. Human tumor antigens for cancer vaccine development. Immunol Rev 1999;170:85–100. 4. Krejci JG, Markiewicz MA, Kwon ED. Immunotherapy for urological malignancies. J Urol 2004;171:870–876. 5. Michael A, Pandha HS. Renal cell carcinoma: tumour markers, T cell epitopes and potential for new therapies. Lancet Oncol 2003;4:215–223. 6. Sondak VK, Sabel MS, Mule JJ. Allogeneic and autologous melanoma vaccines: where have we been and where are we going? Clin Cancer Res 2006;12:2337–2341. 7. Boon T, Coulie PG, Van den Eynde BJ, et  al. Human T cell responses against melanoma. Annu Rev Immunol 2006;24:175–208. 8. Gajewski TF, Meng Y, Harlin H. Immune suppression in the tumor microenvironment. J Immunother 2006;29:233–240. 9. Laheru DA, Jaffee EM. Immunotherapy for pancreatic cancer – science driving clinical progress. Nat Rev 2005;5(6):459–467. 10. Greten TF, Jaffee EM. Cancer vaccines. J Clin Oncol 1999;17:1047–1060.

18  Cancer Vaccines

531

11. Marincola FM, Jaffee EM, Hicklin DJ, Ferrone S. Escape of human solid tumors from T-cell recognition: molecular mechanisms and functional significance. Adv Immunol 2000;74:181–273. 12. Laheru D, Biedrzycki B, Jaffee EM. Immunologic approaches to the management of pancreatic cancer. Cancer J 2001;7(4):324–337. 13. Wolf AM, Wolf D, Steurer M, et al. Increase of regulatory T cells in the peripheral blood of cancer patients. Clin Cancer Res (Advances in Brief) 2003;9:606–612. 14. von Bernstorff W, Voss M, Freichel S, et al. Systemic and local immunosuppression in pancreatic cancer patients. Clin Cancer Res 2001;7(Suppl):925s–932s. 15. Salama P, Phillips M, Grieu F, et  al. Tumor-infiltrating FOXP3+ T regulatory cells show strong prognostic significance in colorectal cancer. J Clin Oncol 2009;27:186–192. 16. Hinz S, Pagerols-Raluy L, Oberg HH, et al. Foxp3 expression in pancreatic carcinoma cells as a novel mechanism of immune evasion in cancer. Cancer Res 2007;67:8344–8350. 17. Hiraoka N, Onozato K, Kosuge T, et al. Prevalence of FOXP3+ regulatory T cells increase during the progression of pancreatic ductal adenocarcinoma and its pre-malignant lesions. Clin Cancer Res 2006;12:5423–5434. 18. Shen LS, Wang J, Shen DF, et al. CD4(+)CD25(+)CD127 (low/−) regulatory T cells express Foxp3 and suppress T cell proliferation and contribute to gastric cancers progression. Clin Immunol 2009;131:109–118. 19. Kabbinavar FF, Schulz J, McCleod M, et al. Addition of bevacizumab to bolus fluorouracil and leukovorin in first line metastatic colorectal cancer: results of a randomized phase II trial. J Clin Oncol 2005;23:3697–3705. 20. Huwitz H, Fehrenbacher L, Novotny W, et al. Bevacizumab plus irinotecan, fluorouracil and leukovorin for metastatic colorectal cancer. N Engl J Med 2004;350:2335–2342. 21. Giantonio BJ, Catalano PJ, Meropol NJ, et al. Bevacizumab in combination with oxaliplatin, fluorouracil and leukovorin (FOLFOX4) for previously treated metastatic colorectal cancer: results from the Eastern Cooperative Oncology Group Study E3200. J Clin Oncol 2007;25:1539–1544. 22. Cunningham D, Humblet Y, Siena S, et al. Cetuximab monotherapy and cetuximab plus irinotecan in irinotecan refractory metastatic colorectal cancer. N Engl J Med 2004;351:337–345. 23. Van Cutsem E, Peeters M, Siena S, et al. Open label phase III trial of panitumumab plus best supportive care compared with best supportive care alone in patients with chemotherapyrefractory metastatic colorectal cancer. J Clin Oncol 2007;25:1658–1664. 24. Yokoi K, Thaker PH, Yzici S, et al. Dual inhibition of epidermal growth factor receptor and vascular endothelial growth factor receptor phosphorylation by AEE788 reduces growth and metastasis of human colon carcinoma in an orthotopic nude mouse model. Cancer Res 2005;65:3716–3725. 25. Bruns CJ, Solorzano CC, Harbison MT, et al. Blockade of the epidermal growth factor receptor signaling by a novel tyrosine kinase inhibitor leads to apoptosis of endothelial cells and therapy of human pancreatic carcinoma. Cancer Res 2000;60:2926–2935. 26. Saltz LB, Lenz HJ, Kindler H, et al. Randomized phase II trial of cetuximab, bevacizumab, and irinotecan compared with fluorouracil alone in irinotecan refractory colorectal cancer: the BOND-2 study. J Clin Oncol 2007;25:4557–4561. 27. Bokemeyer C, Bondarenko I, Makhson A, et al. Fluorouracil, leucovorin, and oxaliplatin with and without cetuximab in the first-line treatment of metastatic colorectal cancer. J Clin Oncol 2009;27:663–671. 28. Van Cutsem E, Lanf I, D’Haens G, et al. KRAS status and efficacy in the first line treatment of patients with metastatic colorectal cancer treated with FOLFIRI with or without cetuximab: the crystal experience. J Clin Oncol 2008;26:5s (suppl: abstract 2). 29. Hecht JR, Mitchell E, Chidiac T, et  al. A randomized phase IIIB trial of chemotherapy, bevacizumab, and panitumumab compared with chemotherapy and bevacizumab alone for metastatic colorectal cancer. J Clin Oncol 2009;27:672–680. 30. Amado RG, Wolf M, Peeters M, et al. Wild type KRAS is required for panitumumab efficacy in patients with metastatic colorectal cancer. J Clin Oncol 2008;26:1626–1634.

532

D. Laheru

31. Allegra CJ, Jessup JM, Somerfield MR, et al. American Society of Clinical Oncology provisional clinical opinion: testing for KRAS gene mutations in patients with metastatic colorectal carcinoma to predict response to anti-epidermal growth factor receptor monoclonal antibody therapy. J Clin Oncol 2009;27(12):2091–2096. 32. Hartung G, Hofheinz RD, Dencausse Y, et  al. Adjuvant therapy with edrecolomab versus observation in stage II colon cancer: a multi-center randomized phase III study. Onkologie 2005;28:347–350. 33. Gjertsen MK, Buanes T, Rosseland AR, et al. Intradermal ras peptide vaccination with granulocyte–macrophage colony stimulating factor as adjuvant: clinical and immunologic responses in patients with pancreatic adenocarcinoma. Int J Cancer 2001;92:441–450. 34. Maki RG, Livingston PO, Lewis JJ, et  al. A phase I pilot study of autologous heat shock protein vaccine HSPPC-96 in patients with resected pancreatic adenocarcinoma. Dig Dis Sci 2007;52:1964–1972. 35. Tolcher AW, Ochoa L, Hammond LA, et al. Cantuzumab mertansine, a maytansinoid immunoconjugate directed to the CanAg antigen: a phase I pharmacokinetic and biologic correlative study. J Clin Oncol 2003;21(2):211–222. 36. Kondo H, Hazama S, Kawaoka T, et al. Adoptive immunotherapy for pancreatic cancer using MUC-1 peptide pulsed dendritic cells and activated T lymphocytes. Anticancer Res 2008;28:379–387. 37. Marshall JL, Gulley JL, Aren PM, et al. Phase I study of sequential vaccinations with FowlpxCEA (6D)-TRICOM alone and sequentially with vaccinia – CEA (6D)-TRICOM with or without granulocyte–macrophage colony stimulating factor in patients with carcinoenbryonic antigen-expressing carcinomas. J Clin Oncol 2005;23:720–731. 38. Imai K, Hirata S, Irie A, et al. Identification of a novel tumor associated antigen, cadherin 3/P – cadherin, as a possible target for immunotherapy of pancreatic, gastric and colorectal cancers. Clin Cancer Res 2008;14:6847–6895. 39. Dranoff G, Jaffee E, Lazenby A, et al. Vaccination with irradiated tumor cells engineered to secrete murine granulocyte–macrophage colony stimulating factor stimulates potent specific and long lasting immunity. Proc Natl Acad Sci U S A 1993;90:3539–3543. 40. Jaffee EM, Hruban R, Biedrzycki B, et al. A novel allogeneic GM-CSF secreting tumor vaccine for pancreatic cancer: a phase I trial of safety and immune activation. J Clin Oncol 2001;19(1):145–156. 41. Argani P, Iacobuzio-Donahue C, Ryu B, et al. Mesothelin is over-expressed in the vast majority of ductal adenocarcinoma of the pancreas: identification of a new pancreatic cancer marker by serial analysis of gene expression. Clin Cancer Res 2001;7(12):3862–3868. 42. Argani, P, Rosty, C, Reiter, RE, et  al. Discovery of new markers of cancer through serial analysis of gene expression (SAGE): prostate stem cell antigen (PSCA) is over-expressed in pancreatic adenocarcinoma. Cancer Res 2001;61:4320–4324. 43. Hassan R, Bera T, Pastan I. Mesothelin: a new target for immunotherapy. Clin Cancer Res 2004;10:3937–3942. 44. Swierczynski SL, Maitra A, Abraham SC, et al. Analysis of novel markers in pancreatic and biliary carcinomas using microarrays. Hum Pathol 2004;35(3):357–366. 45. Thomas AM, Santarsiero LM, Lutz ER, et  al. Mesothelin-specific CD8+ T cell responses provide evidence of in vivo cross priming by antigen presenting cells in vaccinated pancreatic cancer patients. J Exp Med 2004;200:297–306. 46. Morse M, Clay T, Hobeika A, et al. Phase I study of immunization with dendritic cells modified with recombinant fowlpox encoding carcinoembryonic antigen (CEA) and costimulatory molecules. Clin Cancer Res 2005;11:3017–3024. 47. Wang Q, Liu Y, Wang J, et al. Induction of allospecific tolerance by immature dendritic cells genetically modified to express soluble TNF-receptor. J Immunol 2006;177:2175–2185. 48. Kim R, Emi M, Tanabe K, et  al. Tumor driven evolution of immunosuppressive networks during malignant progression. Cancer Res 2006;66:5527–5536. 49. Mahnke K, Enk AH. Dendritic cells: key cells for induction of regulatory T cells? Curr Top Microbiol Immunol 2005;293:133–150.

18  Cancer Vaccines

533

50. Mende I, Engleman EG. Breaking tolerance to tumors with dendritic cell-based immunotherapy. Ann N Y Acad Sci 2005;1058:96–104. 51. Kochi SK, Killeen KP, Ryan US. Advances in the development of bacterial vector technology. Expert Rev Vaccines 2003;2:31–43. 52. Dietrich G, Spreng S, Favre D, et al. Live attenuated bacteria as vectors to deliver plasmid DNA vaccines. Curr Opin Mol Ther 2003;5:10–19. 53. Chambers CA, Kuhns MS, Egen JG, Allison JP. CTLA-4 mediated inhibition in regulation of T cell responses: mechanisms and manipulation in tumor immunotherapy. Annu Rev Immunol 2001;19:565–594. 54. Kwon ED, Foster BA, Hurwitz AA, et al. Elimination of residual metastatic prostate cancer after surgery and adjunctive cytotoxic T lymphocyte associated antigen 4 (CTLA-4) blockade immunotherapy. Proc Natl Acad Sci U S A 1999;96:15074–15079. 55. Maker AV, Yang JC, Sherry RM, et al. Intrapatient dose escalation of anti-CTLA-4 antibody in patients with metastatic melanoma. J Immunother 2006;29:455–463. 56. Reuben JM, Lee BN, Li C, et al. Biologic and immunomodulatory events after CTAL-4 blockade with ticilimumab in patients with advanced malignant melanoma. Cancer 2006;106:2437–2444. 57. Maker AV, Attia P, Rosenberg SA. Analysis of the cellular mechanism of anti-tumor responses and autoimmunity in patients treated with CTLA-4 blockade. J Immunol 2005;175: 7746–7754. 58. Attia P, Phan GQ, Maker AV, et al. Autoimmunity correlates with tumor regression in patients with metastatic melanoma treated with anti-cytotoxic T lymphocyte-4. J Clin Oncol 2005;23: 6043–6053. 59. Movva S, Verschraegen C. The monoclonal antibody to cytotoxic T lymphocyte antigen-4; ipilumimab (MDX-010) a novel treatment strategy in cancer management. Expert Opin Biol Ther 2009;9:231–241. 60. Weber JS, O’Day S, Urba W, et al. Phase I/II study of ipilumimab for patients with metastatic melanoma. J Clin Oncol 2008;26:5950–5956. 61. Hodi FS, Butler M, Oble DA, et al. Immunologic and clinical effects of antibody blockade of cytotoxic T lymphocyte associated antigen 4 in previously vaccinated patients. Proc Natl Acad Sci U S A 2008;105:3005–3010. 62. Laheru D, Lutz E, Burke J. Allogeneic granulocyte macrophage colony stimulating factor secreting tumor immunotherapy alone or in sequence with cyclophosphamide for metastatic pancreatic cancer: a pilot study of safety, feasibility and immune activation. Clin Cancer Res 2008;14:1455–1463. 63. Jones S, Zhang X, Parsons DW, et al. Core signaling pathways in human pancreatic cancers revealed by global genomic analyses. Science 2008;321:1801–1806.

Chapter 19

Optimising the Development of Antibodies as Treatment for Cancer Craig P. Carden, Hendrik-Tobias Arkenau, and Johann S. de Bono

19.1 Introduction Antibodies are proteins secreted by vertebrates, which bind to specific molecular moieties [1]. Millions of different antibodies are made by organisms, and early in development most self-recognising antibodies are removed from the repertoire. The remaining antibodies, accordingly, recognise non-self molecules, and are used by the host to eliminate microorganisms, foreign objects, and malignant cells. Manufacturing these highly specific molecules and using them for cancer treatment represents a major breakthrough. Nonetheless, the challenges required to bring this technology to the clinic are significant and costly. Drug discovery, manufacturing, and clinical development are all expensive and have high risk [2], and optimising these processes are crucial. This chapter will summarise the science behind this technology, the current use of antibodies in anti-cancer treatment, the methodology of antibody anti-cancer drug development, and the ways in which these components may be improved to facilitate the transition from the laboratory to the clinic.

19.2 Antibody Structure and Function Antibodies are protein molecules of MW 100,000–150,000, composed of two light chains and two heavy chains, and two domains: a constant region and a variable region [1]. The variable region determines the specificity of the antibody for a target. Antibodies have different variable regions because different combinations of genes are combined during embryogenesis to form a unique DNA and thus protein

J.S. de Bono (*) Section of Medicine and Drug Development Unit, Institute of Cancer Research, Royal Marsden Hospital, Downs Road, Sutton, Surrey SM2 5PT, UK e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_19, © Springer Science+Business Media, LLC 2011

535

536

C.P. Carden et al.

sequence, with a particular three-dimensional and chemical structure. The constant region, Fc, is recognised by a range of receptors on effector cells [3]. The total repertoire of antibodies an organism makes is that which is left after the T and B cells that recognise normal self are removed by central (thymus and bone marrow) and peripheral mechanisms of tolerance [4]. Humans have five different classes of antibodies with different roles, IgG, IgE, IgM, IgA, and IgD. The specificity of an antibody for a target is determined by the three-dimensional and bonding structure conferred by its protein sequence. The strength of bonding of a naturally occurring antibody (stated in KD, dissociation constant) ranges from 10−5 to 10−8 M [5]. In response to infection, antigen-presenting cells are mobilised, and these cells present foreign antigens to lymphocytes. This causes the clone (or clones) of lymphocytes with an antibody that is appropriately configured to bind to the foreign host to proliferate and produce the antibody. The attached antibody summons cells to attack the invading organism by two main mechanisms: antibodydependent cell cytotoxicity (ADCC) and complement-dependent cytotoxicity (CDC), both mediated via the Fc component of the antibody, as discussed below [6]. These mechanisms are termed indirect mechanisms because they rely on activation of the immune system, an important mechanism of their anti-cancer action. In the setting of treatment of cancer, direct mechanisms are specifically targeted against diverse biological functions of the cancer cell. Most currently produced antibodies are of the IgG class, which are bivalent, increasing their functional affinity and conferring high retention times. Subclasses within IgG have different half lives, with IgG1, IgG2, and IgG3 having half lives of approximately 21 days, and IgG4 of 7 days [7]. In addition, IgG1 and IgG3 activate ADCC significantly more than IgG2 or IgG4, because of their increased affinity for Fc receptors on effector cells [8].

19.3 Mechanisms of Action 19.3.1 Indirect 19.3.1.1 Antibody-Dependent Cell Cytotoxicity The binding of the variable domain of an antibody to an antigen on the surface of target cells is followed by the interaction of the constant domain with effector cells. These cells are generally natural killer or macrophage cells, and when activated they lead to target cell death by mechanisms of cytotoxic granules, phagocytosis, release of nitric oxide and reactive oxygen species, and Fas/Fas–ligand interactions [3]. In addition, they also release cytokines and chemokines that have the potential to inhibit angiogenesis and cell proliferation, and increase immunogenicity by increasing the expression of major histocompatibility antigens [9]. Anti-cancer antibodies that may utilise this method of action include rituximab, cetuximab and trastuzumab.

19  Optimising the Development of Antibodies as Treatment for Cancer

537

19.3.1.2 Complement-Dependent Cytotoxicity The complement pathway is effectively activated by IgG1 and IgG3, on binding of the C1 complement complex to the antibody. Lysis of the cell is triggered by a complement cascade and membrane attack complex formation. In addition, the complement attracts effector cells towards the cell [6, 10]. The importance of CDC relative to ADCC in antibody anti-cancer activity is controversial, and probably it varies for each antibody and tumour.

19.3.2 Direct 19.3.2.1 Cell Surface Receptor Antibodies Growth factor receptors are overexpressed in various cancers, promoting cancer cell growth [11]. Blocking of these receptors by antibodies can lead to inactivation of downstream growth pathways and growth arrest or death. Such antibodies may interfere with ligand binding to receptor, receptor folding, interaction of the receptor with other receptors (dimerisation), internalisation of receptors, or their breakdown [6]. Antibodies in clinical use include those targeting the epidermal growth factor receptor (EGFR), vascular endothelial growth factor receptor(VEGFR), platelet-derived growth factor receptor (PDGFR), hepatocyte growth factor receptor (c-Met) and various other growth factor receptors (see references in relevant sections). 19.3.2.2 Neutralising Antibodies This group of antibodies bind growth factors that are involved in receptor activation and subsequently in downstream signalling. For example, the target of bevacizumab is the vascular endothelial growth factor A, blocking its signal transduction through both the VEGFR-1 and VEGFR-2, as demonstrated by the inhibition of VEGFinduced cell proliferation, survival, migration, and vascular permeability [12, 13]. Although bevacizumab has the ability to bind to human Fc receptors and complement protein C1q, it does not cause ADCC or CDC in either VEGF-producing or targeted cells. Thus, the mechanism of anti-tumour activity of bevacizumab is most likely due to its anti-angiogenesis effect through binding and neutralization of secreted VEGF. 19.3.2.3 Apoptosis Inducing Antibodies Induction of apoptosis is an attractive approach for the development of new targeted agents [14]. Apoptosis can be induced through the tumour necrosis factor-a (TNF)-related apoptosis-inducing ligand (TRAIL)-receptor family, which has five related members, of which two are pro-apoptotic, TRAIL-R1 (DR4) and

538

C.P. Carden et al.

TRAIL-R2 (DR5) [15]. These receptors have a cytoplasmic death domain, and signal to induce the extrinsic cell death pathway; they are broadly expressed on many human tumours. Agonist antibodies specific for TRAIL-R1 and TRAIL-R2 are in development as anti-cancer therapeutics [14]. 19.3.2.4 Immune Modulation Observations that cytotoxic T lymphocyte antigen 4 (CTLA4) plays a key role in regulating immune responses mediated via T cells have led to therapeutic approaches targeting this pathway for tumour immunotherapy [16]. Therapy ­targeting this antigen may enhance immune system activation against tumour antigens by blocking the inhibitory co-stimulatory signals mediated by CTLA4 expressed on activated T cells. It is possible to combine anti-CTLA4 antibodies with vaccines, radiation or chemotherapy, potentially helping to direct immune responses towards target antigens, enhancing clinical efficacy and perhaps reducing treatment-related adverse effects [17–19]. 19.3.2.5 Other Strategies Antibody-directed enzyme prodrug therapy (ADEPT) involves linking a monoclonal antibody that targets cancer cells to a linked enzyme that activates an inactive prodrug [20]. The inactive prodrug is then administered systemically, and the enzyme activates the drug, with a cytotoxic activity localised to the malignant cell. ADEPT has been difficult to translate to the clinic thus far, in large part because of systemic toxicity.

19.4 Target Validation The development of tumour-specific antibodies requires understanding of the target. An optimal target should therefore be a tumour-specific molecule that is predominantly associated with the malignant phenotype and not expressed by healthy tissue or organs. Additionally, the effect on the target should be measurable by reliable and validated techniques and more importantly should be correlated with clinical outcome [21].

19.5 Antibody Technology: Murine, Chimeric, Humanised, and Fully Human Antibodies Antibodies have been known to target tumour cells since 1953, but their success in the clinic has been (and is) crucially dependent on advances in their manufacture. Table 19.1 summarises the most relevant types of antibodies. The initial important

19  Optimising the Development of Antibodies as Treatment for Cancer Table 19.1  Types of antibodies Type of Source of antibody constant region Murine Murine

Source of variable region Murine

Penultimate syllable -o-

Chimeric

Human

Murine

-xi-

Humanized

Human

Largely human, part murine

-zu-

Primatized

Human

Fully human

Human

Part human, part -iprimate -uHuman through xenomouse technology and transgenic techniques

539

Examples in clinical use (target) Ibritumomab tiuxetan (CD20) Tositumomab (CD20) Rituximab (CD20) Cetuximab (EGFR) Trastuzumab (HER2) Alemtuzumab (CD52) Gemtuzumab (CD33) Pertuzumab (HER) Matuzumab (EGFR) Nimotuzumab (EGFR) Not in clinical use Panitumumab (EGFR) Tremelimumab (antiCTLA4 human AB)

advance was in 1975, with the description of hybridoma technology – fusion of myeloma cells with specific antibody-producing B cells – by Kohler and Milstein that lead to provision of a consistent supply of antibodies [22, 23]. Despite early promising results using therapeutic murine monoclonal antibodies, these agents failed largely because of the dissimilarity between murine and human immune systems leading to development of human anti-mouse antibodies (HAMA). HAMA caused formation of complexes after repeated administration, resulting in allergic reactions including anaphylaxis (see Sect. 19.9) [24]. To avoid murine antibody immunogenicity, and to increase immunologic efficiency, particular components of murine antibodies were replaced by human elements [25]. These chimeric antibodies were developed by combining the human IgG molecule with the murine variable regions using transgenic fusion of the immunoglobulin genes and were produced by hybridomas and Chinese hamster ovary (CHO) cells. This technology resulted in fewer HAMA responses, but allergic reactions remain a significant limitation in the use of chimeric antibodies [21]. Humanized antibodies were developed by grafting murine hypervariable amino acid domains into human antibodies, generating antibodies of 95% human origin. Despite their low immunogenecity, these antibodies have significantly less antigen affinity compared to the parent murine monoclonal antibodies. However, an increase in antibody–antigen binding was achieved by mutations of the complementarity determining regions (CDR) [26]. Deimmunised variable domains were also created by replacing immunogenic components with non-immunogenic amino acid sequences, with these subsequently linked to human Fc domains [27]. Primatised antibodies are chimeric combinations of human and primate antibodies,

540

C.P. Carden et al.

but are not in clinical use. Lastly, fully human antibodies require incorporation of human heavy chain genes into the mouse genome [21]. The transgenic mouse is then vaccinated with the antigen, and monoclonal antibodies are produced with little immunogenicity [24, 27].

19.6 Classes of Antibodies Overall, one can distinguish between unconjugated and conjugated antibodies currently in use as cancer treatments. Unconjugated antibodies act by the mechanisms described in Sect. 19.3, and are the significant majority of agents in current clinical use, as discussed below. Conjugated antibodies, antibodies with an associated drug, toxin, or radioisotope are discussed under Sect. 19.13 in this chapter.

19.6.1 Selected Unconjugated Antibodies Currently in Clinical Use 19.6.1.1 Rituximab Rituximab is an IgG1 chimeric monoclonal antibody against the CD20 antigen found on the surface of normal and malignant B lymphocytes [28]. It contains murine variable region sequences and human constant region sequences [29]. Methods of action appear to include, as demonstrated in in vitro studies, cell lysis through ADCC, CDC, and direct signalling interruption, causing apoptosis [30]. Rituximab has been used in the treatment of patients with relapsed or refractory, low-grade or follicular, CD20-positive, B-cell, non-Hodgkin’s lymphoma. Side effects are generally mild, although serious adverse reactions, some with fatal outcomes, have been reported, in particular severe infusion reactions, tumour lysis syndrome, and mucocutaneous reactions [31]. 19.6.1.2 Bevacizumab Bevacizumab, an IgG1 humanized monoclonal antibody, binds the vascular endothelial growth factor-A (VEGF-A) and prevents the interaction between VEGF-A and its receptors (VEGFR-1 and VEGFR-2) on the surface of endothelial cells. Blocking VEGFA in clinical models leads to inhibition of new vessel growth, incorporation of haematopoeitic and endothelial progenitor cells into new blood vessels, and normalisation of vasculature [32]. Substantial experience has now been gained in the use of bevacizumab as a single agent and in combination with chemotherapy [33]. A cytotoxic effect via ADCC has not been observed with bevacizumab [12, 34]. Currently, bevacizumab

19  Optimising the Development of Antibodies as Treatment for Cancer

541

is approved for treatment in advanced colorectal cancer (ACRC), non-small cell lung cancer (NSCLC), and renal and breast cancer [35]. The most serious adverse events associated with bevacizumab in clinical trials were gastrointestinal perforation, wound healing complications, haemorrhage, arterial thromboembolic events, hypertensive crisis, reversible posterior leukoencephalopathy syndrome, nephrotic syndrome, and congestive heart failure. 19.6.1.3 Trastuzumab HER2 (also called c-neu or ErbB-2) is a receptor tyrosine kinase of the ­epidermal growth factor family. In animal models, overexpression or mutation of HER2 causes neoplastic transformation [36]. Trastuzumab is a humanized monoclonal antibody that binds to the extracellular domain of the HER2 receptor, leading to arrest during the G1 phase of the cell cycle and reduced proliferation. Trastuzumab may also disrupt receptor dimerisation and inhibition of downstream signalling through the PI3K pathway. Trastuzumab is also implicated in the suppression of angiogenesis by the induction of antiangiogenic factors and repression of proangiogenic factors [37, 38]. In vivo breast cancer models and clinical trials demonstrate that trastuzumab also has cytotoxic activity. These properties may be due in part to ADCC causing activation of effector cells and cell lysis [39, 40]. Trastuzumab is approved by the FDA as part of a treatment regimen containing doxorubicin, cyclophosphamide, and paclitaxel for the adjuvant treatment of women with node-positive, HER2-overexpressing breast cancer and for the treatment for patients with metastatic HER2-overexpressing breast cancer [41]. The most serious adverse reactions in patients receiving trastuzumab are infusion reactions, pulmonary toxicity, and chronic heart failure with decline in left ventricular cardiac function. 19.6.1.4 Cetuximab Ligands of the epidermal growth factor family bind to the EGFR, also called ErbB1, HER1 [42]. Ligand binding to the extracellular domain of the EGFR leads to activation and homodimerization, resulting in the phosphorylation of the intracellular tyrosine kinase that initiates a series of intracellular signals including the central Ras-/mitogen-activated protein kinase pathway [43]. Cetuximab is a chimeric IgG1 monoclonal antibody that binds to the extracellular domain of the EGFR on normal and tumour cells, where it competitively inhibits the binding of epidermal growth factor and other ligands [44]. Additionally, cetuximab has also been shown to mediate ADCC [45]. It is currently approved for patients with ACRC and squamous cell head and neck cancer [46, 47]. Serious adverse events observed in clinical trials were infusion reactions, skin toxicity, interstitial lung disease, and diarrhoea.

542

C.P. Carden et al.

19.6.1.5 Panitumumab Panitumumab (ABX-EGF) is a fully human monoclonal antibody (IgG2) that is specific to the EGFR [48]. Clinical activity has been demonstrated in patients with EGFR-positive ACRC who have failed prior therapy. The most common side effect is a dose-dependent acneiform rash. Studies thus far indicate a low rate of infusionrelated reactions (1%, grade 3–4). Monoclonal antibodies of the IgG1 isotype activate the complement pathway and mediate ADCC better than their IgG2 counterparts. Whether the different isotypes of these two antibodies, panitumumab (IgG2) and cetuximab (IgG1), have impacted anti-tumour activity through the differing activation of ADCC is currently being investigated. 19.6.1.6 Alemtuzumab Alemtuzumab is an IgG1 humanised monoclonal antibody directed against CD52, a cell surface glycoprotein. The variable and constant regions are of human origin, while the complementarity-determining regions are from a rat monoclonal antibody. Binding of CD52 induces a host immune response that results in lysis of CD52+ cells [49]. Alemtuzumab is used in the treatment of B-cell chronic lymphocytic leukemia (B-CLL) [50]. Serious adverse events have been mainly associated with infusionrelated anaphylactic reactions, syncope, pulmonary infiltrates, ARDS, respiratory arrest, and cardiac arrhythmias.

19.7 Differences Between Small Molecules and Antibodies Antibodies differ from small molecules in a number of ways [42, 51] that are crucial when considering their drug development, as illustrated in Table 19.2.

19.8 Pharmacokinetics of Antibodies Experience with murine antibodies in the clinic in humans indicated limited ADCC, the development of neutralising antibodies, and a half-life of 1 day in the human, compared with a half-life of 5 days of human antibodies in the mouse. This suggested different PK of murine and human antibodies in mice and humans [52]. Chimerised and humanised antibodies, in contrast, have a significantly longer half-life, equivalent to normal IgG, and good induction of ADCC and CDC. It has become clear that a key component determining the catabolism and ­half-life of IgG is the presence of human neonatal FC receptor (FcRn) in the endothelium of arterioles and capillaries in the muscle, skin, adipose tissue, and

19  Optimising the Development of Antibodies as Treatment for Cancer

543

Table 19.2  Comparison of small molecules and antibodies

Characteristic Cost of manufacture Proportion of drugs licensed Administration halflife/frequency of dosing

Small molecule Lower

Antibody Higher

Implications for early clinical trials studying antibodies Higher stakes

4%

18%

Lower stakes

PO/1–2 days/daily or twice daily

IV/weeks/2–4 weekly

PK variability

Variable oral absorption and hepatic elimination Extracellular or intracellular targets, uniform distribution, penetrate large tumours or CNS more easily Less specific, but multiple potential targets Not significant

More consistent between individuals

Extended PK, prolonged interaction with co-administered agents, potential delay Less potential need for extensive PK

Site of action

Specificity

Essentially extracellular targets, non uniform distribution, less penetration of large tumours and CNS

Role of patient selection for appropriate patients

Highly specific

Implications for appropriate patient selection Safety issues (see text)

May be very important, and thus producing different in vitro versus in vivo effects, and variance between animal and human models Side effect profile Generally well Generally well tolerated, Safety issues tolerated potential for immunogenicity Selection of single Similar to small Efficacy Similar to antibodies agent versus molecules as single as single agent, combination agent, more synergy less synergy with approaches with chemotherapy chemotherapy observed so far observed so far Nomenclature: Guidelines on the use of international nonproprietary names (INNs) for pharmaceutical substances, World Health Organisation, 1997 http://whqlibdoc.who.int/hq/1997/WHO_ PHARM_S_NOM_1570.pdf, Accessed 17 January 2009

Immune component to efficacy

liver [8, 52, 53]. Antibody is taken into cells and incorporated into endosomes where it binds with FcRn, and then recycled to the cell surface, where it disassociates. Antibody not bound to FcRn, whether because FcRn is saturated or when the antibody has little affinity to FcRn, is degraded within lysosomes. As would be predicted by this model, alterations in particular amino acids responsible for

544

C.P. Carden et al.

a­ ffinity of the antibody to the FcRn increased its half-life. Because of the low ­affinity of mouse antibody to FcRn, they have a low t1/2. Accordingly, PK in mice of human and non-human antibodies cannot necessarily be extrapolated to humans [54]. Preclinical studies in mice should therefore be complemented by studies to measure monoclonal antibody binding to human FcRn. As with manipulation of the Fc component to improve ADCC and CDC efficiency, genetic engineering of ­antibodies can improve binding to the FcRn and alter half-life (see Sect. 19.15). Fragments of IgG antibody have a very short half-life due to their small size and good tumour penetration. Immunotoxins alter the PK of monoclonal antibodies, also leading to a significantly shorter half-life than that of the parent molecule, due to interactions of the toxin with receptors on cells, in particular to the mannose residues used to cross-link immunotoxin to antibody, which are recognised by hepatic cells [8]. Most immunotoxins in trials in humans have used mouse antibody, with correspondingly short half-lives Tumour cells can act as an antibody sink, increasing the t1/2 until receptors within them are saturated, at which time PK normalises. Accordingly, the bulk of tumour may affect PK at the initial administration of the antibody. Persistence in the circulation, correlations between doses, PK, maximum tolerated dose, and biological effects can and should be studied in phase I trials, with the above factors in mind, because early characterisation of these may allow subsequent study and accommodation of them in later phase studies. In particular, note should be taken of the crucial role of FcRn binding in the PK of antibodies.

19.9 Potential Toxicities Toxicities of antibodies are generally classified as follows: Target effects related [55]. Examples include the following: • Anti-angiogenic antibodies such as bevacizumab can cause hypertension, thrombosis, and bleeding as a direct result of targeting VEGF signalling. • Immunosuppressive antibodies such as infliximab, which targets TNF alpha, can lead to vulnerability to infections such as TB. • Immunomodulatory antibodies such as the anti-CTLA4 antibody may cause autoimmunity including serious colitis, pneumonitis, or hepatitis. • EGFR – skin; colon. The anti-EGFR antibodies cetuximab and panitumumab cause an acneiform rash on face and chest probably as a result of EGFR expression in human skin cells. Diarrhoea is a relatively frequent side-effect for a similar reason. • HER2 – cardiac. Cardiac compromise was seen in 9% of patients receiving monotherapy with trastuzumab, and up to 28% of patients receiving doxorubicin and trastuzumab, probably because of low levels of expression of HER2 on cardiac myocytes, although other causes have not been excluded. Infusion reactions. These non-specific reactions may be related to the effects of antibody on tumour or effector cells.

19  Optimising the Development of Antibodies as Treatment for Cancer

545

• Cytokine release – anaphylactoid. Foreign proteins may prompt release of ­mediators from mast cells and basophils in the absence of IgE antibodies [56]. These reactions represent the majority of reactions to monoclonal antibodies and may be able to be overcome with pre-medications and re-challenge. Extreme cytokine reactions have been seen in the TGN1424 case, as discussed below [57]. • Anaphylaxis – true anaphylaxis is mediated by IgE, characterised by tryptase release, and is possibly more severe than anaphylactoid presentations [56, 58].

19.10 Preclinical Development: Animal–Human Model Transitions Detailed review of preclinical antibody development is beyond the scope of this chapter, and the reader is referred to reviews and regulatory guidance on the topic [59–61] (Box  19.1). Nonetheless, a brief overview of the area is instructive as it critically informs the transition to the clinic. An antigen target in the human will not necessarily be manifest in other species, meaning that interpretation of toxicity and effect needs to be done carefully between animals and humans. Usually the antigen is a protein or glycoprotein, with the likelihood of cross reactivity between species proportional to the phylogenetic distance between them. For example, apes have a higher cross reactivity than old world monkeys, which in turn have a higher cross reactivity than new world monkeys, which have a higher cross reactivity than rodents [62]. Because of this difference, cynologous (Macaque) – old world monkeys – are often used as animal models. Cross reactivity is established before animal work is performed using immunofluorescence or immunohistochemistry [62]. Because of the difficulties using primates for these studies, other methods have been attempted. Surrogate antibodies in rodents are similar antibodies to those to be used in the human studies, but created to react to the rodent antigen [62]. Human antigens in murine models using xenobiotic or nude mice and transplanted tumours is an alternative approach. Although animal models have potentially different pharmacokinetics to humans, data from their use, carefully extrapolated, is critical. A catastrophic example of the difficulties in extrapolation from preclinical studies is discussed below, as this case has had a significant influence on recent efforts to guide the safe transition of preclinical experiments to the clinic.

19.11 TGN1412: A Cautionary Tale TGN1412 is an anti-CD 28 monoclonal antibody proposed to activate human T cells as a “superagonist”, bypassing the requirement of the T-cell antigen receptor for activation. The antibody was engineered by transferring the CDR of a monoclonal mouse anti-human CD28 antibody to human variable chain regions [63].

546

C.P. Carden et al.

The  drug was proposed to be potentially useful in diseases in which T cells are involved in the pathogenesis of chronic inflammation, or in leukaemia. A surrogate model in the mouse demonstrated no significant cytokine release toxicity and preclinical development in cynomolgus and rhesus monkeys [63]. Moderate changes in systemic cytokines were seen in individual animals, but no cytokine release syndrome or long-term side effects were observed, up to the dose of 50 mg/kg, which was defined in the study as the no-observed adverse effect level (NOAEL). The dose for the first in human study took this NOAEL level, applied an allometric correction factor of 3.1 (to reach 16  mg/kg), divided this by 10 as a safety factor (to reach 1.6 mg/kg), and then added an additional safety margin to start at a dose of 0.1 mg/kg, representing a total safety factor of 160. Six healthy volunteers were administered the drug intravenously at a dose of 0.1 mg/kg, all over the space of an hour in March 2006; all of them rapidly became extremely unwell [64]. The trial volunteers initially experienced myalgia, fever, headache, gastrointestinal disturbance, amnesia, and rash, followed in the next few hours by hypotension, fever to 40.0°C, and respiratory failure. Subsequently all six patients were admitted to intensive care for monitoring, with two patients requiring intubation, and four patients requiring non-invasive mechanical ventilation. All six patients had significant deterioration in their renal function, underwent haemodialysis, and developed disseminated intravascular coagulopathy. Four patients improved over the next 4 days, while two patients required intubation and haemodialysis for prolonged periods, experiencing necrosis of fingers and toes and sepsis. All patients had marked lymphopenia and monocytopenia. This extraordinary sequence of events garnered enormous press attention, unsurprisingly, and investigations were carried out by different authorities, including the Medicines and Health Care Regulatory Agency (MHRA) and an Expert Scientific Group of the UK Department of Health (ESG) [63, 65]. The Committee for Medicinal Products for Human Use (CHMP) of the European Medicines Agency (EMEA) subsequently issued guidelines for high-risk medicinal products, which incorporated many of the ESG’s suggestions [66], including the following recommendations: 1 . Improved sharing of information about potential side-effects in the preclinical setting. 2. Increased collaboration between sponsors and regulators. 3. Conservative approach to dose selection based on a “Minimal Anticipated Biological Effect Level” or other models. 4. Careful monitoring at administration. 5. Staggered administration of drug to subjects with appropriate period of observation between dosing of individual subjects. 6. Appropriately qualified personnel who are sufficiently informed about the agent, its target, and mechanism of action to make informed clinical judgements. 7. Consideration of a treatment strategy in advance if adverse events are ­anticipated, including intensive care facilities if required, appropriate standard operating ­procedures, and clear information given to subjects about their response to an adverse reaction.

19  Optimising the Development of Antibodies as Treatment for Cancer

547

These and other factors are further expanded on in the resources listed below. The list is only partial, and the regulatory process is complex; accordingly, readers are advised to consult widely the appropriate regulations and authorities. With this case in mind, we now review phase I trial dose selection, schedule of administration, selection of subjects, toxicity assessment, and correlative studies. • MHRA: Final report on TGN1412 clinical trial; 2006. Available at http://www. mhra.gov.uk. • Final Report of Expert Scientific Group on Phase I Clinical Trials, Norwich; 2006. http://www.dh.gov.uk/en/Publicationsandstatistics/Publications/Publications PolicyAndGuidance/DH_063117. • CHMP Guideline on Strategies to Identify and Mitigate Risks for First-inHuman Clinical Trials with Investigational Medicinal Products; 2007. http:// www.emea.europa.eu/pdfs/human/swp/2836707enfin.pdf. • EMEA Note for Guidance on Preclinical Safety Evaluation of BiotechnologyDerived Pharmaceuticals. http://www.dbtbiosafety.nic.in/guideline%5CICH%20 GUidelines.pdf. • FDA Guidance for Industry Monoclonal Antibodies Used as Reagents in Drug Manufacturing, 2006. http://www.fda.gov/cder/Guidance/3630fnl.pdf. • FDA Points to Consider in the Manufacture and Testing of Monoclonal Antibody Products for Human Use. http://www.fda.gov/Cber/gdlns/ptc_mab.pdf. • FDA Industry and Reviewers Guidance for Industry Estimating the Safe Starting Dose in Clinical Trials for Therapeutics in Adult Healthy Volunteers. http:// www.fda.gov/cber/gdlns/dose.pdf. • Please note that this is a partial list of relevant documents, and up-to-date and ­relevant local guides and regulations should be widely consulted.

19.12 Phase I Trials 19.12.1 Dose Selection The TGN1412 example as discussed above used the NOAEL dose as a starting dose, a conversion factor to the Human Equivalent Dose (HER), and a safety factor of 160. Calculation of the percentage CD28 receptors occupied by TGN1412, however, shows this to be very high at this starting dose – 90.6%, and (possibly) as a result, patients underwent profound cytokine storm [63]. The NOAEL approach to dose selection was developed in the setting of small molecules, and its lack of safety in this setting illustrates potential differences in the applicability of this approach to specific biological agents such as antibodies, particularly because of the less straightforward translation of preclinical work in animals to humans. Alternative approaches have been proposed, and the ESG commentary on the TGN1412 incident specifically calls for “a wider approach to dose calculation,

548

C.P. Carden et al.

based on all relevant information”, and suggests alternative methods of dose selection, such as establishing a “Minimum Anticipated Biological Effect Level” (MABEL) [63, 67]. This approach uses a range of factors to determine a dose at which any biological effect is seen and reduces the dose from this point. It incorporates such factors as those listed below, any of which is sufficient to determine the MABEL, and the lowest of which is used as the dose used in humans: • • • • • • •

The novelty of the agent The degree of species specificity The mechanism of action Dose–response curves of biological effects in human and animal cells Dose–response data from relevant in vivo animal studies Receptor occupancy versus concentration The extent of exposure of targets and target cells in vivo

Hypothetically and in retrospect, the MABEL approach to calculating the dose in the TGN1412 example, using available data at the time, could have been used to determine a no observed effect level (NOEL) of <0.3 mg/kg from preclinical studies in rats and an optimal pharmacological response at 1–5  mg/kg, leading to a MABEL dose between these ranges (e.g. 0.5 mg/kg). From this dose a safety factor of 100 would be introduced to give a starting dose of 0.005 mg/kg (20-fold lower than that used in this study) [63, 66]. While the MABEL approach may increasingly become a relevant standard in the field, more data is needed to justify its broad acceptance. Further evaluation of these different methods to establish an optimised methodology for therapeutic antibody Phase I starting-dose estimation is now required.

19.12.2 Staggering of Treatment of Patients The ESG and EMEA guidelines both support staggered administration of drug with an appropriate interval between dosing of subjects to limit the number of subjects that may be affected by a severe adverse event [63, 66]. Although no PK data will be available from humans, animal studies may provide some guidance. The period between successive doses of an agent should take into account drug, target, and recipient and potential adverse reactions and is usually a course of treatment in duration (3–4 weeks of monitoring of patients). A period of treatment of the first patient at each new dose level, before the next patients are treated, was also recommended (usually <  7 days).

19.12.3 Schedule Selection It is suggested by the above guidelines that schedule may be predicted based upon the IG subtype, animal PK, and any specific toxicities expected. Most therapeutic

19  Optimising the Development of Antibodies as Treatment for Cancer

549

a­ ntibodies are given every 2 or 3 weeks based upon half-lives around this period, and a generally tolerable side-effect profile. Certain therapeutic antibodies, particularly immunity-altering antibodies, may benefit from a longer drug-free period to minimise risk of toxicity. Protocols can be written with an a priori exploration of schedule variation to manage toxicity. Real-time PK and PD should be performed and available to clinical investigators to guide exploration and validation of different schedules.

19.12.4 Selection of Patients ESG recommendations are that the decision whether patients or healthy volunteers are enrolled for a phase I trial be made on a case-by-case basis, with the risk to benefit assessment being crucial [63]. The benefits of phase I trial conducted in volunteers are debatable. Overall, there appears to be little robust evidence to support implementation of this. With regard to cancer patients being entered into phase I trials, selection of patients is key to minimizing risk and increasing the possibility of benefit [68]. We have shown that patients with an albumin <30 g/L, high disease burden and a high LDH have a poor prognosis and are probably not optimal candidates for phase I study due to their short life-expectancy [69]. Moreover, careful consideration of preclinical data regarding disease molecular characterisation, responses to relevant agents in particular tumour types, relevant data regarding other agents in the treatment of the particular cancer, and careful incorporation of potential predictive biomarkers as inclusion criteria may lead to improvements in selection of patients. A recent example of this approach is an expansion cohort of a phase I trial treating patients with Ewings sarcoma with CP 751–871, an IGFR antibody, based on a strong molecular rationale and preclinical models [70].

19.12.5 Trial Conduct Appropriate staffing, supervision, and training must be in place in the setting of the firstin-man introduction of a therapeutic antibody into the clinic, including appropriate availability of intensive care facilities [63]. Because of the potential immunogenicity of antibodies, protocols and the systems that are set up to run them must have clear-cut supportive and emergency care procedures for dealing with toxicities ranging from anaphylaxis to autoimmunity. Triggers for concern should be prospectively identified in the protocol and ongoing review of adverse events incorporated into the trial practice.

19.12.6 Dose Escalation While small molecules are often dose-escalated by the modified Fibonacci or linear escalation strategies, traditional escalation of antibodies has been based on a

550

C.P. Carden et al.

l­ogarithmic scale with increments in half logs, e.g. 0.1, 0.3, 3.0, 10.0, etc. [61]. CPHM Guidance document suggests considering amending the dose escalation increment based on the steepness of the dose/toxicity and dose/effect curves observed in preclinical studies [66].

19.12.7 Pharmacokinetic Sampling Flexible schedules should be used to acquire a detailed PK disposition profile. This may depend on the expected half-life of the antibody, but should take into account uncertainties of translating preclinical data into the human setting, and thus allow for sufficient time points for sampling to cover the potential of an unexpectedly long or short half-life. Again, pre-specified criteria for changing the PK sampling schedule parameters may be appropriate to allow extra sampling if required to characterise unexpected PK dynamics. The acquisition of PK samples when doselimiting toxicity or unexpected severe toxicities occur is also recommended.

19.12.8 Pharmacodynamic Sampling to Biopsy or Not to Biopsy? Pharmacodynamic (PD) sampling should be used to provide correlative data ­regarding tumour effects, using techniques identified in the preclinical setting. PD sampling can be used to identify target activity, pathway modulation, desired biological effect and to correlate these factors to response or otherwise to treatment [71]. In the phase I setting, an analysis of abstracts presented at scientific meetings showed infrequent use of biomarkers overall in phase I studies of different agents [72]. Antibodies may differ from small molecules, however, in the extent to which biomarkers can help select dose and schedule in phase I trials. As we will discuss in Sect. 19.13.3, PK data of receptor occupancy and elimination of antibody have been used to select the RP2D of at least one commonly used antibody, cetuximab, because a concentration of antibody is reached beyond which further increases in drug administered lead to no significant increase in concentration due to saturated FcRn sequestration devices. In addition highly sensitive immunoassays have been developed for assessing receptor occupancy. Therefore, this suggestion of low biomarker use in phase I trials may not translate into the specific setting of antibody trials. It has also been suggested that biopsies in appropriately selected and consented patients are acceptable and safe. A survey of patient willingness for biopsy suggested that patients were frequently happy to have a biopsy for research purposes [73]. A report of 192 biopsies performed for PD studies in the phase I setting describes very low rates of serious complication (<1%) and the success rate of obtaining paired tumour samples was 88% [74]. Alternatives to biopsy are becoming more recognised: surrogate tissues such as peripheral blood lymphocytes or hair follicles, or new technologies such as circulating tumour cell isolation and molecular analysis, and new trial

19  Optimising the Development of Antibodies as Treatment for Cancer

551

designs incorporating biomarker validation all provide opportunity for biomarker exploration within the early stage trial setting [71, 75]. Appropriate timing of posttreatment biopsy, given more prolonged half-life of antibodies, should be considered.

19.12.9 Combinations with Cytotoxic Chemotherapy or Radiotherapy Although addition of chemotherapy or radiotherapy to antibody treatment has not generally impacted the recommended combination antibody dose (at least for reasons of safety) in most combinations of treatment, a phase I trial of the requisite combination should be conducted in most settings. This is because of the low possibility of pharmacokinetic interaction and the more significant chance of increased toxicity [76]. The 2–4 week t1/2 of antibodies also means that antibody will be found at significant concentrations at cycle 2, and scheduling will be most relevant to the first cycle. Dose escalation would generally start at the higher dose of the agent thought to have the most efficacy, and escalate the other component, although the first dose of both agents is often reduced.

19.12.10 Combinations with Other Targeted Agents Again, phase I trials must be performed for reasons of toxicity, as dual blockade of one pathway or blockade of two pathways may well have toxicity not seen with a single agent. A current example of interest is the combination of an EGFR targeted agent and a VEGFR-targeted agent, which in two phase III trials caused unexpected severe toxicity (skin toxicity, diarrhoea, and deep venous thrombosis in the PACCE trial combining bevacizumab and irinotecan +/− panitumumab, and skin toxicity and diarrhoea, with worse PFS in the combination arm in the CAIRO2 trial examining capecitabine, oxaliplatin, and bevacizumab +/− cetuximab) emphasizing the need for early stage trials of combinations [77, 78]. Of particular relevance here is the combination of two therapeutic antibodies that may have potentially interacted at target, metabolism [especially with respect to FcRn recycling (Fig. 19.1)] and ADCC levels: appropriate PK–PD studies should be performed to assess these interactions in early clinical trials.

19.12.11 Optimising Transition from Phase I to Phase II: The “Seamless” Transition Alternatives to considering phase I and subsequent trials as isolated entities should be explored in order to accelerate drug development. Expansion at the ­recommended

552

C.P. Carden et al.

FcRn

A

ds bin ands lig

FcRn

blocks ion diversat

opsonins From NK cell inhibits effector pathway

s ck ge blo ava cle

c c4 5a a c3

Chemoattractant

C

phagocytosis

is) onins lys lls: NK (ops es etc. macrophag is) (phagocytos

Effector ce

a

c3

C1q

MAC promotes apoptosis

B

eron

s,interf etc

cytokine

Merocyte / Endoreticular cell

c3b

phagocytos is

C3b R

c3b

immune activation FcR IIIa

D

T cells

Lysis

receptor internalised Cancer Cell

E

CTLA4

A:

Mechanism of Antibody Recycling by means fo FcRn.

B:

Modes of Antibody Action - Direct.

C:

Indirect Antibody Action - Complement Dependent Cytotoxicity.

D:

Indirect Antibody Action - Antibody Dependent cell Cytotoxicity.

E:

Other Immune Effects: effect of ipilimumab or T cell function.

immune Function

Fig. 19.1 

phase II dose in unselected or in selected patient groups can also allow more rapid recognition of anti-tumour activity and further assessment of tolerability. Should such activity be demonstrated, trials can be designed to allow acquisition of tissue and initial evaluation of the molecular basis of response. An expansion cohort provides an ideal window for evaluation of activity in multiple tumour types.

19.13 Phase II and III Trials 19.13.1 Endpoints and Study Design Compared with the significant differences in approach for studying antibodies and small molecules in the phase I setting, the phase II and III approaches are more similar, largely because the criteria for determining clinical benefit do not differ [61]. Experience with these types of agents has shown them to frequently result in little change in the size of tumours compared with cytotoxic agents, and accordingly there is significant interest in alternative endpoints and trial designs for phase II trials for targeted agents [79–83]. Study designs to facilitate development of these agents include randomised phase II trials and the randomised discontinuation design.

19  Optimising the Development of Antibodies as Treatment for Cancer

553

19.13.2 Other Endpoint and Design Considerations Phase II trials should be designed to assess different exploratory endpoints compared with those for cytotoxic agents and small molecules. In general, there is lesser need to test for drug–drug interactions, compared with many of the small molecule agents, which are metabolised by hepatic P450 enzymes. A recent review of interactions between antibodies and chemotherapy, however, did show occasional effects of the former on the latter (e.g. 25% decreased epirubicin concentration when coadministered with trastuzumab and docetaxel) and vice versa (e.g. 1.5× increase in trastuzumab concentration in combination with paclitaxel as opposed to other chemotherapy agents), suggesting a role for dedicated drug study [76]. There is less need to assess effects on the QT interval on ECG, because antibodies are not thought to cause significant QT prolongation [61]. On the other hand, because of the critical effects of host immune responses to the therapeutic antibody on PK and toxicity, these do need to be assessed, with measurement of human anti-human antibodies.

19.13.3 Selection of Recommended Phase II Dose As opposed to the maximum tolerated dose as an endpoint of phase I trials, it has been suggested that antibody phase I trials should focus on identifying an optimal biological dose to take forward as the recommended phase II dose (RP2D) [84]. This is proposed to be feasible and reasonable on the basis of PK measures such as antigen binding, receptor saturation, and downstream effects, as these can be quantified more easily in many cases for antibodies than they can be for small molecules. An example of PK-guided dose selection was cetuximab: beyond 200 mg/m2 exposure to drug did not further increase with dose escalation, suggesting saturation of the FcRn mechanism [85]. Indeed, most monoclonal antibodies have not reached a dose-limiting toxicity in phase I trials, with toxicities being further defined in phase II and III trials. However, the anti-CTLA4 antibody ipilimumab in combination with melanoma vaccine RP2D was established by virtue of three CTCAE Grade 3 gastrointestinal events, and the agent has been significantly associated with autoimmune reactions [86]. Accordingly, dose selection for the phase II study is probably best made on a combination of these factors.

19.13.4 Selection of Patient Population As with other agents, selection of the appropriate population for a phase II trial of an antibody alone or combination of antibody with another modality will likely depend on preclinical data generated from cell lines, animal models, and the ­clinical experience gained in phase I trials. Inter-patient heterogeneity in a cancer

554

C.P. Carden et al.

type is an important consideration. Some lung cancers, for example, are driven predominantly by Ras, whereas others have mutations of the EGFR itself and are more likely to respond to EGFR-targeted therapies. Patients with colorectal cancer and KRAS mutations are unlikely to respond to cetuximab [87–89]. Methods of establishing dysfunction or hyperfunction of particular oncogenic pathways as well as resistance mechanisms through molecular analysis or imaging need further development and validation [68]. As a consequence, in the future, phase II studies may need to become target-selective (e.g. Akt-activated, homologous recombination-deficient, EGFR-mutated, etc.), not disease site-selective. Various methods of selecting patients for studies are under evaluation and are reviewed elsewhere [68]. One method of particular relevance to the field of therapeutic antibodies is the use of antibodies as diagnostics. Antibodies can relatively easily be radiolabelled, as has been performed with the anti-Epcam molecule ING-1 and trastuzumab, and shown to localise to tumours. Theoretically patients with uptake in tumours on dosing with radiolabelled antibody may be more likely to respond to treatment with the therapeutic antibody than those who do not [90, 91].

19.13.5 Moving from the Metastatic to the Adjuvant Setting Most anti-cancer agents are initially evaluated in the metastatic setting, often initially as a single agent, then in combination with second or third-line ­chemotherapy, before their study in the first-line setting, and finally in the ­adjuvant setting [92]. This process is dictated by ethical constraints. Nonetheless, the benefit of some agents may be most fully realised in earlier stages of disease, particularly in the adjuvant setting, where microscopic residual disease is ­present and cure is a possibility. For example, in studies reported from 2001, the HER2 monoclonal antibody trastuzumab in combination with chemotherapy in the metastatic setting demonstrated response rates of 50% compared with 32% with chemotherapy alone and was associated with an overall survival of 25.1 and 20.3 months, respectively. However, it was not until 2005 when improvements in the adjuvant setting – HR for an event (recurrence, contralateral breast cancer, other cancer, or death) of 0.54 (95% CI 0.43–0.67; P < 0.0001) favouring those patients who received adjuvant trastuzumab – were reported [93, 94].

19.13.6 Single Agent Versus Combination with Chemotherapy In general, chemotherapy can be combined safely with antibodies. The molecular mechanisms for the effect of the combination of cytotoxic chemotherapy are various and include the following: • Normalisation of the vasculature and increased chemotherapy delivery (antiangiogenic)

19  Optimising the Development of Antibodies as Treatment for Cancer

555

• Enhanced cytotoxicity resulting from chemotherapy and ADCC/CDC from antibodies • The ability of some antibodies to overcome resistance to chemotherapy (IGFR antibodies) [95] • Sensitisation to chemotherapy-induced apoptosis [96] The differential efficacy of antibodies, as opposed to small molecules, in combination with chemotherapy is currently under examination, as is the optimal sequencing of treatments.

19.14 Emerging Technologies: Conjugated Antibodies (Immunoconjugates) Immunoconjugates were developed as another strategy to maximise the delivery (and thereby intratumoural concentration) of a variety of anti-cancer drugs to tumours, while sparing non-tumour tissues’ exposure to the agent [97, 98]. Conjugated antibodies function as carriers of various substances including cytotoxic drugs, radio isotopes, and toxins. They can be categorised into three groups: (1) radio-immunoconjugates [99], (2) immunotoxin conjugates [100], and (3) antibody–drug conjugates or tumour-activated prodrugs [101]. Crucial to the efficacy of immunoconjugates is the requirement for a high ­therapeutic ratio, and this is attempted by using effector molecules that are effective at low dose levels and by optimising the linker molecule [97]. The linker, a short molecule containing hydrazone, disulfide, peptide, or other chelating or covalently bonding components, connects the antibody to the active substance (drug, toxin, radionuclide) [101]. The linker is crucial in determining the stability of the compound, and immunoconjugates can be designed with this in mind. For radio-immunoconjugates, for example, stability is required so that the molecule is metabolised and excreted in keeping with the PK properties of the antibody, in order to avoid toxicity to normal organs from the freely circulating radioactive component [101, 102]. Metal chelators and covalent bonding to amino acids both provide such stability and have been used in 90Y-ibritumomab tiuxetan and 131 I-tositumumab, respectively, in the treatment of non-Hodgkin’s lymphoma (see below). For drug immunoconjugates and toxin-immunoconjugates, in comparison, a linker is required that is stable in plasma, but labile on internalisation of the complex into lysosomes in a target cell, thus again avoiding normal organ toxicity [103]. Hydrazine and disulphide bond linkers provide such properties and have been incorporated in practice with gemtuzumab ozagamicin (see below). PK assessment of free drug and antibody in trial subjects can provide information on these crucial properties of immunoconjugates if performed appropriately in early trials. There are a variety of drugs currently being investigated in clinical trials or used as standard treatments.

556

C.P. Carden et al.

19.14.1 Drug-Immunoconjugates Considerable effort is underway to develop promising immunoconjugates – antibodies conjugated to small highly cytotoxic drugs. These studies exemplify the difficulties of developing immunoconjugates because of the need for potency and appropriate stability [104]. Drug classes being explored as drug-immunoconjugates include inhibitors of tubulin polymerisation, DNA alkylating agents, and enediyene antibiotics that catalyse DNA double-strand breaks [105]. Gemtuzumab ozogamicin is an antibody–drug conjugate constructed from calicheamicin, which potently binds to the DNA minor groove, linked covalently to an anti-CD33 monoclonal antibody. CD33 is expressed on cells of myeloid lineage, and is approved by the FDA for the treatment of patients with relapsed acute myeloid leukemia [106–108]. Other antibody–drug conjugates in development include the maytansine-derivative DM1, a potent anti-microtubule agent, which has been linked to an antibody CD56 to form huN901-DM1 [97]. This conjugate was studied and examined in a phase II trial in 30 patients with small-cell lung cancer and other CD56-positive small cell tumours, with one objective and one non-confirmed response, and huC242-DM1, which targets the CanAg antigen present on various tumours [109, 110]. Two phase I studies have examined different schedules of trastuzumab-DM1 in patients with metastatic breast cancer who have progressed on (or soon after) trastuzumab administration [109, 110]. Interestingly, these ­studies have demonstrated response rates of 36 and 53% in trastuzumab-resistant/refractory HER2+ metastatic breast cancer, and an expansion cohort of 15 patients treated at RP2D revealed a median progression-free survival of 9.8 months, with little toxicity [111, 112]. SGN-15 is an anti-Lewis Y antibody conjugated to doxorubicin and a randomised phase II study comparing SGN-15 and docetaxel versus docetaxel alone for NSCLC has been performed, demonstrating little toxicity, and activity of the combination [113]. Several other antibody–drug conjugates have shown striking activity in preclinical models and are advancing towards or have entered clinical trials.

19.14.2 Radio-Immunoconjugates The b-emitters Yttrium-90 (90Y) and Iodine-131 (131I) linked to monoclonal antibodies as radio-immunoconjugates have been most extensively studied, with most promising clinical results in haematological malignancies, which tend to be radiosensitive tumours, and to which it is possible to deliver sufficient dose for tumour eradication [114, 115]. The radio-immunoconjugates 90-Y ibritumomab (90Y-anti CD20) and 131-I tositumumab (131I-anti-CD20) have been approved by the FDA for the treatment of non-Hodgkin’s lymphoma. Solid tumours have been less successful in the clinic as yet due to various factors [116]. Multiple factors such as tumour radiosensitivity, total dose, dose rate, and degree of tumour penetration all affect

19  Optimising the Development of Antibodies as Treatment for Cancer

557

tumour response to radioimmunoconjugates, all of which are more difficult to achieve sufficient levels in solid tumours compared with haematological malignancies [115]. In particular, cumulative doses of radiolabelled antibodies do not usually exceed more than 1,500 cGy and are significantly lower than 5,000 cGy needed to achieve a therapeutic response in most tumours. Additionally, the most critical relevant limiting factor for radio-immunotherapy is the normal organ toxicity of 150–200  cGy. Vascularisation, intratumoural pressure, and other barriers to antibody penetration affect the levels of targeted antibodies impacting on tumours, and these are potentially by size and location of disease [115].

19.14.3 Immunotoxin Conjugates Immunoconjugates have also been difficult to bring to clinical fruition [117]. Given that toxins are often proteins, one of the limiting factors is the development of human anti-toxin immune responses that limit efficacy and repeat dosing. Fragments of antibody conjugated with toxins, often murine single-chain variable-domain fragments, also result in the appearance of neutralising HAMAs [100]. A limited number of immunotoxin conjugates have undergone clinical ­development. BL22, an anti-CD22 Fv fragment fused to a Pseudomonas toxin, has undergone phase I and II testing, and in the latter, patients with hairy cell leukaemia resistant to chemotherapy complete responses were observed in 47% of patients, with an overall response rate of 72% of patients. Toxicities included a reversible haemolytic-uremic syndrome in 6% and immunogenicity in 11% of patients in the phase II trial [118]. SGN-10, a Lewis Y-targeting single-chain variable fragment conjugated to a fragment of pseudomonas exotoxin, showed little efficacy in metastatic carcinoma positive for the Lewis Y antigen [119]. Denileuin diftitox (Ontak) is a diphtheria toxin fragment fused to IL-2 and used in cutaneous T-cell lymphoma. A 30% objective response rate was observed, but 98% of patients developed human anti-toxin antibodies by the second treatment dose [97, 120]. Targeting solid tumours with immunotoxin conjugates remains limited by immunogenicity, although attempts to modulate this are in development (see below).

19.15 Other Modulations of Antibody Function 19.15.1 Multivalent Antibodies Antibodies that have more than one binding site have the potential to bind more strongly to the cell surface, increasing their effect [9, 121, 122]. In addition, ­multivalent antibodies may target two spatially co-localising epitopes, for

558

C.P. Carden et al.

e­ xample a receptor and a co-receptor, or two different receptors. A phase IIa study comparing two doses of an intraperitoneally infused bispecific antibody catumaxomab, which has anti-Epcam and anti-CD3 (T-cell coreceptor) activity, in patients with platinum refractory ovarian cancer was recently reported, demonstrating modest activity [123]. A fully humanised antibody with specificity for EGFR and IGFR has shown interesting preclinical efficacy, with high rates of receptor internalisation and recruitment of ADCC effector cells in xenograft models [95]. Multiple variable Fab fragments can be connected with small ­linking molecules providing structures smaller than a complete antibody with potentially better tumour ­penetration, but also faster elimination from the body.

19.15.2 Antibody Fragments Intact antibodies are large molecules with correspondingly poor tumour penetration. The Fc component can be removed by proteolysis or genetic engineering, leaving the variable region intact, resulting in better penetration and less immunogenicity. These protein molecules can be engineered in a variety of formats, around the framework of a variable region connected to other components by a linking protein [110]. The ­current generation of fragments, however, is poorly soluble, prone to aggregation, and quickly eliminated from the body, lacking the recycling FcRn mechanism.

19.15.3 Intrabodies Intrabodies are antibodies that function within the cytoplasm of a cell [124]. Delivering part of an antibody to the intracellular component of the cell remains technically difficult to achieve, and it involves either of the following: • Transfer of genes from a vector such as an adenovirus to the DNA of the host, with resultant transcription and translation into a functional protein capable of interacting with intracellular processes • Transfer of protein from outside the cell by liposomes, encapsulation by lipid or peptides, or by linkage with peptide transduction domains [125, 126] Such techniques, although intriguing, appear distant away from clinical development.

19.15.4 Modulation of Immunogenicity 19.15.4.1 Enhanced ADCC Animal model studies have suggested that improving the efficacy of ADCC may improve the cytotoxicity and anti-tumour activity of an antibody [127]. With

19  Optimising the Development of Antibodies as Treatment for Cancer

559

respect to rituximab, this is on the basis that individuals with certain ­polymorphisms of one of the effector cell receptors for Fc, FcgR IIIa (i.e. valine/valine or valine/ phenylalanine at position 158 on the gene for the receptor) maintain effector cells with increased FcgR IIIa expression, binding to rituximab, and ADCC activation. This is in contrast to patients with other FcgR polymorphisms (i.e. valine/phenylalanine or phenylalanine/phenylalanine) [128–130]. In addition, polymorphisms in the FcgR IIIa may be important for the response of patients with colorectal cancer to cetuximab [131, 132]. In addition, specific amino acid residues in the Fc region of antibodies have been shown to play crucial roles in FcgR binding, and these can be genetically modified to potentially enhance ADCC [52, 133]. Second-generation anti-CD20 antibodies have been designed which have a higher affinity for the FcgR, and ­preclinical studies show that NK cell increased activation at lower concentrations than rituximab [134–136]. The impact of carbohydrate chains attached to the Fc and FcgR binding also appears to be important, by effecting a conformational change in the CH2 domains. Cell lines having “fructose-free” antibody structure have been developed, and these demonstrated higher activity in murine models of T-cell leukaemia and lymphoma than high-fructose-containing antibodies [133, 137–140].

19.15.4.2 Modulation of CDC As with ADCC there are second-generation antibodies being engineered that have increased activation of CDC [127]. Ofatumumab, for example, a secondgeneration fully human IgG1 anti-CD20 mAb, has high affinity to a region of CD20 close to the membrane, which should facilitate C1q capture and efficient CDC [141]. A phase I–II trial in CLL has shown an impressive clinical response rate of 50% [142]. However, there are data suggesting that complement can interfere with ADCC function by inhibition of FcgR by C3b, suggesting that antibodies that promote ADCC independent of CDC may be more effective, but this awaits ­confirmation [143]. 19.15.4.3 Modulation of Other Immune Components Modulation of other components of the immune response is also potentially important. Of current interest is the anti-CTLA 4 antibody, ipilimumab, which releases inhibition of T cells, and either alone or in combination with vaccines, radiotherapy or chemotherapy is showing promising results in ongoing trials, albeit at the cost of potentially significant autoimmune toxicity. For example, initial reports of phase I trial of ipilimumab in castration-resistant prostate cancer with or without radiotherapy to a target lesion had PSA 50% reduction response rates of 21%, but high rates of severe CTCAE Grade >3 toxicity (colitis, skin, or hepatic) at 28–50% [17].

560

C.P. Carden et al.

These examples demonstrate the potential for new strategies in optimising a­ ntibody function through modulation of the immune system but also emphasise the need for careful and safe trial conduct.

19.16 Conclusions and Future Directions Monoclonal antibodies as therapeutic agents have become a major part of ­treatments in various settings including transplantation, oncology, autoimmune, ­cardiovascular, and infectious diseases. Today there are more than 160 monoclonal antibodies in clinical trials or approved for therapy. This progress is mainly related to new developments and techniques in two fields of research: (a) antibody production and (b) target identification. These advances are associated with substantial development costs, with the cost of taking an antibody to registration being in the order of US one billion dollars, and a lag time from early clinical development to registration of many years [2]. Improving the different components of this process, including trial design, manufacturing, transition from preclinical to clinical studies, appropriate patient and disease site selection, and transition from the later to earlier stages of disease, is a major but critical challenge for industries, academia, and regulatory bodies. Finally, new technologies and approaches to antibody engineering provide exciting opportunities for continuing the expansion of antibodies in the treatment of cancer. Other developments in the field of oncology have implications for the field as well, including novel imaging modalities, high-throughput methods of assessing small volumes of tissue for specific biomarkers predictive of response to targeted agents, and novel use of technologies such as circulating tumour cells to provide a “non-invasive biopsy”. Careful, rigorous and innovative trial design involving ­academia, the pharmaceutical industry, and regulatory authorities, as well as ­optimal trial conduct, are required to enhance the safe and timely introduction of new antibodies into clinical practice.

References 1. Delves, P.J. and Roitt, I.M. 2000. The immune system. First of two parts. N Engl J Med 343:37–49. 2. DiMasi, J.A. and Grabowski, H.G. 2007. Economics of new oncology drug development. J Clin Oncol 25:209–216. 3. Iannello, A. and Ahmad, A. 2005. Role of antibody-dependent cell-mediated cytotoxicity in the efficacy of therapeutic anti-cancer monoclonal antibodies. Cancer Metastasis Rev 24:487–499. 4. Kamradt, T. and Mitchison, N.A. 2001. Tolerance and autoimmunity. N Engl J Med 344:655–664. 5. Foote, J. and Eisen, H.N. 1995. Kinetic and affinity limits on antibodies produced during immune responses. Proc Natl Acad Sci U S A 92:1254–1256.

19  Optimising the Development of Antibodies as Treatment for Cancer

561

6. Zafir-Lavie, I., Michaeli, Y., and Reiter, Y. 2007. Novel antibodies as anticancer agents. Oncogene 26:3714–3733. 7. Lobo, E.D., Hansen, R.J., and Balthasar, J.P. 2004. Antibody pharmacokinetics and pharmacodynamics. J Pharm Sci 93:2645–2668. 8. Tabrizi, M.A., Tseng, C.M., and Roskos, L.K. 2006. Elimination mechanisms of therapeutic monoclonal antibodies. Drug Discov Today 11:81–88. 9. Weiner, L.M. 2007. Building better magic bullets – improving unconjugated monoclonal antibody therapy for cancer. Nat Rev Cancer 7:701–706. 10. Reichert, J.M. and Valge-Archer, V.E. 2007. Development trends for monoclonal antibody cancer therapeutics. Nat Rev Drug Discov 6:349–356. 11. Croce, C.M. 2008. Oncogenes and cancer. N Engl J Med 358:502–511. 12. Kim, K.J., Li, B., Winer, J., Armanini, M., Gillett, N., Phillips, H.S., and Ferrara, N. 1993. Inhibition of vascular endothelial growth factor-induced angiogenesis suppresses tumour growth in vivo. Nature 362:841–844. 13. Lin, Y.S., Nguyen, C., Mendoza, J.L., Escandon, E., Fei, D., Meng, Y.G., and Modi, N.B. 1999. Preclinical pharmacokinetics, interspecies scaling, and tissue distribution of a humanized monoclonal antibody against vascular endothelial growth factor. J Pharmacol Exp Ther 288:371–378. 14. Ashkenazi, A. and Herbst, R.S. 2008. To kill a tumor cell: the potential of proapoptotic receptor agonists. J Clin Investig 118:11. 15. Walczak, H. and Krammer, P.H. 2000. The CD95 (APO-1/Fas) and the TRAIL (APO-2L) apoptosis systems. Exp Cell Res 256:58–66. 16. Haynes, N.M., van der Most, R.G., Lake, R.A., and Smyth, M.J. 2008. Immunogenic anticancer chemotherapy as an emerging concept. Curr Opin Immunol 20:545–557. 17. Beer, T.M., Slovin, S.F., Higano, C.S., Tejwani, S., Dorff, T.B., Stankevich, V., and Lowy, I. 2008. Phase I trial of ipilimumab (IPI) alone and in combination with radiotherapy (XRT) in patients with metastatic castration resistant prostate cancer (MCRPC). J Clin Oncol 26:251s. 18. Hersh, E.M., Weber, J.S., Powderly, J.D., et al. 2008. Disease control and long-term survival in chemotherapy-naive patients with advanced melanoma treates with ipilimumab (MDX010) with or without dacarbazine. J Clin Oncol 26:9022. 19. Weber, J.S., Targan, S., Scotland, R., et al. 2006. Phase II trial of extended dose anti-CTLA-4 antibody ipilimumab (formerly MDX-010) with a multi-peptide vaccine for resected stages IIIC and IV melanoma. J Clin Oncol 24:2510. 20. Bagshawe, K.D. 2006. Antibody-directed enzyme prodrug therapy (ADEPT) for cancer. Expert Rev Anticancer Ther 6:1421–1431. 21. Ross, J.S., Gray, K., Gray, G.S., Worland, P.J., and Rolfe, M. 2003. Anticancer antibodies. Am J Clin Pathol 119:472–485. 22. Kohler, G. and Milstein, C. 1975. Continuous cultures of fused cells secreting antibody of predefined specificity. Nature 256:495–497. 23. Countouriotis, A., Moore, T.B., and Sakamoto, K.M. 2002. Cell surface antigen and molecular targeting in the treatment of hematologic malignancies. Stem Cells 20:215–229. 24. Lonberg, N. 2005. Human antibodies from transgenic animals. Nat Biotechnol 23:1117–1125. 25. Reichert, J.M., Rosensweig, C.J., Faden, L.B., and Dewitz, M.C. 2005. Monoclonal antibody successes in the clinic. Nat Biotechnol 23:1073–1078. 26. Ohno, S., Mori, N., and Matsunaga, T. 1985. Antigen-binding specificities of antibodies are primarily determined by seven residues of VH. Proc Natl Acad Sci U S A 82:2945–2949. 27. Zhang, Q., Chen, G., Liu, X., and Qian, Q. 2007. Monoclonal antibodies as therapeutic agents in oncology and antibody gene therapy. Cell Res 17:89–99. 28. Stashenko, P., Nadler, L.M., Hardy, R., and Schlossman, S.F. 1980. Characterization of a human B lymphocyte-specific antigen. J Immunol 125:1678–1685. 29. Reff, M.E., Carner, K., Chambers, K.S., Chinn, P.C., Leonard, J.E., Raab, R., Newman, R.A., Hanna, N., and Anderson, D.R. 1994. Depletion of B cells in  vivo by a chimeric mouse human monoclonal antibody to CD20. Blood 83:435–445.

562

C.P. Carden et al.

30. Glennie, M.J., French, R.R., Cragg, M.S., and Taylor, R.P. 2007. Mechanisms of killing by anti-CD20 monoclonal antibodies. Mol Immunol 44:3823–3837. 31. Administration, F.-F.D. 2008. Highlights of Prescribing Information Rituxan. FDA-Federal Drugs Administration. 32. Hicklin, D.J. and Ellis, L.M. 2005. Role of the vascular endothelial growth factor pathway in tumor growth and angiogenesis. J Clin Oncol 23:1011–1027. 33. Eskens, F.A. and Sleijfer, S. 2008. The use of bevacizumab in colorectal, lung, breast, renal and ovarian cancer: where does it fit? Eur J Cancer 44:2350–2356. 34. Ferrara, N., Hillan, K.J., Gerber, H.P., and Novotny, W. 2004. Discovery and development of bevacizumab, an anti-VEGF antibody for treating cancer. Nat Rev Drug Discov 3:391–400. 35. Administration, F.-F.D. 2008. FDA Approval for Bevacizumab. 36. McKeage, K. and Perry, C.M. 2002. Trastuzumab: a review of its use in the treatment of metastatic breast cancer overexpressing HER2. Drugs 62:209–243. 37. Hudis, C.A. 2007. Trastuzumab – mechanism of action and use in clinical practice. N Engl J Med 357:39–51. 38. Aird, K.M., Ding, X., Baras, A., Wei, J., Morse, M.A., Clay, T., Lyerly, H.K., and Devi, G.R. 2008. Trastuzumab signaling in ErbB2-overexpressing inflammatory breast cancer correlates with X-linked inhibitor of apoptosis protein expression. Mol Cancer Ther 7:38–47. 39. Arnould, L., Gelly, M., Penault-Llorca, F., Benoit, L., Bonnetain, F., Migeon, C., Cabaret, V., Fermeaux, V., Bertheau, P., Garnier, J., et  al. 2006. Trastuzumab-based treatment of HER2-positive breast cancer: an antibody-dependent cellular cytotoxicity mechanism? Br J Cancer 94:259–267. 40. Valabrega, G., Montemurro, F., and Aglietta, M. 2007. Trastuzumab: mechanism of action, resistance and future perspectives in HER2-overexpressing breast cancer. Ann Oncol 18:977–984. 41. Institute, N.-N.C. 2008. FDA Approval for Trastuzumab. 42. Dassonville, O., Bozec, A., Fischel, J.L., and Milano, G. 2007. EGFR targeting therapies: monoclonal antibodies versus tyrosine kinase inhibitors. Similarities and differences. Crit Rev Oncol Hematol 62:53–61. 43. Mendelsohn, J. 2001. The epidermal growth factor receptor as a target for cancer therapy. Endocr Relat Cancer 8:3–9. 44. Goldstein, N.I., Prewett, M., Zuklys, K., Rockwell, P., and Mendelsohn, J. 1995. Biological efficacy of a chimeric antibody to the epidermal growth factor receptor in a human tumor xenograft model. Clin Cancer Res 1:1311–1318. 45. Kurai, J., Chikumi, H., Hashimoto, K., Yamaguchi, K., Yamasaki, A., Sako, T., Touge, H., Makino, H., Takata, M., Miyata, M., et al. 2007. Antibody-dependent cellular cytotoxicity mediated by cetuximab against lung cancer cell lines. Clin Cancer Res 13:1552–1561. 46. Cunningham, D., Humblet, Y., Siena, S., Khayat, D., Bleiberg, H., Santoro, A., Bets, D., Mueser, M., Harstrick, A., Verslype, C., et al. 2004. Cetuximab monotherapy and cetuximab plus irinotecan in irinotecan-refractory metastatic colorectal cancer. N Engl J Med 351:337–345. 47. Bonner, J.A., Harari, P.M., Giralt, J., Azarnia, N., Shin, D.M., Cohen, R.B., Jones, C.U., Sur, R., Raben, D., Jassem, J., et al. 2006. Radiotherapy plus cetuximab for squamous-cell carcinoma of the head and neck. N Engl J Med 354:567–578. 48. Saadeh, C.E. and Lee, H.S. 2007. Panitumumab: a fully human monoclonal antibody with activity in metastatic colorectal cancer. Ann Pharmacother 41:606–613. 49. Keating, M.J., Cazin, B., Coutre, S., Birhiray, R., Kovacsovics, T., Langer, W., Leber, B., Maughan, T., Rai, K., Tjonnfjord, G., et al. 2002. Campath-1H treatment of T-cell prolymphocytic leukemia in patients for whom at least one prior chemotherapy regimen has failed. J Clin Oncol 20:205–213. 50. Institute, N.-N.C. 2008. FDA Approval for Alemtuzumab. 51. Imai, K. and Takaoka, A. 2006. Comparing antibody and small-molecule therapies for cancer. Nat Rev Cancer 6:714–727. 52. Ghetie, V., Sally, W.E., and Vitetta, E.S. 2004. Pharmacokinetics of antibodies and immunotoxins in mice and humans. In Handbook of Anticancer Pharmacokinetics and Pharmacodynamics, Figg, W. and McLeod, H.L., editors. Totowa, NJ: Humana Press.

19  Optimising the Development of Antibodies as Treatment for Cancer

563

53. Roopenian, D.C. and Akilesh, S. 2007. FcRn: the neonatal Fc receptor comes of age. Nat Rev Immunol 7:715–725. 54. Chapman, K., Pullen, N., Graham, M., and Ragan, I. 2007. Preclinical safety testing of monoclonal antibodies: the significance of species relevance. Nat Rev Drug Discov 6:120–126. 55. Dillman, R.O. 2003. Monoclonal antibody therapy. In Principles of Cancer Biotherapy, Oldham, R.K., editor. Dordrecht, The Netherlands, Kluwer Academic. 56. Patel, D.D. and Goldberg, R.M. 2006. Cetuximab-associated infusion reactions: pathology and management. Oncology (Williston Park) 20:1373–1382; discussion 1382, 1392–1374, 1397. 57. Stebbings, R., Findlay, L., Edwards, C., Eastwood, D., Bird, C., North, D., Mistry, Y., Dilger, P., Liefooghe, E., Cludts, I., et al. 2007. “Cytokine storm” in the phase I trial of monoclonal antibody TGN1412: better understanding the causes to improve preclinical testing of immunotherapeutics. J Immunol 179:3325–3331. 58. Mirick, G.R., Bradt, B.M., Denardo, S.J., and Denardo, G.L. 2004. A review of human antiglobulin antibody (HAGA, HAMA, HACA, HAHA) responses to monoclonal antibodies. Not four letter words. Q J Nucl Med Mol Imaging 48:251–257. 59. Lynch, C.M. and Grewal, I.S. 2008. Preclinical safety evaluation of monoclonal antibodies. Handb Exp Pharmacol 181:19–44. 60. Tabrizi, M.A. and Roskos, L.K. 2007. Preclinical and clinical safety of monoclonal antibodies. Drug Discov Today 12:540–547. 61. Weinberg, W.C., Frazier-Jessen, M.R., Wu, W.J., Weir, A., Hartsough, M., Keegan, P., and Fuchs, C. 2005. Development and regulation of monoclonal antibody products: challenges and opportunities. Cancer Metastasis Rev 24:569–584. 62. Loisel, S., Ohresser, M., Pallardy, M., Dayde, D., Berthou, C., Cartron, G., and Watier, H. 2007. Relevance, advantages and limitations of animal models used in the development of monoclonal antibodies for cancer treatment. Crit Rev Oncol Hematol 62:34–42. 63. Expert Scientific Group. 2006. Final Report of Expert Scientific Group on Phase I Clinical Trials. Norwich: The Stationery Office. 64. Suntharalingam, G., Perry, M.R., Ward, S., Brett, S.J., Castello-Cortes, A., Brunner, M.D., and Panoskaltsis, N. 2006. Cytokine storm in a phase 1 trial of the anti-CD28 monoclonal antibody TGN1412. N Engl J Med 355:1018–1028. 65. Authority, M.a.H.p.R. 2006. Final report on TGN1412 clinical trial. 66. (CHMP), C.f.M.P.f.H.U. 2007. Guideline on Strategies to Identify and Mitigate Risks for First-in-Human Clinical Trials with Investigational Medicinal Products. Committe for Medicinal Products for Human Use (CHMP). 67. Agoram, B.M. 2008. Use of pharmacokinetic/pharmacodynamic modelling for starting dose selection in first-in-human trials of high-risk biologics. Br J Clin Pharmacol 67:8. 68. Carden, C.P., Sarker, D., Postel-Vinay, S., Kaye, S.B., and de Bono, J.S. 2008. Patient selection for phase I trials using predictive biomarkers. Drug Discov Today- In submission. 69. Arkenau, H.T., Olmos, D., Ang, J.E., De Bono, J.S., Judson, I., and Kaye, S.B. 2008. Clinical outcome and prognostic factors for patients treated within the context of a phase I study: the Royal Marsden Hospital experience. Br J Cancer 98:5. 70. Olmos, D., Okuno, S., Schuetze, S.M., Paccagnella, M.L., Yin, D., Gualberto, A., Worden, F.P., Haluska, P., de Bono, J.S., and Scurr, M. 2008. Safety, pharmacokinetics and preliminary activity of the anti-IGF-IR antibody CP-751,871 in patients with sarcoma. J Clin Oncol 26S:10501. 71. Sarker, D. and Workman, P. 2007. Pharmacodynamic biomarkers for molecular cancer therapeutics. Adv Cancer Res 96:213–268. 72. Goulart, B.H., Clark, J.W., Pien, H.H., Roberts, T.G., Finkelstein, S.N., and Chabner, B.A. 2007. Trends in the use and role of biomarkers in phase I oncology trials. Clin Cancer Res 13:6719–6726. 73. Agulnik, M., Oza, A.M., Pond, G.R., and Siu, L.L. 2006. Impact and perceptions of mandatory tumor biopsies for correlative studies in clinical trials of novel anticancer agents. J Clin Oncol 24:4801–4807.

564

C.P. Carden et al.

74. Dowlati, A., Haaga, J., Remick, S.C., Spiro, T.P., Gerson, S.L., Liu, L., Berger, S.J., Berger, N.A., and Willson, J.K. 2001. Sequential tumor biopsies in early phase clinical trials of anticancer agents for pharmacodynamic evaluation. Clin Cancer Res 7:2971–2976. 75. Burzykowski, T., Buyse, M., Yothers, G., Sakamoto, J., and Sargent, D. 2008. Exploring and validating surrogate endpoints in colorectal cancer. Lifetime Data Anal 14:54–64. 76. Seitz, K. and Zhou, H. 2007. Pharmacokinetic drug-drug interaction potentials for therapeutic monoclonal antibodies: reality check. J Clin Pharmacol 47:1104–1118. 77. Hecht, J.R., Mitchell, E., Chidiac, T., Scroggin, C., Hagenstad, C., Spigel, D., Marshall, J., Cohn, A., Suzuki, S., and Grifin, T. 2008. Interim results from PACCE: Irinotecan/bevacizumab +/− panitumumab as first-line treatment for metastatic cancer. In ASCO 2008 Gastrointestinal Cancers Symposium. 78. Punt, C.J., Tol, J., Rodenburg, C.J., Cats, A., Creemers, G., Schrama, J.G., Erdkamp, F.L., Vos, A., Mol, L., and Antonini, N.F. 2008. Randomized phase III study of capecitabine, oxaliplatin and bevacizumab with or without cetuximab in advanced colorectal cancer (ACC), the CAIRO2 study of the Dutch Colorectal Cancer Group. J Clin Oncol 26:LBA4011. 79. Scher, H.I., Halabi, S., Tannock, I., Morris, M., Sternberg, C.N., Carducci, M.A., Eisenberger, M.A., Higano, C., Bubley, G.J., Dreicer, R., et al. 2008. Design and end points of clinical trials for patients with progressive prostate cancer and castrate levels of testosterone: recommendations of the Prostate Cancer Clinical Trials Working Group. J Clin Oncol 26:1148–1159. 80. Thall, P.F. and Wathen, J.K. 2008. Bayesian designs to account for patient heterogeneity in phase II clinical trials. Curr Opin Oncol 20:407–411. 81. Tuma, R.S. 2008. Examining heterogeneity in phase II trial designs may improve success in phase III. J Natl Cancer Inst 100:164–166. 82. Zohar, S. and Chevret, S. 2007. Recent developments in adaptive designs for Phase I/II dosefinding studies. J Biopharm Stat 17:1071–1083. 83. Suman, V.J., Dueck, A., and Sargent, D.J. 2008. Clinical trials of novel and targeted therapies: endpoints, trial design, and analysis. Cancer Invest 26:439–444. 84. Shaked, Y., Emmenegger, U., Man, S., Cervi, D., Bertolini, F., Ben-David, Y., and Kerbel, R.S. 2005. Optimal biologic dose of metronomic chemotherapy regimens is associated with maximum antiangiogenic activity. Blood 106:3058–3061. 85. Baselga, J., Pfister, D., Cooper, M.R., Cohen, R., Burtness, B., Bos, M., D’Andrea, G., Seidman, A., Norton, L., Gunnett, K., et al. 2000. Phase I studies of anti-epidermal growth factor receptor chimeric antibody C225 alone and in combination with cisplatin. J Clin Oncol 18:904–914. 86. Sanderson, K., Scotland, R., Lee, P., Liu, D., Groshen, S., Snively, J., Sian, S., Nichol, G., Davis, T., Keler, T., et  al. 2005. Autoimmunity in a phase I trial of a fully human anticytotoxic T-lymphocyte antigen-4 monoclonal antibody with multiple melanoma peptides and Montanide ISA 51 for patients with resected stages III and IV melanoma. J Clin Oncol 23:741–750. 87. Subramanian, J. and Govindan, R. 2008. Molecular genetics of lung cancer in people who have never smoked. Lancet Oncol 9:676–682. 88. Van Cutsem, E., Lang, I., D’haens, G., Moiseyenko, V., Zaluski, J., Folprecht, G., Tejpar, S., Kisker, O., Stroh, C., and Rougier, P. 2008. KRAS status and efficacy in the first-line treatment of patients with metastatic colorectal cancer (mCRC) treated with FOLFIRI with or without cetuximab: The CRYSTAL experience. J Clin Oncol 26:2. 89. Lievre, A., Bachet, J.B., Boige, V., Cayre, A., Le Corre, D., Buc, E., Ychou, M., Bouche, O., Landi, B., Louvet, C., et al. 2008. KRAS mutations as an independent prognostic factor in patients with advanced colorectal cancer treated with cetuximab. J Clin Oncol 26:374–379. 90. de Bono, J.S., Tolcher, A.W., Forero, A., Vanhove, G.F., Takimoto, C., Bauer, R.J., Hammond, L.A., Patnaik, A., White, M.L., Shen, S., et al. 2004. ING-1, a monoclonal antibody targeting Ep-CAM in patients with advanced adenocarcinomas. Clin Cancer Res 10:7555–7565.

19  Optimising the Development of Antibodies as Treatment for Cancer

565

91. Smith-Jones, P.M., Solit, D.B., Akhurst, T., Afroze, F., Rosen, N., and Larson, S.M. 2004. Imaging the pharmacodynamics of HER2 degradation in response to Hsp90 inhibitors. Nat Biotechnol 22:701–706. 92. Viani, G.A., Afonso, S.L., Stefano, E.J., De Fendi, L.I., and Soares, F.V. 2007. Adjuvant trastuzumab in the treatment of her-2-positive early breast cancer: a meta-analysis of published randomized trials. BMC Cancer 7:153. 93. Slamon, D.J., Leyland-Jones, B., Shak, S., Fuchs, H., Paton, V., Bajamonde, A., Fleming, T., Eiermann, W., Wolter, J., Pegram, M., et al. 2001. Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2. N Engl J Med 344:783–792. 94. Piccart-Gebhart, M.J., Procter, M., Leyland-Jones, B., Goldhirsch, A., Untch, M., Smith, I., Gianni, L., Baselga, J., Bell, R., Jackisch, C., et al. 2005. Trastuzumab after adjuvant chemotherapy in HER2-positive breast cancer. N Engl J Med 353:1659–1672. 95. Lu, D., Zhang, H., Koo, H., Tonra, J., Balderes, P., Prewett, M., Corcoran, E., Mangalampalli, V., Bassi, R., Anselma, D., et al. 2005. A fully human recombinant IgG-like bispecific antibody to both the epidermal growth factor receptor and the insulin-like growth factor receptor for enhanced antitumor activity. J Biol Chem 280:19665–19672. 96. Henson, E.S., Hu, X., and Gibson, S.B. 2006. Herceptin sensitizes ErbB2-overexpressing cells to apoptosis by reducing antiapoptotic Mcl-1 expression. Clin Cancer Res 12:845–853. 97. Payne, G. 2003. Progress in immunoconjugate cancer therapeutics. Cancer Cell 3:207–212. 98. Schrama, D., Reisfeld, R.A., and Becker, J.C. 2006. Antibody targeted drugs as cancer therapeutics. Nat Rev Drug Discov 5:147–159. 99. Milenic, D.E., Brady, E.D., and Brechbiel, M.W. 2004. Antibody-targeted radiation cancer therapy. Nat Rev Drug Discov 3:488–499. 100. Frankel, A.E., Kreitman, R.J., and Sausville, E.A. 2000. Targeted toxins. Clin Cancer Res 6:326–334. 101. Ricart, A.D. and Tolcher, A.W. 2007. Technology insight: cytotoxic drug immunoconjugates for cancer therapy. Nat Clin Pract Oncol 4:245–255. 102. Chakrabarti, M.C., Le, N., Paik, C.H., De Graff, W.G., and Carrasquillo, J.A. 1996. Prevention of radiolysis of monoclonal antibody during labeling. J Nucl Med 37:1384–1388. 103. Wu, A.M. and Senter, P.D. 2005. Arming antibodies: prospects and challenges for immunoconjugates. Nat Biotechnol 23:1137–1146. 104. Chari, R.V. 2008. Targeted cancer therapy: conferring specificity to cytotoxic drugs. Acc Chem Res 41:98–107. 105. Miller, M.L., Roller, E.E., Wu, X., Leece, B.A., Goldmacher, V.S., Chari, R.V., and Ojima, I. 2004. Synthesis of potent taxoids for tumor-specific delivery using monoclonal antibodies. Bioorg Med Chem Lett 14:4079–4082. 106. Sievers, E.L., Appelbaum, F.R., Spielberger, R.T., Forman, S.J., Flowers, D., Smith, F.O., Shannon-Dorcy, K., Berger, M.S., and Bernstein, I.D. 1999. Selective ablation of acute myeloid leukemia using antibody-targeted chemotherapy: a phase I study of an anti-CD33 calicheamicin immunoconjugate. Blood 93:3678–3684. 107. Larson, R.A., Sievers, E.L., Stadtmauer, E.A., Lowenberg, B., Estey, E.H., Dombret, H., Theobald, M., Voliotis, D., Bennett, J.M., Richie, M., et al. 2005. Final report of the efficacy and safety of gemtuzumab ozogamicin (Mylotarg) in patients with CD33-positive acute myeloid leukemia in first recurrence. Cancer 104:1442–1452. 108. Hamblett, K.J., Senter, P.D., Chace, D.F., Sun, M.M., Lenox, J., Cerveny, C.G., Kissler, K.M., Bernhardt, S.X., Kopcha, A.K., Zabinski, R.F., et al. 2004. Effects of drug loading on the antitumor activity of a monoclonal antibody drug conjugate. Clin Cancer Res 10:7063–7070. 109. McCann, J., Fossella, F.V., Villalona-Calero, A., Tolcher, A.W., Fidias, P., Raju, R., Zildjian, S., Guild, R., and Fram, R. 2007. Phase II trial of huN901-DM1 in patients with relapsed small cell lung cancer and CD56+ small cell carcinoma. J Clin Oncol 25:18084.

566

C.P. Carden et al.

110. Tassone, P., Gozzini, A., Goldmacher, V., Shammas, M.A., Whiteman, K.R., Carrasco, D.R., Li, C., Allam, C.K., Venuta, S., Anderson, K.C., et al. 2004. In vitro and in vivo activity of the maytansinoid immunoconjugate huN901-N2¢-deacetyl-N2¢-(3-mercapto-1-oxopropyl)maytansine against CD56+ multiple myeloma cells. Cancer Res 64:4629–4636. 111. Holden, S.N., Beeram, M., Krop, I.E., Burris, H.A., Birkner, M., Girish, S., Tibbitts, J., Lutzker, S.G., and Modi, S. 2008. A phase I study of weekly dosing of trastuzumab-DM1 in patients with advanced HER2+ breast cancer. J Clin Oncol 26:1029. 112. Beeram, M., Burris, H.A., Modi, S., Birkner, M., Girish, S., Tibbitts, J., Holden, S.N., Lutzker, S.G., and Krop, I.E. 2008. A phase I study of trastuzumab-DM1, a first-in-class HER2 antibody-drug conjugate, in patients with advanced HER2+ breast cancer. J Clin Oncol 26:1028. 113. Ross, H.J., Hart, L.L., Swanson, P.M., Rarick, M.U., Figlin, R.A., Jacobs, A.D., McCune, D.E., Rosenberg, A.H., Baron, A.D., Grove, L.E., et  al. 2006. A randomized, multicenter study to determine the safety and efficacy of the immunoconjugate SGN-15 plus docetaxel for the treatment of non-small cell lung carcinoma. Lung Cancer 54:69–77. 114. Knox, S.J., Goris, M.L., Trisler, K., Negrin, R., Davis, T., Liles, T.M., Grillo-Lopez, A., Chinn, P., Varns, C., Ning, S.C., et  al. 1996. Yttrium-90-labeled anti-CD20 monoclonal antibody therapy of recurrent B-cell lymphoma. Clin Cancer Res 2:457–470. 115. Goldenberg, D.M. 2002. Targeted therapy of cancer with radiolabeled antibodies. J Nucl Med 43:693–713. 116. Sharkey, R.M. 2005. The direct route may not be the best way to home. J Nucl Med 46:391–394. 117. Kreitman, R.J. 2001. Toxin-labeled monoclonal antibodies. Curr Pharm Biotechnol 2:313–325. 118. Kreitman, R.J., Wilson, W.H., Stetler-Stevenson, M., Noel, P., FitzGerald, D.J., and Pastan, I. 2007. Phase II trial of CAT-3888 (BL22) in chemo-resistant hairy cell leukemia. J Clin Oncol 25:7095. 119. Posey, J.A., Khazaeli, M.B., Bookman, M.A., Nowrouzi, A., Grizzle, W.E., Thornton, J., Carey, D.E., Lorenz, J.M., Sing, A.P., Siegall, C.B., et al. 2002. A phase I trial of the singlechain immunotoxin SGN-10 (BR96 sFv-PE40) in patients with advanced solid tumors. Clin Cancer Res 8:3092–3099. 120. Turturro, F. 2007. Denileukin diftitox: a biotherapeutic paradigm shift in the treatment of lymphoid-derived disorders. Expert Rev Anticancer Ther 7:11–17. 121. Holliger, P. and Hudson, P.J. 2005. Engineered antibody fragments and the rise of single domains. Nat Biotechnol 23:1126–1136. 122. Chowdhury, P.S. and Wu, H. 2005. Tailor-made antibody therapeutics. Methods 36:11–24. 123. Belau, A., Pfisterer, J., Wimberger, P., Kurzeder, C., Du Bois, A., Sehouli, J., Loibl, S., Burchardi, N., Vergote, I., and Wagner, U. 2007. Randomized, multicenter, two dose level, open-label, phase IIa study with the intraperitoneally infused trifunctional bispecific antibody catumaxomab (anti-EpCAM x anti-CD3) to select the better dose level in platinum refractory epithelial ovarian cancer patients. J Clin Oncol 25:5556. 124. Boldicke, T. 2007. Blocking translocation of cell surface molecules from the ER to the cell surface by intracellular antibodies targeted to the ER. J Cell Mol Med 11:54–70. 125. Lo, A.S., Zhu, Q., and Marasco, W.A. 2008. Intracellular antibodies (intrabodies) and their therapeutic potential. Handb Exp Pharmacol 181:343–373. 126. Williams, B.R. and Zhu, Z. 2006. Intrabody-based approaches to cancer therapy: status and prospects. Curr Med Chem 13:1473–1480. 127. Taylor, R.P. and Lindorfer, M.A. 2008. Immunotherapeutic mechanisms of anti-CD20 monoclonal antibodies. Curr Opin Immunol 20:444–449. 128. Cartron, G., Dacheux, L., Salles, G., Solal-Celigny, P., Bardos, P., Colombat, P., and Watier, H. 2002. Therapeutic activity of humanized anti-CD20 monoclonal antibody and polymorphism in IgG Fc receptor FcgammaRIIIa gene. Blood 99:754–758. 129. Weng, W.K. and Levy, R. 2003. Two immunoglobulin G fragment C receptor polymorphisms independently predict response to rituximab in patients with follicular lymphoma. J Clin Oncol 21:3940–3947.

19  Optimising the Development of Antibodies as Treatment for Cancer

567

130. Hatjiharissi, E., Xu, L., Santos, D.D., Hunter, Z.R., Ciccarelli, B.T., Verselis, S., Modica, M., Cao, Y., Manning, R.J., Leleu, X., et  al. 2007. Increased natural killer cell expression of CD16, augmented binding and ADCC activity to rituximab among individuals expressing the Fc{gamma}RIIIa-158 V/V and V/F polymorphism. Blood 110:2561–2564. 131. Zhang, W., Gordon, M., Schultheis, A.M., Yang, D.Y., Nagashima, F., Azuma, M., Chang, H.M., Borucka, E., Lurje, G., Sherrod, A.E., et al. 2007. FCGR2A and FCGR3A polymorphisms associated with clinical outcome of epidermal growth factor receptor expressing metastatic colorectal cancer patients treated with single-agent cetuximab. J Clin Oncol 25:3712–3718. 132. Bibeau, F., Lopez-Crapez, E., Di Fiore, F., Thezanas, S., Ychou, M., Blanchard, F., Lamy, A., Penault-Llorca, F., Freborg, T., Michel, P., et al. 2009. Impact of Fc(gamma)RIIa-Fc(gamma) RIIIa polymorphosms and KRAS mutations on the clinical outcome of patients with metastatic colorectal cancer treated with cetuximab plus irinotecan. J Clin Oncol 27:8. 133. Siberil, S., Dutertre, C.A., Fridman, W.H., and Teillaud, J.L. 2007. FcgammaR: the key to optimize therapeutic antibodies? Crit Rev Oncol Hematol 62:26–33. 134. Bowles, J.A., Wang, S.Y., Link, B.K., Allan, B., Beuerlein, G., Campbell, M.A., Marquis, D., Ondek, B., Wooldridge, J.E., Smith, B.J., et al. 2006. Anti-CD20 monoclonal antibody with enhanced affinity for CD16 activates NK cells at lower concentrations and more effectively than rituximab. Blood 108:2648–2654. 135. Masuda, K., Kubota, T., Kaneko, E., Iida, S., Wakitani, M., Kobayashi-Natsume, Y., Kubota, A., Shitara, K., and Nakamura, K. 2007. Enhanced binding affinity for FcgammaRIIIa of fucose-negative antibody is sufficient to induce maximal antibody-dependent cellular cytotoxicity. Mol Immunol 44:3122–3131. 136. de Romeuf, C., Dutertre, C.A., Le Garff-Tavernier, M., Fournier, N., Gaucher, C., Glacet, A., Jorieux, S., Bihoreau, N., Behrens, C.K., Beliard, R., et al. 2008. Chronic lymphocytic leukaemia cells are efficiently killed by an anti-CD20 monoclonal antibody selected for improved engagement of FcgammaRIIIA/CD16. Br J Haematol 140:635–643. 137. Radaev, S. and Sun, P.D. 2001. Recognition of IgG by Fcgamma receptor. The role of Fc glycosylation and the binding of peptide inhibitors. J Biol Chem 276:16478–16483. 138. Mori, K., Kuni-Kamochi, R., Yamane-Ohnuki, N., Wakitani, M., Yamano, K., Imai, H., Kanda, Y., Niwa, R., Iida, S., Uchida, K., et al. 2004. Engineering Chinese hamster ovary cells to maximize effector function of produced antibodies using FUT8 siRNA. Biotechnol Bioeng 88:901–908. 139. Yamane-Ohnuki, N., Kinoshita, S., Inoue-Urakubo, M., Kusunoki, M., Iida, S., Nakano, R., Wakitani, M., Niwa, R., Sakurada, M., Uchida, K., et  al. 2004. Establishment of FUT8 knockout Chinese hamster ovary cells: an ideal host cell line for producing completely defucosylated antibodies with enhanced antibody-dependent cellular cytotoxicity. Biotechnol Bioeng 87:614–622. 140. Niwa, R., Sakurada, M., Kobayashi, Y., Uehara, A., Matsushima, K., Ueda, R., Nakamura, K., and Shitara, K. 2005. Enhanced natural killer cell binding and activation by low-fucose IgG1 antibody results in potent antibody-dependent cellular cytotoxicity induction at lower antigen density. Clin Cancer Res 11:2327–2336. 141. Teeling, J.L., Mackus, W.J., Wiegman, L.J., van den Brakel, J.H., Beers, S.A., French, R.R., van Meerten, T., Ebeling, S., Vink, T., Slootstra, J.W., et al. 2006. The biological activity of human CD20 monoclonal antibodies is linked to unique epitopes on CD20. J Immunol 177:362–371. 142. Coiffier, B., Lepretre, S., Pedersen, L.M., Gadeberg, O., Fredriksen, H., van Oers, M.H., Wooldridge, J., Kloczko, J., Holowiecki, J., Hellmann, A., et al. 2008. Safety and efficacy of ofatumumab, a fully human monoclonal anti-CD20 antibody, in patients with relapsed or refractory B-cell chronic lymphocytic leukemia: a phase 1–2 study. Blood 111:1094–1100. 143. Wang, S.Y., Racila, E., Taylor, R.P., and Weiner, G.J. 2008. NK-cell activation and antibodydependent cellular cytotoxicity induced by rituximab-coated target cells is inhibited by the C3b component of complement. Blood 111:1456–1463.

Chapter 20

Oligonucleotide Therapeutics Cy A. Stein, Britta Hoehn, and John Rossi

20.1 Introduction The idea of sequence-specific gene silencing by synthetic oligonucleotides targeting mRNA is at least 40  years old, but it was only in the mid-1980s when technical advances made the chemical synthesis of oligonucleotides possible that practical steps could be taken toward its implementation. The result was a deluge of experimental data in a variety of systems [1], most of which employed the phosphorothioate (PS) backbone modification, and much of which was ultimately, and unfortunately, uninterpretable. The reason for uninterpretability is somewhat complicated. A PS oligonucleotide contains a sulfur atom that has been substituted for a nonbridging oxygen atom at each phosphorus in the oligonucleotide chain. These molecules were produced [2] because phosphodiester oligonucleotides (containing linkages identical to what is found in normal DNA) could not be used to silence gene expression either in tissue cultures or in  vivo because they were very sensitive to nuclease digestion, especially to 3¢-exonucleases [3] and were also rapidly cleared from the plasma through the kidneys. In contrast, phosphorothioates are degraded relatively slowly by nucleases [4] and are also cleared by the kidneys relatively slowly because of their low-affinity binding to plasma proteins (predominantly albumin) [5–7]. Further, because sulfur is immediately beneath oxygen in the periodic table, the PS linkage retains the same negative charge as the PO linkage, thus bestowing the property of extreme aqueous solubility. Importantly, the biophysical behavior of PS and PO oligonucleotides in solution are governed by their backbone charge and not by their sequence. In addition, the PS linkage retains the property of being a substrate for the RNase H, a ubiquitous, predominately nuclear enzyme that cleaves the mRNA strand of an mRNA–DNA duplex [8] and apparently functions naturally to eliminate Okazaki fragments. This ostensibly permits gene silencing to occur via

C.A. Stein (*) Department of Oncology, Albert Einstein-Montefiore Cancer Center, Montefiore Medical Center, 111 E. 210 St, Bronx, NY 10467, USA e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_20, © Springer Science+Business Media, LLC 2011

569

570

C.A. Stein et al.

a “pseudocatalytic mechanism” requiring only submicromolar concentrations of PS oligonucleotides for efficacy, at least when they are transfected into cells by employing carrier lipids or other vehicles. For all practical purposes, only PO and PS oligonucleotides elicit RNase H activity. (Interestingly, whether the RNAse H mechanism of gene silencing is correct or not, there is mounting evidence that it may be responsible for only a small percentage of total gene silencing. Silencing may actually occur predominately due to the activity of Ago2, the same enzyme with slicer function that cleaves the mRNA strand of an siRNA–mRNA duplex in the RISC complex) [9]. However, despite these important properties of PS oligonucleotides, which have led to their being featured in several important clinical cancer trials, the down side to this chemical modification is that PS oligonucleotides are biochemically fundamentally different from PO oligonucleotides. These differences are most significant in their ability to hybridize to a complementary mRNA strand, which is greatly diminished with respect to an isosequential PO oligonucleotide, and with respect to off-target effects, which are greatly enhanced. Diminished hybridization, reflected in a lower melting temperature (Tm) of the PS-oligomRNA duplex [4], directly correlates with diminished antisense activity and may be part of the reason why some clinical trials with PS oligonucleotides have not been successful, as will be discussed, although other clinically important aspects of study design and disease characteristics may also explain this observation. Fortunately, the more recent development of sugar-modified locked nucleic acids (LNAs) as antisense oligonucleotides appears to have solved the diminished hybridization problem of the PS oligonucleotide. Each LNA incorporated into a PS oligonucleotide (four are commonly employed, two at each molecular terminus) may raise the Tm of the LNA–mRNA duplex by 4–8°C [10, 11], and hence potentiate gene silencing [12–15]. Clinical trials with an anti-Bcl-2 LNA oligonucleotide in chronic lymphocytic leukemia (CLL) are currently underway, but data are not yet available. The off-target effects of PS oligonucleotides fall into two basic categories: their ability to bind to heparin-binding proteins and their immunostimulatory properties, which can induce splenomegaly, B-lymphocyte proliferation, and cytokine production. Rodents appear to be particularly sensitive to this phosphorothioate class effect [16]. Neither category of off-target effect will be dealt with extensively in this review, but if these effects are not rigorously controlled for, particularly in experiments in immunosuppressed mice, it may be difficult, if not impossible, to differentiate sequence-specific effects from off-target effects. This has led to a great difficulty in interpreting experimental data [17], as previously noted. The problem is further compounded by the fact that the extent of immune-stimulation by PS oligonucleotides (with or without CpG motifs) in a mouse is very different from that in a human. Unfortunately, this means that preclinical animal models employing PS oligonucleotides may have had little or no predictive value for response in clinical trial patients. Nevertheless, despite these problems, some extremely interesting data have emerged from recent large, randomized, multicenter phase III trials with antisense oligonucleotides, particularly in melanoma and CLL. These trials,

20  Oligonucleotide Therapeutics

571

in addition to a few provocative phase II–III trials in other cancers, form the basis for the majority of this review. Phase I trials and later-phase trials performed in populations other than cancer patients will not be discussed.

20.2 Clinical Trials of Oblimersen Oblimersen (G3139 or Genasense®) is an 18-mer PS oligonucleotide that is complementary to codons 1–6 of the Bcl-2 mRNA [18] and was first synthesized approximately 15 years ago [19]. Relative to all other PS oligonucleotides of similar length studied, it is a molecule with unusual and entirely unexplained properties – its measured Tm, when in a duplex with its complementary mRNA, is substantially higher than predicted. The molecule has been studied extensively in phase I and phase II trials. A compilation of these studies has recently been published [20]. Indeed, in the opinion of the author, sufficient data to support the antisense activity of oblimersen as the sole mechanism of action in any clinical trial is presently lacking, although the drug has demonstrated evidence of clinical activity, as described below. Webb et al. [21] began a phase I trial with oblimersen in nine patients with lymphoma who had relapsed after at least two chemotherapy interventions and showed overexpression of Bcl-2 protein based on lymph node biopsies. Oblimersen was administered as a continuous subcutaneous infusion for 2  weeks at doses ranging from 4.6 to 73.6 mg/m2/day. The maximum-tolerated dose (MTD) was not reached. One patient had a complete response (CR) and three patients had stable disease. Enrollment continued and a total of 21 patients were treated with oblimersen at doses up to 195.8 mg/m2/day [22]. The MTD was 147.2 mg/m2/day. A CR was attained in one patient, a minor response in two patients, and stable disease in nine patients. In 7 of 16 evaluable patients, Bcl-2 protein was decreased in cells from lymph nodes (two patients) and in samples of peripheral blood or bone marrow (five patients).

20.2.1 Phase III Trial of Oblimersen in Chronic Lymphocytic Leukemia There is substantial evidence to indicate that in CLL cells, Bcl-2 silencing may lead to significant cellular apoptosis. Therefore, a randomized phase III trial of fludarabine (F) plus cyclophosphamide (C) with or without oblimersen was initiated in patients with relapsed or refractory CLL [23]. A total of 241 patients were stratified and randomized according to three criteria: responsive vs. refractory to prior fludarabine therapy, number of prior regimens (1–2 vs. ³3), and duration of response to last therapy (>6 months vs. £6 months). All patients were required to have received ­treatment with at least one prior chemotherapy regimen that included at least two cycles of fludarabine. Using standard definitions, patients were

572

C.A. Stein et al.

c­ onsidered relapsed after prior treatment with fludarabine if they achieved at least a PR lasting more than 6 months; patients who did not achieve at least a PR lasting more than 6  months after their last fludarabine treatment were considered to be refractory. In the FC group, patients received fludarabine 25 mg/m2/day intravenously (IV) followed by cyclophosphamide 250 mg/m2/day IV on days 1–3. In the oblimersen/ FC group, patients were given oblimersen 3 mg/kg/day by continuous IV infusion on days 1–7, with FC administered at the above doses on days 5–7. Cycles were 28 days in duration, and up to six cycles were administered. The primary end point of the study was the between-treatment difference in the proportion of patients who achieved CR + nodular partial response (nPR, which is the clinical equivalent of CR; heretofore CR + nPR will be referred to as CR). Demographic characteristics between the two groups were well balanced. The median number of prior treatment regimens in both groups was three, and previous therapy was balanced between-treatment groups [24]. Twenty patients (17%) in the oblimersen/FC group achieved a CR, as opposed to 8 (7%; P = 0.025) in the FC group. Moreover, these CRs were significantly more durable in the oblimersen/FC group. At 36  months of follow-up, the duration of complete remission in FC-treated patients was 22 months, whereas the median had not been reached but was estimated to exceed 36 months for patients in the oblimersen/FC group (P = 0.031). With 54 months of follow-up, 12/20 (60%) of oblimersen/FC patients with CR remained alive, including five who remained in complete remission [25]. Of the eight FC-treated patients with CR, only three were alive at 5  years, and all three had relapsed. Maximum benefit was seen in fludarabinesensitive patients, e.g., those who had a PR or better for more than 6 months after prior fludarabine treatment. In this population, there was a fourfold increase in the CR rate in the oblimersen/FC group as compared to the FC control (25% vs. 6%; P = 0.016). With 5 years of follow-up, among all patients who achieved a response, including both complete and partial responses (N = 103), there was an 18-month median survival benefit for patients in the oblimersen/FC group (HR = 0.60; P = 0.038) [24]. In nonresponding patients, there was no difference in survival outcome between groups. With respect to nonhematologic toxicities, nausea, pyrexia, and fatigue (primarily grade 1–2) were the most commonly occurring and affected more patients in the oblimersen/FC group than in the FC group [23]. Grade 3–4 occurrences of nausea, pyrexia, and fatigue were limited (8% vs. 2% of patients; 3% of patients in both groups; and 6% vs. 4% of patients, respectively). Importantly, in a population in which infection and immunosuppression are the most common cause of death, the incidence of grade 4 neutropenia (7% vs. 11%) was not increased with the addition of oblimersen to the FC regimen. Grade 4 thrombocytopenia was more frequent in the oblimersen/FC group, but was not associated with an increased incidence of grade 3–4 bleeding events (4% vs. 2%). In approximately 3% of patients, oblimersen administration was associated with first-cycle reactions (cytokine release with or without tumor lysis).

20  Oligonucleotide Therapeutics

573

These data demonstrate the importance of achieving a CR for long-term survival in CLL, a point that was initially contested by the FDA. Although oblimersen has, to date, not yet been approved in this indication, the FDA is currently reconsidering that decision on the basis of the recently reported 5-year survival data.

20.2.2 Bcl-2 Silencing and Chemosensitization While there is no doubt that oblimersen can silence Bcl-2 expression in tissue cultures, its ability to do this in vivo, and most importantly, the extent to which Bcl-2 silencing actually chemosensitizes malignant cells, as it is predicted to do, are matters of great debate. Bcl-2 is far from the only antiapoptotic protein present in the vast majority of malignant cells. Even in follicular lymphoma cells bearing the t14:18 chromosomal translocation, which produces a fused Bcl-2/immunoglobulin mRNA in about 65–70% of cases [26], it is not clear whether the Bcl-2 protein is necessary and sufficient for the maintenance of the neoplastic phenotype. In some tumors, elevated expression of Bcl-2 protein in tumor cells may merely be an epiphenomenon, despite documented clinical correlations between the expression (or “overexpression”) of Bcl-2 protein and a poor prognosis in cancer patients with tumors [27–30]. Blagosklonny [31] when addressing this question, noted that Bcl-2 expression in colorectal, breast, and lung carcinomas was associated with an “increased apoptotic index, lower risk of distant metastases, and improved prognosis.” Furthermore, concordant with much experimental data, cell lines may “…become resistant due to a strong selection during establishment of cells in culture, overexpression of Bcl-2 simply cannot further increase resistance and [the] effects of Bcl-2 are undetectable.” Tumor cells may also become resistant to cytotoxic therapeutics by downregulating proteins in the apoptotic cascade. An example in melanoma is Apaf-1, which is downstream of Bcl-2, and in whose absence, the level of Bcl-2 protein expression would appear to be irrelevant [32]. Further complicating any potential value of Bcl-2 silencing is the role of this protein in melanoma. Its role in the pathogenesis and prognosis of clinical melanoma is controversial [33] because the protein can be found in normal melanocytes, benign nevi, and primary melanomas, in addition to melanoma metastases [34]. Interestingly, in some studies, Bcl-2 expression was decreased in melanoma cells vs. normal melanocytes [35–38]. However, this finding has not been confirmed in other studies [34, 39, 40], in which minimal differences in the expression of Bcl-2 have been observed. To add to the confusion, in advanced melanoma, about one-third of the data suggest an increase in Bcl-2 expression, while one-third suggest a decrease [33], although the function that the Bcl-2 protein actually serves, rather than just the amount of Bcl-2 protein present, would probably be more important. However, one study that was insufficiently powered has demonstrated that if lymph node deposits express Bcl-2, advanced melanoma patients have a poorer prognosis than those who do not [33]. In toto, conflicting data render the role of Bcl-2 in advanced melanoma unclear.

574

C.A. Stein et al.

20.2.3 Clinical Trials in Advanced Melanoma Dacarbazine (DTIC), the only approved chemotherapy drug for advanced ­melanoma, was combined with oblimersen in a phase I/II trial (N = 14) in patients with advanced disease [41]. Oblimersen was administered via continuous IV infusion for 14 days each month. The initial dose was 0.6 mg/kg/day, increasing to a dose maximum of 6.5 mg/kg/day. Dacarbazine 200 mg/m2 was given IV on days 5–9. Six patients also received the same total daily dose of oblimersen administered as twice-daily subcutaneous injections on days 1–7 and dacarbazine 800 mg/m2 IV on day 5. The maximum decrease in Bcl-2 expression in patients’ biopsy specimens was highly variable, and no conclusions could be drawn due to insufficient sampling. Responses included one CR, two PRs, and three minor responses, including two in patients whose disease stabilized for a period of at least 1 year. These data led to the initiation of the largest phase III trial (GM301) in advanced melanoma to date. Between July 2000 and February 2003, 771 chemotherapy-naïve patients with advanced melanoma were randomly assigned to receive treatment with dacarbazine alone 1,000 mg/m2/day IV for 60 min or oblimersen 7 mg/kg/day by continuous IV infusion for 5 days followed by the same dose of dacarbazine [42]. Patients were stratified according to ECOG performance status (0 vs. 1–2), presence or absence of liver metastasis, and disease site/serum LDH level. This latter category included two groups, patients with nonvisceral disease (skin, subcutaneous tissue, or lymph node disease) and normal LDH, and patients with visceral disease (excluding liver) or elevated LDH [baseline serum level more than 1.1 times the upper limit of normal (ULN)] [42]. The primary end point of the study was an intent-to-treat (ITT) comparison of overall survival between the two treatment groups. Secondary end points included progression-free survival, overall and durable response (i.e., response ³6 months in duration), and duration of response. The baseline characteristics of the groups were well balanced. With a minimum follow-up of 24 months, the median overall survival in the oblimersen/DTIC group was 9 months, compared with 7.8 months in the DTIC-alone group (HR = 0.87; 95% CI 0.75–1.01; P = 0.077) [42]. Overall response rates (CRs + PRs) were 13.5% for patients treated with oblimersen/DTIC and 7.5% for patients receiving DTIC alone (P = 0.007). Durable responses were also increased in the oblimersen/DTIC group (7.3% vs. 3.6%; P = 0.03). Eleven patients (2.8%) in the oblimersen/DTIC group achieved a CR in comparison to three patients (0.8%) in the DTIC-alone group. Median ­progression-free survival was also significantly longer among patients who received oblimersen/DTIC than among those treated with DTIC (2.6 months vs. 1.6 months, HR = 0.75; P < 0.001). Outcome data were subsequently analyzed according to the LDH stratification category. Serum LDH has long been recognized as an important independent biomarker of poor prognosis in malignant melanoma [43] and, in the GM301 study, an interaction between treatment and baseline serum LDH was observed. Patients with LDH values £1.1 × ULN who received oblimersen/DTIC (approximately two-thirds [508] of the 771 subjects) were observed to have significantly better treatment

20  Oligonucleotide Therapeutics

575

o­ utcomes for all efficacy end points. These included overall survival (median, 11.4 months vs. 9.7 months; P = 0.02), progression-free survival (median, 3.1 months vs. 1.6  months, P < 0.001), overall response (17.2% vs. 9.3%; P = 0.009), complete response (3.4% vs. 0.8%), and durable response (9.6% vs. 4.0%; P = 0.01) [42]. On the other hand, significant differences between-treatment groups were not observed for patients with elevated baseline LDH (LDH > 1.1 × ULN). Recent data demonstrate that the extent to which pretreatment LDH level is increased, even within the “normal” range, is predictive of prognosis in advanced melanoma [44]. For example, a retrospective examination of data obtained from EORTC study 18951 (N = 330) demonstrates a monotonic progression to improved prognosis in advanced melanoma patients as the value of LDH decreases, similar to what has been observed in the GM301 trial. For patients with baseline LDH £ 0.8 × ULN in study GM301 (N = 274), the median survival at 24 months in the oblimersen/DTIC group vs. the DTIC group was 12.3 and 9.9 months, respectively (HR = 0.64, P < 0.001). A confirmatory trial (AGENDA, GM307) of 300 patients, similar in design to study GM301 but with a double-blind design and limited to patients with baseline LDH £ 0.8 × ULN, is currently ongoing, with recruitment expected to be completed in early 2009. The results from this study should provide important prospective confirmatory data for the previously discussed observations in the GM301 trial. But why should overall prognosis in advanced melanoma depend on pretreatment levels of serum LDH? LDH is a ubiquitous enzyme, but its expression is frequently elevated in neoplastic cells because of their shift to glycolysis secondary to relatively poor vascularization and diminished oxygen delivery. Cells dying via the process of necrosis will release LDH, but LDH is not commonly released after apoptosis. Tumor cell survival and the rate of necrosis of tumor cells may often depend on the balance between their rate of proliferation vs. the rate of vascularization of the growing tumor. Therefore, it is possible that high LDH levels in patients may reflect disease that is still growing, but is, at least in part, poorly vascularized. These types of tumors are frequently highly resistant to chemotherapy due to poor oxygen delivery, as well as possibly poor drug delivery (hence the lack of response to treatment). Hypoxia can ultimately be an important survival factor for some tumor cells. For example, hypoxia can induce genetic instability that can select for tumor cells with increased metastatic potential [45–47] and, via c-met protooncogene activation, lead to cells that are more aggressive and invasive [48, 49]. Diminished blood flow and low pH can also compromise the functions of tumor-infiltrating immune effector cells and cytokines. Clinical studies [50] have demonstrated that the presence of hypoxic regions within tumors correlates with poor prognosis and increased metastatic risk regardless of treatment – viz., what is observed in advanced melanoma. Thus, tumor hypoxia leads to necrosis (and thus spillage of LDH) and also ultimately to more aggressive tumors and a poorer prognosis. These ideas predict that the size of the tumor is not the critical factor in either serum LDH levels or prognosis (which it was not in the GM301 trial), but rather that the balance between oxygen supply to the tumor and its intrinsic growth rate is critical.

576

C.A. Stein et al.

In the GM301 trial, neutropenia and thrombocytopenia were the most significant adverse effects, but were not associated with an increase in serious infections or bleeding. In the oblimersen/DTIC group, the incidence of grade 3–4 neutropenia with infection was 4.3% for the combination vs. 2.8% for DTIC alone. Grade 1–2 bleeding events (primarily epistaxis or hematuria) were also increased in the combination-treatment group to 13.7% from 9.2% observed for the DTIC group, but more grade 3–4 bleeding events (mostly gastrointestinal) occurred with DTIC (3.1% vs. 2.2%). These rates are substantially lower than those associated with other drugs and drug regimens used for the treatment of advanced melanoma [51–53]. An increased rate of catheter-related events (venous thrombosis, infection, occlusion) was observed in the oblimersen/DTIC group (19.1% vs. 8.6%). Lower rates of adverse events resulting in treatment discontinuation or death and serious adverse events were observed in patients without elevated baseline LDH values [42].

20.2.4 Other Trials of Oblimersen Oblimersen was added to a regimen of etoposide and carboplatin in a randomized (3:1) trial in 56 assessable patients with small-cell lung cancer [54]. In each 21-day cycle, patients in one group received oblimersen 7 mg/kg/day on days 1–8, carboplatin on day 6, and etoposide on days 6–8. Patients in the control group received the same carboplatin and etoposide regimen beginning on day 1 of each cycle. Treatment groups were balanced with respect to baseline characteristics. Response rates were nearly identical in the two treatment groups, and survival at 1 year was actually worse with oblimersen (24%, 95% CI 12–40%) than without oblimersen (47%, 95% CI = 21–73%). The incidence of grade 3–4 hematologic toxicity was also somewhat increased with the addition of oblimersen (88% vs. 60%, P = 0.05). The authors offer several possible explanations for the lack of improved efficacy with the oblimersen-containing regimen, one plausible explanation being that oblimersen does not adequately suppress Bcl-2 levels in patients with small-cell lung cancer, as demonstrated in the phase I study undertaken to determine the regimen for this phase II study. In acute myelogenous leukemia (AML), Bcl-2 expression may contribute to a lower CR rate and shorter patient survival [55, 56]. In a phase I study, Marcucci et al. enrolled 29 untreated patients with AML [55]. All patients were over 60 years of age, had either intermediate or adverse cytogenetics, and initially received induction therapy with oblimersen 7 mg/kg/day by continuous IV infusion on days 1–10 + cytarabine by continuous IV infusion on days 4–10 + daunorubicin IV at one of two doses on days 4–6. CR was achieved in 14 patients (48%), and an incomplete remission was achieved in three patients (10%). Levels of normalized Bcl-2 mRNA expression in bone marrow mononuclear cells were found to be decreased from baseline (P = 0.03) in patients with CR, but increased from baseline (P = 0.05) in nonresponding patients. Expression of Bcl-2 protein in bone marrow mononuclear cells after 72  h of oblimersen demonstrated a small (about 20%), but ­statistically significant decrease (P = 0.004) in patients

20  Oligonucleotide Therapeutics

577

with CR vs. nonresponding patients. However, given recent data that the gymnotic (i.e., naked) delivery of oligonucleotides to cells is a very slow process requiring 6 days or more to produce antisense effects, it is possible that 72 h was an insufficiently long time point for meaningful measurement of the Bcl-2 protein. All patients developed pancytopenia. Toxicities were independent of the daunorubicin dose, as well as reversible and/or “not directly attributable” to oblimersen. A phase II trial of oblimersen + gemtuzumab ozogamicin (Mylotarg; a humanized anti-CD33 monoclonal antibody conjugated to calicheamicin) was performed in patients ³60 years of age with AML at first relapse [57]. Oblimersen 7 mg/kg/ day was administered as a continuous IV infusion on days 1–7 and 15–21, with gemtuzumab given IV on days 4 and 18. A total of 48 patients were enrolled at 18 centers, but the study was eventually terminated due to slow accrual. Based on an ITT analysis, five patients (10%) achieved a CR and seven patients (15%) achieved a CR without platelet recovery (CRp), for an overall ITT response rate of 25%. (These findings are similar to those previously reported for single-agent Mylotarg in a more favorable patient population.) [58] For the CR + CRp patients, median relapse-free survival was 3.75 months (95% CI 3.3–6.3 months), and median survival was not reached at 6 months. The probability of surviving at 6 months was 0.80, 0.86, and 0.17 for the CR, CRp, and nonresponding patients, respectively. A total of 13 patients (27%) withdrew before completing therapy, the most common reason being toxicity (6 of 13 patients). Of 16 patients who died within 30 days of last dose of study medication, five did so from treatment toxicity. Nausea was the most common nonhematologic event (79% of patients) and febrile neutropenia the most common hematologic event (50% of patients). A Phase III trial (CALGB 10201) in which 503 untreated older patients with AML were randomized to induction treatment with cytosine arabinoside + daunorubicin followed by high-dose cytarabine consolidation therapy, with or without oblimersen 7  mg/kg/day (days 1–10 for induction, days 1–8 for consolidation) showed no differences in CR rates, overall survival, disease-free survival, or toxicity [59]. Further trials of oblimersen in AML are not planned. Another hematologic malignancy in which oblimersen was not successful in phase III was multiple myeloma. In a phase II trial [60], 33 patients relapsing after prior chemotherapy or transplantation received oblimersen 5–7 mg/kg/day for 7 days by continuous IV infusion. On day 4, patients received dexamethasone 40 mg orally for 4 days and thalidomide 200 mg/day increasing to 400 mg/day, if tolerated, for the study duration. Responding and stable patients received maintenance dosing for up to 2 years, and the cycles were repeated every 35 days. A total of 24 of 33 patients (73% [50% historically for the combination of dexamethasone + thalidomide] [20]) had responses, including two CRs, four near CRs, 12 PRs, and six minor responses. The median duration of response was 13  months and the median overall survival was 17.4  months. A rise in polyclonal IgM (from a median of 35.5 to 94  mg/dL) was found to be predictive of response and was suggested to be due to immunostimulation by the oligonucleotide. Of seven assessable patients, three demonstrated a decrease in Bcl-2 protein in malignant cells, but there was no correlation between Bcl-2 protein levels and expression and response in this limited number of patients. The most

578

C.A. Stein et al.

c­ ommon grade 3 toxicities were neutropenia (n = 8), thrombocytopenia (n = 5), infection (n = 5), and hypocalcemia (n = 6). Grade 4 events were limited to neutropenia in four patients and increased serum creatinine in one patient. The dosing scheme in the phase II trial was not pursued. Instead a randomized, multinational phase III trial of dexamethasone 40 mg/day orally for 4 days during weeks 1–3 (Cycle 1) or during week 1 (all other cycles) ± oblimersen 7 mg/kg/day by continuous intravenous infusion beginning 3 days before dexamethasone treatment in weeks 1 and 3 (Cycle 1) and in week 1 (all other cycles) was conducted in a total of 224 patients with relapsed or refractory disease [61]. The primary end point was a comparison of time to disease progression between the two groups. At baseline, an imbalance was observed between the treatment groups in several important prognostic factors [62]. ECOG Performance Status at baseline was significantly worse in the oblimersen/dexamethasone group (P = 0.03). In addition, more patients in the oblimersen/ dexamethasone group were categorized as having Durie–Salmon stage III, IIIa, or IIIb disease than in the dexamethasone group (70% vs. 61%, respectively). Imbalances between the two groups in baseline laboratory parameters also suggested that patients in the oblimersen/dexamethasone group were more seriously impaired than those in the dexamethasone group (ANC <1,000/mm3: 5 and 2%, respectively; platelet count <50,000/mm3: 7 and 3%, respectively; creatinine >2.0 mg/dL: 5 and 0%, respectively; and lactate dehydrogenase >ULN: 27 and 14%, respectively). There was no statistically significant difference between the groups in time to tumor progression [62]. The oblimersen/dexamethasone regimen was generally well tolerated, with fatigue, fever, and nausea as the most commonly observed adverse events. Failure to show an advantage over standard treatment (dexamethasone) may be attributable to significant differences between-treatment groups at baseline that favored the dexamethasone group and/or the fact that many patients in this heavily pretreated population were refractory to dexamethasone.

20.3 OGX-011 This oligonucleotide is targeted to the mRNA of clusterin, an antiapoptotic protein that apparently promotes chemo- and radioresistance through inhibition of the function of the pro-apoptotic bax protein [63]. The compound has a phosphorothioate backbone and is further modified by the presence of 2¢-methoxyethyl (MOE) substituents on the four 3¢ and 5¢ terminal ribose sugar moieties. The MOE modification appears to dramatically increase the tissue half-life of this oligomer, in part, by increasing its stability vs. nucleases. There is also some evidence that MOE “gap-mers” may have fewer off-target effects, in addition to diminished immunostimulatory properties. In a phase I study in combination with docetaxel [63], serum clusterin levels in the 640 mg dosing group declined approximately 35% after Cycle 1. However, declines in clusterin expression in peripheral blood mononuclear cells

20  Oligonucleotide Therapeutics

579

could not be assessed because of the wide variability in pretreatment ­expression. This trial was followed by a randomized phase II trial of OGX-11 plus docetaxel vs. docetaxel plus prednisone in chemotherapy-naive patients with metastatic hormonerefractory prostate cancer [64]. Eighty-two patients at 12 centers were randomized to each arm. The docetaxel dose was 75 mg/m2, and the OGX-11 dose was 640 mg. In the initial 56 patients, the toxicity due to OGX-011 included grade 1–2 fever and rigors in 37 and 67% of the patients, respectively. Based on a recent press report by Oncogenex, the median survival for patients in the OGX-011 arm was 27.5 months, but only 16.9 months in the control arm. This is certainly an encouraging signal with respect to proceeding to a large, phase III randomized trial in this indication.

20.4 AP 12009 AP 12009 is an antisense PS oligonucleotide targeted to the TGF-b2 mRNA. The justification for targeting TGF-b2 as an important anticancer target has been previously made by Hau et al. [65]. In brief, TGF-b2 is widely overexpressed in human tumors and is negatively correlated with prognosis. The protein blocks the proliferation and cytotoxic activity of T- and NK cells and is a potent immunosuppressant, while at the same time acting, in gliomas, as a growth and angiogenic factor. However, similar to the oblimersen story, it is unclear to what extent the in vivo mechanism of action of AP 12009 is related to these observations. In early phase I/II trials, the drug was delivered by convection-enhanced delivery directly into the tumors of patients with grade 3 (anaplastic astrocytoma) or grade 4 (glioblastoma multiforme) disease via an implanted catheter either for four or seven days continuously. In another trial, multiple cycles of drug were administered. Twentyfour patients were enrolled, receiving a total of 48 cycles. Many of the patients had been pretreated with temozolamide. Seven showed stable disease after 28 days. One patient had a CR after one cycle of AP 12009 without further therapy; a second patient (who received a total of 12 cycles) also had a CR and was still in remission after 4.5 years. No treatment-related deaths, grade 4 events, or catheter-related infections were observed. Two adverse events were grade 3, and the MTD was not reached after more than a 100-fold dose escalation. Plasma levels of AP 12009 after intracerebral infusion were below the limit of detection, and no laboratory abnormalities were observed. One serious event (brain edema) was considered possibly drug related. All told, the drug appeared to be extremely well tolerated [66]. A phase IIb international, open-label trial in 134 patients with high-grade (3 or 4) glioma was designed to compare (1:1:1) low (10 mM) and high (80 mM) doses of AP 12009 vs. standard chemotherapy (temozolamide or procarbazine + lomustine + vincristine) [67]. The test drug was administered weekly via convection-enhanced delivery for 6  months. Six serious adverse events possibly related to the study drug, and 37 procedure-related serious adverse events (92% grade 1 or 2) were reported. “Several long-term tumor responses were observed by local MRI reading;” response rates by central reading have not yet been presented, to our knowledge.

580

C.A. Stein et al.

20.5 Affinitak This molecule is a 20-mer PS oligonucleotide targeted to the 3¢ untranslated region of the PKC-a mRNA, whose translation product was believed to be a very important signal transduction protein. The compound was evaluated in several phase I and phase II trials [68–72]. A total of 55 patients with non-small-cell lung cancer received 80 mg/m2 cisplatin and either gemcitibine 1,000 or 1,250 mg/m2 + Affinitak 2  mg/kg/day for 14  days via continuous IV infusion, repeated every 3  weeks. Sixteen of 48 (33%) evaluable patients achieved a response (1 CR, 15 PRs). The median overall duration of response was 7 months (95% CI 4.2–7.8 months), and the median duration of stable disease was 4 months (95% CI 3–5.5 months). Based on these data, a large, multicenter, randomized phase III trial in non-small-cell lung cancer was performed. The details of this trial have apparently not been published, but it is understood that Affinitak did not add anything to the gemcitibine + cisplatin combination, and it is no longer being clinically pursued.

20.6 Conclusions The results of phase III studies of oblimersen in melanoma, CLL, and multiple myeloma suggest that oblimersen may not be as active in patients who have advanced disease and have received multiple prior chemotherapy regimens. Despite the evidence of clinical benefit at this point, our understanding of the mechanism of action of oblimersen, to date the only clinically active anticancer antisense oligonucleotide, is far from complete. While this is of little consequence to the advanced cancer patient, it is far from an optimal situation for those who view antisense as a platform technology. Does this mean that oblimersen is a one-off, a clinical oddity not to be repeated? Will increasing the Tm of the oligonucleotide– mRNA duplex by inclusion of LNA lead to improved clinical efficacy? Will recent advances in our understanding of the uptake of oligonucleotides by cancer cells suggest improved dosing schedules? Are siRNAs too “clean” to be active anticancer agents, and how can they be distributed efficiently to targeted cells? There are a large number of questions that need to be answered, but we believe that additional significant clinical advances can only be achieved rationally by a more complete understanding of the fundamental properties of these highly pleiotropic, biologically active compounds.

20.7 RNAi and siRNAs The field of oligonucleotide-based therapy experienced a revival with the discovery of RNA interference (RNAi) in 1998 [73]. RNAi is a conserved endogenous mechanism, which is triggered by double-stranded (ds) RNAs leading to target-specific

20  Oligonucleotide Therapeutics

581

inhibition of gene expression by promoting mRNA degradation or translational repression. There are two RNAi pathways that are guided either by small-interfering RNAs (siRNAs), which are perfectly complementary to the mRNA or by microRNAs (miRNAs), which bind imperfectly to their target mRNA [74]. A breakthrough in the field of siRNA therapeutic agents was achieved by Elbashir et  al. [75], who demonstrated that synthetic, exogenously applied dsRNAs of 21 nucleotides in length can induce silencing in mammalian cells. In addition to the siRNA design of 21-mer duplex with 3¢-overhangs at both sides, Dicer-substrate formats such as 27-mers or short hairpin (sh) RNAs have been developed that elicit a more potent gene-silencing effect at lower concentrations as compared to conventional 21-mer siRNAs [76–78]. It is remarkable how quickly after its discovery RNAi has been established as the method of choice for targeted inhibition of gene expression in mammalian systems. Because RNAi uses a natural pathway for gene silencing, it generally results in a greater potency of knockdown than antisense oligonucleotides or ribozymes. Preclinical results have confirmed the effectiveness of RNAi and have generated serious optimism about the potential for siRNA drugs. As with the other oligonucleotide-based approaches, the applications of siRNAs as a therapeutic agent face most of the above mentioned challenges. Some of these challenges, however, have already been addressed in the course of antisense oligonucleotide and ribozyme development. Many of the standard stabilizing oligonucleotide modifications that have been already explored for antisense strategies were employed in siRNA designs. SiRNA properties can be beneficially improved by the introduction of certain chemical modifications at distinct positions in the sequence, including thermal stability of the duplex, resistance against degradation, specificity for the target mRNA, reduction of off-target effects, biodistribution, and cellular uptake [79]. In a systematic study, Jackson and coworkers reported that many individual nucleotides in the antisense strand may be modified with 2¢-O-Me groups without loss of the silencing potential. A similar study has been performed with 2¢-fluoro (2¢-F) and 2¢-O-MOE [80]. An additional advantage of using 2¢-O-Me nucleotides is a reduction in off-target effects [81], as well as avoidance of the interferon responses [82]. The strategic placement of these modifications is crucial. Modifications at the 5¢-end of the guide strand can inhibit the silencing effect [83], while modifications at the 5¢-end of the passenger strand can improve stability as well as guide strand selection and targeting specificity [84, 85]. Incorporation of 3¢-S-phophorothiolate [86], boranophosphates [87], 4¢-thioriboses [88], and LNAs [89, 90], were also reported to enhance target-binding affinity and increase silencing potency. Preclinical studies have demonstrated the safe use and the potential for ­therapeutic benefit of RNAi-mediated gene silencing [91, 92]. SiRNAs are in earlystage clinical trials for the treatment of viral infections, cancer, and ocular diseases. Phase I studies are planned for numerous other diseases, including neurodegenerative diseases, asthma/allergies, and inflammatory diseases [93]. The most advancedstage testing for a siRNA-based drug is for the treatment of viral infection and was developed by Alnylam Pharmaceuticals (Cambridge, MA, USA). The siRNA

582

C.A. Stein et al.

ALN-RSV01 was designed against the respiratory syncytial virus (RSV), which causes severe respiratory illness, primarily in infants [94]. The unmodified siRNAs, administered by inhalation, showed significant viral reduction in experimentally infected adult volunteers compared to the placebo group in a phase II GEMINI study and is now being tested in patients with naturally acquired RSV infection. Other examples of antiviral applications have been proposed for severe acute respiratory syndrome (SARS) [95], herpes simplex virus 2 [96], and HIV-1 [97, 98]. Serious concerns about the rapid development of drug-resistant HIV variants make the use of multiple-drug combinations inevitable. Recently, a pilot study of safety and feasibility of stem cell therapy for lymphoma patients with AIDS was initiated using a lentivirus vector encoding three anti-HIV RNAs [99]. The combinatorial approach involves a shRNA targeting tat/rev, an RNA TAR decoy, and an antichemokine receptor 5 (CCR5) ribozyme. The lead product of Intradigm (Palo Alto, CA, USA) targets angiogenesis (http://www.intradigm.com) by an RNAi nanoplex particle ICS-283 comprised of a nanoparticle and two siRNAs, one against vascular endothelial growth factor (VEGF) and the other against the VEGF’s main receptor (VEGFR2). The product is in preclinical development for a variety of cancer indications, and the company expects to initiate clinical evaluation in 2009. Two ongoing clinical trials also aim at angiogenesis in age-related macular deficiency (AMD). Bevasiranib (previously known as Cand5) was developed against VEGF and AGN 211745 (previously known as Sirna-027) against its receptor (VEGFR1). Early clinical studies showed that the therapeutic reagents were well tolerated and could prevent neovascularization in the eye after intravitreal injection. AGN 211745 is being investigated in a phase II study in combination with ranibizumab, and patients are currently being enrolled in a phase III study to evaluate the safety and effectiveness of bevasiranib. Controversially, a report was recently published suggesting that the suppression of neovascularization in two animal models is a generic property of siRNAs through TLR3 activation, independent of the sequence [100]. This example clearly demonstrates that preclinical studies need to be carefully conducted to prove safety and a specific siRNA-mediated silencing effect. Encouraged by earlier achievements of oligonucleotide-based therapeutics, some RNAi strategies may have been rushed into clinical trials. It is crucial to understand the basic mechanism of RNAi and its diverse related effectors to avoid toxic side effects and to develop rationally designed biopharmaceuticals.

References 1. Stein CA, Cheng YC: Antisense oligonucleotides as therapeutic agents: is the bullet really magical? Science 261:1004–1012, 1993 2. Stec WJ, Zon G, Egan W, et  al: Automated solid-phase synthesis, separation and stereochemistry of phosphorothioate analogs of oligodeoxyribonucleotides. J Am Chem Soc 106:6077–6079, 1984

20  Oligonucleotide Therapeutics

583

3. Eder PS, DeVine RJ, Dagle JM, et  al: Substrate specificity and kinetics of degradation of ­antisense oligonucleotides by a 3¢ exonuclease in plasma. Antisense Res Dev 1:141–151, 1991 4. Stein CA, Subasinghe C, Shinozuka K, et al: Physicochemical properties of phosphorothioate oligodeoxynucleotides. Nucleic Acids Res 16:3209–3221, 1988 5. Watanabe TA, Geary RS, Levin AA: Plasma protein binding of an antisense oligonucleotide targeting human ICAM-1 (ISIS 2302). Oligonucleotides 16:169–180, 2006 6. Geary RS, Watanabe TA, Truong L, et al: Pharmacokinetic properties of 2¢-O-(2-methoxyethyl)modified oligonucleotide analogs in rats. J Pharmacol Exp Ther 296:890–897, 2001 7. Geary RS, Yu RZ, Watanabe T, et  al: Pharmacokinetics of a tumor necrosis factor-alpha phosphorothioate 2¢-O-(2-methoxyethyl) modified antisense oligonucleotide: comparison across species. Drug Metab Dispos 31:1419–1428, 2003 8. Walder RY, Walder JA: Role of RNase H in hybrid-arrested translation by antisense oligonucleotides. Proc Natl Acad Sci USA 85:5011–5015, 1988 9. Stein CA, Hansen B, Lai J, et al: Efficient gene silencing by delivery of locked nucleic acid antisence oligonucleotides, unassisted by transfection reagents. Nucl. Acids Res. 2009, doi: 10,1093/nar/gkp841 10. Koshkin AA, Singh SK, Nielsen P, et al: LNA (locked nucleic acids): synthesis of the adenine, cytosine, guanine, 5-methylcytosine, thymine and uracil bicyclonucleoside monomers, oligomerisation and unprecedented nucleic acid recognition. Tetrahedron 54:3607–3630, 1998 11. Singh SK, Nielsen P, Koshkin AA, et al: LNA (locked nucleic acids): synthesis and highaffinity nucleic acid recognition. Chem Commun (Camb) 4:455–456, 1998 12. Grünweller A, Wyszko E, Bieber B, et  al: Comparison of different antisense strategies in mammalian cells using locked nucleic acids, 2¢-O-methyl RNA, phosphorothioates and small interfering RNA. Nucleic Acids Res 31:3185–3193, 2003 13. Fluiter K, Frieden M, Vreijling J, et al: On the in vitro and in vivo properties of four locked nucleic acid nucleotides incorporated into an anti-H-Ras antisense oligonucleotide. Chembiochem 6:1104–1109, 2005 14. Elayadi AN, Braasch DA, Corey DR: Implications of high-affinity hybridization by locked nucleic acid oligomers for inhibition of human telomerase. Biochemistry 41:9973–9981, 2002 15. Braasch DA, Liu Y, Corey DR: Antisense inhibition of gene expression in cells by oligonucleotides incorporating locked nucleic acids: effect of mRNA target sequence and chimera design. Nucleic Acids Res 30:5160–5167, 2002 16. Monteith DK, Henry SP, Howard RB, et al: Immune stimulation – a class effect of phosphorothioate oligodeoxynucleotides in rodents. Anticancer Drug Des 12:421–432, 1997 17. Gekeler V, Gimmnich P, Hofmann HP, et  al: G3139 and other CpG-containing immunostimulatory phosphorothioate oligodeoxynucleotides are potent suppressors of the growth of human tumor xenografts in nude mice. Oligonucleotides 16:83–93, 2006 18. Klasa RJ, Gillum AM, Klem RE, et al: Oblimersen Bcl-2 antisense: facilitating apoptosis in anticancer treatment. Antisense Nucleic Acid Drug Dev 12:193–213, 2002 19. Kitada S, Takayama S, De Riel K, et al: Reversal of chemoresistance of lymphoma cells by antisense-mediated reduction of bcl-2 gene expression. Antisense Res Dev 4:71–79, 1994 20. Gjertsen BT, Bredholt T, Anensen N, et al: Bcl-2 antisense in the treatment of human malignancies: a delusion in targeted therapy. Curr Pharm Biotechnol 8:373–381, 2007 21. Webb A, Cunningham D, Cotter F, et  al: BCL-2 antisense therapy in patients with nonHodgkin lymphoma. Lancet 349:1137–1141, 1997 22. Waters JS, Webb A, Cunningham D, et  al: Phase I clinical and pharmacokinetic study of bcl-2 antisense oligonucleotide therapy in patients with non-Hodgkin’s lymphoma. J Clin Oncol 18:1812–1823, 2000 23. O’Brien S, Moore JO, Boyd TE, et al: Randomized phase III trial of fludarabine plus cyclophosphamide with or without oblimersen sodium (Bcl-2 antisense) in patients with relapsed or refractory chronic lymphocytic leukemia. J Clin Oncol 25:1114–1120, 2007

584

C.A. Stein et al.

24. O’Brien S, Moore JO, Boyd TE, et al: 5-year survival in patients with relapsed or refractory CLL in randomized Phase III trial of fludarabine plus cyclophosphamide with or without oblimersen: the Oblimersen CLL Study Group. J. Clin. Oncol. 27:5208–5212, 2009 25. Rai KR, Moore J, Wu J, et  al: Effect of the addition of oblimersen (Bcl-2 antisense) to fludarabine/cyclophosphamide for relapsed/refractory chronic lymphocytic leukemia (CLL) on survival in patients who achieve CR/nPR: five-year follow-up from a randomized phase III study. J Clin Oncol 26:374s, 2008 (suppl; abstr 7008) 26. Weiss LM, Warnke RA, Sklar J, et al: Molecular analysis of the t(14;18) chromosomal translocation in malignant lymphomas. N Engl J Med 317:1185–1189, 1987 27. Reed JC, Kitada S, Takayama S, et al: Regulation of chemoresistance by the bcl-2 oncoprotein in non-Hodgkin’s lymphoma and lymphocytic leukemia cell lines. Ann Oncol 5:61–65, 1994 28. Schmitt CA, Rosenthal CT, Lowe SW: Genetic analysis of chemoresistance in primary murine lymphomas. Nat Med 6:1029–1035, 2000 29. Gazitt Y, Hu WX: Fas (APO-1/CD95)-mediated apoptosis is independent of bcl-2: a study with cell lines overexpressing bcl-2 and with bcl-2 transfected cell lines. Int J Oncol 12:211– 220, 1998 30. Gleave ME, Miayake H, Goldie J, et al: Targeting bcl-2 gene to delay androgen-independent progression and enhance chemosensitivity in prostate cancer using antisense bcl-2 oligodeoxynucleotides. Urology 54:36–46, 1999 31. Blagosklonny MV: Paradox of Bcl-2 (and p53): why may apoptosis-regulating proteins be irrelevant to cell death? Bioessays 23:947–953, 2001 32. Soengas MS, Capodieci P, Polsky D, et al: Inactivation of the apoptosis effector Apaf-1 in malignant melanoma. Nature 409:207–211, 2001 33. Bush JA, Li G: The role of Bcl-2 family members in the progression of cutaneous melanoma. Clin Exp Metastasis 20:531–539, 2003 34. Leiter U, Schmid RM, Kaskel P, et al: Antiapoptotic bcl-2 and bcl-xL in advanced malignant melanoma. Arch Dermatol Res 292:225–232, 2000 35. Tang L, Tron VA, Reed JC, et al: Expression of apoptosis regulators in cutaneous malignant melanoma. Clin Cancer Res 4:1865–1871, 1998 36. Ramsay JA, From L, Kahn HJ: bcl-2 protein expression in melanocytic neoplasms of the skin. Mod Pathol 8:150–154, 1995 37. Saenz-Santamaria MC, Reed JA, et al: Immunohistochemical expression of BCL-2 in melanomas and intradermal nevi. J Cutan Pathol 21:393–397, 1994 38. Tron VA, Krajewski S, Klein-Parker H, et al: Immunohistochemical analysis of Bcl-2 protein regulation in cutaneous melanoma. Am J Pathol 146:643–650, 1995 39. Plettenberg A, Ballaun C, Pammer J, et al: Human melanocytes and melanoma cells constitutively express the Bcl-2 proto-oncogene in situ and in cell culture. Am J Pathol 146:651–659, 1995 40. Cerroni L, Soyer HP, Kerl H: bcl-2 protein expression in cutaneous malignant melanoma and benign melanocytic nevi. Am J Dermatopathol 17:7–11, 1995 41. Jansen B, Wacheck V, Heere-Ress E, et al: Chemosensitisation of malignant melanoma by BCL2 antisense therapy. Lancet 356:1728–1733, 2000 42. Bedikian AY, Millward M, Pehamberger H, et al: Bcl-2 antisense (oblimersen sodium) plus dacarbazine in patients with advanced melanoma: the Oblimersen Melanoma Study Group. J Clin Oncol 24:4738–4745, 2006 43. Manola J, Atkins M, Ibrahim J, et al: Prognostic factors in metastatic melanoma: a pooled analysis of Eastern Cooperative Oncology Group trials. J Clin Oncol 18:3782–3793, 2000 44. Agarwala S, Gilles E, Wu J, et al: LDH correlation with survival in advanced melanoma from two large, randomized trials: Oblimersen (GM 301) and EORTC 18951. Eur. J. Cancer 45:1807–1814, 2009 45. Cairns RA, Kalliomaki T, Hill RP: Acute (cyclic) hypoxia enhances spontaneous metastasis of KHT murine tumors. Cancer Res 61:8903–8908, 2001 46. Postovit LM, Adams MA, Lash GE, et al: Oxygen-mediated regulation of tumor cell invasiveness. Involvement of a nitric oxide signaling pathway. J Biol Chem 277:35730–35737, 2002

20  Oligonucleotide Therapeutics

585

47. Rofstad EK, Rasmussen H, Galappathi K, et al: Hypoxia promotes lymph node metastasis in human melanoma xenografts by up-regulating the urokinase-type plasminogen activator receptor. Cancer Res 62:1847–1853, 2002 48. Bottaro DP, Liotta LA: Out of air is not out of action. Nature 423:593–595, 2003 49. Pennacchietti S, Michieli P, Galluzzo M, et al: Hypoxia promotes invasive growth by transcriptional activation of the met protooncogene. Cancer Cell 3:347–361, 2003 50. Höckel M, Vaupel P: Tumor hypoxia: definitions and current clinical, biologic, and molecular aspects. J Natl Cancer Inst 93:266–276, 2001 51. Avril MF, Aamdal S, Grob JJ, et al: Fotemustine compared with dacarbazine in patients with disseminated malignant melanoma: a phase III study. J Clin Oncol 22:1118–1125, 2004 52. Chapman PB, Einhorn LH, Meyers ML, et al: Phase III multicenter randomized trial of the Dartmouth regimen versus dacarbazine in patients with metastatic melanoma. J Clin Oncol 17:2745–2751, 1999 53. Eton O, Legha SS, Bedikian AY, et al: Sequential biochemotherapy versus chemotherapy for metastatic melanoma: results from a phase III randomized trial. J Clin Oncol 20:2045– 2052, 2002 54. Rudin CM, Salgia R, Wang X, et al: Randomized phase II study of carboplatin and etoposide with or without the bcl-2 antisense oligonucleotide oblimersen for extensive-stage small-cell lung cancer: CALGB 30103. J Clin Oncol 26:870–876, 2008 55. Marcucci G, Stock W, Dai G, et al: Phase I study of oblimersen sodium, an antisense to Bcl2, in untreated older patients with acute myeloid leukemia: pharmacokinetics, pharmacodynamics, and clinical activity. J Clin Oncol 23:3404–3411, 2005 56. Banker DE, Radich J, Becker A, et al: The t(8;21) translocation is not consistently associated with high Bcl-2 expression in de novo acute myeloid leukemias of adults. Clin Cancer Res 4:3051–3062, 1998 57. Moore J, Seiter K, Kolitz J, et al: A phase II study of Bcl-2 antisense (oblimersen sodium) combined with gemtuzumab ozogamicin in older patients with acute myeloid leukemia in first relapse. Leuk Res 30:777–783, 2006 58. Larson RA, Boogaerts M, Estey E, et al: Antibody-targeted chemotherapy of older patients with acute myeloid leukemia in first relapse using Mylotarg (gemtuzumab ozogamicin). Leukemia 16:1627–1636, 2002 59. Marcucci G, Moser B, Blum W, et al: A phase III randomized trial of intensive induction and consolidation chemotherapy ± antisense oligonucleotide in untreated acute myeloid leukemia patients >60 years old. J Clin Oncol 25:360s, 2007 (suppl; abstr 7012) 60. Badros AZ, Goloubeva O, Rapoport AP, et al: Phase II study of G3139, a Bcl-2 antisense oligonucleotide, in combination with dexamethasone and thalidomide in relapsed multiple myeloma patients. J Clin Oncol 23:4089–4099, 2005 61. Chanan-Chan AA, Niesvizky R, Hohl RJ, et al: Randomized multicenter phase 3 trial of high-dose dexamethasone (dex) with or without oblimersen sodium (G3139; Bcl-2 antisense; Genasense) for patients with advanced multiple myeloma (MM). Blood 104:413a, 2004 (abstr 1477) 62. Data on file. Genta Incorporated. Berkeley Heights, NJ 63. Chi K, Siu L, Hirte H, et al: A phase I study of OGX-011, a 2¢-methoxyethyl phosphorothioate antisense to clusterin, in combination with docetaxel in patients with advanced cancer. Clin Cancer Res 14:833–839, 2007 64. Chi K, Hotte S, Yu E, et al: A randomized phase II study of OGX-011 in combination with docetaxel and prednisone or docetaxel and prednisone alone in patients with metastatic hormone refractory prostate cancer (HRPC). J Clin Oncol 25:252s, 2007 (suppl; abstr 5069) 65. Hau P, Jachimczak P, Schlingensiepen R, et al: Inhibition of TGF-b2 with AP 12009 in recurrent malignant gliomas: from preclinical to phase I/II studies. Oligonucleotides 17:201–212, 2007 66. Schlingensiepen KH, Fischer-Blass B, Schmaus S, et al: Antisense therapeutics for tumor treatment: the TGF-beta2 inhibitor AP 12009 in clinical development against malignant tumors. Recent Results Cancer Res 177:137–150, 2008

586

C.A. Stein et al.

67. Bogdahn U, Oliushine VE, Parfenov VE, et al: Results of G004, a phase IIb study in recurrent glioblastoma patients with the TGF-b2 targeted compound AP 12009. J Clin Oncol 24:71s, 2006 (suppl; abstr 1553) 68. Nemunaitis J, Holmlund JT, Kraynak M, et al: Phase I evaluation of ISIS 3521, an antisense oligodeoxynucleotide to protein kinase C-alpha, in patients with advanced cancer. J Clin Oncol 17:3586–3595, 1999 69. Yuen AR, Halsey J, Fisher GA, et  al: Phase I study of an antisense oligonucleotide to ­protein kinase C-alpha (ISIS 3521/CGP 64128A) in patients with cancer. Clin Cancer Res 5:3357–3363, 1999 70. Cripps MC, Figueredo AT, Oza AM, et al: Phase II randomized study of ISIS 3521 and ISIS 5132 in patients with locally advanced or metastatic colorectal cancer: a National Cancer Institute of Canada clinical trials group study. Clin Cancer Res 8:2188–2192, 2002 71. Tolcher AW, Reyno L, Venner PM, et al: A randomized phase II and pharmacokinetic study of the antisense oligonucleotides ISIS 3521 and ISIS 5132 in patients with hormone-refractory prostate cancer. Clin Cancer Res 8:2530–2535, 2002 72. Villalona-Calero MA, Ritch P, Figueroa JA, et al: A phase I/II study of LY900003, an antisense inhibitor of protein kinase C-a, in combination with cisplatin and gemcitabine in patients with advanced non-small cell lung cancer. Clin Cancer Res 10:6086–6093, 2004 73. Fire A, Xu S, Montgomery MK, et al: Potent and specific genetic interference by doublestranded RNA in Caenorhabditis elegans. Nature 391, 806–811, 1998 74. Rana TM: Illuminating the silence: understanding the structure and function of small RNAs. Nat Rev Mol Cell Biol 8:23–36, 2007 75. Elbashir SM, Harborth J, Lendeckel W, et al: Duplexes of 21-nucleotide RNAs mediate RNA interference in cultured mammalian cells. Nature 411:494–498, 2001 76. Rose SD, Kim DH, Amarzguioui M, et al: Functional polarity is introduced by Dicer processing of short substrate RNAs. Nucleic Acids Res 33:4140–4156, 2005 77. Kim DH, Behlke MA, Rose SD, et  al: Synthetic dsRNA Dicer substrates enhance RNAi potency and efficacy. Nat Biotechnol 23:222–226, 2005 78. Siolas D, Lerner C, Burchard J, et  al: Synthetic shRNAs as potent RNAi triggers. Nat Biotechnol 23:227–231, 2005 79. Corey DR: Chemical modification: the key to clinical application of RNA interference? J Clin Invest 117:3615–3622, 2007 80. Prakash TP, Allerson CR, Dande P, et al: Positional effect of chemical modifications on short interference RNA activity in mammalian cells. J Med Chem 48:4247–4253, 2005 81. Fedorov Y, Anderson EM, Birmingham A, et  al: Off-target effects by siRNA can induce toxic phenotype. RNA 12:1188–1196, 2006 82. Morrissey DV, Lockridge JA, Shaw L, et al: Potent and persistent in vivo anti-HBV activity of chemically modified siRNAs. Nat Biotechnol 23:1002–1007, 2005 83. Chiu YL, Rana TM: siRNA function in RNAi: a chemical modification analysis. RNA 9:1034–1048, 2003 84. Chen PY, Weinmann L, Gaidatzis D, et  al: Strand-specific 5¢-O-methylation of siRNA duplexes controls guide strand selection and targeting specificity. RNA 14:263–274, 2008 85. Kubo T, Zhelev Z, Ohba H, et al: Modified 27-nt dsRNAs with dramatically enhanced stability in serum and long-term RNAi activity. Oligonucleotides 17:445–464, 2007 86. Gaynor JW, Brazier J, Cosstick R: Synthesis of 3¢-S-phosphorothiolate oligonucleotides for their potential use in RNA interference. Nucleosides Nucleotides Nucleic Acids 26:709–712, 2007 87. Hall AH, Wan J, Shaughnessy EE, et al: RNA interference using boranophosphate siRNAs: structure-activity relationships. Nucleic Acids Res 32:5991–6000, 2004 88. Hoshika S, Minakawa N, Matsuda A: RNA interference induced by siRNAs modified with 4¢-thioribonucleosides. Nucleic Acids Symp Ser (Oxf) 49:77–78, 2005 89. Mook OR, Baas F, de Wissel MB, et al: Evaluation of locked nucleic acid-modified small interfering RNA in vitro and in vivo. Mol Cancer Ther 6:833–843, 2007

20  Oligonucleotide Therapeutics

587

90. Elmén J, Thonberg H, Ljungberg K, et al: Locked nucleic acid (LNA) mediated improvements in siRNA stability and functionality. Nucleic Acids Res 33:439–447, 2005 91. de Fougerolles A, Vornlocher HP, Maraganore J, et al: Interfering with disease: a progress report on siRNA-based therapeutics. Nat Rev Drug Discov 6:443–453, 2007 92. Behlke MA: Progress towards in vivo use of siRNAs. Mol Ther 13:644–670, 2006 93. Kim DH, Rossi JJ: Strategies for silencing human disease using RNA interference. Nat Rev Genet 8:173–184, 2007 94. Bitko V, Musiyenko A, Shulyayeva O, et  al: Inhibition of respiratory viruses by nasally administered siRNA. Nat Med 11:50–55, 2005 95. Li BJ, Tang Q, Cheng D, et  al: Using siRNA in prophylactic and therapeutic regimens against SARS coronavirus in Rhesus macaque. Nat Med 11:944–951, 2005 96. Palliser D, Chowdhury D, Wang QY, et al: An siRNA-based microbicide protects mice from lethal herpes simplex virus 2 infection. Nature 439:89–94, 2006 97. Jacque JM, Triques K, Stevenson M: Modulation of HIV-1 replication by RNA interference. Nature 418:435–438, 2002 98. Coburn GA, Cullen BR: Potent and specific inhibition of human immunodeficiency virus type 1 replication by RNA interference. J Virol 76:9225–9231, 2002 99. Rossi JJ, June CH, Kohn DB: Genetic therapies against HIV. Nat Biotechnol 25:1444–1454, 2007 100. Kleinman ME, Yamada K, Takeda A, et al: Sequence- and target-independent angiogenesis suppression by siRNA via TLR3. Nature 452:591–597, 2008

Chapter 21

Anticancer Drug Development in Pediatric Patients Lia Gore and Margaret Macy

21.1 Introduction The incidence of pediatric cancer is relatively rare, particularly when compared to rates in the adult oncology. Despite the relative low numbers, death due to cancer remains the primary cause of death due to disease in the pediatric population. Pediatric cancer cure rates have markedly improved in the past several decades with well over 70% of pediatric cancer patients achieving cure. However, there remains a subset of patients who have a dismal prognosis either due to tumor type or stage at presentation or relapse. As a result, there is a great need for novel therapeutics and innovative approaches in this population. The development of new therapies is being extensively explored in the adult oncology population; however, the advancement of such drugs in pediatrics has been much slower.

21.2 Historical Perspective Traditionally, research in children has lagged significantly behind adults due in large part to the ethical and regulatory guidelines protecting children as vulnerable research subjects. The Belmont Report issued by the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research, in 1979, outlined ethical principles in human subject research, including children. These recommendations led to specific regulations regarding children as research subjects and the protection of their welfare and rights beginning in the early 1980s. In 1991, the Code of Federal Regulations was published which included the federal policy

M. Macy (*) The Children’s Hospital, Center for Cancer and Blood Disorders, 13123 East 16th Avenue, Aurora, CO 80045, e-mail: [email protected] M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_21, © Springer Science+Business Media, LLC 2011

589

590

L. Gore and M. Macy

for the protection of Human Subject Research and further delineated these guidelines. Current federal regulations state that pediatric research can only be approved if the trial offers direct benefit to the patient or does not offer direct benefit but the risk to the child is minimal or only a minor increase over minimal risk [1]. Minimal risk is defined as the degree and risk of harm or discomfort “ordinarily encountered in daily life or during the performance of routine physical or psychological examinations or tests” [1]. Pediatric research that does not provide direct benefit to the participating children and presents more than a minor increase over minimal risk is prohibited by federal regulations. Since published, the Code of Federal Regulations has formed the basis of the regulatory structure on which pediatric research is conducted. Due to these regulations, the participation of children in research trials has a foundation on which to build. However, drug testing in children continues to remain a challenge as society is concerned about exposing children to agents that may cause harm. Paradoxically, though, without testing in a pediatric population, children may also be harmed by medications that have not been tested in pediatrics. It is common for children to be prescribed medication that does not have adequate pediatric dosing data available. Of note, off label use has been found to be highest in pediatric/neonatal intensive care units and in oncology wards [2]. Given the complex ethical issues of conducting research in children, typically the safety and tolerability of a new anticancer agent is first established in adults. This approach protects children from potential harm before the novel drug is fairly well assessed; however, it also delays the development in pediatrics and the access young patients have to potentially promising therapies. With the large numbers of cancer therapeutics in adult trials, these agents are moving into the pediatric arena; however, often the agent is in late phase II and III trials, or even has FDA approval, before it even reaches pediatric phase I trials. Since 1997, federal regulations have been instituted to help develop pediatric labeling. The Food and Drug Administration Modernization Act, reauthorized as the Best Pharmaceutics Act provides an additional 6 months or marketing exclusivity if drug manufacturers conduct studies of that drug in pediatrics [3]. The Pediatric Research Equity Act mandates that if the sponsors of a new drug are seeking approval for indications which exist in pediatrics, the new drug must be studied in pediatrics, unless waiver or deferral by the FDA. These federal mandates, renewed and further enhanced in 2007, have significantly helped pediatric drug development with increased approval of drugs for children, improved pediatric research infrastructure and enhanced knowledge about the pharmacokinetics and safety of drugs in pediatrics. The European Medicines Agency (EMEA) now requires that a Pediatric Investigational Plan (PIP) be developed early during any new drug application, which will significantly enhance the inclusion of children in new agent evaluations. Sponsors are increasingly interested in developing drugs for the pediatric population and newer governmental regulations will facilitate this; however, there still remains a lag from drug development in adults to that in pediatrics.

21  Anticancer Drug Development in Pediatric Patients

591

21.3 The Difference Between Children and Adults It is imperative that new agents are tested in pediatric populations in addition to the testing in adults which is already under study. Pediatricians often state that a “child is not a small adult” and it is well known that children may have widely different pharmacokinetic and pharmacodynamic effects from drugs. Furthermore, an infant is not just a small child. Pediatrics includes patients who vary greatly in body size, making population modeling difficult in these patients. Extrapolating data from an adult trial may not be relevant due to these myriad differences as well as differences that may or may not exist between different classes of drugs. The mechanism of drug delivery is important to consider. For example, intramuscular injections are unpredictable in young infants as skeletal muscle blood flow is reduced in neonates [4]. Absorption of oral drugs may vary widely from infancy to children and young adults. Infants have higher gastric pHs until about age 2 when their parietal cells fully mature [5]. Gastrointestinal motility also changes during development. Typically, the time to achieve maximal plasma levels in infants is prolonged compared to older children, in part due to a slower rate of absorption [6]. Differences in intestinal enzymes responsible for metabolism or transport have also been described. For example, there is an age-dependent decrease in glutathione-S-transferase throughout childhood, which likely in part results in the age-dependent decrease in oral clearance of busulfan [7]. Disease states can also affect intestinal absorption with changes in biliary function or alterations in microbial colonization. Clearance of drug varies widely, as metabolizing enzymes and renal function are discrepant across ages as well as disease states. In addition to the obvious changes in size and growth velocity that occurs throughout childhood, children are known to have different volumes of distribution as they move from infancy through adolescence. The overall composition of their body changes with total body water initially 80–90% of body weight in infancy to 55–60% by late adolescence [8]. Infants also tend to have lower concentrations of binding proteins, leading to higher free fraction of drug in these patients [9]. This plasma protein-binding effect can result in many drugs having an altered apparent volume of distribution in neonates and infants than in older children [5]. As mentioned above, metabolism of drugs differs with age. There is an age-dependent increase in plasma clearance of drugs metabolized in the liver [6]. Additionally, the expression and activity of hepatic enzymes change significantly over the first year of life, with each enzyme having different maturational timelines [10]. Maturational changes of the cytochrome P450 system does impact the pharmacokinetics of many drugs, particularly in the infant population where many of the enzyme are not fully expressed. Ginsberg and colleagues in a review of 45 drugs, primarily metabolized through the CYP pathway, demonstrated that while premature and term infants have prolonged half lives compared to adults, they reach adult levels by 2–6  months of age and can actually have shorter half lives for some substrates [11]. Other enzymes have been less well studied, although it has been shown that the glucoronosyltransferase (UGT) has decreased activity in infants when compared to adolescents and adults [4].

592

L. Gore and M. Macy

Renal function changes greatly in the first year of life with maturation of GFR to adult levels occurring by about 12  months of age [10]. Additionally, tubular secretion and reabsorption also do not reach adult values until about 1 year of life; an important consideration if plasma clearance of a drug is dependent on renal elimination [5]. Another consideration in developing antineoplastic agents in pediatrics is the role the drug will have on actively growing cells. Traditional cytotoxic therapies are generally nonspecific in their capacity to kill rapidly growing cells, whether they are tumor cell or other normal healthy cells and tissues in the body, such as those in skin, hair, GI tract, and bone marrow. The side effect profiles of these drugs typically are due to these nonspecific killing effects. Pediatric patients tend to tolerate these side effects as well if not better than their adult counterparts. Often, they endure anemia at much lower hemoglobin levels than adults would, without significant complaints. As a result, pediatric oncologists have increased dosing and/or dose intensity and compressed the timing of these agents with improvement in survival without significant increase in side effects. This may not be the case with the new targeted agents, where dose intensity and dose maximization may be less biologically relevant. Agents which target growth pathways or angiogenesis may have more effects in children or effects that are not seen in adults. How these agents effect actively dividing cells in the growth plates, brain, muscles may not be known until they are actually tested in children with significant longitudinal observations. While pediatric patients differ from adults physiologically, psychologically, and socially, their malignancies are usually very different from those seen in adults also. “Pediatric” tumors are often mesenchymal in origin compared to the epithelially derived carcinomas that predominate in adults. These differences in origin often result in different pathways involved in the pathogenesis. Typically “pediatric” tumors like Wilm’s tumors, hepatoblastoma, or neuroblastoma, have portions if not all of the tumor which recapitulate primitive or embryonic tissues. Immature markers are seen on many pediatric tumors, including leukemias, suggesting a role for dysregulation or failure of maturation. Chemotherapy agents that work in adults such as the taxanes have not shown particular promise in pediatrics to date [12] suggesting that even with agents which demonstrate antitumor activity in adults, do not necessarily translate similarly to the pediatric realm. The biology of the tumor pathogenesis can also be quite different in pediatric patients. For example, acute lymphoblastic leukemia (ALL) in pediatrics has an overall cure rate of approximately 80% [13] while adult rates are closer to 40–50% at best [14], and often significantly worse. This discrepancy is due in part to different chemotherapy regimens, as adults have difficulty tolerating the intensive chemotherapy used in most pediatric protocols, but are likely also due to different biology, including higher incidences of poorer prognosis diseases. For example, Philadelphia chromosome-positive ALL accounts for 25% of all cases in adults, compared to less than 5% of children with ALL [15–18]. Similarly, good prognosis translocation like TEL-AML1 are the opposite, being much more common in children than adults [19, 20].

21  Anticancer Drug Development in Pediatric Patients

593

21.4 Drug Development in Pediatrics There are several “pediatric-specific” issues that are required for a pediatric trial to move forward. Due to the ethical constraints involved in pediatric research, it is important to realize that the consent process differs. Children under the age of 18 are unable to grant legal independent consent for participation in a study. Instead the consent must be obtained from the parent(s) or legal guardian(s). However, for children who can comprehend the concept of participating in a study, typically around age 7 or older, must give a written assent to participate. Without assent, the patient cannot be enrolled on the study even if the legally responsible adults desire. The number of children with cancer, newly diagnosed or relapsed, is substantially smaller than adults. Therefore, the incentive to develop a drug for a pediatric indication is less, as there are fewer patients who could benefit. Additionally, due to the relatively small number of children with cancer and the high proportion of those who are cured, there are even fewer who are eligible for early phase trials. As a result, if an industry sponsor wishes to move forward with pediatric testing there still remains a limit to the number of early phase agents which can be evaluated. To address the issue of limited numbers, there has been great effort in pediatric oncology to develop multi-institutional trials through collaborative consortia. These consortia allow for multiple institutions to enroll patients while centralizing data collection and monitoring, as well as regulatory compliance and reporting. In the USA, Children’s Oncology Group (COG) phase I consortium, the Pediatric Brain Tumor Consortium (PBTC), New Approaches to Neuroblastoma Therapy (NANT) consortium, Pediatric Oncology Experimental Therapeutic Investigators Consortium (POETIC), Saint Jude Research Consortium and Therapeutic Advances in Leukemia Consortium (TACL) are responsible for the majority of the pediatric phase I trials currently open. The ability to centralize the data regulatory management decreases cost and improves consistency across centers. Traditionally, objectives endpoint of phase I trials include the determination of the maximally tolerated dose (MTD) and dose-limiting toxicity (DLT) of the agent or combination studies. In adult phase I trials, typically “first in human” studies, the starting dose is usually very low, often one-tenth the lethal dose in animals requiring multiple dose escalations to reach the MTD. Limited pediatric patient numbers do not permit such extensive dose escalation levels in a realistic time frame. As the agents have already been evaluated in adults, a typical starting dose for pediatric trials is approximately 80% of the adult MTD [21]. These and other guidelines were developed based on principals of early phase clinical trials in adults to establish an American and European consensus about the practice of pediatric phase I trials [21]. This approach not only decreases the number of patients required but also decreases the number of patients who might be enrolled at a potentially ineffective dose, although again, as biologically relevant dosing is used increasingly over maximum tolerated dosing, this may be less important. In a recent study, Lee and colleagues sought to evaluate the actual efficiency of starting at a dose close to the adult MTD and demonstrated that the pediatric MTD is strongly correlated to the adult MTD,

594

L. Gore and M. Macy

usually within two dose levels of the adult MTD. There is rarely an occasion where a pediatric MTD is actually lower than the recommended adult dose. Defining the pharmacokinetic profile of a new agent is imperative in pediatric phase I trials. As, described above, age influences in the metabolism, distribution and absorption of drugs the pharmacokinetics should be established over the range of pediatric ages in which the drug will be used. Often due to the limited number of pediatric patients, only a single phase I trial is done. Resources and patients are limited for phase I trials; therefore, often only a single trial is done with a given new agent in pediatrics, unlike in adults where several phase I trials are often performed. Given the differences in pediatric population pharmacokinetics, it is critical that the pediatric phase I trial collects as much data about the agent in children as possible. Without solid pharmacokinetic data from a pediatric phase I trial, it can be extremely difficult to design a safe and efficient phase II trial. Typically, the practice of first defining the toxicity then identifying the appropriate dose based on side effect profile and disease response has been effective in the study of new agents, particularly cytotoxic agents. Historically, pediatric phase I trials have followed a standard 3 + 3 dose escalation design. This design has been used extensively and seeks to limit the number of patients exposed to a possibly toxic drug while concomitantly establishing safety data. In pediatrics, this method has proven safe with a very low toxic death rate (4%) [22]. However, given the limited numbers of patients and the amount of time required to evaluate patients in a certain cohort, this practice has come into question more recently. Skolnik and colleagues propose a “rolling six” design to shorten the duration of pediatric studies, eliminating some of the delays associated with suspended enrollment while patients are being evaluated for toxicity. In their trial model, 2–6 patients (rather than three in the traditional 3 + 3) are enrolled on a dose concurrently. Decisions about dose level assignments are based on evaluable patients, the number of DLTs and the number of patients at risk for developing DLTs. This design still determines MTD as the dose below which two or more DLTs are observed in up to six patients [23]. In the era of targeted therapeutics, it can be argued that idea of obtaining a MTD is less appropriate in evaluating a new agent’s contribution to antitumor effects. The concentration at which an agent demonstrates biologic effects against the relevant target may be markedly different (typically lower) from the concentrations at which toxicity is seen. Many of the small molecule tyrosine kinase inhibitors exert their target-specific effects at low concentrations, but have “off-target” effects when used at higher concentrations. Often it is these off-target effects that result in the toxicity profile. There may be no benefit to increasing the dose to reach MTD; instead we should be extrapolating the idea of biologically efficacious dose, the dose at which the targeted protein or pathway is downregulated or inhibited, to pediatrics as well. It is important to note that drugs used in pediatric phase I trials are often given with some therapeutic intent. This in part is due to the drugs already having been studied in adults, suggesting possible patient populations for which that drug might show effect. Also, given the federal regulations, patients participating on these trial should have the potential of experiencing some benefit beyond that of

21  Anticancer Drug Development in Pediatric Patients

595

“the greater population” if they are to assume greater than minimal risk. Although they are not designed to analyze efficacy, with the range of tumors, doses, and small sample size, pediatric patients on phase I trials demonstrate an objective response rate of 4–9.6% [22, 24, 25]. This is higher than the 3–7% objective response rate observed in classic single agent adult trials [26–28]. Of course as these trials are not designed to report efficacy, benefit from the drug from either symptom alleviation or disease stabilization is not necessarily reported. Given the limited numbers of patients as well as financial cost of early phase trials and drug development, the use of preclinical data in trial design is very important. The response and efficacy data from adult trials should help drive selection of agents to move into pediatric trials. However, there must be a prioritization of agents based on novel and relevant mechanisms of action and preclinical, particularly xenograft data. Preclinical lab data based on xenograft modeling has been successfully used in pediatric phase I trial design [29]. The National Cancer Institute has established the Pediatric Preclinical Testing Program (PPTP) to systematically evaluate new agents with activity in pediatric tumors, in an attempt to delineate which therapies should be moved quickly into phase I trials [30]. Preclinical testing may also provide important data regarding chemotherapy to combine with the novel agent. Combination studies in tissue cultures and animal models can help predict potential pitfalls or suggest successful combinations that might show efficacy when single agent data, in a heavily pretreated population may not. Pediatric tumors often have an associated molecular/cytogenetic fusion protein suggesting a high percentage of nonrandom chromosomal rearrangements. These fusion products have helped define certain mutations in receptors or proteins that are important in oncogenesis or tumor suppression. These fusion proteins mean that for a large number of pediatric tumors, there is a specific molecular target that may be affected pharmacologically. For example, the Bcr-Abl fusion in CML and Ph+ ALL was one of the first proteins successfully targeted by the tyrosine kinase inhibitor imatinib. Similarly, the tremendous promise of the insulin-like growth factor 1 receptor (IGF1R) targeted agents (antibodies and small molecules) that has gained momentum over the past several years has been noted in typically pediatric tumors like Ewing sarcoma, which has a defined pathogenic abnormality, the EWS-FLI1 fusion, a direct downstream target of IGF1R. In fact many of the exciting early responses to these agents have been in “pediatric” patients, and this has guided not only the practical application of these therapies to specific patients but has provided a proof of concept for targeted therapy as well.

21.5 The Role of Combination Studies A challenge in the development of biologically targeted molecules as anticancer therapy is that it can be difficult to determine the effect of the agent in patients with refractory disease. The majority of patients in whom these drugs are first tested

596

L. Gore and M. Macy

have relapsed at least once and have often been heavily pretreated with chemotherapy [31]. Current curative treatment with chemotherapy always involves a combination of agents. New agents would most likely be used in combination with already effective drugs rather than alone. Response rates in phase I trials are higher when the investigational drug is studied in combination with a known active anticancer agent [22]. Subsequently, it can be argued that phase I studies should not only focus on the agent alone but also the MTD and toxicities of the agent when combined with chemotherapy. In this setting, appropriate starting doses for a novel agent are typically lower than the previously determined MTD as a single agent. The design of these clinical trials can be difficult and require either two separate trials; one for the single agent MTD and DLTs and the second for the MTD/DLT in combination. Another and more time efficient option is to have a “window period” or brief single agent portion to establish the single agent MTD before evaluating the agent in combination. One difficulty with this approach is determining the toxicity of a new agent added to conventional chemotherapy, and assessment methods to attempt to isolate these affects are in development in new cooperative group protocols in particular. Combining agents for synergistic effect has been common practice in oncology for many years. It is important to determine the effects of new drugs in combination with drugs already in use to treat pediatric tumors preclinically to decrease the numbers of studies and patients needed in early phase trials. For example, clofarabine (Clolar®), a second-generation purine nucleoside analog designed to decrease side effects while improving efficacy. Like fludarabine and cladribine, it requires intracellular phosphorylation for to generate the cytotoxic triphosphate. However, clofarabine inhibits both DNA polymerase, like fludarabine, and ribonucleotide reductase, like cladribine leading to inhibition of DNA synthesis and apoptosis [32,  33]. Clofarabine has shown efficacy as a single agent both in adults and ­pediatric leukemia patients. Cytarabine (ara-C) also demonstrates good efficacy in both AML and ALL. The combination of purine nucleoside analogs with cytarabine demonstrate activity both preclinically and clinically, leading to the initial hypothesis that clofarabine might have similar activity which was demonstrated in preclinical studies [34–38]. It is thought that clofarabine increases the levels of intracellular AraC triphosphate by decreasing dNTP levels which negatively feedback and inhibit the AraC triphosphate formation [39]. The initial phase I/II study of clofarabine in combination with AraC demonstrated good antileukemia activity against AML and MDS with an overall response rate of 41%. In the phase II single agent pediatric ALL study, overall response rate was 30%. In concordance with the preclinical data, many patients who had responses demonstrated increased leukemic blast levels of AraC triphosphate [40]. Some of these novel targeted agents may be dependent on the cytotoxic chemotherapy to elicit a significant effect. They may require specific cell cycle timing or alterations in DNA conformation. Topotecan, a topoisomerase I inhibitor, stabilizes the bond between the topoisomerase I and single DNA strand, preventing repair of the single-strand break [41, 42]. As a single agent, it demonstrates some antitumor activity in pediatric tumors, although appears to be relatively small with an objective

21  Anticancer Drug Development in Pediatric Patients

597

response rate of less than 20% [43–46]. However, when given following alkylating agents (i.e., cyclophosphamide) or radiation, it can stabilize the DNA strand breaks created, leading to increased cytotoxicity [47, 48]. Pediatric trials have shown increased response rates with the combination of cyclophosphamide and topotecan when compared to single agent topotecan alone [49, 50]. Similarly, irinotecan, another camptothecin-derived topoisomerase I inhibitor has been widely used in pediatrics. It demonstrates excellent preclinical activity against several pediatric tumors, including medulloblastoma, neuroblastoma, and rhabdomyosarcoma [51–54]. As a single agent, irinotecan has demonstrated a response rate of 11–23% in relapsed pediatric tumors [55–57]. Temozolamide, a methylating agent has been used in combination with irinotecan, partly due to nonoverlapping toxicities, myelosupression in temolozomide and diarrhea with irinotecan [29]. The combination of irinotecan and temozolamide initially showed preclinical synergistic activity, and has subsequently shown activity in early phase II trials [58–60]. Alternatively, the action of the agent may have a noncytotoxic effect but instead change the tumor microenvironment to improve chemotherapy delivery. For example, inhibition of VEGF receptors has been shown to potentiate chemosensitivity through normalization of blood vessels and subsequent improved delivery of the cytotoxic agent [61–63]. Additionally, the presence of the VEGF receptor inhibitors itself can potentiate chemosensitivity through mechanisms not completely understood [64, 65]. Another alternative is that the combination of the new agent may enhance an already active combination of chemotherapy. In an attempt to systematically and safely evaluate new drugs in a relapse population, the COG has defined a set backbone of chemotherapy for early relapsed ALL and is now testing the addition of new agents to this backbone for efficacy when compared to the backbone alone [66]. This design is beneficial, as it ensures that the relapsed patients are getting active therapy with the conventional chemotherapy backbone alone but also having the addition of a potentially efficacious new agent in a controlled setting. This method continues to offer the hope for some direct benefit while generating data that is easy to compare to historical controls as well as other new agents. Determining which agents to combine with a novel therapy can be difficult. As mentioned above, establishing a standard backbone for certain diseases can be beneficial as it allows for consistent evaluation of the new agent itself. Preclinical testing of the investigational agent in combination with standard chemotherapy can help drive study design. If preclinical data can suggest combinations that are synergistic or that requires certain sequencing this will further help in characterizing which agents should move forward and in what capacity. The role of patient-directed therapy continues to play a role in pediatric ­oncology. The addition of imatinib to the standard treatment in patients with Philadelphiapositive ALL has shown significant improvement in early (3  year) survival [67]. MLL rearrangement in infant leukemia and AML has now become another target, with the new generation of trials incorporating lestaurtinib, a FLT3 small molecule inhibitor for those patient expressing high levels of Flt3, or specific Flt3 mutations.

598

L. Gore and M. Macy

Other pediatric tumors with known fusion proteins or specific mutations, may also provide good targets for patient-directed therapy, although the “drugability” of some of these targets remains challenging. These agents may clarify which patients may have good responses when the entire patient group may not see a result as seen with colorectal cancer patients treated with EGFR inhibitors. Patients with kRas mutational status are unresponsive or resistant to EGFR inhibitor therapy [68, 69].

21.6 Conclusion Pediatric drug development will likely never contribute significantly financially or in terms of patient volume to the oncology world. However, it can uniquely contribute to the knowledge of biology, proof of principle, and biologic target to then be used as a paradigm for drug development in larger adult trials. It is common in adult developmental therapeutics to perform multiple phase I trials at varying doses and schedules. In pediatrics, due to the limited number of patients, only one or two phase I trials can logistically be performed. Therefore, the value of that trial has to be significant. It is imperative that vigorous pharmacokinetic and pharmacodynamic and other biologic markers are obtained. Additionally, a high percentage of patient participation in the biology component of the trial also expands the benefit of and data gathered from the single trial. It is with these main principles that early phase pediatric trials must be designed to provide the most beneficial and pertinent information for the number of patients available.

References 1. Department of Health and Human Services, Protection of Human Subjects. Revised June 23, 2005. 2. Cuzzolin, L., A. Atzei, and V. Fanos, Off-label and unlicensed prescribing for newborns and children in different settings: a review of the literature and a consideration about drug safety. Expert Opin Drug Saf, 2006. 5(5): pp. 703–18. 3. Best Pharmaceuticals Act, in PL107-109, 2002. 4. Miller, R.P., R.J. Roberts, and L.J. Fischer, Acetaminophen elimination kinetics in neonates, children, and adults. Clin Pharmacol Ther, 1976. 19(3): pp. 284–94. 5. Kearns, G.L., Impact of developmental pharmacology on pediatric study design: overcoming the challenges. J Allergy Clin Immunol, 2000. 106(3 Suppl): pp. S128–38. 6. Kearns, G.L., et al., Developmental pharmacology – drug disposition, action, and therapy in infants and children. N Engl J Med, 2003. 349(12): pp. 1157–67. 7. Gibbs, J.P., et al., Up-regulation of glutathione S-transferase activity in enterocytes of young children. Drug Metab Dispos, 1999. 27(12): pp. 1466–9. 8. McLeod, H.L., et al., Disposition of antineoplastic agents in the very young child. Br J Cancer Suppl, 1992. 18: pp. S23–9. 9. Strolin Benedetti, M. and E.L. Baltes, Drug metabolism and disposition in children. Fundam Clin Pharmacol, 2003. 17(3): pp. 281–99. 10. Bartelink, I.H., et al., Guidelines on paediatric dosing on the basis of developmental physiology and pharmacokinetic considerations. Clin Pharmacokinet, 2006. 45(11): pp. 1077–97.

21  Anticancer Drug Development in Pediatric Patients

599

11. Ginsberg, G., et al., Evaluation of child/adult pharmacokinetic differences from a database derived from the therapeutic drug literature. Toxicol Sci, 2002. 66(2): pp. 185–200. 12. Zwerdling, T., et al., Phase II investigation of docetaxel in pediatric patients with recurrent solid tumors: a report from the Children’s Oncology Group. Cancer, 2006. 106(8): pp. 1821–8. 13. Irken, G., et al., Treatment outcome of adolescents with acute lymphoblastic leukemia. Ann Hematol, 2002. 81(11): pp. 641–5. 14. Kantarjian, H., et  al., Long-term follow-up results of hyperfractionated cyclophosphamide, vincristine, doxorubicin, and dexamethasone (Hyper-CVAD), a dose-intensive regimen, in adult acute lymphocytic leukemia. Cancer, 2004. 101(12): pp. 2788–801. 15. Gaynon, P.S., et al., Expression of BCR-ABL, E2A-PBX1, and MLL-AF4 fusion transcripts in newly diagnosed children with acute lymphoblastic leukemia: a Children’s Cancer Group initiative. Leuk Lymphoma, 1997. 26(1–2): pp. 57–65. 16. Rambaldi, A., et al., Molecular diagnosis and clinical relevance of t(9;22), t(4;11) and t(1;19) chromosome abnormalities in a consecutive group of 141 adult patients with acute lymphoblastic leukemia. Leuk Lymphoma, 1996. 21(5–6): pp. 457–66. 17. Schlieben, S., et al., Incidence and clinical outcome of children with BCR/ABL-positive acute lymphoblastic leukemia (ALL). A prospective RT-PCR study based on 673 patients enrolled in the German pediatric multicenter therapy trials ALL-BFM-90 and CoALL-05-92. Leukemia, 1996. 10(6): pp. 957–63. 18. Tuszynski, A., et al., Detection and significance of bcr-abl mRNA transcripts and fusion proteins in Philadelphia-positive adult acute lymphoblastic leukemia. Leukemia, 1993. 7(10): pp. 1504–8. 19. Shurtleff, S.A., et al., TEL/AML1 fusion resulting from a cryptic t(12;21) is the most common genetic lesion in pediatric ALL and defines a subgroup of patients with an excellent prognosis. Leukemia, 1995. 9(12): pp. 1985–9. 20. McLean, T.W., et al., TEL/AML-1 dimerizes and is associated with a favorable outcome in childhood acute lymphoblastic leukemia. Blood, 1996. 88(11): pp. 4252–8. 21. Smith, M., et al., Conduct of phase I trials in children with cancer. J Clin Oncol, 1998. 16(3): pp. 966–78. 22. Lee, D.P., J.M. Skolnik, and P.C. Adamson, Pediatric phase I trials in oncology: an analysis of study conduct efficiency. J Clin Oncol, 2005. 23(33): pp. 8431–41. 23. Skolnik, J.M., et al., Shortening the timeline of pediatric phase I trials: the rolling six design. J Clin Oncol, 2008. 26(2): pp. 190–5. 24. Kim, A., et al., Characteristics and outcome of pediatric patients enrolled in phase I oncology trials. Oncologist, 2008. 13(6): pp. 679–89. 25. Shah, S., et al., Phase I therapy trials in children with cancer. J Pediatr Hematol Oncol, 1998. 20(5): pp. 431–8. 26. Horstmann, E., et  al., Risks and benefits of phase 1 oncology trials, 1991 through 2002. N Engl J Med, 2005. 352(9): pp. 895–904. 27. Italiano, A., et al., Treatment outcome and survival in participants of phase I oncology trials carried out from 2003 to 2006 at Institut Gustave Roussy. Ann Oncol, 2008. 19(4): pp. 787–92. 28. Roberts, T.G., Jr., et al., Trends in the risks and benefits to patients with cancer participating in phase 1 clinical trials. JAMA, 2004. 292(17): pp. 2130–40. 29. Furman, W.L., et al., Direct translation of a protracted irinotecan schedule from a xenograft model to a phase I trial in children. J Clin Oncol, 1999. 17(6): pp. 1815–24. 30. Houghton, P.J., et al., The pediatric preclinical testing program: description of models and early testing results. Pediatr Blood Cancer, 2007. 49(7): pp. 928–40. 31. Carlson, L., et al., Pediatric phase I drug tolerance: a review and comparison of recent adult and pediatric phase I trials. J Pediatr Hematol Oncol, 1996. 18(3): pp. 250–6. 32. Parker, W.B., et al., Effects of 2-Chloro-9-(2-deoxy-2-fluoro-{beta}-D-arabinofuranosyl)adenine on K562 cellular metabolism and the inhibition of human ribonucleotide reductase and DNA polymerases by its 5’-triphosphate. Cancer Res, 1991. 51(9): pp. 2386–94.

600

L. Gore and M. Macy

33. Xie, C. and W. Plunkett, Metabolism and actions of 2-Chloro-9-(2-deoxy-2-fluoro-{beta}-D -arabinofuranosyl)-adenine in human lymphoblastoid cells. Cancer Res, 1995. 55(13): pp. 2847–52. 34. Estey, E., et al., Fludarabine and arabinosylcytosine therapy of refractory and relapsed acute myelogenous leukemia. Leuk Lymphoma, 1993. 9(4–5): pp. 343–50. 35. Gandhi, V., et  al., Biochemical modulation of arabinosylcytosine for therapy of leukemias. Leuk Lymphoma, 1993. 10(Suppl): pp. 109–14. 36. Gandhi, V., et  al., Fludarabine potentiates metabolism of cytarabine in patients with acute myelogenous leukemia during therapy. J Clin Oncol, 1993. 11(1): pp. 116–24. 37. Chow, K.U., et al., In AML cell lines Ara-C combined with purine analogues is able to exert synergistic as well as antagonistic effects on proliferation, apoptosis and disruption of mitochondrial membrane potential. Leuk Lymphoma, 2003. 44(1): pp. 165–73. 38. Han, T., et  al., Quantitation of synergism of arabinosylcytosine and cladribine against the growth of arabinosylcytosine-resistant human lymphoid cells. J Cancer Res Clin Oncol, 2005. 131(9): pp. 609–16. 39. Cooper, T., et al., Biochemical modulation of cytarabine triphosphate by clofarabine. Cancer Chemother Pharmacol, 2005. 55(4): pp. 361–8. 40. Faderl, S., et al., Results of a phase 1-2 study of clofarabine in combination with cytarabine (ara-C) in relapsed and refractory acute leukemias. Blood, 2005. 105(3): pp. 940–7. 41. Eng, W.K., et al., Evidence that DNA topoisomerase I is necessary for the cytotoxic effects of camptothecin. Mol Pharmacol, 1988. 34(6): pp. 755–60. 42. Hsiang, Y.H. and L.F. Liu, Identification of mammalian DNA topoisomerase I as an intracellular target of the anticancer drug camptothecin. Cancer Res, 1988. 48(7): pp. 1722–6. 43. Pratt, C.B., et  al., Phase I study of topotecan for pediatric patients with malignant solid tumors. J Clin Oncol, 1994. 12(3): pp. 539–43. 44. Langler, A., et al., Topotecan in the treatment of refractory neuroblastoma and other malignant tumors in childhood – a phase-II-study. Klin Padiatr, 2002. 214(4): pp. 153–6. 45. Blaney, S.M., et al., Phase II trial of topotecan administered as 72-hour continuous infusion in children with refractory solid tumors: a collaborative Pediatric Branch, National Cancer Institute, and Children’s Cancer Group Study. Clin Cancer Res, 1998. 4(2): pp. 357–60. 46. Hawkins, D.S., et al., Topotecan by 21-day continuous infusion in children with relapsed or refractory solid tumors: a Children’s Oncology Group study. Pediatr Blood Cancer, 2006. 47(6): pp. 790–4. 47. Coggins, C.A., et al., Enhancement of irinotecan (CPT-11) activity against central nervous system tumor xenografts by alkylating agents. Cancer Chemother Pharmacol, 1998. 41(6): pp. 485–90. 48. Mattern, M.R., et al., Synergistic cell killing by ionizing radiation and topoisomerase I inhibitor topotecan (SK&F 104864). Cancer Res, 1991. 51(21): pp. 5813–6. 49. Kushner, B.H., et al., Pilot study of topotecan and high-dose cyclophosphamide for resistant pediatric solid tumors. Med Pediatr Oncol, 2000. 35(5): pp. 468–74. 50. Saylors, R.L., III, et  al., Cyclophosphamide plus topotecan in children with recurrent or refractory solid tumors: a Pediatric Oncology Group phase II study. J Clin Oncol, 2001. 19(15): pp. 3463–9. 51. Vassal, G., et al., Therapeutic activity of CPT-11, a DNA-topoisomerase I inhibitor, against peripheral primitive neuroectodermal tumour and neuroblastoma xenografts. Br J Cancer, 1996. 74(4): pp. 537–45. 52. Houghton, P.J., et al., Therapeutic efficacy of the topoisomerase I inhibitor 7-ethyl-10-(4-[1piperidino]-1-piperidino)-carbonyloxy-camptothecin against human tumor xenografts: lack of cross-resistance in vivo in tumors with acquired resistance to the topoisomerase I inhibitor 9-dimethylaminomethyl-10-hydroxycamptothecin. Cancer Res, 1993. 53(12): pp. 2823–9. 53. Vassal, G., et al., Potent therapeutic activity of irinotecan (CPT-11) and its schedule dependency in medulloblastoma xenografts in nude mice. Int J Cancer, 1997. 73(1): pp. 156–63. 54. Hare, C.B., et  al., Therapeutic efficacy of the topoisomerase I inhibitor 7-ethyl-10-(4[1-piperidino]-1-piperidino)-carbonyloxy-camptothecin against pediatric and adult central nervous system tumor xenografts. Cancer Chemother Pharmacol, 1997. 39(3): pp. 187–91.

21  Anticancer Drug Development in Pediatric Patients

601

55. Bisogno, G., et al., Phase II study of a protracted irinotecan schedule in children with refractory or recurrent soft tissue sarcoma. Cancer, 2006. 106(3): pp. 703–7. 56. Cosetti, M., et al., Irinotecan for pediatric solid tumors: the Memorial Sloan-Kettering experience. J Pediatr Hematol Oncol, 2002. 24(2): pp. 101–5. 57. Vassal, G., et  al., Phase II trial of irinotecan in children with relapsed or refractory rhabdomyosarcoma: a joint study of the French Society of Pediatric Oncology and the United Kingdom Children’s Cancer Study Group. J Clin Oncol, 2007. 25(4): pp. 356–61. 58. Kushner, B.H., et al., Irinotecan plus temozolomide for relapsed or refractory neuroblastoma. J Clin Oncol, 2006. 24(33): pp. 5271–6. 59. Wagner, L.M., et  al., Phase I trial of temozolomide and protracted irinotecan in pediatric patients with refractory solid tumors. Clin Cancer Res, 2004. 10(3): pp. 840–8. 60. Wagner, L.M., et  al., Temozolomide and intravenous irinotecan for treatment of advanced Ewing sarcoma. Pediatr Blood Cancer, 2007. 48(2): pp. 132–9. 61. Dickson, P.V., et al., Bevacizumab-induced transient remodeling of the vasculature in neuroblastoma xenografts results in improved delivery and efficacy of systemically administered chemotherapy. Clin Cancer Res, 2007. 13(13): pp. 3942–50. 62. Tong, R.T., et al., Vascular normalization by vascular endothelial growth factor receptor 2 blockade induces a pressure gradient across the vasculature and improves drug penetration in tumors. Cancer Res, 2004. 64(11): pp. 3731–6. 63. Wildiers, H., et al., Effect of antivascular endothelial growth factor treatment on the intratumoral uptake of CPT-11. Br J Cancer, 2003. 88(12): pp. 1979–86. 64. Andersson, M.K. and P. Aman, Proliferation of Ewing sarcoma cell lines is suppressed by the receptor tyrosine kinase inhibitors gefitinib and vandetanib. Cancer Cell Int, 2008. 8: p. 1. 65. Rowe, D.H., et al., Anti-VEGF antibody suppresses primary tumor growth and metastasis in an experimental model of Wilms’ tumor. J Pediatr Surg, 2000. 35(1): pp. 30–2; discussion 32–3. 66. Raetz, E.A., et al., Outcomes of children with first marrow relapse: results from Children’s Oncology Group (COG) study AALL01P2. ASH Annual Meeting Abstracts, 2006. 108(11): p. 1871. 67. Schultz, K.R., et al., Improved early event free survival (EFS) in children with Philadelphia chromosome-positive (Ph+) acute lymphoblastic leukemia (ALL) with intensive imatinib in combination with high dose chemotherapy: Children’s Oncology Group (COG) study AALL0031. ASH Annual Meeting Abstracts, 2007. 110(11): p. 9a. 68. Lievre, A., et  al., KRAS mutations as an independent prognostic factor in patients with advanced colorectal cancer treated with cetuximab. J Clin Oncol, 2008. 26(3): pp. 374–9. 69. Lievre, A., et  al., KRAS mutation status is predictive of response to cetuximab therapy in colorectal cancer. Cancer Res, 2006. 66(8): pp. 3992–5.

Chapter 22

Clinical Trials in Special Populations S. Percy Ivy, Merrill J. Egorin, Chris H. Takimoto, and Jeannette Y. Wick

22.1 Introduction In oncology, special populations are generally excluded from studies of investigational agents because dosing or scheduling information is limited or unknown. Patients who are members of special populations may be considered too frail to tolerate therapy. Special population study subjects may have hepatic or renal dysfunction from a variety of etiologies or may have poor performance status. Children and young adults represent another patient group with limited access to new agents. They are often excluded due to unique concerns about arresting or impeding their growth or development; their unique risks create an ethical dilemma in many cases. These populations are addressed in the following pages. This chapter’s emphasis is on ways to approach the study of special populations and on issues of greatest concern for investigators.

22.2 Organ Dysfunction Kidney failure, liver failure, and cancer have at least two things in common: they all increase in incidence with age, and they may pose a challenge to clinicians when the patient receives treatment. The issue is more problematic when patients with renal or kidney dysfunction develop cancer. Chemotherapeutic agents are primarily metabolized or excreted by the kidneys or liver, although some hepatically reduced drug metabolites are excreted by the kidneys. Standard chemotherapeutic dosing schedules are often developed in studies that exclude patients with organ dysfunction [1]. In general, The US Food and Drug Administration (FDA) approves many S.P. Ivy (*) Investigational Drug Branch, Cancer Therapy Evaluation Program, Division of Cancer Treatment and Diagnosis, National Cancer Institute, 6130 Executive Blvd, Suite 7131, Rockville, MD 20852, USA e-mail: [email protected]

M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_22, © Springer Science+Business Media, LLC 2011

603

604

S.P. Ivy et al.

oncology drugs (and their manufacturers market them) with scant pharmacokinetic (PK)/pharmacodynamic (PD) information in patients with organ dysfunction. In diseases other than cancer, drug doses are often adjusted empirically by extending dosing intervals, decreasing doses, or both to avoid excess drug exposure. In addition, the FDA encourages use of pharmacologically guided dosing, whereby drug concentrations in plasma or serum are monitored and doses and their frequencies are manipulated to achieve a therapeutic target range. This is a standard procedure in a number of medical specialties, such as neurology, cardiology, and infectious disease. Cancer patients who have renal or hepatic organ dysfunction may require dose reductions or modifications. Gathering organ dysfunction data in patients with cancer has traditionally been a postmarketing function. Little or no information has been available concerning use of the newest, potentially most effective chemotherapies for patients with organ dysfunction, and clinicians have tended to make empiric dose adjustments for these and the tried-and-true standard therapies. The number of cancer patients with impaired hepatic or renal function eligible for protocols specifically evaluating organ dysfunction is limited. Researchers report that accrual of full cohorts of cancer patients with organ dysfunction can be an arduous task when studies involve one or a few institutions; the conduct of such studies can take years [2]. Involving well-­coordinated, multicenter groups with access to experienced phase I investigators can, and does, shorten accrual time significantly. The National Cancer Institute’s (NCI’s) Organ Dysfunction Working Group (discussed below), the Southwest Oncology Group (SWOG) Early Therapeutics Committee, and the Cancer and Leukemia Group B (CALGB) Pharmacology and Experimental Therapeutics Committee all have specific goals for gathering information in patients with organ dysfunction and accruing patients efficiently. Each has been able to involve as many as 12 centers, reducing the accrual time and study completion to as little as 15 months [2]. Some chemotherapies require dose reductions to prevent excessive toxicity. These adjustments are accompanied by a risk of suboptimal disease treatment unless organ impairment is associated with decreased clearance, and decreasing doses still produce the same area under the curve as that of full doses administered to patients who have normal organ function. The risk of suboptimal treatment is most pronounced if the organ impairment is tumor-related, such as the hepatic infiltration seen in colorectal cancer. Given the recognition that hepatic and renal impairment can affect chemotherapeutic drug metabolism and clearance unpredictably, the National Cancer Institute (NCI)-sponsored Organ Dysfunction Working Group, CALGB, and SWOG are leading structured efforts to assess promising or recently approved agents in patients who have organ dysfunction. Each program performs phase I dose-escalation trials in different cohorts of patients who are defined by their degree of organ dysfunction. The clinical PK data collected are then used to develop formal guidelines for dosing in these specific populations [2].

22  Clinical Trials in Special Populations

605

22.3 FDA Regulatory Guidance The kidneys, as excretory, biosynthetic, and metabolic organs, have a vital role in normal physiology. In the USA, approximately one in nine adults has chronic kidney disease (CKD), and a minority suffer from its terminal complication, end-stage renal disease (ESRD) [3, 4]. CKD is usually silent until its late stages, and without aggressive screening, its detection may not occur until immediately before symptomatic kidney failure develops. A diagnosis of renal impairment has serious repercussions for many drugs excreted by the kidneys. Most obviously, renal impairment may decrease excretion of drugs or their metabolites, leading to accumulation and potential toxicity. Additionally, renal impairment has the potential to affect drug absorption, hepatic metabolism, plasma protein binding, and drug distribution. As renal impairment progresses from mild to moderate or severe, these changes may become more pronounced. Hepatic dysfunction is also seen in the oncology patient population, and important hepatic functions – metabolizing drugs or promoting the excretion of unchanged drugs or metabolites in the bile – are crucial for many drugs. Hepatic disease can alter PK (drug absorption and disposition) or PD (effectiveness and adverse effects). Drug accumulation or, less often, failure to form an active metabolite can occur if the liver’s metabolic activities or excretory pathways are altered; biliary excretion of unchanged drugs or metabolites can cause these problems, too. In the USA, drug-induced liver injury (DILI) is now the leading cause of acute liver failure (ALF), exceeding all other causes combined [5, 6]. Drugs have the potential to cause substantial hepatic injury–contributing to illness, disability, hospitalization, life-threatening liver failure, the need for liver transplantation, and sometimes death. The FDA has issued guidance documents related to the conduct of studies in patients who have organ dysfunction specifying: • When studies of PK in patients with organ dysfunction should be performed and when they may be unnecessary • How to design and conduct PK studies in patients with organ dysfunction • Inclusion criteria for patient populations to be studied • Appropriate design and conduct of PK studies in ESRD patients treated with dialysis (hemodialysis or peritoneal dialysis) • Study analysis and result reporting • Representation of organ dysfunction study results in approved product labeling The FDA makes the hepatic guidance document, Guidance for Industry: Pharmacokinetics in Patients with Impaired Hepatic Function: Study Design, Data Analysis, and Impact on Dosing and Labeling, is available at http://www.fda.gov/ cber/gdlns/imphep.pdf. The FDA renal guidance document, Guidance for Industry: Pharmacokinetics in Patients with Impaired Renal Function – Study Design, Data Analysis, and Impact on Dosing and Labeling, is available at http://www.fda.gov/ cder/guidance/1449fnl.pdf.

606

S.P. Ivy et al.

Patients with kidney dysfunction are relatively stable medically. As a result, studies done in patients with kidney dysfunction often have low dropout rates, and study participants may receive multiple cycles of chemotherapy even though renal function continues to decrease (as it often does in patients enrolled in studies). Abnormal liver dysfunction also often continues to decline, and it may do so precipitously. In multiple phase I studies of antitumor agents conducted in patients with organ dysfunction, a very high proportion of patients with severe hepatic disease cannot complete one cycle of therapy [7].

22.4 Pharmacologic Outcomes vs. Toxicity Of note, there are two ways to assess a drug’s performance and efficacy and establish a recommended dose. The FDA prefers the first method, establishing pharmacologic outcomes, which requires evaluation in all disease states, including cancer, and for all drugs. The second, toxicity-guided dosing, is the traditional way that oncology researchers evaluate drugs. Observing dose-limiting toxicities (DLT) assures that drugs are not given at doses whereby their toxicity prevents its further use or dose escalation. Once DLTs occur in populations with acceptable organ function, researchers establish a maximum tolerated dose (MTD), and that dose is used and thus confirmed in subsequent studies. In the organ dysfunction population, assessing dose-related toxicity end points can be difficult because they are rarely established definitively in phase I trials; instead, they require confirmation in subsequent clinical trials. Extrapolating dosing recommendations from small numbers of patients is not ideal, even when accepted phase I study designs (discussed below) are employed. Researchers attempt to strengthen clinical dosing recommendations using supporting PK data demonstrating relatively consistent systemic drug exposures across cohorts of patients with organ dysfunction [2].

22.5 Hepatic Impairment Liver disease may be caused by countless disease states and triggers, some of which are listed in Table 22.1. Regardless, hepatic dysfunction usually manifests itself in one of three distinct clinical patterns: hepatocellular, cholestatic (obstructive), or mixed. • Predominate features of hepatocellular disease include liver injury, inflammation, and necrosis; this pattern follows viral hepatitis or alcoholic liver disease. • Cholestatic disease is characterized by bile flow inhibition; its most common causes include gall stones, malignant obstruction, primary biliary cirrhosis, and exposure to some drugs. • Mixed pattern disease features both hepatocellular and cholestatic injury, such as cholestatic forms of viral hepatitis and many drug-induced liver diseases.

22  Clinical Trials in Special Populations

607

Table 22.1  Causes of liver disease Abscess Alcohol abuse Autoimmune disease (systemic lupus erythematosus, autoimmune hepatitis, primary biliary cirrhosis, and primary sclerosing cholangitis) Bile duct obstruction (resulting from gallstones, tumors, or improperly performed surgery) Budd–Chiari syndrome Cancers, tumors, and cysts (primary or metastatic cancers, echinococcosis or hydatid cyst disease) Drugs (partially listed below) and toxins (e.g., carbon tetrachloride and pyrrolizidine alkaloids) Acetaminophen overdose Alpha-methyldopa Amiodarone Anabolic steroids Anticonvulsants Antidiabetic agents Cancer chemotherapy Estrogens (birth control pills) Halothane HMG-CoA reductase inhibitors (statins) Isoniazid (INH) Methotrexate Phenytoin Phenothiazines Valproic acid Hepatitis infection Hereditary genetic aberrations (primary hemochromatosis, alpha-1-antitrypsin deficiency, Wilson’s disease, congenital disorders of bilirubin metabolism, Gilbert syndrome) Ischemic hepatitis Pancreatic inflammation Source: Refs. [5–9]

The pattern of onset and prominence of symptoms can rapidly suggest a d­ iagnosis, particularly if major risk factors are considered, such as the patient’s age and gender and exposure or risk behavior history. Liver disease can also alter kidney function, leading to drug/metabolite accumulation even if these drugs/metabolites are not hepatically eliminated in some cases. Yet, the specific effects of diseases on hepatic function are often poorly described and highly variable, particularly as they relate to potential changes to PK and PD. Among cancer patients, infiltrative liver disease is common, although most studies do not differentiate between liver dysfunction caused by cancer and liver dysfunction from other causes. Treating patients who have chemosensitive tumors (e.g., breast cancer, germ cell tumors, lymphoma) may offer the best approach to correct tumor-related hepatic dysfunction [8, 9]. Researchers and clinicians have attempted to establish reliable measurements of hepatic function (see Table 22.2), and they have also studied categorical clinical and laboratory variables. These include ascites or encephalopathy, nutritional status,

608

S.P. Ivy et al. Table 22.2  Indices used to measure hepatic function Bilirubin Albumin Prothrombin time, partial thromboplastin time, international normalized ratio (INR) Antipyrine Indocyanine green (ICG) Monoethylglycine-xylidide (MEGX) Galactose Source: Refs. [5–9]

Table 22.3  Child–Pugh classification of liver dysfunction Points scored for observed findings 1 2 3 Encephalopathy grade None 1 or 2 3 or 4 Ascitesa Absent Slight Moderate Serum bilirubin (mg/dL) <2 2–3 >3 Serum albumin (g/dL) >3.5 2.8–3.5 <2.8 Prothrombin time (sec prolonged) <4 4–6 >6 Encephalopathy grading (EEG required for Gr. 2, 3, 4) 0 Normal consciousness, personality, neurological exam, EEG 1 Restless, sleep disturbed, irritable/agitated, tremor, impaired handwriting, 5 cps waves 2 Lethargic, time-disoriented, inappropriate, asterixis, ataxia, slow triphasic waves 3 Somnolent, stuporous, place-disoriented, hyperactive reflexes, rigidity, slower waves 4 Unrousable coma, no personality/behavior, decerebrate, slow 2–3 cps delta activity Child’s A (mild dysfunction) 5–6 points Child’s B (moderate dysfunction) 7–9 points Child’s C (severe dysfunction) 10–15 points Ascites: slight – asymptomatic, moderate – requires intervention

a

peripheral edema, histological evidence of fibrosis, or a combination of variables such as the Child-Pugh classification for alcoholic cirrhosis and portal hypertension, the Mayo risk scores for primary biliary cirrhosis and primary sclerosing cholangitis [10], and the Mayo End-Stage Liver Disease (MELD) score [11]. No single measurement or group of measurements has proven sufficiently robust to estimate how hepatic impairment may affect a drug’s PK and/or PD in specific patients. Thus, clinically useful measurements of hepatic function to predict drug PK and PD are not generally available, and oncologists must rely on clinical studies, careful observation and possibly, dose titration to guide initial dosing. The NCI Organ Dysfunction Working Group uses the Child-Pugh Classification of Liver Dysfunction as one measure of hepatic dysfunction (see Table 22.3). The Child-Pugh score includes variables that reflect the hepatic synthetic (albumin and prothrombin) and elimination (bilirubin) functions. It incorporates a combination of three biochemical elements (i.e., prothrombin time, albumin level, bilirubin level) and two clinical features (i.e., presence of ascites, encephalopathy) to assess the primary

22  Clinical Trials in Special Populations

609

Table 22.4  NCI stratifications for hepatic function Group

Group A

Group B

Group C

Group D

Liver function Total bilirubin

Normal

Mild

Moderate >1.5x −  3x ULN

Severe >3x ULN

£ULN

Group E Liver transplant Any

B1: £ULN B2: >1.0x −  1.5x ULN B1: >ULN SGOT/AST Any Any Any £ULN B2: Any Patients entering an NCI-sponsored study would be stratified into five groups or cohorts (A: normal, B: mild dysfunction, C: moderate dysfunction, D: severe dysfunction, E: liver transplant) according to their hepatic function as outlined here

functions of the liver. Originally derived from patients undergoing ­portosystemic shunting for variceal hemorrhage, further study determined its utility in estimating risk in other patients undergoing surgery or treatment [12, 13]. In protocols developed for the hepatic dysfunction populations, cohorts are stratified as described in Table 22.4. However, the Child-Pugh score is not without limitations. Albumin concentration can be altered by transvascular escape or clearance that often occurs secondary to disease state, nutrition, sepsis [14], or ascites [15], and renal insufficiency, hemolysis, and sepsis can increase bilirubin [16]. Sepsis often decreases the international normalized ratio (INR) via coagulation activation [17]. Sepsis or renal insufficiency can cause metabolic encephalopathy [18]. Therefore, these prognostic markers emanate from sources other than the liver and may reflect multiorgan health rather than strictly liver function. The NCI Organ Dysfunction Working Group acknowledges that this model is not perfect, but it does capture data for use in determining dosing modifications and recommendations. The FDA guidance recommends that drug manufacturers consider conducting PK studies in patients with impaired hepatic function if: • The drug or active metabolite is subject to substantial hepatic metabolism or excretion • The drug/metabolite’s hepatic metabolism and/or excretion is less than 20%, but evidence indicates that it has a narrow therapeutic range. • One or more of the hepatic pathways of elimination might become important in the event of renal failure, or • The drug’s metabolism is unknown and no data suggest that hepatic elimination routes are minor, in which case the drug should be considered extensively metabolized Studies that have been conducted in patients with hepatic dysfunction as well as those conducted in patients with renal dysfunction (discussed below) are described briefly in Table 22.5. In some cases, these studies have been used as licensing trials or included in the product labeling (see Table 22.5).

89 Adults with varying solid tumors and liver function were stratified into four groups according to serum total bilirubin and AST

Retrospective case study of 24 adults with advanced renal failure

256 Patient with varying renal function, 10 of whom had CrCl £30 mL/min

55 Patients with hepatic dysfunction Miller et al. [80] or renal dysfunction:   Cohort 1, AST ³3× upper limit of normal   Cohort 1a, albumin less than 2.5 g/dL   Cohort 2, direct bilirubin of 1–7 mg/dL   Cohort 3, creatinine of 1.6–5.0 mg/dL

Imatinib

Bortezomib

Bortezomib

Erlotinib

Jagannath et al. [79]

Chanan-Khan et al. [78]

Ramanathan et al. [77]

60 Adults with advanced solid tumors and varying renal function

Imatinib

Gibbons et al. [76]

Table 22.5  Studies conducted in patients with renal and hepatic dysfunction Agent Study size Study Results Daily imatinib doses up to 800 or 600 mg were well tolerated by patients with mild and moderate renal dysfunction, respectively, despite their having increased imatinib exposure The maximal recommended dose of imatinib for patients with mild liver dysfunction is 500 mg/dL. Dosing guidelines for patients with moderate and severe liver dysfunction remain undetermined The standard 1.3 mg/m2 dose appeared well tolerated in patients with mild to moderate renal dysfunction Patients with CrCl £50 mL/min (n = 52 patients) had similar rates of discontinuation and similar adverse event profiles Patients with renal dysfunction tolerate 150 mg of erlotinib daily and seem to have an erlotinib clearance similar to patients without organ dysfunction Patients with hepatic dysfunction should be treated at a reduced dose (i.e., 75 mg daily) consistent with their reduced clearance

Not included in product labeling

Included in product labeling

Included in product labeling

Not included in product labeling

Labeling

610 S.P. Ivy et al.

60 Adult cancer patients with variable hepatic function or liver transplantation

37 Patients stratified into four renal dysfunction groups based on measured 24-h urinary CrCl

Oxaliplatin

34 Patients stratified by 24-h urinary CrCl into four renal dysfunction groups: group A (control, CrCl, ³60 mL/min), B (mild, CrCl, 40–59 mL/ min), C (moderate, CrCl, 20–39 mL/min), and D (severe, CrCl, <20 mL/min) 37 Adult cancer patients with variable renal function

Oxaliplatin

Oxaliplatin

Oxaliplatin

Takimoto et al. [84]

Synold et al. [83]

Takimoto et al. [82]

Takimoto et al. [81]

Oxaliplatin at 130 mg/m2 every 3 weeks is well tolerated by patients with mild to moderate degrees of renal dysfunction. These data strongly support the recommendation that dose reductions of single-agent oxaliplatin are not necessary in patients with a CrCl greater than 20 mL/min Oxaliplatin at 130 mg/m2 every 3 weeks was well tolerated in all patients with impaired liver function. Dose reductions of single-agent oxaliplatin are not indicated in patients with hepatic dysfunction Dose reductions of single-agent oxaliplatin are not necessary in patients with CrCl >20 mL/min

Oxaliplatin PK is altered in patients with renal impairment, but a corresponding increase in oxaliplatin-related toxicities is not observed

(continued)

Unclear

Unclear

22  Clinical Trials in Special Populations 611

60 Patients grouped into five classes of liver function

40 Patients with multiple myeloma and renal function. Renal deterioration was graded based on creatinine clearance in four stages

Oxaliplatin

Ibandronate

Bergner et al. [86]

Study Doroshow et al. [85]

Table 22.5  (continued) Agent Study size Results Oxaliplatin was well tolerated at 130 mg/ m2 every 21 days in patients with all levels of liver dysfunction, and there was no apparent alteration in the clearance of either total or ultrafilterable platinum species from plasma, even in patients with severe hepatic functional abnormalities In patients with compromised renal function, there was no evidence of acute nephrotoxicity with ibandronate

Labeling Unclear

612 S.P. Ivy et al.

22  Clinical Trials in Special Populations

613

22.6 Renal Dysfunction Oncologists should consider renal dysfunction resulting from preexisting comorbidities, nephrotoxicity from prior treatment, or complications of the cancer itself when administering drugs that are primarily metabolized or excreted by the kidneys. The PK of renally cleared drugs can be affected by (a) the kidney’s altered metabolic capacity, (b) altered renal excretion pursuant to altered renal blood flow or cancerous involvement of the organ, or (c) production of toxic compounds that damage the kidneys. There is a paucity of literature with guidelines for antineoplastic agent dosing in patients with renal impairment. Most dose reductions are based on data that are limited and primarily empiric. Renal dysfunction causes a cascade of pathological and physiological alterations in every organ system of the body, including the liver. Thus, renal decline not only alters the renal elimination of drugs, but may also affect the nonrenal disposition of drugs. Many studies have shown that loss of renal function can result in decreased hepatic clearance of drugs. The mechanism by which this occurs remains unclear, but studies have shown that as the kidneys fail, key enzymatic systems in the liver, intestine, and kidney become inhibited, thus affecting metabolism of some drugs [19–22]. In chronic renal failure (CRF), downregulation of selected isoforms of phase I liver metabolism (oxidation, reduction, or hydrolysis), and specifically the hepatic cytochrome P450 (CYP450), occurs, probably due to decreased gene expression and accumulation of circulating factors that modulate CYP450 activity (uremic toxins). Phase II metabolic reactions in the liver (glutathione S-transferases, UDPglucuronosyltransferases, N-acetyltransferases, amino acid N-acyltransferases, sulfotransferases) are also reduced in CRF. Although most evidence comes from in vitro and in vivo animal studies, it appears that in humans, CRF is associated with a decrease in the expression of specific liver CYP450 isoforms secondary to reduced mRNA levels [20, 22]. Phase II reactions in CRF have not been examined as extensively, but studies suggest that phase II enzyme activities, such as glucuronidation and acetylation, may also be suppressed in renal failure, probably due to the presence of uremic toxins [23, 24]. Reduced nonrenal clearance in patients with renal failure has also been observed for drugs that are poorly metabolized, and reduced CYP450 activity does not explain this phenomena [25–27]. Thus, in patients who have renal dysfunction, other mechanisms may contribute to altered drug clearance. CRF affects intestinal drug disposition in several ways. It can decrease intestinal first-pass metabolism or drug excretion mediated by decreased P-glycoprotein; in these cases, drug bioavailability increases. CRF also downregulates intestinal CYP450 via reduced gene expression [21, 22].

614

S.P. Ivy et al.

22.6.1 Measuring Renal Failure While few markers satisfactorily stratify patients with liver disease according to degrees of impairment of drug biotransformation and elimination, glomerular filtration rate (GFR) is often an adequate surrogate marker of renal drug elimination and is quantitatively exact. The methods employed to determine GFR directly – inulin clearance or exogenous filtration markers (e.g., cold iothalamate and iohexol, and hot radionuclides) – are generally too cumbersome for clinical practice. Most renal researchers will use normograms or formulas to estimate the GFR for stratification of patients into organ dysfunction cohorts to guide dosing of renally eliminated drugs. Accuracy becomes particularly crucial in the drugs with low therapeutic indices [28, 29]. In oncology, GFR-based dosing of carboplatin is an accepted practice in order to achieve a targeted area under the curve [30–33]. The NCI has used four formulas to calculate GFR and creatinine clearance (CrCl). Several methods can be used to estimate the degree of renal function: • GFR as discussed above. • Serum creatinine (SCr) can be used to stratify the degree of renal function and suggest dose reduction of some drugs; this method has some limitations based on the formula used to estimate it. • Measuring CrCl via 24-h urine collection avoids some of the pitfalls of estimated SCr. It accounts for muscle mass variations and associated creatinine generation, but it is prone to overestimating GFR since the kidneys secrete creatinine in addition to filtering it. Urine collection, prompt processing, and analysis can also be cumbersome in the clinical setting. More than 25 different formulas to estimate renal function have been tried using SCr levels, although only a few are in common clinical use (see Table 22.6). The National Kidney Foundation of the USA has recommended using either the formula proposed by Donald W. Cockcroft and M. Henry Gault (C&G) or the Modification Table 22.6  Formulas used to estimate creatinine clearance Originator(s) Formula The Cockcroft– CrCl (male) = ([140 − age] × weight in kg)/(SCr × 72) Gault method CrCl (female) = CrCl (male) × 0.85 The Jelliffe formula CrCl (mL/min) = {98 − [0.8 × (age − 20)]} × [1 − (sex × 0.1)]/ [(SCr × 0.814)/72 × (BSA/1.73)], where actual body weight (ABW) is measured in kg and SCr is measured in mmole/L The Martin formula GFR (mL/min) = {163 × ABW × [1 − (0.00496 × age)]  × [1 − (0.252 × sex)]}/SCr The Wright formula GFR (mL/min) = {[6550 − (38.8 × age)] × [1 − (0.168 × sex)] × BSA}/SCr MDRD 186 × (SCr)−1.154 × (age)−0.203 × 0.742 (if the subject is female) or ×1.212 (if the subject is black) Source: Refs. [34–37] BSA is measured in square meters, height in centimeters, age in years (nearest 10 years for Jelliffe formula) and sex = 0 (male) or 1 (female)

22  Clinical Trials in Special Populations

615

of Diet in Renal Disease (MDRD) [34, 38, 39]. Both formulas use SCr, age, and gender to estimate renal clearance, but the C&G formula is shorter, easier to calculate, and has been in use for approximately 20 years longer. The FDA recommends and the NCI Organ Dysfunction Working Group employs the C&G formula, although other methodologies are under study and may lead to improved drug-dosing study designs in patients with renal impairment. Historically, the FDA and the NCI have categorized their patients differently. • According to the NCI, renal dysfunction is historically classified as severe, moderate, mild, or “normal” when CrCl is 0–19, 20–39, 40–59, or >60  mL/min, respectively. • According to the FDA, renal dysfunction is classified as severe, moderate, mild, “normal” when CrCl is 0–30, 30–50, 50–80, or >80 mL/min, respectively. Thus, many of patients considered “normal” by the NCI will have FDA-characterized mild renal dysfunction. Mchayleh et al. [40] recently analyzed renal function in 12,575 patients entered onto CTEP-sponsored phase I clinical trials from 1979 to 2005 to evaluate the percentage of patients with acceptable renal function according to three different formulae (Cockroft–Gault, Jelliffe, and Levey), and GFR according to MDRD. Distributions of CrCl and GFR were defined, and patients were classified as having normal renal function or severe, moderate, or mild renal dysfunction. Approximately 40% of patients entered into CTEP-sponsored trials had mild renal dysfunction if the FDA criteria were applied. Since 1995, the percentage of patients having renal dysfunction has decreased. The investigators found no obvious increases in hematologic or other toxicities in the group of patients who had CrCl between 59 and 79 mL/min, compared with those with CrCl greater than 80 mL/min. These findings make a case for studying only those patients whose renal dysfunction is classified as moderate (30–50 mL/min) or severe (0–29 mL/min) in renal dysfunction studies [40]. They also support recommending modified renal dysfunction criteria to the FDA and NCI. Studies have demonstrated that the C&G formula underestimates CrCl in patients with a high GFR [41–44]. Additionally, C&G can overestimate GFR in severely renally impaired patients [41]. Compared to exogenous markers, including 51 Cr-EDTA, 125iodine iothalamate, C&G lacks precision [42–45], although it appears equivalent to or more precise than other formulas [39, 41, 42, 44, 45]. In infants, different normalization variables, such as extracellular fluid volume, may need to be measured [46]. C&G also neglects to consider disease states, such as diabetes or cancer, which may affect SCr or GFR [43, 45, 47]. Clinicians should not use C&G if a patient’s life could be in jeopardy if the proper dose is not given, or if a more accurate test is available. Despite these shortcomings, C&G may still be one of the more accurate ways to estimate renal function quickly and conveniently for a variety of clinical indications. The MDRD is more precise than C&G, but it, too, has limitations (i.e., not factoring in certain subgroups defined by age, sex, race, diabetes, renal transplant status, and body mass index, leading to extreme values for serum albumin concentration)

616

S.P. Ivy et al.

and has not proven to replicate the precision of exogenous markers [39]. Rather than using estimated CrCl unadjusted for BSA or race (and empirically adjusted for sex), MDRD automatically estimates BSA-indexed GFR, in units of mL/min/1.73 m2, as opposed to the C&G formula. The National Institutes of Health (NIH) and pharmaceutical company sponsors prefer using the MDRD for studies of therapeutic interventions in patients with CKD, and when selecting candidates for living kidney donation [39, 46–48]. Extensive experimental data have correlated GFR with BSA [49–53], but methods that rely on the association still engender some controversy. Regardless, BSAindexing of GFR avoids overdiagnosis of renal impairment in small patients and underdiagnosis of renal impairment in large individuals [38, 39, 42], and it could eliminate a source of variability between groups in randomized studies of interventions in patients with CKD. Many studies that use the FDA-mandated C&G formula now perform BSAindexing of CrCl estimates to ensure that the subsequent reclassification of patients from one GFR range to another will not alter the study’s conclusions. Many methods of determining GFR and renal function are available to investigators. In the future, other serum markers may become useful. Cystatin C, a 13-kDa cysteine protease that is produced at a constant rate by all nucleated cells, looks promising. Several studies have found cystatin C more useful than SCr, but cystatin C is affected by high-dose steroids, which may limit its use in some types of cancer treatment and following transplantation [54, 55].

22.7 Pharmacokinetics and Pharmacodynamics in Organ Dysfunction Pharmacokinetics (PK) addresses the effect of the body on a drug; it reflects the rates of absorption if given by routes other than intravenously, drug metabolism, distribution to tissues, and excretion. Pharmacodynamics (PD) considers the effect of the drug on the body, which would ideally be beneficial but in the case of antineoplastics is likely to include some toxic effects as well. The ratio of the dose causing significant damage to a patient to the dose required to produce therapeutic response is called the therapeutic index. In oncology, most drugs have relatively low therapeutic indices (from 1.5 to 2). Researchers do not generally see and therefore do not expect significant perturbation in PK or PD among patients with mild organ dysfunction. More significant organ impairment–especially liver or kidney dysfunction – can amplify drug exposure, resulting in potentially greater benefit but also potentially greater toxicity without increasing the dose. This increased exposure may result from changes in the way a drug is metabolized or excreted. Implicit in organ dysfunction studies is the understanding that drug exposure and toxicity are related. This could be from increased exposure, or increases in end organ sensitivity due to hepatic or renal disease. The latter possibility is much less likely or common, but needs to be acknowledged. The studies involve rigorous PK sampling and careful clinical data collection inherent to any phase I trial. Studies

22  Clinical Trials in Special Populations

617

are generally done as a series of independent phase I studies, each with a cohort of patients of a specific subgroup (e.g., normal organ function, mild organ dysfunction, moderate organ dysfunction, and severe organ dysfunction). The dose is then escalated individually for each cohort. This design places an emphasis on safety, and it is best suited for situations where a potentially toxic drug may have altered clearance due to organ impairment. To assess various subgroups accurately, approximately 50–60 patients are needed. Patients with mild dysfunction are often started at the same dose as that recommended for and used in patients with normal hepatic and renal function. Initially, however, researchers will usually employ a lower dose in patients with moderate organ dysfunction. Gradual dose escalation follows, assuming it is well tolerated. The severe-dysfunction cohort is started at an even lower dose. This type of study design is best suited for agents with limited toxicities or that are not expected to have major changes in drug clearance due to organ impairment. Historically, organ dysfunction studies have focused on finding appropriate doses for each organ dysfunction subgroup. Recently, the CALGB applied a different approach for their study of sorafenib in patients who had hepatic or renal dysfunction; they focused on defining sorafenib PK in a patient population rather than individual subgroups. All patients received a single standard dose of sorafenib, followed by intensive PK sampling. Using this design, most patients are evaluable for the primary PK endpoint, but less information is available concerning toxicity. Even if the PK of an agent is unaltered in patients with organ dysfunction, individuals with liver or kidney disease may still have an increased sensitivity to a particular drug. For example: • Low serum albumin may create a higher free fraction of the drug, increasing the agent’s dose-limiting toxicity (DLT); this is actually a PK change, but we currently have no method of measuring it in most patients. • Increased anemia secondary to renal dysfunction may lead to less bone marrow reserve and greater likelihood of toxicity. Because liver function tests are notoriously unreliable as a measure of drug metabolizing capability (discussed above), alternative approaches are being developed. Analysis of the metabolism of a surrogate drug can elucidate how a specific antineoplastic will behave. Many investigators use low-dose midazolam as a surrogate to measure hepatic drug clearance for CYP3A metabolic studies. This method requires an extra day of treatment, additional blood draws, and hospitalization. Researchers have also used intravenous 14C-erythromycin as a surrogate for CYP3A4 activity. As this agent is metabolized, the 14C-labeled carbon dioxide is liberated. Patients exhale into a balloon coated with material that traps the carbon dioxide, and the radioactivity present in the balloon is then measured. This approach correlates with docetaxel clearance, and it is also related to the degree of myelosuppression that a patient may develop. However, this test is costly and requires the handling of radioactive materials. Furthermore, patients may not breathe into the balloon adequately, and multiple samples may be required for optimal accuracy, thereby increasing time and expense. These factors may limit its advantages over midazolam sampling for phenotyping hepatic drug metabolizing activity [7].

618

S.P. Ivy et al.

22.8 The Dose Escalation Process In studies examining antineoplastic use in patients with organ dysfunction, researchers usually begin dose escalation in new patient cohorts using a modified Fibonacci design, ultimately increasing the dose by a minimum of 33% in each successive cohort. The typical study design includes a cohort of patients who have normal organ function to serve as PK controls. Cohort expansion usually follows the standard 3 + 3 with expansion to six patients design (three patients per cohort with up to six-patients if a dose-limiting toxicity is observed). Up to an additional 12 patients are accrued at the recommended dose to confirm that the PK and toxicities observed are appropriate. An additional difference from a standard phase I study design is that the toxicity events in one cohort may influence enrollment of patients into other cohorts with greater degrees of organ dysfunction. For example, if unacceptable toxicity is observed in two or more patients in one dose cohort, no further escalation above this dose level should occur in the more severely impaired patient cohorts. Accrual into the different organ dysfunction groups may occur concurrently or separately, but in the rare cases where major hepatotoxicity is a concern, some study designs have delayed entry of patients with the most severe degrees of organ dysfunction until data are accumulated in the less severely impaired groups. Finding, screening, and accruing patients are the rate-limiting steps in the conduct of studies in patients with renal dysfunction. Eligible patients with organ dysfunction severe enough to qualify for study entry, but healthy enough to participate in a clinical trial are rare [56–58]. Ultimately, the dose recommendation will be defined by the degree of impairment and the availability of data in the cohorts that include patients with the most severe organ dysfunction. Researchers who conduct these studies are hopeful that in the near future, they will have access to this type of information before novel agents are licensed and labeled, thus improving care for patients with hepatic and renal dysfunction [56–58].

22.9 Another Challenge: The Elderly The elderly have a cancer rate almost ten times that of people younger than 65 and are 16 times more likely to die of cancer [59, 60]. The elderly comprise only a small proportion of patients entering most cancer clinical trials (average age of about 55 years [61]). However, surviving cancer in late life is becoming increasingly common, and more than 60% of patients diagnosed with cancer today will live at least 5 years [59]. Survival rates are misleading because by the time they are calculated, they are out of date, and today’s survivors were diagnosed and treated 8 or more years ago. Thus, these rates fail to reflect current improvements in care and recent advances [62]. Oncologists’ caseloads involve elderly patients, and oncologists must

22  Clinical Trials in Special Populations

619

be familiar with seniors’ polypharmacy, organ decline, and psychosocial issues. Currently, the likelihood of receiving full-dose chemotherapy (and other treatments) decreases with age [63, 64]. A precise age-based definition for geriatric patients is lacking; however, most oncologists see elders as part of a continuum: • • • • •

Healthy, with admirable or exemplary aging Healthy, with usual aging Healthy, but vulnerable Frail At end of life

Based on this assessment of functionality, clinicians strive to minimize ­treatment-related toxicity while maintaining efficacy when prescribing them for older cancer patients. The National Cancer Center Network (NCCN) has issued a series of guidelines that delineate important treatment and management considerations based on current evidence related to the geriatric cancer patient. It is available at http://www.nccn.org/professionals/physician_gls/PDF/senior.pdf. The Senior Adult Oncology guidelines promote the Comprehensive Geriatric Assessment as an appropriate way to assess elders’ functional status. This ­multidisciplinary tool evaluates elderly cancer patients thoroughly, allowing the team to develop a coordinated treatment plan [65]. The NCCN’s senior ­guidelines also address prevention of neutropenic infections with filgrastim and peg-­ filgrastim, prevention of anemia with epietin or darbepoietin, and prevention and early management of mucositis. Researchers suspect that cancer’s increased incidence with age may be related to the following: • • • • •

Longer exposure to possible carcinogens (mutational load) Increased cell susceptibility to carcinogens Decreased ability to repair DNA Dysregulated cellular proliferation Reduced immune surveillance

Acknowledging these factors, the next step is to determine whether cancer should be treated differently in younger and older patients. This is a role for clinical trials. We know that the risk and severity of some toxicities increase with age; these are addressed clearly in the NCCN guidelines. Additionally, nutrient absorption declines with age, and bioavailability of oral drugs may be reduced as this occurs. Hepatic and renal function, as discussed earlier, also decline [66]. In general, data obtained in young adults should not be applied to elders without considerable analysis. Clearly, studies of older individuals – either during the drug development process or shortly after drug approval – are needed, considering the high proportion of older cancer patients receiving therapy. Some clinical trials have addressed the relative efficacy of cancer treatment in the elderly or provide information related to the short- and long-term complications of treatment including decline in function [67]. Most experts believe that phase II

620

S.P. Ivy et al.

studies are the appropriate venue for collecting this information [68]. Older patients are less likely to be enrolled in phase I trials, a fact that many experts decry. They acknowledge, however, that requiring phase I trials specifically for older patients would delay drug development unnecessarily. Members of the Geriatric Oncology Consortium, a group dedicated to accelerating the pace of study of cancer in the elderly, suggest that phase II trials represent an opportunity to elucidate and compare antineoplastic agents’ efficacy and toxicity in older and younger patients and categorize age-related PK changes [68]. Table 22.7 describes studies that have been conducted in elderly populations.

22.10 The Pediatric Clinical Trial: An Additional Challenge The timeline for conducting pediatric phase I trials has remained unchanged for 40 years. Several factors contribute to clinical-trial timelines in pediatric patients that are longer than those in adult populations. Evaluation of new agents in pediatric patients usually begins only after phase I adult trials are completed. In addition, the number of pediatric patients who meet eligibility criteria of any study is limited, and few pharmaceutical companies undertake pediatric cancer drug development initiatives [69]. On January 4, 2002, a legislative initiative, the Best Pharmaceuticals for Children Act (BCPA), was approved, establishing a process for studying on-patent and off-patent drugs for use in pediatric populations. The Act also seeks to improve pediatric therapeutics by promoting collaboration on scientific investigation, ­clinical study design, weight of evidence, and ethical and labeling issues. The Eunice Kennedy Shriver National Institute of Child Health and Human Development leads BPCA efforts on behalf of the NIH, in part because of its record of success in the Pediatric Pharmacology Research Unit (PPRU) Network, a group of 13 sites and a data-coordinating center that have been conducting pediatric clinical trials since 1994. Efforts to shorten the overall timeline for study completion have increased during the past 5 years and have been impacted most notably by legislative initiatives, including the BCPA [70]. On July 29, 2008, President George W. Bush signed the Caroline Pryce Walker Conquer Childhood Cancer Act. Named in memory of Caroline Pryce Walker, Congresswoman Deborah Pryce’s daughter who died of neuroblastoma in 1999 at age nine, the bill authorizes $30 million annually for 5 years for collaborative pediatric cancer clinical trials research to create a population-based national childhood cancer database and to further public awareness about treatment and research for children with cancer and their families. Both adult and pediatric phase I studies determine a recommended phase II dose and use DLT as the primary end point. Pediatric studies are complicated by numerous factors, including children’s lower blood volume that may have PK repercussions, drug palatability and formulation issues, different cancer trajectories in younger patients, unpredictable patterns of response to treatment, and the

Despite elderly patients’ receiving less chemotherapy, adjuvant vinorelbine and cisplatin improves survival in patients older than 65 years with acceptable toxicity. Adjuvant chemotherapy should not be withheld from elderly patients Response rates obtained in other studies could not be confirmed in this small cohort of older women with breast cancer; trial prematurely terminated The easy self-management, favorable toxicity profile, and synergy with other compounds makes oral etoposide suitable for further clinicalpharmacological studies in elderly patients

Pepe et al. [87]

Basso et al. [88]

12 Eligible patients with a median age of 74 years

17 Elderly patients with advanced progressive non-small cell lung cancer (NSCLC)

Gemcitabine added to vinorelbine in metastatic breast cancer

Oral etopside

Sorio et al. [89]

Results

Study

Pretreatment characteristics and survival was compared for 327 young (£65 years) and 155 elderly (>65 years) patients

Adjuvant vinorelbine and cisplatin in NSCLC

Table 22.7  Cancer clinical trials in the elderly Agent Study size

(continued)

Not included in product labeling

Not included in either product’s labeling

Not included in either product’s labeling

Labeling

22  Clinical Trials in Special Populations 621

Combination gemcitabine and cisplatin appears to be effective and tolerated in elderly patients with advanced NSCLC, regardless of aging and condition of comorbidities Twice a month continuousinfusion CPT-11 combined with FU is a valid therapeutic alternative for elderly patients in good general condition Accrual of older patients was not increased by this intervention. More intense and multifaceted approaches will be needed to change physician (and patient) behavior and to increase accrual of older persons to clinical trials

Moscetti et al. [90]

Sastre et al. [91]

85 Patients ³72 years old with metastatic colorectal cancer

126 Cancer patients age 65 years and older

Irinotecan (CPT-11) and fluorouracil (FU)

None

Kimmick et al. [92]

Results

Study

46 Patients with NSCLC aged over 65 years or older

Study size

Gemcitabine and cisplatin

Table 22.7  (continued) Agent

N/A

Labeling

622 S.P. Ivy et al.

22  Clinical Trials in Special Populations

623

parent’s understating of and biases about treatment. As adult trials do, almost all pediatric phase I oncology trials use the traditional 3 + 3, phase I cancer trial design, ultimately defining the MTD as the dose level at which none or one of six participants (0–17%) experience a DLT, when at least two of three to six ­participants (33–67%) experience a DLT at the next highest dose. The large majority of dose levels are ultimately expanded to accrue six patients. In the 3 + 3 design, accrual is suspended after enrollment of each cohort of three patients. When a participant becomes inevaluable for toxicity, most commonly because of early disease progression, the cohort is reopened to a single patient. Delays associated with patient accrual, replacement of inevaluable patients, time to event (i.e., DLT), and time associated with data submission and review consume ­considerable time [71]. A review of pediatric phase I oncology trials reveals that they are as safe as those conducted in adults; among 1,066 pediatric patients reported from 47 studies, the toxic death rate was 0.5%, which was identical to that observed in adult phase I trials [72]. Researchers are now considering alternative study designs in an effort to make life-saving drugs available to children more quickly. One such design, the rolling six design, allows accrual of two to six patients concurrently onto a dose level. Decisions as to which dose level to enroll a patient are based on three factors: • The number of patients currently enrolled and evaluable • The number of patients experiencing DLTs • The number of patients still at risk of developing a DLT at the time of entry of new patient The rolling six design shortens the overall duration of the study, because while the 3 + 3 method requires that accrual be suspended between cohorts, the rolling six design significantly decreases the number of times a study is suspended to accrual. Further, it decreases the likelihood that patients who are eligible for a pediatric phase I study are unable to participate because of study suspension to accrual [73]. A special concern in the pediatric cancer patient is quality of life and ability to function during treatment, which can be lengthy; during extended survival; and if cure is achieved. A multidisciplinary team must design interventions to help patients cope with cancer and all of its tangential issues, including acute toxicities and treatment’s long-term sequelae. In response to such multidisciplinary needs, clinical trials in children are increasingly expanding their focus to include biobehavioral and psychosocial investigations [74, 75]. Among the most pressing and heavily studied issues are that of the neurocognitive consequences of treatment such as the intellectual impairment secondary to cranial radiation [74]. The next logical step is to study other outcomes and remedial interventions targeting learning disabilities, coping, family dynamics during the cancer experience, sociometric assessment of child and family functioning, and behavioral and appropriate psychopharmacologic intervention for pain and discomfort [75]. For more information about pediatric studies in oncology, see Chap. 21.

624

S.P. Ivy et al.

22.11 Conclusions Special populations such as the elderly, the very young, and patients with impaired renal and hepatic function present substantial challenges to the practicing oncologist. Historically, these patients have been underrepresented in most oncology drug development clinical trials, leading to a paucity of data regarding how best to treat these patients. Recently, concerted efforts have been implemented to conduct specific studies to define the PK and clinical use of new cancer therapeutic agents in these special populations. These trials have their own unique challenges in their design, conduct, and most importantly, patient accrual. Nonetheless, the relevance of this information for the practicing oncologist justifies this important effort to define optimal individualized treatment regimens for special populations of patients with cancer.

References 1. Donelli MG, Zucchetti M, Munzone E, et al. Pharmacokinetics of anticancer agents in patients with impaired liver function. Eur J Cancer. 1998;34:33–46. 2. Takimoto CH, Mita AC. Design, conduct, and interpretation of organ impairment studies in oncology patients. J Clin Oncol. 2006;24:3509–10. 3. National Kidney Foundation. K/DOQI Clinical Practice Guidelines for Chronic Kidney Disease: Evaluation, Classification, and Stratification. Available at http://www.kidney.org/ Professionals/Kdoqi/guidelines_ckd/toc.htm. Accessed October 20, 2008. 4. Coresh J, Astor BC, Greene T, Eknoyan G, Levey AS. Prevalence of chronic kidney disease and decreased kidney function in the adult US population: Third National Health and Nutrition Examination Survey. Am J Kidney Dis. 2003;41:1–12. 5. Lee WM, Seremba E. Etiologies of acute liver failure. Curr Opin Crit Care. 2008; 14:198–201. 6. Lee WM. Etiologies of acute liver failure. Semin Liver Dis. 2008;28:142–52. 7. Egorin M. The effects of organ dysfunction on drug dosing. Clin Adv Hematol Oncol. 2006;4:116–8. 8. Mano MS, Cassidy J, Canney P. Liver metastases from breast cancer: Management of patients with significant liver dysfunction. Cancer Treat Rev. 2005;31:35–48. 9. Ghobrial IM, Wolf RC, Pereira DL, et al. Therapeutic options in patients with lymphoma and severe liver dysfunction. Mayo Clin Proc. 2004;79:169–75. 10. Wiesner RH. Liver transplantation for primary biliary cirrhosis and primary sclerosing cholangitis: Predicting outcomes with natural history models. Mayo Clin Proc. 1998;73:575–88. 11. Kamath PS, Wiesner RH, Malinchoc M, et al. A model to predict survival in patients with end-stage liver disease. Hepatology. 2001;33:464–70. 12. Child CG, Turcotte JG. Surgery and portal hypertension. In: Child CG, ed. The Liver and Portal Hypertension. Philadelphia, PA: Saunders; 1964:50–64. 13. Pugh RNH, Murray-Lyon IM, Dawson JL, Pietroni MC, Williams R. Transection of the esophagus in bleeding oesophageal varices. Br J Surg. 1973;60:648–52. 14. Fleck A, Raines G, Hawker F, et al. Increased vascular permeability: A major cause of hypoalbuminaemia in disease and injury. Lancet. 1985;1:781–4. 15. Henriksen JH, Parving HH, Christiansen L, Winkler K, Lassen NA. Increased transvascular escape rate of albumin during experimental portal and hepatic venous hypertension in the pig. Relation to findings in patients with cirrhosis of the liver. Scand J Clin Lab Invest. 1981;41:289–99.

22  Clinical Trials in Special Populations

625

1 6. Moseley RH. Sepsis and cholestasis. Clin Liver Dis. 2004;8:83–94. 17. Plessier A, Denninger MH, Consigny Y, et al. Coagulation disorders in patients with cirrhosis and severe sepsis. Liver Int. 2003;23:440–8. 18. Kunze K. Metabolic encephalopathies. J Neurol. 2002;249:1150–9. 19. Touchette MA, Slaughter RL. The effect of renal failure on hepatic drug clearance. DICP. 1991;25:1214–24. 20. Dreisbach AW, Lertora JJ. The effect of chronic renal failure on hepatic drug metabolism and drug disposition. Semin Dial. 2003;16:45–50. 21. Sun H, Frassetto L, Benet LZ. Effects of renal failure on drug transport and metabolism. Pharmacol Ther. 2006;109:1–11. 22. Pichette V, Leblond FA. Drug metabolism in chronic renal failure. Curr Drug Metab. 2003;4:91–103. 23. Uchida N, Kurata N, Shimada K, et al. Changes of hepatic microsomal oxidative drug metabolizing enzymes in chronic renal failure (CRF) rats by partial nephrectomy. Jpn J Pharmacol. 1995;68:431–9. 24. Leblond F, Guévin C, Demers C, Pellerin I, Gascon-Barré M, Pichette V. Downregulation of hepatic cytochrome P450 in chronic renal failure. J Am Soc Nephrol. 2001;12:326–32. 25. Martin DE, Chapelsky MC, Ilson B, et al. Pharmacokinetics and protein binding of eprosartan in healthy volunteers and in patients with varying degrees of renal impairment. J Clin Pharmacol. 1998;38:129–37. 26. Kovacs SJ, Tenero DM, Martin DE, Ilson BE, Jorkasky DK. Pharmacokinetics and protein binding of eprosartan in hemodialysis-dependent patients with end-stage renal disease. Pharmacotherapy. 1999;19:612–9. 27. McTaggart F, Buckett L, Davidson R, et  al. Preclinical and clinical pharmacology of Rosuvastatin, a new 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitor. Am J Cardiol. 2001;87:28B–32B. 28. Aronoff GR, Berns JS, Brier ME, et al. Drug Prescribing in Renal Failure: Dosing Guidelines for Adults, 4th ed. Philadelphia, PA: American College of Physicians-American Society of Internal Medicine; 1999. 29. Kasiske BL, Keane WF. Laboratory assessment of renal disease: Clearance, urinalysis, and renal biopsy. In: Brenner BM, Rector FC, eds. Brenner and Rector’s the Kidney, 6th ed. Philadelphia, PA: WB Saunders; 2000:1129–70. 30. Calvert AH, Newell DR, Gumbrell LA, et al. Carboplatin dosage: Prospective evaluation of a simple formula based on renal function. J Clin Oncol. 1989;7:1748–56. 31. Chatelut E, Canal P, Brunner V, et  al. Prediction of carboplatin clearance from standard morphological and biological patient characteristics. J Natl Cancer Inst. 1995;87:573–80. 32. Dooley MJ, Poole SG, Rischin D, et  al. Carboplatin dosing: Gender bias and inaccurate estimates of glomerular filtration rate. Eur J Cancer. 2002;38:44–51. 33. Calvert AH, Egorin MJ. Carboplatin dosing formulae: Gender bias and the use of creatininebased methodologies. Eur J Cancer. 2002;38:11–16. 34. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron. 1976;16:31–41. 35. Jelliffe RW. Creatinine clearance: Bedside estimate. Ann Intern Med. 1973;79:604–5. 36. Wright JG, Boddy AV, Highley MS, et al. Estimation of glomerular filtration rate in cancer patients. Br J Cancer. 2001;84:452–9. 37. Martin L, Chatelut E, Boneu A, et al. Improvement of the Cockcroft–Gault equation for predicting glomerular filtration in cancer patients. Bull Cancer. 1998;85:631–6. 38. National Kidney Foundation. K/DOQI clinical practice guidelines for kidney disease: Evaluation, classification, and stratification. Kidney Disease Outcome Quality Initiative. Am J Kidney Dis. 2002;39 (Suppl. 1):S1–S266. 39. Levey AS, Bosch JP, Lewis JB, et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Ann Intern Med. 1999;130:461–70. 40. Mchayleh WM, Sehgal R, Potter DM, et al. Analysis of renal function in patients entered onto CTEP-sponsored phase I studies since 1979. Proc Am Soc Clin Oncol. 2008;A2519.

626

S.P. Ivy et al.

41. Lamb EJ, Webb MC, Simpson DE, et  al. Estimation of glomerular filtration rate in older patients with chronic renal insufficiency: Is the modification of diet in renal disease formula an improvement? J Am Geriatr Soc. 2003;51:1012–7. 42. Poole SG, Dooley MJ, Rischin D. A comparison of bedside renal function estimates and measures glomerular filtration rate (Tc^99m DTPA clearance) in cancer patients. Ann Oncol. 2002;13:949–55. 43. Perlemoine C, Rigalleau V, Baillet L, et al. Cockcroft’s formula underestimates glomerular filtration rate in diabetic subjects treated by lipid-lowering drugs. Diabetes Care. 2002;11:2106–7. 44. Van Den Noortgate NJ, Janssens WH, Delanghe JR, et al. Serum cystatin C concentration compared with other markers of glomerular filtration in the old. J Am Geriatr Soc. 2002;50:1278–82. 45. Burkhardt H, Bojarsky G, Gretz N, et al. Creatinine clearance, Cockcroft–Gault formula and cystatin C: Estimators of true glomerular filtration rate in the elderly? Gerontology. 2002;48:140–6. 46. Hsu C-Y, Chertow GM, Curhan GC. Methodological issues in studying the epidemiology of mild to moderate chronic renal insufficiency. Kidney Int. 2002;61:1567–76. 47. Coresh J, Toto RD, Kirk KA, et al. Creatinine clearance as a measure of GFR in screenees for the African-American Study of kidney disease and hypertension pilot study. Am J Kidney Dis. 1998;32:32–42. 48. Parving H-H, Lehnert H, Brochner-Mortensen J, et al. The effect of irbesartan on the development of diabetic nephropathy in patients with type 2 diabetes. N Engl J Med. 2001;345:870–8. 49. Nyengaard JR, Bendtsen TF. Glomerular number and size in relation to age, kidney weight, and body surface in normal man. Anat Rec. 1992;232:194–201. 50. MacKay EM. Kidney weight, body size, and renal function. Arch Intern Med. 1932;50:590–4. 51. McCance RA, Widdowson EM. The correct physiological basis on which to compare infant and adult renal function. Lancet. 1952;2:860–2. 52. White AJ, Strydom WJ. Normalisation of glomerular filtration rate measurements. Eur J Nucl Med. 1991;18:385–90. 53. Kasiske BL, Umen AJ. The influence of age, sex, race, and body habitus on kidney weight in humans. Arch Pathol Lab Med. 1986;110:55–60. 54. Xu H, Lu Y, Teng D, Wang J, Wang L, Li Y. Assessment of glomerular filtration rate in renal transplant patients using serum cystatin C. Transplant Proc. 2006;38:2006–8. 55. Mendiluce A, Bustamante J, Martin D, et al. Cystatin C as a marker of renal function in kidney transplant patients. Transplant Proc. 2005;37:3844–7. 56. Margolin K, Synold T, Longmate J, Doroshow JH. Methodologic guidelines for the design of high-dose chemotherapy regimens. Biol Blood Marrow Transplant. 2001;7:414–32. 57. Dent SF, Eisenhauer EA. Phase I trial design: Are new methodologies being put into practice? Ann Oncol. 1996;7:561–6. 58. Eisenhauer EA, O’Dwyer PJ, Christian M, Humphrey JS. Phase I clinical trial design in cancer drug development. J Clin Oncol. 2000;18:684–92. 59. Jemal A, Tiwari RC, Murray T, et al. Cancer statistics, 2004. CA Cancer J Clin. 2004;54:8–29. 60. Hurria A. Incorporation of geriatric principles in oncology clinical trials. J Clin Oncol. 2007;8:5350–1. 61. Hutchins LF, Unger JM, Crowley JJ, et al. Underrepresentation of patients 65 years of age or older in cancer-treatment trials. N Engl J Med. 1999;341:2061–7. 62. Reis LAG, Kosary CL, Hankey BF, et al., eds. SEER Cancer Statistics Review, 1973–1994 [NIH Pub. No. 97-2789]. Bethesda, MD: National Cancer Institute; 1997. 63. Matesich SM, Shapiro CL. Second cancers after breast cancer treatment. Semin Oncol. 2003;30:740–8. 64. Kushner BH, Heller G, Cheung NK, et al. High risk of leukemia after short-term dose-­intensive chemotherapy in young patients with solid tumors. J Clin Oncol. 1998;16:3016–20. 65. Balducci L. Management of cancer in the elderly. Oncology. 2006;20:135–52.

22  Clinical Trials in Special Populations

627

66. Balducci L. Cancer chemotherapy in the older person. In: Balducci L, Ershler WB, DeGaetano G, eds. Blood Disorders in the Elderly. Cambridge, MA: Cambridge University Press; 2008:225–55. 67. Muss HB, Woolf S, Berry D, et al. Adjuvant chemotherapy in older and younger women with lymph node-positive breast cancer. JAMA. 2005;293:1073–81. 68. Balducci L, Tam-McDevitt J, Hauser R, Simon J. Long overdue: Phase II studies in older cancer patients: Where does the FDA stand? J Clin Oncol. 2008;26:1387–88. 69. Adamson PC, Weiner SL, Simone JV, et al. Making Better Drugs for Children with Cancer. Washington, DC: National Academies Press; 2005 70. National Institutes of Health, Eunice Kennedy Shriver, National Institute of Child Health and Human Development. The Best Pharmaceuticals for Children Act. Title V – Best Pharmaceuticals for Children Amendments of 2007. Available at http://bpca.nichd.nih.gov/ index.cfm. Accessed July 23, 2008. 71. Lee DP, Skolnik JM, Adamson PC. Pediatric phase I trials in oncology: An analysis of study conduct efficiency. J Clin Oncol. 2005;23:8431–41. 72. Ross L. Phase I research and the meaning of direct benefit. J Pediatr. 2006;149:S20–4. 73. Skolnik JM, Barrett JS, Jayaraman B, Patel D, Adamson PC. Shortening the timeline of pediatric phase I trials: The rolling six design. J Clin Oncol. 2008;26:190–5. 74. Armstrong FD, Reaman GH. Psychological research in childhood cancer: The Children’s Oncology Group perspective. J Pediatr Psychol. 2005;30:89–97. 75. Reaman GH. Pediatric cancer research from past successes through collaboration to future transdisciplinary research. J Pediatr Oncol Nurs. 2004;21:123–7. 76. Gibbons J, Egorin MJ, Ramanathan RK, et al. Phase I and pharmacokinetic study of imatinib mesylate in patients with advanced malignancies and varying degrees of renal dysfunction: A study by the National Cancer Institute Organ Dysfunction Working Group. J Clin Oncol. 2008;26:570–6. 77. Ramanathan RK, et al. Phase I and pharmacokinetic study of imatinib mesylate in patients with advanced malignancies and varying degrees of liver dysfunction: A study by the National Cancer Institute Organ Dysfunction Working Group. J Clin Oncol. 2008;26:563–9. 78. Chanan-Khan AA, Kaufman JL, Mehta J, et al. Activity and safety of bortezomib in multiple myeloma patients with advanced renal failure: A multicenter retrospective study. Blood. 2007;109:2604–6. 79. Jagannath S, Barlogie B, Berenson JR, et  al. Bortezomib in recurrent and/or refractory multiple myeloma. Initial clinical experience in patients with impaired renal function. Cancer. 2005;103:1195–200. 80. Miller AA, Murry DJ, Owzar K, et al. Phase I and pharmacokinetic study of erlotinib for solid tumors in patients with hepatic or renal dysfunction: CALGB 60101. J Clin Oncol. 2007; 25:3055–60. 81. Takimoto CH, Graham MA, Lockwood G, et al. Oxaliplatin pharmacokinetics and pharmacodynamics in adult cancer patients with impaired renal function. Clin Cancer Res. 2007;13:4832–9. 82. Takimoto CH, Remick SC, Sharma S, et  al. Dose-escalating and pharmacological study of oxaliplatin in adult cancer patients with impaired renal function: A National Cancer Institute Organ Dysfunction Working Group Study. J Clin Oncol. 2003;21:2664–72. 83. Synold TW, Takimoto CH, Doroshow JH, et al. Dose-escalating and pharmacologic study of oxaliplatin in adult cancer patients with impaired hepatic function: A National Cancer Institute Organ Dysfunction Working Group study. Clin Cancer Res. 2007;13:3660–6. 84. Takimoto CH, Remick SC, Sharma S, et  al. Administration of oxaliplatin to patients with renal dysfunction: A preliminary report of the National Cancer Institute Organ Dysfunction Working Group. Semin Oncol. 2003;30(4 Suppl 15):20–5. 85. Doroshow JH, Synold TW, Gandara D, et  al. Pharmacology of oxaliplatin in solid tumor patients with hepatic dysfunction: A preliminary report of the National Cancer Institute Organ Dysfunction Working Group. Semin Oncol. 2003;30(4 Suppl 15):14–9.

628

S.P. Ivy et al.

86. Bergner R, Henrich DM, Hoffmann M, et al. Renal safety and pharmacokinetics of ibandronate in multiple myeloma patients with or without impaired renal function. J Clin Pharmacol. 2007;47:942–50. 87. Pepe C, Hasan B, Winton TL, et  al. Adjuvant vinorelbine and cisplatin in elderly patients: National Cancer Institute of Canada and Intergroup Study JBR.10. J Clin Oncol. 2007;25: 1553–61. 88. Basso U, Fratino L, Brunello A, et al. Which benefit from adding gemcitabine to vinorelbine in elderly (³70 years) women with metastatic breast cancer? Early interruption of a phase II study. Ann Oncol. 2007;18:58–63. 89. Sorio R, Toffoli G, Crivellari D, et al. Oral etoposide in elderly patients with advanced non small cell lung cancer: A clinical and pharmacological study. J Chemother. 2006;18:188–91. 90. Moscetti L, Nelli F, Padalino D, et al. Gemcitabine and cisplatin in the treatment of elderly patients with advanced non-small cell lung cancer: Impact of comorbidities on safety and efficacy outcome. J Chemother. 2005;17:685–92. 91. Sastre J, Marcuello E, Masutti B, et al. Irinotecan in combination with fluorouracil in a 48-h continuous infusion as first-line chemotherapy for elderly patients with metastatic colorectal cancer: A Spanish Cooperative Group for the Treatment of Digestive Tumors study. J Clin Oncol. 2005;23:3545–51. 92. Kimmick GG, Peterson BL, Kornblith AB, et al. Improving accrual of older persons to cancer treatment trials: A randomized trial comparing an educational intervention with standard information: CALGB 360001. J Clin Oncol. 2005;23:2201–7.

Part VI

Chapter 23

NCI-Sponsored Clinical Trials Andriana Papaconstantinou and Janet E. Dancey

23.1 Introduction The National Cancer Institute (NCI) was established under the National Cancer Institute Act of 1937 as the Federal Government’s principal agency for cancer research and training. Its mandate is to conduct and foster cancer research; review and approve grant applications to support promising research projects on causes, prevention, diagnosis, and treatment of cancer; collect, analyze, and disseminate the results of cancer research conducted in the USA and in other countries, and provide training and instruction in the diagnosis and treatment of cancer [1]. To fulfill its mission NCI supports and conducts preclinical studies and clinical trials in the areas of cancer prevention, detection, and treatment, as well as studies which seek to enhance the understanding of basic cancer biology. Within the NCI, clinical trials are mostly supported by the Division of Cancer Prevention (DCP) and the Division of Cancer Treatment and Diagnosis (DCTD). The DCP is focused on early cancer detection, cancer risk, and chemoprevention. The DCTD is focused on the preclinical and clinical development of agents and therapies for cancer treatment and diagnosis. There are several types of clinical trials supported by the NCI [2, 3]: • Prevention trials which test medications, vitamins, or other supplements that may lower the risk of developing certain types of cancers. • Screening trials which study ways to detect cancers before symptoms begin or to test whether diagnosing a cancer before it causes symptoms decreases the chance of dying from the disease. • Diagnostic trials which study tests or procedures that can be used to diagnose a cancer more accurately. J.E. Dancey (*) Investigational Drug Branch, Cancer Therapy Evaluation Program, Division of Cancer Treatment and Diagnosis, National Cancer Institute, 6130 Executive Blvd, EPN 7131, Rockville, MD 20892, USA e-mail: [email protected]

M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0_23, © Springer Science+Business Media, LLC 2011

631

632

A. Papaconstantinou and J.E. Dancey

• Treatment trials which study the effectiveness of a new treatment or new ways to use a standard treatment. • Quality-of-life (also called supportive care) trials which explore ways to improve the comfort and the quality of life of cancer patients and survivors. • Genetics studies which usually focus on the role of the genetic makeup on the detection, diagnosis, or response to treatment (these studies are often correlative to clinical trials). As of March 31, 2008 the Physician Data Query (PDQ®) database, NCI’s Comprehensive Trials Database [3], included 7,162 clinical trials sponsored by the NCI that were accepting participants. Of those, 5,411 were treatment trials, 80 were screening, 119 were prevention, and 475 were diagnostic trials (www.cancer.gov/ clinicaltrials/search). The majority of NCI clinical trials, treatment and diagnostic, are overseen by the DCTD [4]. DCTD has six major programs that work together to identify novel agents in the laboratory bench and bring them to patients in the clinic. Figure 23.1 outlines the missions of each of DCTD’s six programs.

Fig. 23.1  Division of Cancer Treatment and Diagnosis (DCTD) of the National Cancer Institute (NCI): organization and mission of its programs

23  NCI-Sponsored Clinical Trials

633

In 2005, the National Cancer Advisory Board’s (NCAB) Clinical Trials Working Group (CTWG) issued 22 strategic recommendations to restructure the conduct of NCI-supported clinical trials to improve efficiency, decrease redundancy and administrative burden, and better coordinate activities to enhance the development and delivery of promising treatments for patients [5]. The implementation of the CTWG recommendations led to the formation of the Clinical Trials Advisory Committee (CTAC) which was established to advise the NCI director on NCI’s clinical trials program and to oversee the implementation of the CTWG initiatives; the Clinical Trials Operations Committee (CTOC) to coordinate clinical trials programs across NCI and make recommendations to improve cost-effectiveness and decrease redundancy among NCI trial components; and the Coordinating Center for Clinical Trials (CCCT) to provide project management for the implementation of the CTWG initiatives, and coordinate new disease-specific steering committees for the prioritization of phase 3 trials and the Investigational Drug Steering Committee (IDSC) to provide input on prioritization for phase 1 and 2 trials, and working groups. The Translational Research Working Group (TRWG) was formed in 2005 to review NCI’s intramural and extramural research portfolio and recommend ways of integrating translational research efforts [6]. TRWG goals are to establish partnerships with industry, government and academia, improve the process for translational research studies, and institute a coordinated NCI management effort to reduce duplication, and accelerate and optimize the selection of early translational research studies. The main steps of new agent development are discovery, preclinical assessment, and clinical study development. Decisions and support at various stages of agent development are the responsibility of the following groups: the Drug Development Group (DDG) which oversees agents within the Developmental Therapeutics Program (DTP), Cancer Therapy Evaluation Program (CTEP) portfolios, and the Joint Development Committee (JDC) which governs the joint early therapeutics development program within DCTD and the Center for Cancer Research (CCR) [7]. The program, called the NCI Experimental Therapeutics Program (NExT), is responsible for deciding which agents to develop and managing resources. The major responsibilities for the evaluation of anticancer therapies are given to the CTEP and the DTP. These two programs support the majority of NCI intramural and extramural drug development efforts and are described in greater detail in this chapter.

23.2 Agent Discovery and Development 23.2.1 The Drug Development Group The DDG oversees the development and preclinical and clinical decision-making for anticancer agents to take place under an NCI Investigational New Drug (IND)

634

A. Papaconstantinou and J.E. Dancey

application, regardless of whether the source of this agent is academia, industry, or the NCI [8]. The DDG prioritizes the use of DCTD resources to support the agent’s preclinical development by DTP. The Biological Resources Branch Oversight Committee (BRB-OC) oversees the development of biologics approved by DDG. Initial presentation of an agent to the DDG requires an identified CTEP or DTP staff member to act as a liaison. The NCI liaison coordinates with the originator, who supplies an application summarizing the tasks and support being requested. For an agent to proceed in preclinical development the criteria of DTP’s screening program (discussed below) must be met [9]. These screening steps are also referred to a Stage I (early screening) and Stage IB (late screening) of the DDG program. Stage IIA activities include range-finding toxicology, GMP synthesis, formulation and Stage IIB activities focus on IND-directed toxicology, clinical lot manufacture. Stage III development focuses on clinical trials to be sponsored under NCI-held IND and coordinated through CTEP. At each transition from Stage I forward, DDG prioritizes agents for the development based on novelty, activity, need for NCI involvement as well as cost/benefit. In general, the agent should demonstrate activity in the in  vitro antiproliferative screens and antitumor activity in  vitro and in vivo. If not active in the in vitro screen, then there must be compelling evidence to advance the agent to in  vivo studies, as might be the case for antiangiogenic agents. In addition, the supply of the agent must be assured. Results from DDG IIa, IIb activities may be used to support IND filing by the originator of the agent (whether academic, industry, or NCI). In addition, the NCI/CTEP files its own IND for agents that are candidates for NCI clinical development (i.e., have received DDG III approval).

23.2.2 The Developmental Therapeutics Program: Organization and Resources The Cancer Chemotherapy National Service Center (NSC) was established in 1955 by Congress to serve as the leading resource for anticancer agent ­discovery and development [10]. By 1976, the functions of the NSC were incorporated into the DTP. Since 1955, DTP has been instrumental in the discovery and development of 40 U.S.-licensed chemotherapeutic agents. Among these agents are paclitaxel (Taxol®) and bortezomib (Velcade®). About 85% of DTP’s ­budget is reserved for funding investigator-initiated research through grant mechanisms [11]. The remaining 15% are reserved for intramural agent ­development programs. Before an agent can be tested in humans, regulatory authorities and review bodies need to confirm that the dose and schedule of the agent to be employed in a clinical study are safe and effective. DTP’s major role is to conduct and to support drug discovery and preclinical testing prior to the initiation of human clinical trials (Table 23.1).

23  NCI-Sponsored Clinical Trials

635

Table 23.1  Drug discovery and development resources Resource Web site Animal production, tumor cell lines, http://dtp.nci.nih.gov/branches/btb/services. animal tumors html In vitro screening, COMPARE http://dtp.nci.nih.gov/screening.html analysis New screening assays http://spheroid.ncifcrf.gov/STB/stb_index.cfm http://dtp.nci.nih.gov/webdata.html Web-accessible data and tools, including human tumor cell line assay, yeast assay, COMPARE results http://dtp.nci.nih.gov/repositories.html NCI/DTP repositories (chemicals and biologics as well as natural products and radiolabeled materials) Rapid Access to NCI Discovery Resources http://dtp.nci.nih.gov/docs/rand/rand_index.html (RAND) In vivo testing http://dtp.nci.nih.gov/branches/btb/services.html Compound GMP bulk production, analytical http://dtp.nci.nih.gov/branches/prb/prb_ operations.html method development, formulation, and clinical dosage form manufacturing http://dtp.nci.nih.gov/ptresources.html Pharmacokinetics, metabolism, and toxicology studies, methods for plasma analyses Drug Development Group (DDG) http://dtp.nci.nih.gov/docs/ddg/ddg_descript.html http://dtp.nci.nih.gov/docs/raid/raid_index.html Rapid Access to Intervention Development (RAID) Rapid Access to Preventive Intervention and http://prevention.cancer.gov/programs-resources/ Development (RAPID) programs/rapid/about http://edrn.nci.nih.gov/ Early Detection Research Network (EDRN) program for early cancer detection biomarkers National Cooperative Drug http://dtp.nci.nih.gov/branches/gcob/gcob_web3. Discovery Group (NCDDG) html International Cooperative http://dtp.nci.nih.gov/branches/gcob/gcob_ Biodiversity Groups (ICBG) web14.htmlhttp://www.fic.nih.gov/programs/ research_grants/icbg/index.htm Molecular Target Drug Discovery http://dtp.nci.nih.gov/branches/gcob/gcob_web9. program html BRB grants and contracts for biologics http://web.ncifcrf.gov/research/brb/grants.asp Other grants and contracts for drug http://dtp.nci.nih.gov/branches/gcob/gcob_index. development html

DTP’s activities include screening, preclinical efficacy studies, production of large quantities of agents to support ongoing trials, formulation, and toxicology studies. DTP consists of nine branches, whose missions are outlined in Fig. 23.2. The DTP screening process combines two approaches in agent discovery [12]. The Empirical approach seeks evidence that a compound is toxic to tumor cells

636

A. Papaconstantinou and J.E. Dancey

Fig. 23.2  Developmental Therapeutics Program (DTP): organization and mission of its branches

in  vitro or when growing in animals. In this approach, the agent’s structure, mechanism of action, and relationship between its pharmacologic and pharmacodynamic characteristics is not considered. The Rational approach starts with a particular molecular target in mind. Then, an agent’s structure is optimized based on the ­relation of its pharmacology to its effect on the tumor. In this approach, agents may be set aside if they fall into a recognized class, or if results from the in vitro screen suggest a well-studied mechanism of action. In that case, additional data would be needed to generate enthusiasm for further studies. The NCI has a screening program that allows empirical evidence of antitumor activity to be combined with the knowledge of mechanism of action, facilitating rational use of screening studies. The schema for the screening, which has been used since 2007, is presented in Fig. 23.3 [9]. All screened agents are given an NSC number to assist in their identification. The first step in the screening process, added to the schema in 2007 to increase compound throughput and minimize the time spent on inactive agents, is agent evaluation against 60 human cell lines at a single dose of 10  mM. If an agent is active, it proceeds to step 2, the 60-cell line multidose screen. These in  vitro cell line screens consist of 60 human tumor cell lines, representing leukemia and melanoma as well as breast, lung, colon, brain, ovarian, and prostate and renal cancers [10, 13, 14]. The 60-cell line panel screen has remained relatively constant in methodology with some changes in composition since its inception. As a result, screening data from 10,000’s of compounds have been obtained over time.

23  NCI-Sponsored Clinical Trials

637

Fig. 23.3  Drug Screening Schema of the Developmental Therapeutics Program (DTP)

Results from the dose-dependent 60-cell line screen are translated into clues about the agent’s mechanism of action using a computer-based pattern recognition algorithm, “COMPARE”. Use of this algorithm has allowed NCI to combine the empirical approach with a more targeted outcome to predict mechanism of action of an agent [11, 15]. COMPARE is accessible for free through the DTP Web site [16]. DTP has been storing information about both the patterns of activity of agents as well as the expression of molecular targets from the 60-cell line screen on its Web site [16]. The database contains screening results from about 70,000 agents of which approximately 43,000 are available to the public. In addition, genetic mutations, mRNA, and protein target expression profiles across the 60 cell lines have been characterized. For example, cDNA microarray chips have been employed to analyze the gene expression profile of the 60-cell line assay. RNA data from microarray analyses on untreated cells have been derived by multiple groups on multiple microarray platforms. Early datasets include a cDNA array and an early version of Affymetrix arrays. Newer microarray datasets include Affymetrix U95A arrays performed in triplicate (Novartis), Affymetrix U95A-E (GeneLogic) and Affymetrix U133 arrays (GeneLogic). These arrays provide basal RNA expression on most genes. Additionally, DTP curates a structural searchable dataset. A full list of all the web-accessible data and tools for investigators can be obtained on the DTP Web site. The database allows the mechanism of action of individual agents

638

A. Papaconstantinou and J.E. Dancey

to be predicted based on correlations to activity of other agents with known ­mechanism and to target expression within the cell line screen. Agents selected through this screening process because of their predicted potency, differential activity, or selectivity against a specific disease category, are then evaluated in animal models (steps 3, 4, and 5 in Fig. 23.3). Step 3 comprises acute toxicity testing (one mouse/dose level) to determine the maximum tolerated dose (MTD). Initially, agents were evaluated in in  vivo mouse xenograft models. However, because of the large number of molecules advancing to the development after the 60-cell line in vitro screen, a hollow fiber assay (step 4 in Fig. 23.3) was developed in 1995 and activity in this assay became a prerequisite for testing in the in  vivo xenograft models [12]. In this assay, hollow fibers are prepared from 12 tumor cell lines and implanted into mice. Each mouse receives three intraperitoneal implants and three subcutaneous implants. Starting on day 3 or 4 following the implantation mice are treated intraperitoneally with the agents for 4 days. Two dose levels are examined; the high dose is equal to [MTD × 1.5]/4 and the low dose is equal to the 0.67 × high dose. The fibers are then collected from the mice and subjected to the stable endpoint MTT assay. The hollow fiber assay provides a rapid and efficient method of assessing the activity potential of agents in the in vivo xenograft models. Fifty or more agents can be screened per week in a 10-day assay. Agents that meet the activity criteria of this assay are advanced to step 5 (Fig. 23.3). After these criteria are met, the agent can move into preclinical (Stages IIA and IIB) and clinical development (Stage III) as outlined in Fig. 23.4. Tasks in Stage IIA (early preclinical) include review of the pharmacokinetic/pharmacodynamic data demonstrating in  vivo antitumor activity; initial and dose range-finding toxicology, where evidence that the toxicity of the agent at doses above those that show useful activity, is reversible; and assurance that the agent is available and can be formulated for clinical

Fig. 23.4  Drug Development Group (DDG): Preclinical and Clinical Development Program

23  NCI-Sponsored Clinical Trials

639

use [12, 17]. These tasks are performed by DDG without external review or with BRB-OC involvement for biologics. If an agent successfully meets the criteria of this stage, it moves onto Stage IIB (late preclinical development). This stage involves the review of the agents by the DDG and two extramural experts to ascertain that the candidate agent addresses important scientific and clinical needs. Stage IIB involves IND-directed toxicology studies and the elaboration of a final good manufacturing practice (GMP) batch of agent, and manufacture of formulated and vialed material. Pharmacology and Toxicology studies are overseen by the Toxicology and Pharmacology Branch (TPB). Activities of the TPB include [18]: • Development of methods for drug quantitation in human and mouse, rat, and dog plasma; • Pharmacokinetics, including the elimination of the drug in rodent species, dogs, or nonhuman primates and the establishment of the compound’s plasma concentration or area-under the curve (AUC), correlates with efficacy; • Characterization of the compound’s metabolism in vitro and in vivo; preliminary toxicity studies to establish the MTD and the dose-limiting toxicities (DLTs); • IND-directed toxicity studies which may include single or multiple dose schedules, continuous administration, repeated administration studies, and special studies, such as neurotoxicity, immunotoxicity, etc.; • Bone marrow toxicity studies. The Pharmaceutical Resources Branch (PRB) is responsible for providing highquality substances and formulation of compounds that can be used in preclinical or clinical evaluation. These compounds are to be used either in clinical trials under a CTEP-sponsored IND or are available to academics via RAID (discussed below). The main activities of the PRB include [19]: • Evaluation of the synthetic route of the compound and resynthesis of small or medium batches suitable for use in phase 1 clinical trials (i.e., with a purity of >97% and with major impurities of <1%) under GMP conditions; • Development of analytical methods for the evaluation of the compounds; • Development of formulations that minimize vehicle-related discomfort to the patients while meeting stability criteria; • Manufacturing of clinical dosage forms which can then be distributed by the Pharmaceutical Management Branch (PMB) of CTEP. Production of biopharmaceutical compounds (monoclonal antibodies, growth factors, cytokines) under GMP practices for phase 1 and 2 clinical trials is the responsibility of the Biological Resources Branch’s (BRB) Biopharmaceutical Development Program (BDP) facility [20]. To support the joint DCTD-CCR program, DTP has identified laboratory resources needed, and is working to enhance its capabilities for pharmacodynamic studies, in vivo toxicology studies, virus toxicity testing, and animal model efficacy studies. A National Clinical Target Validation Laboratory (NCTVL) was established in DTP’s Toxicology and Pharmacology Branch to assess the pharmacodynamic effects of therapeutics on cellular targets, perform target validation studies, and evaluate the

640

A. Papaconstantinou and J.E. Dancey

effects of the anticancer agents in the early clinical trials [7]. Novel methodologies are employed with the goal of demonstrating the therapeutic effect of small molecule anticancer agents on specific molecular pathways of interest. NCTVL uses patient specimens, collected in advance of patient entry into clinical trials to develop qualitycontrolled methodologies for correlative clinical studies that assess anticancer agent efficacy. In addition, resources from NCTVL are available to all extramural investigators, including those with trials that are part of the clinical centers, Specialized Programs of Research Excellence (SPOREs), or the cooperative groups.

23.2.3 Programs to Assist Academics and Industry in Preclinical Drug Development Efforts With the advancement of bioinformatics, molecular biology and chemistry, and the development of high-throughput screening methods, the preclinical screening process for active cancer agents has been completely transformed. Oftentimes, researchers cannot afford to invest in equipment that allows them to perform the preclinical screening. In an effort to assist academics and private sector investigators in the discovery and development of anticancer agents, the NCI has developed a variety of programs. 23.2.3.1 Rapid Access to NCI Discovery Resources Program The Rapid Access to NCI Discovery Resources (RAND) program initiated in 2001 assists academics and nonprofit investigators in the discovery of anticancer agents by providing services in the following areas [21]: • Recombinant production and characterization of molecular target proteins • Development and implementation of target-based high-throughput screening assays • Computer-assisted agent design • Synthesis of chemical analogs for library generation, structure-activity studies, and lead optimization • Bioassay-directed natural product isolation and characterization • Early formulation, as well as pharmacokinetic, pharmacology, and toxicology studies (establishment of efficacious and maximum tolerated doses). RAND is not a grant program. Rather researchers may submit applications twice a year, on April 1, and October 1 to access services that take place either in DTP contract laboratories or in-house laboratories. The applicant submits a letter of intent (LOI) via e-mail prior to the deadline. The LOI should include the hypothesis and the outline of the services to be requested. The application should include the rationale for the agent to be tested or the strategy to be used as well as the ­hypothesis and requested services. The applicant does not need to estimate the costs of the services;

23  NCI-Sponsored Clinical Trials

641

this is a central function of staff in the RAND review process. The ­application is reviewed by a panel of experts from academia and industry. NCI staff can advise the panel but cannot vote. Applications are reviewed based on the strength of the hypothesis, originality of agent or method, and the cost/benefit consideration. Once a compound has been selected and prescreened further preclinical development is facilitated by the Rapid Access to Intervention Development (RAID) program (discussed below). Normally, NCI does not acquire intellectual property rights to inventions made by its employees with research materials under RAND. If an NCI contractor is in a position to file an invention report and elects to retain rights under the Bayh-Dole Act, the contractor will, as provided by their contract, offer the principal investigator a first option to negotiate a license to the invention.

23.2.3.2 Rapid Access to Intervention Development Program The RAID program, similarly to the previously mentioned RAND program, is not a grant application mechanism but rather a resource program for academic and nonprofit investigators [22]. This program allows investigators from academia or nonprofit organizations that want to conduct their own clinical trials to take advantage of the resources for preclinical development available from the DTP. Applications are accepted twice a year, February 1 and August 1, and are reviewed by a panel of experts from academia. The application is judged based on the strength of the scientific hypothesis, originality of agent or method, and the cost/ benefit consideration. The components of the applications and the application process are described at length on the RAID Web site at . If an optimal compound has been selected via the RAND or another discovery path, RAID facilitates further preclinical development. Similarly to RAND, RAID services are provided free of charge. The goal of RAID is to provide “clinical proof of principle that a new molecule or approach is a viable candidate for expanded clinical evaluation.” Services supported by RAID include but are not limited to: • • • • • • • • • • •

Development of pharmacology studies Performance of pharmacology studies with a predetermined assay Establishment or optimization of dose and schedule for in vivo activity Acquisition of bulk substance (GMP or non-GMP) Large-scale synthesis and formulation of substance Development of analytical methods for bulk substances Production and stability assurance of dosage forms Range-finding initial toxicology studies IND-directed toxicology with correlative pharmacology and histopathology Planning of clinical trials Regulatory affairs to satisfy FDA requirements by the investigators seeking to test novel substances in the clinic • IND filing advice.

642

A. Papaconstantinou and J.E. Dancey

The output of RAID activities are made fully available to the originating investigator for support of an IND application and clinical trials. RAID is not meant to assist industry in the development of anticancer agents. Industrial collaborations with NCI for preclinical agent development are made available mainly through DDG. However, RAID applicants collaborating with a corporation of any size remain eligible, as long as the agent in development is not licensed by the company. Furthermore, a RAID application can be for a licensed product of the academic laboratory, as long as the licensee is a Small Business Innovation Research (SBIR)-defined “small company” with fewer than 500 employees. Since its inception in 1998, the RAID program has approved 104 projects, through which 13 small molecules and 11 biological agents later entered clinical trials.

23.2.3.3 National Cooperative Drug Discovery Group Program DTP’s National Cooperative Drug Discovery Group (NCDDG) program was initiated in 1983 [23]. The program supports multidisciplinary and innovative approaches to the discovery of novel synthetic and natural-source derived anticancer drugs by partnering NCI-funded academic researchers with private sector organizations to fund cooperative agreements (U19s). The NCDDG program has assisted in the development of four FDA-approved anticancer agents: topotecan, NSC 609699; Gliadel Wafers, NSC 714372; denileukin diftitox, NSC 733971; and cetuximab, NSC 714692.

23.2.3.4 International Cooperative Biodiversity Groups Program Funding under the International Cooperative Biodiversity Groups (ICBG) program was first made available in 1992 [24]. Funding is provided to groups that are conducting research on the discovery of natural products with anticancer potential, and are focused on conserving the biodiversity and sustaining economic growth of the countries that have these natural-source drugs. Natural products make up 60–65% of all anticancer drug products. Awards have been made for research conducted in nine different countries in Latin America, Africa, Southeast and Central Asia, and the Pacific Islands. Discovery of drugs from natural sources under this program is not limited to anticancer drugs. Because of the nature of this program, funding is provided by nine different components of the NIH as well as the Biological Sciences Directorate of the National Science Foundation, and the Foreign Agriculture Service and Forest Service of the U.S. Department of Agriculture.

23.2.3.5 Rapid Access to Preventive Intervention and Development Program Similarly, to RAND and RAID, Rapid Access to Preventive Intervention and Development (RAPID) is not a grant program, but rather a mechanism that makes

23  NCI-Sponsored Clinical Trials

643

resources from the NCI’s DCP available to academics for preclinical and early clinical preventive drug development [25]. RAPID’s objectives are to expedite the advancement of novel compounds to phase 1 trials; provide preclinical and clinical development requirements of phase 1 trials, to allow advancement to phase 2 trials. It fulfills its objectives by providing the following services: • In vitro and in vivo preclinical pharmacology and efficacy studies • Acquisition of bulk substance (GMP and non-GMP); analytical methods for agent • Formulation of substance, production of dosage form, large-scale synthesis • Stability testing program for dosage forms • IND-directed toxicology testing • Pharmacokinetics and safety studies of early phase 1 clinical trials • Consultation for planning of clinical trials • Consultation for regulatory affairs and IND filing An easy to use listing of NCI resources available for discovery and development of anticancer therapies or chemoprevention along with Web links is provided in the table below. These resources include programs, grants, and contract mechanisms.

23.3 Clinical Therapeutics Development and Resources Through its programs and new initiatives NCI provides an array of resources for the development of screening, diagnostic, prevention, treatment, quality of life, and imaging trials. Screening and prevention trials are overseen by the DCP [26]. DCTD oversees the treatment, diagnostic, and imaging clinical trials. The NCIsupported clinical trials take place either intramurally at the NIH Clinical Center in Bethesda, Maryland, or extramurally at any of the hundreds of academic or private hospitals, cancer centers, or community-based medical practices located in the USA, Puerto Rico, Canada, and Europe [27]. Today, the majority of NCI funding is dedicated to trials of its extramural partners. Hundreds of clinical trials are supported each year through a variety of mechanisms that include: Single Research Project Grants, Program Project Grants, cooperative agreements, and contracts. The major components of NCI’s extramural clinical research program include: • NCI-designated Cancer Centers and Comprehensive Cancer Centers which are major academic and research institutions that have interdisciplinary programs in cancer research [28] • SPOREs, which bring together basic scientists and clinical researchers to design and implement research programs that may impact cancer prevention, diagnosis, detection, and treatment [29] • Clinical Trials Cooperative Groups, which are networks of research institutions organized according to region or medical specialty that collaborate to conduct large-scale, multisite clinical trials in cancer treatment, prevention, and quality of life often involving thousands of patients [30]

644

A. Papaconstantinou and J.E. Dancey

• Community Clinical Oncology Programs (CCOPs), which provide small-scale community-based medical facilities and physicians with opportunities to participate in cancer prevention and treatment clinical trials [31]. These programs are interrelated since for example every Cancer Center is a participant in at least one Cooperative Group, and the Cooperative Groups are research bases for participants in CCOPs. Within DCTD, the Cancer Diagnosis Program (CDP) not only oversees clinical diagnostic trials, but also coordinates and funds resources and research on diagnostics tests and technologies [32]. CDP has developed the Program for Assessment of Clinical Cancer Tests (PACCT) to expedite the incorporation of new technologies and understanding of cancer biology into clinical practice [33]. DCTD’s Cancer Imaging Program (CIP) sponsors clinical imaging trials, and works for the advancement of imaging technologies that will enhance diagnosis and treatment options for patients [34]. CIP provides grants that foster industry-academic partnerships for the development of biomedical imaging systems and methods to improve early detection, screening, diagnosis of cancer, and assess response to therapy [35]. Furthermore, CIP supports eight In Vivo Cellular and Molecular Imaging Centers (ICMICs) that can provide cutting-edge research in cancer imaging [36]. Radiation Research Program (RRP) supports clinical, translational, and basic research to improve radiation therapy and technologies. Cancer treatment trials are sponsored by the CTEP.

23.3.1 The Cancer Therapy Evaluation Program: Organization and Structure CTEP is responsible for administering, coordinating, and funding clinical trials, as well as sponsoring other clinical research. The program fosters collaborations within the cancer research community and works extensively with the pharmaceutical and biotechnology industries. The CTEP plans, reviews, and coordinates clinical trials for investigational new agents, through the formulation of a clinical development plan and the preparation and submission of an IND application for each agent [37]. CTEP is the liaison to the FDA for both extramural clinical researchers as well as industry collaborators for trials conducted under a CTEP IND. CTEP manages, tracks, and reviews clinical protocols and monitors and ensures regulatory compliance of the clinical trials. CTEP also coordinates the distribution of investigational agents from industry collaborators for use in NCIsponsored clinical trials. As of April 2008, CTEP held INDs for about 110 agents and was sponsoring about 750 trials in approximately 2,000 institutions with 10,000 investigators–collaborators. CTEP funds institutions to perform early clinical trials through multiyear U01 cooperative agreements (referred to as grants) for phase 1 trials, and N01 contracts for phase 2 trials and to perform later stage trials through the Clinical Trials cooperative Group Program through U10 cooperative

23  NCI-Sponsored Clinical Trials

645

Fig. 23.5  Cancer Therapy Evaluation Program (CTEP): organization and mission of its branches. CTA clinical trials agreement, CRADA cooperative research and development agreement, CDA confidential disclosure agreement, MTA material transfer agreement

agreements. CTEP also has multiple collaborative agreements with industry ­partners for the codevelopment of promising new agents. CTEP currently has 38 active Clinical Trials Agreements (CTAs) and 32 Cooperative Research and Development Agreements (CRADAs) as well as 22 Clinical Supply Agreements (CSAs). These agreements define the interactions between NCI, its industry collaborators, and the clinical trials groups, whether it is the cooperative groups or the comprehensive cancer centers, for the conduct of clinical trials. The goals of industry collaborations are to engage CTEP’s assistance to expedite studies leading to commercialization of the agent and to facilitate investigations of relevant scientific questions about the anticancer agent. CTEP’s clinical trial functions are handled by its six branches and by the Protocol Information Office (Fig.  23.5) as described below. 23.3.1.1 Clinical Grants and Contracts Branch The Clinical Grants and Contracts Branch (CGCB) manages the Clinical Oncology, Surgical Oncology, and the Cancer Nutrition grant programs [38]. These programs include grants and cooperative agreements for clinical trials of chemotherapeutic or biological agents, trials using surgical procedures to prevent, diagnose, stage, treat

646

A. Papaconstantinou and J.E. Dancey

cancer, and trials that assess nutritional status of the cancer patient and study its relevance to the disease. The AIDS Malignancy Program is also administratively monitored by CGCB. Furthermore, CGCB participates in the planning, implementation and management of the clinical contracts program, the Clinical Trials Cooperative Groups, and other clinical consortia groups.

23.3.1.2 Clinical Investigations Branch Responsibilities of the Clinical Investigations Branch (CIB) include scientific and administrative coordination of the NCI-sponsored Clinical Trials Cooperative Group Program [39]. CIB is divided into Surgery, Medicine, and Pediatrics groups with further divisions based on disease types. CIB organizes, develops, and reviews programs to compare the effectiveness of specific types and methods of cancer therapy, including curative and adjuvant therapy; facilitates and organizes clinical research studying the role of surgical and radiation therapy; collaborates with other CTEP branches and NCI staff on disease-oriented issues related to clinical trials sponsored by DCTD; oversees and manages the collection, storage, and use of clinical specimens, including tumor specimens; serves as the liaison with international trials organizations and coordinates planning of phase 3 trials with these organizations and collaborates with the Clinical Trials Cooperative Groups, consortia, cancer centers, and other entities for the development of research standards and harmonization of research methodologies.

23.3.1.3 Clinical Trials Monitoring Branch The Clinical Trials Monitoring Branch (CTMB) is responsible for setting the guidelines and standards for conducting clinical trials that provide good data quality and are in compliance with regulatory requirements for clinical research [40]. This branch is also responsible for on-site monitoring of all CTEP-sponsored clinical trials, as well as monitoring select DCP-sponsored cancer prevention trials.

23.3.1.4 Investigational Drug Branch The Investigational Drug Branch (IDB) implements and monitors a comprehensive cancer therapy clinical program that sponsors clinical trials of investigational anticancer agents [41]. This branch develops, implements, and oversees all aspects of development for chemotherapeutics and biological agents under CTEP sponsorship, including phase 1, 2, and 3 trials and translational research IDB physicians are responsible for monitoring the clinical trials for safety, efficacy, clinical pharmacology, for preparing reports of adverse events (AEs) for all INDs and BLAs and for providing FDA with annual reports.

23  NCI-Sponsored Clinical Trials

647

23.3.1.5 Pharmaceutical Management Branch Responsibilities of the PMB include authorization and distribution of all CTEPsponsored IND agents to eligible investigators; management of availability of the agent; provision of pharmaceutical information; coordination, authorization, and processing of all requests for Special Exception and Treatment Referral Center (TRC) IND agent use; registration and maintenance of records for all investigators participating in CTEP-sponsored clinical trials; provision and management of high priority double-blind randomized clinical trials; and distribution of agents for preclinical use [42]. 23.3.1.6 Regulatory Affairs Branch Regulatory Affairs Branch (RAB) is composed of the Drug Regulatory Affairs Section and the Agreement Coordination Group [43]. Responsibilities of this branch include preparation and submission of INDs to the FDA to initiate clinical trials with anticancer agents and compliance with FDA regulations. The Agreement Coordination Group is responsible for developing, negotiating, and implementing research and confidentiality. 23.3.1.7 Protocol and Information Office Protocol and Information Office’s (PIO) mission is to coordinate all administrative aspects related to clinical trial development so that clinical trial protocols are developed in an efficient and timely manner [44]. PIO collects, processes, tracks, and monitors all letters of intent (LOIs) and protocol-related information between CTEP and its extramural collaborators.

23.3.2 CTEP-Sponsored Clinical Development of Investigational Agents and Resources Once an agent is approved for clinical development by the DDG or the JDC, the clinical trial development process begins (Stage III, Fig. 23.4). CTEP coordinates the establishment of the clinical development plan, and the preparation and submission of an IND for each investigational agent. Development plans for new agents are usually a collaborative effort between DCTD and the industry collaborators. Companies wishing for collaboration with NCI for the sponsorship of clinical trials must enter into a collaborative agreement. There are two main types of agreements, CTAs and CRADAs. CTAs are not covered by congressional legislation and have no funding associated with them, but address the intellectual property disposition if a patent is filed as a result of the clinical studies [45]. CRADAs are more elaborate

648

A. Papaconstantinou and J.E. Dancey

agreements than CTAs and offer the collaborator the broadest range of studies that can be conducted under the research plan from preclinical laboratory studies through postmarketing clinical trials. Under this agreement, there is also the option of exclusive or nonexclusive licensing of intellectual property arising from the CRADA. Once a collaborative agreement is established, CTEP focuses on the clinical development plan and the initiation of clinical trials as described below. 23.3.2.1 Clinical Trials Solicitations and Letters of Intent for Early Clinical Trials Development of the clinical plan is mainly the responsibility of the IDB senior investigators [45]. Input, both scientific and clinical, for the design and prioritization of phase 1 and 2 trials, is provided by the recently formed IDSC. Following the formulation and finalization of a collaborative agreement, the IDB prepares agentspecific solicitations for clinical investigators requesting the submission of protocol LOIs to conduct specific clinical trials within the scope of the development plan. Solicitations are sent to principal investigators of phase 1 U01 grants, phase 2 N01 contracts, Cancer Centers, Cooperative Groups, to qualified investigators with unique expertise relevant to the agent or the study population, to NCI-funded disease-specific consortia, and the pediatric cooperative group and consortium. The purpose of the LOI is to declare the investigator’s interest in conducting a phase 1 or 2 trial with a specific investigational agent and for a specific disease. The LOI procedure was developed to maximize the efficiency and fairness of study selection without requiring investigators to prepare a lengthy protocol that may or may not be approved. Approval of an LOI by CTEP and, if relevant, its industry collaborator, is followed by the provision of documents to the investigator to assist with protocol development. The rules for submission of LOIs are as follows: • Agents at the beginning of development (early phase 1): In advance of IND filing, CTEP will announce availability of the agent, issue a request for proposals for phase 1 studies, and provide a deadline for submission • Agents at the beginning of phase 2 studies: In late phase 1, CTEP will issue a request for proposals for initial phase 2 trials, including a deadline for submission • Unsolicited LOIs may be submitted by investigators wanting to address novel hypotheses that may be supported by published or unpublished results at any time. LOIs are reviewed by the IDB Staff and by CTEP Protocol Review Committee (PRC). Review criteria and prioritization are based on the strength of scientific rationale, study design, and feasibility of conducting trial, whether they fit within the development plans of CTEP and its industry collaborator and whether they are nonduplicative of ongoing or planned clinical trials. A consensus review detailing the CTEP decision and comments on the proposed study is sent to the principal investigator within 30 days of submission of the LOI. Following approval by CTEP and its industry collaborator, documents to assist in protocol preparation, such as protocol templates, investigator brochure, and drug-specific ­pharmaceutical

23  NCI-Sponsored Clinical Trials

649

and safety information are sent to Principal Investigator who has 30–60 days to respond and submit a protocol. On average, each year CTEP receives approximately 400–500 LOIs, of which one third are approved. 23.3.2.2 Protocol Submission and Review Protocol submissions are sent directly to the PIO, which upon receipt assigns it an NCI protocol number. Within approximately 2 weeks of receipt, protocols submitted to CTEP are evaluated by the CTEP PRC [45]. PRC is composed of CTEP staff, additional staff from other NCI divisions, and is chaired by the Associate Director, CTEP. It meets weekly and usually reviews 10–20 protocols, LOIs, and concepts for each session. Each initial draft protocol is assigned a minimum of five reviewers; the number may be higher for complicated multimodality protocols. The draft protocol is also sent to the industry collaborator for review. At a minimum, the protocol and the informed consent are reviewed by two physicians, a biostatistician, pharmacist, and regulatory affairs and clinical trial monitoring branch staff to ensure that all scientific, safety, trial design, regulatory and administrative issues are assessed. The PRC discusses the protocol after hearing the views of all the reviewers and makes a decision that based on its science and safety the study is: Accepted as Written; Accepted with Recommendations; Acceptance deferred pending revisions; or Disapproved. After the PRC meeting, the primary reviewer generates a consensus review, which states the concerns of all members of the PRC. The consensus review is sent to the protocol source within 30 days of receipt. Revisions by the investigator, if required, must be sent back to PIO and then distributed to the primary CTEP reviewers for their reassessment. The consensus reviewer can then choose whether to send the revised protocol back to the full PRC for rereview or approval, approve with recommendations or disapprove the revised protocol without PRC rereview. Final approved versions of the protocol are sent to the industry collaborator at the same time that they are submitted to the FDA. The PMB of CTEP is responsible for registering all the investigators participating in NCI-sponsored clinical trials. Any amendments to approved protocol documents are submitted to the PIO and must be reviewed by CTEP staff. Study status updates must also be reported to CTEP via PIO. 23.3.2.3 IND Submission To conduct clinical trials with experimental agents, any organization must submit an IND to the FDA [45]. Within the CTEP, it is the responsibility of RAB to prepare and submit the IND. The IND provides the experimental rationale for human testing, including results of animal toxicology studies, manufacturing data, purity and stability information, and an initial plan of clinical investigation. All information in the NCI IND is made fully and exclusively available to the collaborator following the execution of the appropriate collaborative agreements.

650

A. Papaconstantinou and J.E. Dancey

23.3.2.4 Data Reporting Investigators conducting studies with NCI IND agents are responsible for the timely and accurate reporting of data from the trials to CTEP [45]. The information informs CTEP of the progress and development of the agent. Two types of data are required: individual patient demographic, treatment, adverse event and outcome data, and study summaries. In addition to the above-mentioned requirements for reporting, investigators must report data to the Clinical Trials Monitoring Service (CTMS), and/or Clinical Data Update System (CDUS) according to the study type and category of sponsorship. CTMS is required for nearly all early phase 1 studies [45]. For each patient on trial, data are recorded on the DCTD Case Report Forms or on its electronic version and the forms are submitted biweekly to CTMS. This submission must include case report updates for all patients actively on the study and data for all new patients, in addition to the physician’s judgment on whether the medical events in the patient’s course were agent-induced. All evaluations regarding adverse events must be reported using the DCTD Common Terminology Criteria for Adverse Events (CTCAE) [46]. The CDUS is the primary clinical trial data resources of DCTD and DCP. There are two CDUS data sets: the abbreviated CDUS, which requires quarterly submission of protocol administrative information and patient-specific demographic data; and the complete CDUS which also contains patient administrative, treatment, adverse event, and response information. Reporting requirements can be found on CTEP’s Web site (http://ctep.cancer.gov/protocolDevelopment/ default.htm). 23.3.2.5 Safety Data Reporting Each investigator conducting a study with an agent supplied by DCTD is required to report all adverse events to CTEP (for guidelines see http://ctep.cancer.gov/ protocolDevelopment/default.htm#adverse_events_adeers). Routine reporting occurs through CTMS or CDUS as outlined previously. Investigators using an agent under a CTEP IND are also responsible for identifying if an AE requires expedited reporting to CTEP via the Adverse Event Expedited Reporting System (AdEERS). For expedited AEs, the report is reviewed by an IDB physician and may result in the request for additional information. Each AE report is classified according to its drug-related attribution and the patient’s underlying disease. Based on the assessment, a decision is made concerning the need for further action. For example, findings on an AE report may indicate that the safety of patients enrolled in ongoing trials is affected. In this case, CTEP must take steps to notify the investigator community, the FDA and the collaborating pharmaceutical company, simultaneously. Other actions taken may include a request to update the protocol and informed consents to include the new risk, or altering existing research by modifying the protocols or suspending one or more trials.

23  NCI-Sponsored Clinical Trials

651

23.3.2.6 Data and Safety Monitoring It is the policy of the NIH that each Institute and Center (IC) has a plan that ­provides oversight and review of the conduct of the research, safety and efficacy data, and progress toward achieving the goals of the study [47]. The data and safety monitoring (DSM) plans consist of plans for monitoring the progress of the trial and the safety of the patients; plans to assure compliance in reporting the AEs; plans to report to the NCI grant program any action resulting in the trial being temporarily closed or permanently closed; and plans for assuring accuracy in data reporting and protocol compliance. In general, clinical trial DSM activities are performed by boards (DSMBs) or data monitoring committees (DMCs) composed of experts relevant to the study that regularly assess the trial and offer recommendations about continuation of the study to the institution, awardee, sponsor, or institute. DMCs are responsible for monitoring all phase 3 trials of the cooperative groups and all randomized phase 2 and 3 studies in cancer prevention of the cooperative groups which are funded as CCOP Research Bases [48]. DMC members are appointed for a fixed term from the Group Chair and are approved by the CTEP Associate Director. In addition, for agents with a CTEP IND, CTEP IDB physicians review all routine AE and AdEERS submissions and CTEP has an on-site audit program which evaluates protocol compliance. During this visit, the auditors review administrative information about data accuracy and management, patient medical records, study flow sheets that are signed by the physician examining the patient, and data about the protocol development. The auditors also verify that the patient signed the most current Institutional Review Board (IRB)-approved version of the informed consent; verify all the patients’ eligibility and compliance with the protocol; and verify completeness of the documentation. Audits are randomly timed, but occur about once every 3 years (except for the CTMS-monitored phase 1 trials, which are monitored more often). Major problems needing action are immediately conveyed to the group chair (for the Cooperative groups), Cancer Center director (or institution equivalent); and CTEP. The development of a national cancer clinical trials information infrastructure to facilitate electronic clinical research data management is underway. NCI’s cancer Bioinformatics Grid (caBIG) will be used as the interface with the goal of improving cost effectiveness and comparability of results across trials and centers [49]. The caBIG Clinical Trials Suite, a software package, is being developed and will contain tools that enable • Management of Adverse Events [Cancer Adverse Event Reporting System (caAERS)] • Exchange of clinical data [(Cancer Data Exchange system (caXchange)] • Schedule and management of a participant’s treatment and care events on a ­calendar [Patient Study Calendar (PSC)] • Storing, viewing, and sharing of clinical trials data [Clinical Trials Object Data System (CTODS)] • Connection to all caBIG compatible systems [caGrid] • Integration with Clinical Trials Data Management Systems (CDMS)

652

A. Papaconstantinou and J.E. Dancey

An easy to use listing of NCI resources available for clinical trial development of anticancer therapies or chemoprevention along with Web links is provided in Table 23.2. These resources include grants, contract mechanisms, services, guides, and forms.

23.3.3 Phase 3 Treatment Trials For phase 3 treatment studies, CTEP requests that investigators submit a written concept for the proposed trial [45]. Concepts describe the proposed study, including the hypothesis to be investigated, its rationale, and study design. Phase 3 treatment trial concepts are usually submitted by physicians of the Cooperative Groups registered with CTEP that have submitted a signed FDA Form 1572. Because of the large patient sample size required for these trials, the phase 3 trials are usually conducted by Cooperative Groups. Under the new initiatives of the CTWG, the disease-specific steering committees are responsible for the evaluation and prioritization of phase 3 trial concepts [5]. These committees now evaluate phase 3 trials concepts in place of the CTEP PRC. The membership of these committees represent the various stakeholders in the oncology community in the specific disease areas, including representatives from NCI, and investigators from the Cooperative Groups, CCOPs, Cancer Centers, SPOREs, and the patient advocacy community, as well as investigators with expertise in translational science, drug development, and cancer symptom management. These committees are also responsible for organizing periodic State-of-the-Science meetings to identify areas in need of studies, strategies and concepts to test, and to facilitate innovation. Based on these meetings, the steering committees develop strategic priorities for future phase 3 trials and share the ideas with the relevant oncology centers and institutions. The Clinical Trials Cooperative Group Program consists of 12 Cooperative Groups funded mainly by DCTD. Many of the Cooperative Groups have a specialty area (e.g., pediatrics, gynecologic oncology). Others are multimodality and cover a broad range of activities and some are single modality (e.g., radiation therapy). A listing of these groups along with Web links for additional information is presented in Table 23.3. Certain cooperative groups have entered into collaborations and formed intergroups that focus on specific diseases. One example of an international intergroup is the Gynecologic Cancer Intergroup (GCIG) which was formed by the European Organization of Research and Cancer Gynecological Cooperative Group (EORTC GCCG), the NCIC-CTG, the Nordic Society of Gynecologic Oncology (NSGO), and the Scottish Gynaecological Cancer Trials Group (SGCTG), but has expanded to include 16 international Cooperative Groups [50]. There is growing interest from the Cooperative Groups to conduct clinical trials in collaboration with International Institutions. To that effect, CTEP has worked with the Office for Human Research Protections and the Food and Drug Administration

Resources for investigators Investigator’s Handbook (a manual for participants of CTEP-sponsored clinical trials) Investigator Registration: forms and policies Funding opportunities (RFAs, PAs, RFPs) NIH grants (opportunities and assistance) Agents/Drugs: Policy and guidelines Reporting guidelines (Adverse Events, AdEERs, data, etc.) Monitoring guidelines

Clinical trials programs Clinical Trials Cooperative Group Program Community Clinical Oncology Program (CCOP) Specialized Programs of Research Excellence (SPOREs) Cancer Centers Program (includes Comprehensive Cancer Centers)

Table 23.2  Clinical trials resources Resource Clinical trials sources Clinical Trials Support Unit (CTSU): An initiative sponsored by NCI to support phase 3 trials NCI’s PDQ database of cancer trials Cancer Information Service (CIS): A source of cancer information for the public and health professionals Listing of NIH’s clinical trials and instructions for investigators wishing to register their trials on this database

http://ctep.cancer.gov/protocolDevelopment/default.htm#monitoring (continued)

http://ctep.cancer.gov/investigatorResources/default.htm#investigator_registration http://ctep.cancer.gov/investigatorResources/default.htm#funding_opps http://grants.gov/ http://ctep.cancer.gov/protocolDevelopment/default.htm#agents_drugs http://ctep.cancer.gov/protocolDevelopment/default.htm#adverse_events_adeers

http://ctep.cancer.gov/investigatorResources/default.htm#investigators_handbook

http://cancercenters.cancer.gov/cancer_centers/index.html

http://ctep.cancer.gov/industryCollaborations/default.htm#collaborations http://prevention.cancer.gov/programs-resources/programs/ccop http://spores.nci.nih.gov

http://clinicaltrials.gov/

http://www.cancer.gov/clinicaltrials/search http://cis.nci.nih.gov/

http://www.ctsu.org/

Web site

23  NCI-Sponsored Clinical Trials 653

Guidelines and tools for protocol development Letter of Intent: forms, guidelines, recent solicitations Concept submission form CTEP Review Types and Decision Tree List of Codes and Values (Disease, CTCAE, MedDRA codes; NSC numbers, Cooperative Group and Institution Codes) Protocol Templates, Applications and Guidelines Protocol Authoring Handbook Guidelines for agreements, correlative studies, slow accrual for trials Model agreements (CTAs, CRADAs, CDAs, MTAs)

Table 23.2  (continued) Resource

http://ctep.cancer.gov/protocolDevelopment/templates_applications.htm http://ctep.cancer.gov/protocolDevelopment/default.htm http://ctep.cancer.gov/protocolDevelopment/default.htm#protocol_development http://ctep.cancer.gov/industryCollaborations2/model_agreements.htm

http://ctep.cancer.gov/protocolDevelopment/default.htm#lois_concepts http://ctep.cancer.gov/protocolDevelopment/default.htm#lois_concepts http://ctep.cancer.gov/protocolDevelopment/default.htm http://ctep.cancer.gov/protocolDevelopment/codes_values.htm

Web site

654 A. Papaconstantinou and J.E. Dancey

23  NCI-Sponsored Clinical Trials Table 23.3  DCTD-Sponsored Cooperative Groups Cooperative Group ACOSOG: American College of Surgeons Oncology Group ACRIN: American College of Radiology Imaging Network CALGB: Cancer and Leukemia Group B COG: Children’s Oncology Group ECOG: Eastern Cooperative Oncology Group EORTC: European Organization for Research and Treatment of Cancer GOG: Gynecologic Oncology Group NCCTG: North Central Cancer Treatment Group NCIC CTG: National Cancer Institute of Canada Clinical Trials Group NSABP: National Surgical Adjuvant Breast and Bowel Project RTOG: Radiation Therapy Oncology Group SWOG: Southwest Oncology Group

655

Website https://www.acosog.org/ http://www.acrin.org/ http://www.calgb.org/ http://www.childrensoncologygroup.org/ http://ecog.dfci.harvard.edu/ http://www.eortc.be/default.htm http://www.gog.org/ http://ncctg.mayo.edu/ http://www.ctg.queensu.ca/ http://www.nsabp.pitt.edu/ http://www.rtog.org/ http://www.swog.org/

and has developed a set of guidelines for these collaborations [51]. NCI maintains an international liaison office in Brussels, Belgium, which assist in the coordination of collaborative clinical trials between NCI and clinical trial centers in Europe.

23.4 Conclusion Since its inception, the NCI has supported preclinical studies and clinical trials in the areas of cancer prevention, detection, and treatments, as well as studies which seek to enhance the understanding of basic cancer biology. As described above, a broad variety of NCI initiatives and programs are available for interested laboratory, translational, and clinical investigators within academia and industry to assist their research activities with the goal to reduce cancer morbidity and mortality.

References 1. The National Cancer Institute: More Than 70 Years of Excellence in Cancer Research Web site. http://www.cancer.gov/aboutnci/excellence-in-research. Accessed April 10, 2008 2. National Cancer Institute FactSheet, Clinical Trials Questions and Answers Web site. http:// www.cancer.gov/cancertopics/factsheet/Information/clinical-trials. Accessed March 31, 2008 3. PDQ® – NCI’s Comprehensive Cancer Database Web site. http://www.cancer.gov/cancertopics/ pdq/cancerdatabase#clinical_trial. Accessed March 31, 2008 4. National Cancer Institute, Division of Cancer Treatment and Diagnosis, DCTD Program Accomplishments 2006 Web site. http://www.dctd.cancer.gov. Accessed March 31, 2008

656

A. Papaconstantinou and J.E. Dancey

5. Clinical Trials Working Group, National Cancer Institute Web site. http://integratedtrials.nci. nih.gov/. Accessed April 10, 2008 6. Translational Research Group, National Cancer Institute Web site. http://www.cancer.gov/ trwg. Accessed April 16, 2008 7. Division of Cancer Treatment and Diagnosis: Developmental Therapeutics Program, New Initiatives Web site. http://dctd.cancer.gov/ProgramPages/DTP-ongoinginitiatives.htm. Accessed April 10, 2008 8. NCI Drug Development Group (DDG) Web site. http://www.dtp.nci.nih.gov/docs/ddg/ddg_ descript.html. Accessed April 10, 2008 9. Developmental Therapeutics Program NCI/NIH, Discovery Services, Screening Web site. http://www.dtp.nci.nih.gov/screening.html. Accessed April 10, 2008 10. Monga M, Sausville EA: Developmental therapeutics program at the NCI: molecular target and drug discovery process. Leukemia 16:520–6, 2002 11. Collins JM: The NCI Developmental Therapeutics Program. Clin Adv Hematol Oncol 4:271–3, 2006 12. Sausville EA, Johnson JI: History of the National Cancer Institute Drug Discovery Program, in Budman DR, Calvert AH, Rowinsky EK (eds): Handbook of Anticancer Drug Development. Philadelphia, Lippincott Williams and Wilkins, 2003, pp 25–42 13. Holbeck SL: Update on NCI in vitro drug screen utilities. Eur J Cancer 40:785–93, 2004 14. Shoemaker RH: The NCI60 human tumour cell line anticancer drug screen. Nat Rev Cancer 6:813–23, 2006 15. Zaharevitz DW, Holbeck SL, Bowerman C, et al: COMPARE: a web accessible tool for investigating mechanisms of cell growth inhibition. J Mol Graph Model 20:297–303, 2002 16. Developmental Therapeutics Program, Discovery Services: Web-accessible Data and Tools Web site. http://dtp.nci.nih.gov/webdata.html. Accessed April 10, 2008 17. Takimoto CH: Anticancer drug development at the US National Cancer Institute. Cancer Chemother Pharmacol 52 Suppl 1:S29–33, 2003 18. Developmental Therapeutics Program, Toxicology and Pharmacology Branch Drug Development Process and Procedures Web site. http://www.dtp.nci.nih.gov/branches/tpb/ process.html. Accessed April 22, 2008 19. Developmental Therapeutics Program, Pharmaceutical Resources Branch Web site. http:// www.dtp.nci.nih.gov/branches/prb. Accessed April 22, 2008 20. Developmental Therapeutics Program, Biological Resources Branch Web site. http://web. ncifcrf.gov/research/brb. Accessed April 22, 2008 21. National Cancer Institute DTP: Rapid Access to NCI Discovery Resources (R·A·N·D) Web site. http://dtp.nci.nih.gov/docs/rand/rand_index.html 22. Developmental Therapeutics Program, Rapid Access to Intervention Development (RAID) Web site. http://www.dtp.nci.nih.gov/docs/raid/raid_index.html. Accessed April 10, 2008 23. Grants and Contracts Operations Branch (GCOB), National Cooperative Drug Discovery Groups Web site. http://dtp.nci.nih.gov/branches/gcob/gcob_web3.html. Accessed April 22, 2008 24. National Institutes of Health, International Cooperative Biodiversity Groups (ICBG) (2007) Web site. http://www.fic.nih.gov/programs/research_grants/icbg/index.htm. Accessed April 10, 2008 25. Division of Cancer Prevention, Rapid Access to Preventive Intervention Development (RAPID) Program Web site. http://prevention.cancer.gov/programs-resources/programs/ rapid/about. Accessed April 18, 2008 26. Division of Cancer Prevention, History and Mission Web site. http://prevention.cancer.gov/ about/mission. Accessed April 10, 2008 27. National Cancer Institute, NCI-Supported Cancer Clinical Trials: Facts and Figures (2006) Web site. http://www.cancer.gov/clinicaltrials/facts-and-figures. Accessed April 10, 2008 28. Cancer Centers Program, National Cancer Institute Web site. http://cancercenters.cancer.gov/ index.html. Accessed April 23, 2008

23  NCI-Sponsored Clinical Trials

657

29. SPOREs: Specialized Programs of Research Excellence, National Cancer Research Web site. http://spores.nci.nih.gov/. Accessed April 22, 2008 30. NCI’s Clinical Trials Cooperative Group Program, National Cancer Institute Web site. http:// www.nci.nih.gov/cancertopics/factsheet/NCI/clinical-trials-cooperative-group. Accessed April 23, 2008 31. Community Clinical Oncology Program (CCOP), Division of Cancer Prevention Web site. http://prevention.cancer.gov/programs-resources/programs/ccop. Accessed April 23, 2008 32. Cancer Diagnosis Program, Division of Cancer Treatment and Diagnosis Web site. http:// www.dctd.cancer.gov/ProgramPages/CDP.htm. Accessed April 22, 2008 33. Program for the Assessment of Clinical Cancer Trials, Cancer Diagnosis Program, DCTD Web site. http://dctd.cancer.gov/ProgramPages/CDP-initiative-PACCT.htm. Accessed April 22, 2008 34. Cancer Imaging Program, Division of Cancer Treatment and Diagnosis Web site. http://dctd. cancer.gov/ProgramPages/CIP.htm. Accessed April 22, 2008 35. Industry-Academic Partnerships for Development of Biomedical Imaging Systems and Methods that are Cancer-Specific Web site. http://dctd.cancer.gov/MajorInitiatives/04CIPPart nerships.htm. Accessed April 22, 2008 36. In Vivo Cellular and Molecular Imaging Centers, Division of Cancer Treatment and Diagnosis Web site. http://dctd.cancer.gov/MajorInitiatives/05CIPICMICs.htm. Accessed April 22, 2008 37. Ansher SS, Scharf R: The Cancer Therapy Evaluation Program (CTEP) at the National Cancer Institute: industry collaborations in new agent development. Ann N Y Acad Sci 949:333–40, 2001 38. Clinical Grants and Contracts Branch (CGCB), CTEP Web site. http://ctep.cancer.gov/ branches/cgcb/default.htm. Accessed January 26, 2009 39. Clinical Investigations Branch (CIB), CTEP Web site. http://ctep.cancer.gov/branches/cib/ default.htm. Accessed January 26, 2009 40. Clinical Trials Monitoring Branch (CTMB), CTEP Web site. http://ctep.cancer.gov/branches/ ctmb/. Accessed on January 26, 2009 41. Investigational Drug Branch (IDB), CTEP Website. http://ctep.cancer.gov/branches/idb/ default.htm. Accessed January 26, 2009 42. Pharmaceutical Management Branch (PMB), CTEP Web site. http://ctep.cancer.gov/branches/ pmb/default.htm. Accessed on January 26, 2009 43. Regulatory Affairs Branch (RAB), CTEP Web site. http://ctep.cancer.gov/branches/rab/ default.htm. Accessed on January 26, 2009 44. Protocol and Information Office (PIO), CTEP Web site. http://ctep.cancer.gov/branches/pio/ default.htm. Accessed January 26, 2009 45. Investigator’s Handbook: A Manual for participants in Clinical Trials of Investigational Agents Sponsored by DCTD, NCI Web site. http://ctep.cancer.gov/investigatorResources/ default.htm#investigators_handbook. Accessed January 26, 2009 46. CTC v2.0 and Common Terminology Criteria for Adverse Events v3.0 (CTCAE), CTEP Web site. http://ctep.cancer.gov/protocolDevelopment/electronic_applications/ctc.htm. Accessed January 26, 2009 47. Data Safety and Monitoring Plan for Clinical Trials, CTEP Web site. http://cancertrials.nci. nih.gov/clinicaltrials/conducting/dsm-guidelines. Accessed January 26, 2009 48. Protocol Development, CTEP Web site. http://ctep.cancer.gov/protocolDevelopment/default. htm. Accessed January 26, 2009 49. Cancer Biomedical Informatics Grid (caBIG), Clinical Trials Compatibility Framework, National Cancer Institute Website. https://cabig.nci.nih.gov/tools/toolsuite_view. Accessed April 24, 2008 50. Gynecologic Cancer Intergroup Web site. http://www.gcig.igcs.org/. Accessed January 26, 2009 51. Cooperative Group Guidelines. CTEP Web site. http://ctep.cancer.gov/investigatorResources/ default.htm#guidelines_policies. Accessed January 26, 2009

Index

A Accelerated titration method, 120 Adaptive signature design, predictive marker, 263–265, 267–272 Adverse events (AE), 129–131, 418, 441, 467, 541, 542 Affinity, 39 Anti-angiogenic agents angiogenesis and its mediators degradation and reformation, basement membrane, 350 endogenous inhibitors or receptors, 373–374 endothelial cells and endogenous mediators, 354–355 extracellular matrix, 353–354 growth factors, 353 immune system, 354 small-molecule vascular targeting agents, 374–375 viral gene therapy, 375–376 antibody therapies vs. VEGF, 365–368 approved agents, 347–348 biologic markers microvascular density (MVD), 381 proteomics analysis, 382 von Willebrand factor, 381 clinical trial design cytotoxic agents, 363 hypoxia inducible factor 1-alpha, 364 molecular target expression, 363 xenograft models, 362 combination therapy chemotherapy, 377–378 radiation therapy, 378–379 dose escalation phase I clinical trial methods and design, 360–361 toxicity, 361–362

matrix metalloproteinase inhibitors, 364–365 mechanism-based toxicities and clinical markers hair depigmentation, 384 VEGF-inhibiting therapy, 383 monoclonal antibodies and multi-targeted tyrosine kinase agents, 347, 349 multi-targeted receptor tyrosine kinase inhibitors other, 372–373 vs. preclinical activity, 368–369 randomized discontinuation trial (RDT), 368 sorafenib, 370–371 sunitinib, 371–372 non-selective anti-angiogenic treatments, 365 preclinical screening assays and models advantages and disadvantages, 357–359 Matrigel, 356 orthotropic vs. heterotopic implant, 357 starting dose and administration schedule determination, 359–360 surrogate markers, 380 tumor regression, 379 Anti-angiogenic mechanisms, 101, 362 Antibodies, targeted therapeutics chimeric, 404 monoclonal, FDA approval, 406–407 receptor inhibition, 406 Antibody development, cancer treatment animal–human model transitions, 545 conjugated antibodies drug-immunoconjugates, 556 immunotoxin conjugates, 557 radio-immunoconjugates, 556–557 definition, 535 direct mechanisms of action

M. Hidalgo et al. (eds.), Principles of Anticancer Drug Development, Cancer Drug Discovery and Development, DOI 10.1007/978-1-4419-7358-0, © Springer Science+Business Media, LLC 2011

659

660 Antibody development, cancer treatment (cont.) apoptosis, 537–538 cell surface receptor, 537 immune modulation, 538 neutralisation, 537 other strategies, 538 indirect mechanisms of action cell cytotoxicity, 536 complement-dependent cytotoxicity, 537 other modulations antibody fragments, 558 immunogenicity, 558–560 intrabodies, 558 multivalent antibodies, 557–558 pharmacokinetics, 542–544 phase II and III trials endpoints and study design, 552 metastatic-adjuvant setting, 554 other endpoint, 553 patient population selection, 553–554 selection, recommended phase II dose, 553 single agent vs. chemotherapy combination, 554–555 phase I trials cytotoxic chemotherapy or radiotherapy, 551 dose escalation, 549–550 dose selection, 547–548 other targeted agents, 551 patient selection, 549 pharmacodynamic sampling, 550–551 pharmacokinetic sampling, 550 schedule selection, 548–549 seamless transition, 551–552 staggering, patient treatment, 548 trial conduct, 549 potential toxicities, 544–545 vs. small molecules, 542 structure and function, 535–536 TGN1412, 545–547 types of, 538–539 unconjugated antibodies alemtuzumab, 541 bevacizumab, 540–541 cetuximab, 541 panitumumab, 541 rituximab, 540 trastuzumab, 541 validation, 538 Antibody-directed enzyme prodrug therapy (ADEPT), 538 Antigen-presenting cells (APCs), 520

Index Antitumour response prediction analysis and reporting class prediction, 267 compound covariate method (CCM), 267 leave-one-out cross-validated class prediction model, 269–272 random forest algorithm and neural networks, 269 weighted flexible compound covariate method (WFCCM), 268–269 assay selection, 258–259 centralized laboratory, 259–260 statistical approaches to quality control coefficient of variation (CV), 260–261 intraclass correlation coefficient (ICC), 261–263 variance component analysis, 261–263 statistical design, predictive marker adaptive signature design, 264–265 biomarker-adaptive threshold design, 264 power and sample size analysis, 265–266 tissue collection issues, 258 Apoptosis protein caspases, 433 inhibitor direct proapoptosis activation, 443–444 peptidomimetics and small molecule, 442 survivin and XIAP, 441 prosurvival signal inhibition, 439–441 Atomic absorption spectroscopy (AAS), 76–77 Atomic spectroscopy atomic absorption spectroscopy (AAS), 76–77 inductively coupled plasma spectrometry (ICP), 78 B Bayesian designs, 152–153 BCL-2 proteins apoptosis-targeted therapy, 439–441 lymph node analysis, 441 SPC2996, 439 Bioanalytical methods sample preparation total drug measurements, 67–68 unbound drug concentrations, 68–71 sample separation atomic spectroscopy, 76–78 gas chromatography, 75–76 liquid chromatography, 72–75 small molecule anticancer drugs, 64–66

Index validation requirements accuracy, 81–83 calibration curves, 79–80 linearity, 79–80 matrix effect in LC-MS/MS based methods, 85 precision, 82 range of reliable response, 79–80 recovery, 85 reproducibility, 83 selectivity (specificity), 80 sensitivity, 77, 80–81 stability, 83–85 Bioavailability, 38, 45 Biomarker-adaptive threshold design, 263, 264 Biomarkers anti-angiogenic agents microvascular density (MVD), 381 proteomics analysis, 382 von Willebrand factor, 381 assay development and validation, 259 definition, 215 drug development studies, 260 enrollment of enriched patient population, 172 example, 174 indirect method, 173 labeled retrospective validation, 172 molecular, 472–473, 476 PD biomarkers (see Pharmacodynamic (PD) biomarkers) phase III trials, 171–174 prognostic or predictive, 172 statistical design, predictive marker adaptive signature design, 264–265 biomarker-adaptive threshold design, 264 power and sample size analysis, 265–266 surrogate, 380 therapy, 258–259 Biostatistics aims, endpoints and data analysis, 4–5 data description and displays categorical Variables, 11–12 confidence intervals (see Confidence intervals) continuous variables, 8–11 time to event variables, 12–13 farnesyltransferase inhibitor (FTI) tipifarnib, 3–4 hypothesis testing errors types, 20 evidence evaluation through p-values, 17–20 from research question to statistical hypothesis, 17

661 multivariable regression analyses Cox proportional hazards regression, 33–34 definition, 30 example, 30 logistic regression, 31–33 one-and two-sample tests, comparison chi-square test, 23–25 Fisher’s exact test, 25 means, 21–23 paired data testing, 25–26 proportions, 20–21 survival times, 26–27 sample size calculations precision-based calculations, 27–28 test-based calculations, 28–29 variable types categorical variables, 7 continuous variables, 5–6 time to event variables, 7 transformation, 7–8 Bivariate analysis, 153, 154 Blinding. See Masking Box and whisker plot (boxplot), 9 C Cancer chemoprevention agent selection characteristics, 470 combination strategies, 473 infection-related cancers and vaccines, 473–474 mechanisms, 470 molecular biomarkers, 472–473 carcinogenesis gene chip analysis, 465 intraepithelial neoplasia (IEN), 464 single nucleotide polymorphisms (SNPs), 464 celecoxib, 463 definitive and intermediate endpoint selection, 474 finasteride, 463 population selection cancer risk modeling, 467–468 convergent trial design, 468–469 hereditary cancer syndromes, 469–470 premalignant-malignant cells transition, 466 risk-benefit ratio optimization COX-2 inhibitor, 476 UGT1A1*28 polymorphism, 477

662 Cancer therapy evaluation program (CTEP) investigational agents and resources clinical trials solicitations and letters, 648–649 data and safety monitoring, 651–652 data reporting, 650 IND submission, 649 protocol submission and review, 649 safety data reporting, 650 organization and structure clinical grants and contracts branch (CGCB), 645–646 clinical investigations branch (CIB), 646 clinical trials monitoring branch (CTMB), 646 investigational drug branch (IDB), 646 organization and mission, 645 pharmaceutical management branch (PMB), 647 protocol and information office’s (PIO), 647 regulatory affairs branch (RAB), 647 phase 3 treatment trials, 652–655 Cancer vaccines immunotherapy clinical trials active, 524 antigen-based vaccines, 525–527 passive, 523 vascular endothelial growth factor-A (VEGF-A), 523 VEGF–EGFR pathway inhibition, 524 whole tumor cell vaccines, 527–528 immunotherapy targets immune checkpoints, 529 immunologic recognition and eradication, 530 viral particles and bacterial delivery systems, 528 mechanisms, tumor immune evasion local processes, 520–521 systemic processes, 521–523 T-cell activation vs. suppression, 520, 522 tumor antigen recognition vs. tumor escape, 519 Categorical Variables, 7 Cervical cancer, chemoradiation acute toxicity, 505 vs. radiation therapy, survival rates, 495, 504 RTOG 0417, 506 Chemoprevention agents characteristics, 470 combination strategies, 473 infection-related cancers and vaccines

Index immunization strategy, 474 neoplastic development, 473 mechanisms celecoxib-related cardiovascular adverse events, 472 COX-2 inhibitors, 471 molecular biomarkers cyclin D1 genotype, 472 rapamycin inhibition, 473 safety and effectiveness, 472 Chemoradiation, combined modality therapy cervical cancer acute toxicity, 505 vs. radiation therapy, survival rates, 495, 504 RTOG 0417, 506 head and neck cancer epidermal growth factor receptor (EGFR) expression, 501 locoregional control and organ preservation, 500 overall survival rates, vs. radiation alone, 493, 499 primary endpoint, local-regional control, 499 nonsmall cell lung cancer (NSCLC) CALGB 8433, 502 median progression-free and overall survival rates, 503 RTOG 0324, 504 sequential vs. concurrent chemotherapy, 494, 503 Chemotherapy agents, 592 Chemotherapy-anti-angiogenics metronomic approaches, 378 normalization window, 377 Chi-square test, 23–25 Chronic kidney disease (CKD), 605 Clearance (CL), 38, 51–52, 191 Clinical drug therapy, 278 Clinical pharmacology definition, 37 goal, PK study, 43–44 modelling terms, 39–40 number of patients, 42–43 pharmacodynamic (PD) terms, 39 pharmacokinetics (PK) definition, 44 drug distribution, 45 drug elimination, 47–49 drug metabolism, 45–47 models, 49–61 (see also Pharmacokinetics) non-compartmental type, 55–56

Index non-linear type, 54 physiologically-based pharmacokinetic (PBPK) models, 56–59 population pharmacokinetics, 60–61 rate of drug absorption, 44 pharmacokinetic (PK) terms, 38 sampling intensity, 42–43 sampling schedule and study design, 40–42 Clinical trials, oblimersen advanced melanoma dacarbazine, 574 neutropenia and thrombocytopenia, 576 proliferation vs. vascularization, rate of, 575 Bcl-2 silencing and chemosensitization, 573 chronic lymphocytic leukemia demographic characteristics, 572 responsive vs. refractory, 571 others Bcl-2 protein, 576 dexamethasone, 578 gymnotic delivery, 577 response rates, 576 Clinical trials, special populations cancer clinical trials, elderly populations, 620–622 dose escalation process, 618 FDA regulatory guidance abnormal liver dysfunction, 606 hepatic dysfunction, 605 hepatic impairment Child-Pugh classification, 608 indices, 607–608 international normalized ratio (INR) via coagulation activation, 609 liver disease, causes of, 606–607 Mayo end-stage liver disease (MELD) score, 608 NCI stratifications, 609 renal and hepatic dysfunction, 609–612 neutropenic infection prevention, 619 organ dysfunction chemotherapeutic agents, 603 hepatic infiltration, 604 pharmacokinetics and pharmacodynamics, 616–617 pediatric factors, 623 recommended phase II dose, 620 rolling six design, 623 pharmacologic outcomes vs. toxicity, 606 renal dysfunction creatinine clearance estimation, 614 cystatin C, 616

663 drug elimination, 614 extracellular fluid volume, 615 methods, 614 phase II metabolic reactions, 613 survival rates, 618 Clonogenic assays, 101 Coefficient of variation (CV), 191, 260–262 Combined modality therapy (CMT) chemoradiation cervical cancer, 504–506 head and neck cancer, 499–502 nonsmall cell lung cancer (NSCLC), 502–504 esophageal cancer disease control and survival, 507 locoregional relapse, 506 median survival rates, concurrent chemoradiation, 496, 506 radiation therapy vs. chemoradiation, 507 radiation therapy techniques and modalities biological basis, 486–488 radiation treatment planning, 484–486 radiotherapy delivery, 483 rationale, chemotherapy-radiation capecitabine, 498 cervical cancer, survival rates, 488, 495 cisplatin, 497 esophageal cancer, median survival rates, 488, 496 5-flourouracil, 497 head and neck cancer, overall survival rates, 488, 493 phase III clinical trials, gastrointestinal tumors and glioblastoma, 488–492 sequential vs. concurrent chemoradiation, 488, 494 systemic therapies, 488 temozolomide, 498 in vitro methods, 488 Compartmental modelling, 39 Compound covariate method (CCM), 267 Computer Assisted Drug Design, 98 Confidence intervals for means and differences in means, 14–15 for proportions and comparisons of proportions, 15–16 for time-to-event parameters, 16 width, 14 CONSORT Statement, 185 Continual reassessment method (CRM), 120 Continuous variables, 5–6 Cox proportional hazards regression, 33–34 CTEP. See Cancer therapy evaluation program

664 Cytotoxic agents, early clinical trials administration schedules, 336–337 correlative studies, 338 DLT definition, 340–341 dose escalation, 337, 339–340 efficacy-based trials active/inactive drug, 343 growth modulation index, 342 response criteria in solid tumors (RECIST), 341 vs. noncytotoxic agents, 336 pharmacokinetics and drug-drug interaction, 341 starting dose, 336–337, 339 D Data analysis and reporting measurements for proportions number needed to be treated, 180 Odds ratio, 180–181 relative risk reduction, 180 risk and relative risk, 180 measurements for time to event outcomes, 181 subgroup analyses, 182–183 univariable and multivariable testing, 181–182 Data Safety Monitoring Board (DSMB). See Independent Data Safety Monitoring Committee Death effector domain (DED), 433 Decision theoretic designs, 153 Developmental therapeutics program (DTP) activities, 635 COMPARE, 637 drug discovery and preclinical testing, 634–635 drug screening schema, 636–637 organization and mission, 636 Disease progression, 144, 154 Dose-dependent pharmacokinetics. See Nonlinear pharmacokinetics Dose-limiting toxicity, 337 Dose-limiting toxicity (DLT), 117–119, 121, 321 Dose–response relationship, 37 Dotplot, 9–11 Drug absorption, 44, 45 Drug distribution, 45, 46, 199 Drug dose, 37 Drug elimination, 47–48 Drug exposure, 37 Drug-induced liver injury (DILI), 605

Index Drug interval/holiday, 336 Drug metabolism, 45–48, 197, 208 Drug scheduling and administration sequencing, 195–196 DTP. See Developmental therapeutics program E Early drug trials (phase I/early phase II), 277 Efficacy, 39, 102–103 Efficacy models, 102 Elimination half-life, 38, 40 Elimination rate constant, 38, 50 Empirical vs. rational discovery and development strategies, 93–95 End of phase 2 (EOP2) meetings, 320 Endogenous inhibitors or receptors ABT-510, 373 PI-88 activity, 374 End-stage renal disease (ESRD), 605 Epidermal growth factor receptor (EGFR), 537 Equilibrium dialysis, 69 Equivalence studies, 171 European medicines agency (EMEA), 590 F Factorial designs, 169–170, 321 Farnesyltransferase inhibitor (FTI) tipifarnib, 3–4 Fisher’s exact test, 25 Fixed dose escalation, 118 Food and drug administration modernization act, 590 G Gas chromatography (GC), 75–76 Glomerular filtration rate (GFR), 614 H Head and neck cancer, chemoradiation epidermal growth factor receptor (EGFR) expression, 501 locoregional control and organ preservation, 500 overall survival rates, vs. radiation alone, 493, 499 primary endpoint, local-regional control, 499 HED calculation. See Human equivalent dose (HED) calculation

Index Hepatic impairment, clinical trials Child-Pugh classification, 608 indices, 607–608 international normalized ratio (INR) via coagulation activation, 609 liver disease, causes of, 606–607 Mayo end-stage liver disease (MELD) score, 608 NCI stratifications, 609 renal and hepatic dysfunction, 609–612 Histogram, 9–11 Human equivalent dose (HED) calculation, 124–128, 547 Hypothesis testing errors types, 20 evidence evaluation through p-values, 17–20 from research question to statistical hypothesis, 17 I Image-guided radiation therapy (IGRT), 485 Imaging studies early pharmacodynamics/response assessment examples, 287–290 overview, 287 surrogate endpoints, 290–291 imaging vs. tissue/blood assays, 276 modalities magnetic resonance (MR), 278–279 magnetic resonance spectroscopy (MRS), 279–280 optical, 281 others, 282 radionuclide, 280–281 ultrasound, 281 molecular imaging data analysis and reporting approaches, 292 standardization, 291–292 overview, 275–276 roles clinical drug therapy, 278 early drug trials (phase I/early phase II), 277 late phase II/III drug trials, 277–278 therapeutic targets expression, 283–285 overview, 282–283 resistance factors, 286–287 Immunogenicity modulations complement-dependent cytotoxicity (CDC), 559

665 enhanced antibody-dependent cell cytotoxicity (ADCC), 558–559 other immune components, 559–560 Immunotherapy clinical trials, cancer vaccines active, 524 antigen-based vaccines CEA, 526 heat shock proteins (HSP), 525 mucin-1, 526 mutated k-ras vaccines, 525 passive, 523 VEGF–EGFR pathway inhibition, 524 whole tumor cell vaccines granulocyte/macrophage colonystimulating factor (GM-CSF), 527 serial analysis of gene expression (SAGE), 528 IND. See Investigational new drug Independent Data Safety Monitoring Committee, 178 Inductively coupled plasma spectrometry (ICP), 78 In silico screening, 98 Intensity-modulated radiation therapy (IMRT), 485 Intent to treat (ITT) analysis, 179 International Cooperative Biodiversity Group (ICBG) program, 642 Interquartile range (IQR), 9 Intraclass correlation coefficient (ICC), 260–263 Intrinsic activity, 39 Investigational new drug (IND) application, 307–310 requirements fulfillment, 314 responsibilities of its holder annual report, 316 charging to recover cost, 318 clinical trial monitoring, 318–319 Data Monitoring Committees (DMCs), 319 Informed Consent, 317–318 safety reporting, 316 submission chemistry, manufacturing and control, 310–311 nonclinical pharmacology/toxicology, 311–314 In vitro pure chemical screens, 98 K Kaplan–Meier curve, 12–13

666 L Large volume screening methods ancillary needs, 97–98 historical perspective, 95–96 positive and negative results management, 100–102 types, 98–100 Late phase II/III drug trials, 277–278 Leave-one-out cross-validated class prediction model, 269–272 Linear pharmacokinetics, 39 Liquid chromatography electrical conductivity, 74 fluorescence, 74 mass spectrometry (MS), 74–75 ultraviolet and visible spectroscopy (UV/VIS), 73 Liquid-liquid extraction, 67, 68 Logistic regression, 31–33 M Magnetic resonance (MR) imaging, 278–279 Magnetic resonance spectroscopy (MRS), 279–280 Marketing and postmarketing role, US FDA considerations, 323–326 general efficacy requirements, 321–323 NDA classification and content, 321 Masking, 30, 168–169 Maximal tolerated dose (MTD), 337, 339–340, 593, 639 Maximum concentration, 38 Maximum response, 39 Medical Dictionary for Regulatory Activities (MedDRA), 129, 131 MEK inhibitors CI-1040, 431 molecular determinants, 432 pharmacodynamic analysis, 431 three-dimensional structures, 430 Microdialysis, 70–71 “3 + 3” Modified dose escalation Fibonacci design, 118–121, 618 Molecular and chemical descriptors bioengineering, 93 cancer therapeutic types, 90–93 natural products, 92 synthetic structures, 92 MTD confirmation phase, 118 Multiple randomizations, 165–166 Multi-stage phase II design, 148

Index Multi-targeted receptor tyrosine kinase inhibitors other, 372–373 vs. preclinical activity, 368–369 randomized discontinuation trial (RDT), 368 sorafenib, 370–371 sunitinib, 371–372 Multivariable regression analyses Cox proportional hazards regression, 33–34 definition, 30 example, 30 logistic regression, 31–33 Mutational activation, 94 N National Cancer Center Network (NCCN), 619 National Cancer Institute (NCI)-sponsored clinical trials components, 643 CTEP investigational agents and resources, 647–652 organization and structure, 644–647 phase 3 treatment trials, 652–655 DCTD organization and mission, 632 drug development group (DDG) early and late screening, 634 preclinical and clinical development program, 638 DTP activities, 635 COMPARE, 637 drug discovery and preclinical testing, 634–635 drug screening schema, 636–637 organization and mission, 636 NCI experimental therapeutics program (NExT), 633 pharmaceutical resources branch (PRB), activities, 639 programs, preclinical drug development International Cooperative Biodiversity Group (ICBG), 642 National Cooperative Drug Discovery Group (NCDDG), 642 Rapid Access to Intervention Development (RAID), 641–642 Rapid Access to NCI Discovery Resources (RAND), 640–641

Index Rapid Access to Preventive Intervention and Development (RAPID), 642–643 toxicology and pharmacology branch (TPB), activities, 639 types of, 631 National Cooperative Drug Discovery Group (NCDDG) program, 642 Natural killer (NK) cells, 520 NCI 60, 97, 99–100 Neural networks, 97, 269 New approaches to neuroblastoma therapy (NANT), 593 NOAEL determination. See No observed adverse effect level determination Noncompartmental modelling, 39 Non-compartmental pharmacokinetics, 55–56 Non-eukaryotic cell models, 100 Non-inferiority studies, 171 Nonlinear pharmacokinetics, 39 Nonreceptor kinase inhibition mTOR-targeting agents clinical antitumor activity, 425–426 downstream effects, 426 rapamycin, 423 toxicities, 424 rapamycin pathway cellular translation machinery, 423 phosphatase and tensin homolog, 422 raptor, 423 Nonsmall cell lung cancer (NSCLC), chemoradiation CALGB 8433, 502 median progression-free and overall survival rates, 503 RTOG 0324, 504 sequential vs. concurrent chemotherapy, 494, 503 No observed adverse effect level (NOAEL) determination, 123–127, 546, 547 Nucleic acids, targeted therapeutics antisense oligonucleotide design, 408 delivery systems, 409–410 RNA interference, 409 O Objective response rate, 142–143, 154 Oblimersen, clinical trials advanced melanoma dacarbazine, 574 neutropenia and thrombocytopenia, 576

667 proliferation vs. vascularization, rate of, 575 Bcl-2 silencing and chemosensitization, 573 chronic lymphocytic leukemia demographic characteristics, 572 responsive vs. refractory, 571 others Bcl-2 protein, 576 dexamethasone, 578 gymnotic delivery, 577 response rates, 576 Odds ratio, 32, 180 Oligonucleotide therapeutics affinitak, 580 AP 12009, 579 clinical trials, oblimersen advanced melanoma, 574–576 Bcl-2 silencing and chemosensitization, 573 chronic lymphocytic leukemia, 571–573 others, 576–578 heparin-binding proteins, 570 immunostimulatory properties, 570 OGX-011 docetaxel vs. docetaxel, 579 pro-apoptotic bax protein, 578 pseudocatalytic mechanism, 570 RNAi and siRNAs, 580–582 Oncology drug development, 190–193 Oncology Drugs Advisory Committee (ODAC), 326–327 Optical imaging, 281 Organ dysfunction, clinical trials chemotherapeutic agents, 603 hepatic infiltration, 604 pharmacokinetics and pharmacodynamics 14 C-labeled carbon dioxide, 617 therapeutic index, 616 Orphan Drug program, 328–329 P Paired data testing, 25–26 PD biomarkers. See Pharmacodynamic (PD) biomarkers Pediatric brain tumor consortium (PBTC), 593 Pediatric Initiatives, 329–330 Pediatric investigational plan (PIP), 590 Pediatric patients adult oncology population, 589 children vs. adults

668 Pediatric patients (cont.) acute lymphoblastic leukemia (ALL), 592 age-dependence, 591 antineoplastic agent, 592 drug delivery mechanism, 591 drug development dose-limiting toxicity (DLT), 593 insulin-like growth factor 1 receptor (IGF-1R), 595 maximally tolerated dose (MTD), 593 pharmacokinetic profile, 594 preclinical testing, 595 drug testing, 590 minimal risk, definition, 590 role of, combination anticancer therapy, 595 chemotherapy delivery, 597 clofarabine, 596 drugability, 598 irinotecan, 597 topotecan, 596 Pediatric pharmacology research unit (PPRU) network Pediatric preclinical testing program (PPTP), 595 Pediatric tumors, 592 Pharmacodynamic (PD) biomarkers circulating tumor cells and tumor cell-derived materials, 240–242 incorporation of PD markers dose/regimen selection, 247–248 lead/back-up compound selection, 249 proof of concept, 245–247 proof of drug mechanism, 242–245 surrogate marker of clinical benefit, 249–250 rapamycin, tumor biopsies, 229–231 role in oncology and drug development, 215–217 specific drug, 217–219 surrogate tissues buccal mucosa, 235–237 hair, 234–235 normal tissue vs. solid tumor tissues, 232–234 skin, 229–230 white blood cells and platelets, 237–239 tumor-derived tissue and methodologies analytical techniques for PD endpoints, 228–231 anatomical sites to be biopsied, 222–223

Index operational and planning aspects, 227 processing and preservation, 224–227 tissue acquisition methods, 220–222 tissue heterogeneity, 223–224 Pharmacodynamics (PD), 37, 229–231, 287–291, 616–617 Pharmacokinetic-pharmacodynamic relationships clinical development dose-escalation schemes, 194 parameter estimates, 195 starting dose selection, 193–194 preclinical development, 193 Pharmacokinetics (PK) commonly used terms, 44 compartmental modelling multi-compartment model, 52–54 one-compartment model, 50–52 definition, 37, 44 drug distribution, 45 drug elimination, 47–49 drug metabolism, 45–47 goal, 43 linear, 39 non-compartmental type, 55–56 non-linear type, 39, 54 physiologically-based pharmacokinetic (PBPK) models, 56–59 population pharmacokinetics, 60–61 rate of drug absorption, 44 Pharmacokinetic studies, early anticancer drug development dose adaptation feedback controlled-dosing, 209–210 therapeutic drug monitoring, 209 oncology drug development, 190–193 pharmacokinetic–pharmacodynamic relationships clinical development, 193–195 preclinical development, 193 pharmacokinetic variability sources age, 197–200 body size and body composition, 196–197 drug interactions, 204–207 drug scheduling and administration sequencing, 195–196 inherited genetic factors, 208–209 pathophysiological changes, 200–203 sex dependence, 203 Pharmacokinetic variability sources age absorption changes, 197–198

Index changes in hepatic metabolism, 199–200 changes in renal function, 199 volume of distribution changes, 198–199 body size and body composition, 196–197 drug interactions complementary and alternative medicine co-administration, 206–207 non-chemotherapeutic drugs co-administration, 204–206 drug scheduling and administration sequencing, 195–196 inherited genetic factors, 208–209 pathophysiological changes disease effects, 200 hepatic impairment effects, 201–202 renal impairment effects, 200–201 serum proteins effects, 203 sex dependence chemotherapeutic drugs co-administration, 204 Pharmacology models, 102 Phase I clinical trials design options and dose escalation accelerated titration method, 120 continual reassessment method (CRM), 120 “fixed dose” escalation, 118–121 “3 + 3” modified Fibonacci design, 118–121 rolling six design, 120–121 ethical considerations risk-benefit ratio, 135–137 therapeutic intent, 134–135 goal, 117 informed consent, 136–137 phase I evaluation and end-points correlative studies, 132–133 optimal biologic dose for novel, non-toxic agents, 133–134 radiographic evaluation, 132 reporting of toxicities, 129–131 starting doses and schedules selection modern method, 124–128 other methods, 128 preclinical pharmacology studies, 121–122 preclinical toxicology studies, 122 traditional method, 123–124 Phase I evaluation and end-points correlative studies, 132–133 objectives, 128

669 optimal biologic dose for novel, non-toxic agents, 133–134 radiographic evaluation, 132 reporting of toxicities adverse events (AE), 129–130 serious adverse events (SAE), 130–131 Phase III clinical trials biomarkers enrollment of enriched patient population, 172 example, 174 indirect method, 173 labeled retrospective validation, 172 prognostic or predictive, 172 data analysis and reporting measures of effectiveness, 179–181 subgroup analyses, 182–183 univariable and multivariable testing, 181–182 endpoints measurement criteria, 166–167 primary vs. secondary, 166 surrogate or substitute, 167–168 equivalence and non-inferiority design, 171 factorial designs, 169–170 Independent Data Safety Monitoring Committee, 178 masking, 168–169 multiple arm studies, 169 phase II/III design, 177–178 population, 164–165 randomization multiple, 165–166 stratification, 166 statistical considerations hypothesis testing and confidence intervals, 174–175 interim analyses, 176–177 sample size, 175–176 termination of study, 178–179 transparency and consistency CONSORT Statement, 185 trial registries, 183–184 Phase II trials challenges, 154–155 elements, 155 endpoints disease progression, 144 objective response rate, 142–143 others, 145 toxicity, 143–144 factors influencing design, 142 hypothesis-testing framework, 155–157

670 Phase II trials (cont.) multi-stage, 148 single stage, 145–146 two-stage, 146–148 randomized trial concurrent “comparator” arm, 148–149 discontinuation trials, 150–151 experimental arms, 149–150 phase II/III trials, 151 screening trials, 150 sample sizes, 155, 158 theoretical frameworks Bayesian designs, 152–153 bivariate analysis, 153 decision theoretic designs, 153 Physiologically-based pharmacokinetic (PBPK) models, 56–59 Physiologically based pharmacokinetics (PBPK), 39–40 Platelet-derived growth factor receptor (PDGFR), 537 Population pharmacokinetic modelling, 39 Population pharmacokinetics, 60–61 Population selection, cancer chemoprevention cancer risk modeling cyclin D1 genotypes, 467 estimated event rate, 468 Gail model, 497 convergent trial design angiogenic switch, 468 early-phase trial designs, 469 hereditary cancer syndromes celecoxib vs. placebo, 470 familial adeomatous polyposis (FAP), 469 nonsteroidal anti-inflammatory agent, 470 premalignant-malignant cells transition, 466 Potency, 39 Precision-based sample size calculations, 27–28 Preclinical models cancer drug development, in vivo models, 96 COMPARE algorithm, 97 empirical drug discovery, 89, 91 empirical vs. rational discovery and development strategies, 93–95 large volume screening methods ancillary needs, 97–98 historical perspective, 95–96 positive and negative results management, 100–102 types, 98–100

Index molecular and chemical descriptors bioengineering, 93 cancer therapeutic types, 90–93 natural products, 92 synthetic structures, 92 targeted cancer drug discovery, 89, 90 in vivo evaluation clinical correlation, in vivo screening and model results, 107–109 mouse models, 103–107 overview, 102–103 styles of, 104 Preclinical screening assays and models, anti-angiogenic activity advantages and disadvantages, 357–359 Matrigel, 356 orthotropic vs. heterotopic implant, 357 Premarketing development role, US FDA end of phase 2 (EOP2) meetings, 320 IND application, 307–310 IND holder responsibilities, 314–319 IND requirements fulfillment, 314 IND submission, 310–314 ongoing responsibilities, 319–320 special protocol assessments (SPA), 320 Prognostic (predictive) biomarker, 172, 225 Progression based endpoints. See Disease progression Progression free rate (PFR), 342 Protein precipitation, 67, 68, 70 p-values, 17–20 R Radiation therapy-anti-angiogenics sequencing and timing, 378 toxicity, 379 Radiation therapy techniques and modalities biological basis diffusion-limited hypoxia DNA damage ionizing radiation oxygenated vs. hypoxic setting repopulation radiation treatment planning dose clouds, 485–486 dose painting, 485 image-guided radiation therapy (IGRT), 486 intensity-modulated radiation therapy (IMRT), 485 simulation technologies, 494 Radionuclide imaging, 280–281

Index Raf inhibitors sorafenib, 429 VEGF signal transduction, 430 Random forest algorithm, 269 Randomization multiple, 165–166 stratification, 166 Randomized phase II trials concurrent “comparator” arm, 148–149 discontinuation trials, 150–151 experimental arms, 149–150 phase II/III trials, 151 screening trials, 150 Rapid Access to Intervention Development (RAID) program, 641–642 Rapid Access to NCI Discovery Resources (RAND) program, 640–641 Rapid Access to Preventive Intervention and Development (RAPID) program, 642–643 Rational drug development strategies, 94 Receptor kinase inhibition epidermal growth factor receptor (EGFR) cetuximab, 415 gefitinib, 415 intracellular signals, 413 lapatinib, 416 monoclonal antibodies, 414 single-agent trastuzumab, 415 hepatocyte growth factor (HGF) and c-MET ARQ197, 419 MP470, 421 PF-02341066, 420 insulin-like growth factor-I receptor (IGF-1R), 416–418 molecular determinants intracellular signal transduction pathways, 422 sensitivity, anitibodies, 421 Relative sensitivity, 77 Renal dysfunction, clinical trials creatinine clearance estimation, 614 cystatin C, 616 drug elimination, 614 extracellular fluid volume, 615 methods, 614 phase II metabolic reactions, 613 Repeatability and reproducibility (R&R) study, 260 Response, 31, 37, 39 Response criteria in solid tumors (RECIST), 341

671 Response Evaluation Criteria in Solid Tumors (RECIST), 132, 142–143, 154, 291, 341, 342, 370, 379 Rolling six design, 120–121, 594, 623 S Sample preparation, bioanalytical methods total drug measurements liquid-liquid extraction, 68 protein precipitation, 67, 70 solid phase extraction, 67–68 unbound drug concentrations equilibrium dialysis, 69 microdialysis, 70–71 protein precipitation, 67, 70 ultracentrifugation, 69 ultrafiltration, 69 Sample separation, bioanalytical methods atomic spectroscopy atomic absorption spectroscopy (AAS), 76–77 inductively coupled plasma spectrometry (ICP), 78 gas chromatography (GC), 75–76 liquid chromatography electrical conductivity, 74 fluorescence, 74 mass spectrometry (MS), 74–75 ultraviolet and visible spectroscopy (UV/VIS), 73 Serious adverse events (SAE), 130–131, 541, 542, 579 Short-term/prolonged daily administrations, 336 Single stage phase II trials, 145–146 Small-molecule vascular targeting agents auristatin, 374 combretastatin A-4 phosphate, 375 Solid phase extraction, 67–68 Special protocol assessments (SPA), 320 Starting doses and schedules selection modern method animal doses to human equivalent doses conversion, 125, 128 human equivalent dose (HED) calculation, 124–128 most appropriate species selection, 124, 126, 127 no observed adverse effect level (NOAEL) determination, 123–127 safety factor, 127–128 other methods, 128 preclinical pharmacology studies, 121–122

672 Starting doses and schedules selection (cont.) preclinical toxicology studies, 122 traditional method, 123–124 Stereotactic body radiation therapy (SBRT), 483 STI, targeted therapeutics clinical trials design maximum-tolerated dose, 444 pharmacodynamic response, 445 combination therapy EGFR inhibitors, 447 vascular endothelial cell growth factor receptor, 448 endpoints efficacy evaluation, 446 imaging techniques, 447 patient selection, 445–446 Stratification, 166 Surrogate markers, 249–250, 338, 379 Survival times, 26–27 T Targeted therapeutics antibodies chimeric, 404 monoclonal, FDA approval, 406–407 receptor inhibition, 406 apoptosis protein caspases, 433 inhibitor, 441–443 prosurvival signal inhibition, 434–441 cell cycle BI-2536, 410 cyclin-dependent kinase inhibitors, 407–408 flavopiridol, 407 mitosis-specific inhibitors, 408–409 ON01910, 410 pharmacologic CDK inhibitors (CDKI), 407 phases, 410 PLK-1 protein instability, 411 clinical development, STI clinical trials design, 444–445 combination therapy, 447–448 endpoints, 446–447 patient selection, 445–446 intracellular targets and compound, 404–405 mitogen-activated protein kinase conformational changes, 427 MEK inhibitors, 430–432 Raf inhibitors, 429–430 RAS inhibitors, 427–428

Index nucleic acids antisense oligonucleotide design, 408 delivery systems, 409–410 RNA interference, 409 signal transduction and protein kinases bortezomib inhibition, 412 extracellular ligand-binding domain, 411 nonreceptor kinase inhibition, 422–426 receptor kinase inhibition, 412–422 small molecule inhibitors, 403 SRC kinase inhibitors AZM475271, 433 dasatinib, 432 tyrosine kinase inhibitors chimeric LNA–DNA molecules, 408 DFG, 407 structure, 408 Test-based sample size calculations, 28–29 Time to event variables, 7 Time to maximum concentration, 38 Tissue acquisition methods core biopsies, 220–221 fine needle aspirate (FNA) biopsy, 220–222 third-space collections (ascites, pleural fluid), 222 Tissue management and preservation cell suspensions and FNAs: special handling characteristics DNA/mRNA collection, 226 protein analysis, 226 viable cell collection, 227 surgical and core biopsies processing frozen tissues, 225 paraffin-embedded tissues for IHC, 225–226 Toxicity, 43, 117, 122, 129, 143–144, 311 Tumor-derived tissue and methodologies, PD studies analytical techniques for PD endpoints DNA analysis, 228 messenger RNA analysis, 228 protein analysis, 229–231 anatomical sites to be biopsied, 222–223 cell suspensions and FNAs DNA/mRNA collection, 226 protein analysis, 226 viable cell collection, 227 operational and planning aspects, 227 surgical and core biopsies processing frozen tissues, 225 paraffin-embedded tissues for IHC, 225–226

Index tissue acquisition methods core biopsies, 220–221 fine needle aspirate (FNA) biopsy, 220–222 third-space collections (ascites, pleural fluid), 222 tissue heterogeneity, 223–224 Two-stage phase II trials, 146–148 Tyrosine kinase inhibitors, targeted therapeutics chimeric LNA–DNA molecules, 408 DFG, 407 structure, 408 U Ultracentrifugation, 69 Ultrafiltration, 69 Ultrasound imaging, 281 Unbound drug concentrations equilibrium dialysis, 69 microdialysis, 70–71 protein precipitation, 67, 70 ultracentrifugation, 69 ultrafiltration, 69 US Food and Drug Administration (FDA) marketing and postmarketing role considerations, 323–326 general efficacy requirements, 321–323 NDA classification and content, 321

673 premarketing development role end of phase 2 (EOP2) meetings, 320 IND application, 307–310 IND holder responsibilities, 314–319 IND requirements fulfillment, 314 IND submission, 310–314 ongoing responsibilities, 319–320 special protocol assessments (SPA), 320 regulatory considerations agency use of consultants, 326–327 diagnostic tests, 327–328 Orphan Drug program, 328–329 Pediatric Initiatives, 329–330 V Variance component analysis, 260–263 Vascular endothelial growth factor-A (VEGF-A), 523 Vascular endothelial growth factor receptor (VEGFR), 537 Viral gene therapy adenovirus dl922/947, 375 viable tumor rim, 376 Volume of distribution, 38, 45, 51, 198–199 W Weighted flexible compound covariate method (WFCCM), 268–271