Principles and Practice of Stereotactic Radiosurgery

Principles and Practice of Stereotactic Radiosurgery Lawrence S. Chin, MD • William F. Regine, MD Editors Principles...

109 downloads 1185 Views 14MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Principles and Practice of Stereotactic Radiosurgery

Lawrence S. Chin, MD • William F. Regine, MD Editors

Principles and Practice of Stereotactic Radiosurgery

Editors Lawrence S. Chin, MD Professor and Chairman Department of Neurosurgery Boston University School of Medicine Boston, MA, USA

William F. Regine, MD Professor and Chairman Department of Radiation Oncology University of Maryland Medical School Baltimore, MD, USA

ISBN: 978-0-387-71069-3 e-ISBN: 978-0-387-71070-9 DOI: 10.1007/978-0-387-71070-9 Library of Congress Control Number: 2007931622 © 2008 Springer Science+Business Media, LLC. 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. 9 8 7 6 5 4 3 2 1 springer.com

Foreword hen ﬁrst asked to write a foreword to Principles and Practice of Stereotactic Radiosurgery, I hesitated. There have been so many books and peer-reviewed papers written on this subject that I questioned whether another book would add much. However, after Larry and Bill shared the contents of this book with me, I had to change my mind. From my point of view, this book signals the completion of decades of hard work. Pioneering Gamma Knife surgery during the 1970s and 1980s was often a lonely endeavor, with long ﬂights to innumerable meetings on all continents in order to speak about what we were doing in Stockholm. These talks were usually met with polite skepticism, sometimes even outright hostility, initially from neurosurgeons and later by other specialties as well. During the 1970s and 1980s, a large number of foreign colleagues came through Stockholm. The neurosurgical department at the Karolinska Hospital had a good reputation in stereotactic and functional neurosurgery, and many of the visitors later became prominent proponents of radiosurgery. In the mid-1980s, the adapted linear accelerator was pioneered by Federico Colombo in Italy and Osvaldo Betti in Argentina. Later, others joined the ranks. Nevertheless, it would take until the time of the ﬁrst U.S. Gamma Knife installation in 1987 for the concept of noninvasive brain surgery to gain credibility. Slowly, the veracity of our claims from the 1970s began to take hold. By then, we already knew what the next steps would be for us; namely, the further reﬁnement of the Gamma Knife in parallel with the incorporation of stereotactic principles, concepts of precision and accuracy, and imaging into the practice of radiotherapy in the rest of the body. In 1989, we called this stereotactic radiation therapy, or SRT. We believed that there was a gray zone between radiosurgery and conventional radiotherapy that was worthy of attention. The idea was to use increased precision as a way to allow higher doses and maybe fewer fractions in radiotherapy. This could, we thought, improve the treatment of lesions too large for radiosurgery and too small for radiotherapy. I tried to establish a collaboration with one of the major suppliers of linear accelerators in order to explore this gray zone between radiosurgery and conventional radiotherapy, but there was no interest at all at the time. With the rate of development seen over the past 10 years, one wonders what lies ahead for radiation medicine. My guess is that we will see a somewhat slower rate of development in the radiation delivery systems themselves but an increasing emphasis on the integration of radiation delivery systems with software systems such as planning, imaging, and cancer registry systems. On the clinical side, we will see the continued reemergence of radiosurgery in the treatment of functional brain disorders, including epilepsy, movement disorders, obsessivecompulsive states, and possibly severe endogenous depression. In ophthalmology, there is already exploratory work being done in, for example, glaucoma, macular degeneration, endocrine orbitopathy, and uveal melanomas. We will also see the application of stereotactically guided radiation therapy for disorders that currently are not part of standard practice. These will include the precise targeting of intra- and extraaxial spine lesions, as well as disease in the paranasal sinuses and the larynx. Radiation therapy for, for example, lung and prostate cancer will beneﬁt from the increased precision, allowing higher doses to be delivered despite the close proximity of heart muscle and colon. This book is a very good illustration of the term helicopter perspective. It is particularly impressive in that it really approaches the whole spectrum of disease in a very thorough manner. The title of the book is actually quite humble, belying as it does the fact that all available treatment modalities are represented, compared, and put in perspective. It epitomizes the word comprehensive!

W

v

vi

foreword

This textbook contains a wealth of information and truly encompasses the whole ﬁeld of radiosurgery, regardless of technology and regardless of which disease the reader wants to learn more about, be it in the brain, in the spine, in the eye, or elsewhere in the body. For residents and newcomers to the ﬁeld and for the experienced clinician, this volume will represent an invaluable source of information as you strive to design the best therapeutic approach to your individual patients. This is a book that deserves a prominent—and easy to reach—place on our bookshelves. Dan Leksell, MD

Preface he practice of stereotactic radiosurgery has developed from an unprecedented degree of collaboration among practitioners of neurosurgery, radiation oncology, and medical physics. Because the patients and the diseases that we treat are at the intersection of surgery, radiation, and medical therapy, we felt that a full description of this ﬁeld required a comprehensive and global approach to the subject. Not only would we discuss the main diseases that comprise the typical intracranial radiosurgery practice, such as brain metastases and AVMs, we would also cover the fast-growing ﬁeld of extracranial radiosurgery, as well as more unusual indications such as epilepsy and psychiatric disease. We also wanted to avoid a bias toward Gamma Knife radiosurgery, which tends to dominate most publications. Therefore, we made sure that all major stereotactic radiosurgery techniques were represented. In selecting contributing authors, we felt it critical to enlist the help of our international colleagues who have been at the vanguard of expanding radiosurgery indications. After all, stereotactic radiosurgery was invented by a Swedish neurosurgeon. We organized this book into ﬁve main sections, with the ﬁrst few providing important background for the rest of the book. Part One covers the history of radiosurgery, basics of neuroimaging, and a general overview of key concepts in radiosurgery. Part Two concentrates on the principles of radiation physics and radiobiology that explain the noninvasive, yet powerful, nature of stereotactic radiation treatments. Other topics covered include treatment planning and the designing of a radiosurgery unit. We think this portion of the book will be of particular interest to medical physicists, as it is intended to be a practical guide for the running and maintenance of a radiosurgery center. Part Three contains reviews of the major techniques of stereotactic radiosurgery by physicians who are considered by most to be the leading ﬁgure in their disciplines. We hope you ﬁnd their insights as valuable to your practice as we did. Part Four includes eighteen chapters that describe the major disease types treated by practitioners of stereotactic radiosurgery. In each chapter, we asked the authors to provide case reports of actual patients that illustrated the approach, treatment plan, and outcome of their treatment, thus providing a blueprint to follow for those new to the specialty. One of the more unusual aspects of this book is the inclusion of “perspective” chapters that follow a main topic chapter. We felt that having minichapters written by experts in the ﬁeld who might have a differing viewpoint would provide the most balanced approach to diseases that often have more than one effective treatment. The last part of this book presents topics related to patient care and the often ignored but critical socioeconomic side of stereotactic radiosurgery. The diverse subjects tackled include complication management, cost-effectiveness and quality of life, building a radiosurgery practice, and nursing issues. We also included a few topics that have controversial aspects: regulatory and reimbursement issues, medicolegal pitfalls, and radiosurgery semantics. In these chapters, the reader will ﬁnd that some author opinion is unavoidable but does not necessarily reﬂect the views of the editors and the publisher. Our mantra for this book was to be comprehensive and balanced, but we recognize that there will always be disagreements on many of the topics discussed in this book. We hope that this book will be informative but also stimulate a healthy and constructive dialog among its readers. We must continually examine our results in this critical manner to provide the best care for our patients. This book has been the culmination of several years of planning and execution by a large number of very talented individuals. First, we are indebted to the authors of the individual chapters, who provided their time and expertise in the creation of this project. We would like to thank the editors and staff at Springer who brought dedication and excellence to this project: Beth Campbell, Paula Callaghan, Barbara Lopez-Lucio, and Brad Walsh. We thank Barbara Chernow who rounded this book into its ﬁnal form. We thank our assistants Debbie

T

vii

viii

preface

Redmon, Yvette Green, and Michele Murphy who kept our practices humming while dealing with manuscripts, contributors, editors, and Fed-Ex. Our professional lives owe a debt to the mentors who brought us into neurosurgery, radiation oncology, and the world of radiosurgery, Buz Hoff, Martin Weiss, Michael Apuzzo, Steven Giannotta, Howard Eisenberg, Simon Kramer, Larry Kun, and Jay Loefﬂer. Most importantly, we thank our wives Rita and Julie, along with our children, and the rest of our family and friends for their constant love and support. Lastly, we thank our patients, colleagues, trainees, and students who provided the inspiration for this book. Lawrence S. Chin, MD William F. Regine, MD

Contents Foreword by Dan Leksell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

PART I

The Fundamentals

1

The History of Stereotactic Radiosurgery . . . . . . . . . . . . . . . . . . . . . . . . . Michael Schulder and Vaibhav Patil

2

Neuroimaging in Radiosurgery Treatment Planning and Follow-up Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Clark C. Chen, Paul H. Chapman, Hanne Kooy, and Jay S. Loefﬂer

3

Techniques of Stereotactic Radiosurgery . . . . . . . . . . . . . . . . . . . . . . . . . Chris Heller, Cheng Yu, and Michael L.J. Apuzzo

PART II

v vii xv

3

9

25

Radiation Biology and Physics

4

The Physics of Stereotactic Radiosurgery . . . . . . . . . . . . . . . . . . . . . . . . . Siyong Kim and Jatinder Palta

5

Radiobiological Principles Underlying Stereotactic Radiation Therapy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . David J. Brenner

33

51

6

Experimental Radiosurgery Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ajay Niranjan and Douglas Kondziolka

61

7

Treatment Planning for Stereotactic Radiosurgery . . . . . . . . . . . . . . . . . David M. Shepard, Cedric Yu, Martin Murphy, Marc R. Bussière, and Frank J. Bova

69

8

Designing, Building and Installing a Stereotactic Radiosurgery Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lijun Ma and Martin Murphy

PART III

91

Stereotactic Radiosurgery Techniques

9

Gamma Knife Radiosurgery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ajay Niranjan, Sait Sirin, John C. Flickinger, Ann Maitz, Douglas Kondziolka and L. Dade Lunsford

107

10

Linear Accelerator Radiosurgery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . William A. Friedman

129

ix

x 11

contents

Proton Beam Radiosurgery: Physical Bases and Clinical Experience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Georges Noel, Markus Fitzek, Loïc Feuvret, and Jean Louis Habrand

141

12

Robotics and Radiosurgery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cesare Giorgi and Antonio Cossu

163

13

CyberKnife Radiosurgery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . John R. Adler Jr., Alexander Muacevic, and Pantaleo Romanelli

171

PART IV

Treatment of Disease Types

14

Brain Metastases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . John H. Suh, Gene H. Barnett, and William F. Regine

181

15

Metastatic Brain Tumors: Surgery Perspective . . . . . . . . . . . . . . . . . . . . Raymond Sawaya and David M. Wildrick

193

16

Brain Metastases: Whole-Brain Radiation Therapy Perspective . . . . . . Roy A. Patchell and William F. Regine

201

17

High-Grade Gliomas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . David Roberge and Luis Souhami

207

18

Malignant Glioma: Chemotherapy Perspective . . . . . . . . . . . . . . . . . . . . Roger Stupp and J. Gregory Cairncross

223

19

Meningioma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carlos A. Mattozo and Antonio A.F. de Salles

233

20

Meningioma: Surgery Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lawrence S. Chin, Pulak Ray, and John Caridi

249

21

Intracranial Meningioma: Fractionated Radiation Therapy Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Leland Rogers, Dennis Shrieve, and Arie Perry

257

22

Meningioma: Systemic Therapy Perspective . . . . . . . . . . . . . . . . . . . . . . . Steven Grunberg

271

23

Acoustic Schwannoma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . William M. Mendenhall, Robert J. Amdur, Robert S. Malyapa, and William A. Friedman

275

24

Acoustic Neuroma: Surgical Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . Indro Chakrabarti and Steven L. Giannotta

283

25

Acoustic Neuromas and Other Benign Tumors: Fractionated Stereotactic Radiotherapy Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . David W. Andrews, Greg Bednarz, Beverly Downes, and Maria Werner-Wasik

289

contents

xi

26

Pituitary Tumors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kintomo Takakura, Motohiro Hayashi, and Masahiro Izawa

299

27

Pituitary Adenomas: Surgery Perspective . . . . . . . . . . . . . . . . . . . . . . . . . William T. Couldwell and Martin H. Weiss

309

28

Pituitary and Pituitary Region Tumors: Fractionated Radiation Therapy Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jonathan P.S. Knisely and Paul W. Sperduto

317

Pituitary and Pituitary Region Tumors: Medical Therapy Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mansur E. Shomali

327

29

30

Pediatric Radiosurgery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Andrew Reisner, Nicholas J. Szerlip, and Lawrence S. Chin

31

Pediatric Brain Tumors: Conformal Radiation Therapy Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thomas E. Merchant

331

341

32

Pediatric Brain Tumors: Chemotherapy Perspective . . . . . . . . . . . . . . . . Amar Gajjar

351

33

Pineal Region Tumors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gregory P. Lekovic and Andrew G. Shetter

355

34

Pineal Region Tumors: Surgery Perspective . . . . . . . . . . . . . . . . . . . . . . . Alfred T. Ogden and Jeffrey N. Bruce

365

35

Pineal Tumors: Fractionated Radiation Therapy Perspective . . . . . . . . Steven E. Schild

371

36

Pineal Region Tumors: Chemotherapy Perspective . . . . . . . . . . . . . . . . . Barry Meisenberg and Lavanya Yarlagadda

377

37

Skull Base Tumors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stefanie Milker-Zabel, Young Kwok, and Jürgen Debus

383

38

Skull Base Tumors: Surgery Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . James K. Liu, Oren N. Gottfried, and William T. Couldwell

393

39

Skull Base Tumors: Fractionated Stereotactic Radiotherapy Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . René-Olivier Mirimanoff and Alessia Pica

401

40

Head and Neck Tumors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Daniel T.T. Chua, Jonathan Sham, Kwan-Ngai Hung, and Lucullus Leung

411

41

Head and Neck Tumors: Surgery Perspective . . . . . . . . . . . . . . . . . . . . . Gregory Y. Chin and Uttam K. Sinha

421

xii 42

contents

Head and Neck Malignancies: Chemotherapy and Radiation Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mohan Suntharalingam, Kathleen Settle, and Kevin J. Cullen

425

43

Spinal Tumors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Robert L. Dodd, Iris Gibbs, John R. Adler Jr., and Steven D. Chang

431

44

Spine Tumors: Surgery Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gabriel Zada and Michael Y. Wang

443

45

Spinal Metastases: Fractionated Radiation Therapy Perspective . . . . . Eric L. Chang and Almon S. Shiu

455

46

Arteriovenous Malformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bruce E. Pollock

459

47

Arteriovenous Malformations: Surgery Perspective . . . . . . . . . . . . . . . . Ricardo J. Komotar, Elena Vera, J. Mocco, and E. Sander Connolly Jr.

473

48

Cerebral Arteriovenous Malformations: Endovascular Therapy Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Felipe C. Albuquerque, David Fiorella, and Cameron G. McDougall

479

49

Cavernous Malformations and Other Vascular Diseases . . . . . . . . . . . . Ajay Niranjan, David Mathieu, Douglas Kondziolka, John C. Flickinger, and L. Dade Lunsford

491

50

Cerebral Cavernous Malformations: Surgical Perspective . . . . . . . . . . . Robert L. Dodd and Gary K. Steinberg

503

51

Cavernous Malformations and Other Vascular Abnormalities: Observation-Alone Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sepideh Amin-Hanjani and Frederick G. Barker II

513

52

Trigeminal Neuralgia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lawrence S. Chin, Shilpen Patel, Thomas Mattingly, and Young Kwok

519

53

Trigeminal Neuralgia: Surgical Perspective . . . . . . . . . . . . . . . . . . . . . . . . David B. Cohen, Michael Y. Oh, and Peter J. Jannetta

527

54

Trigeminal Neuralgia: Medical Management Perspective . . . . . . . . . . . Neil C. Porter

535

55

Movement Disorder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sangjin Oh, Ajay Niranjan, and William J. Weiner

541

56

Movement Disorders: Deep-Brain Stimulation Perspective . . . . . . . . . . John Y.K. Lee, Joshua M. Rosenow, and Ali R. Rezai

549

57

Movement Disorder: Medical Perspective . . . . . . . . . . . . . . . . . . . . . . . . Sangjin Oh and William J. Weiner

559

58

Psychiatric and Pain Disorders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jason Sheehan, Nader Pouratian, and Charles Sansur

563

contents

xiii

59

Intractable Epilepsies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jean Régis, Fabrice Bartolomei, and Patrick Chauvel

573

60

Epilepsy: Surgery Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Keith G. Davies and Edward Ahn

583

61

Ocular and Orbital Lesions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gabriela Šimonová, Roman Liscˇák, and Josef Novotný Jr.

593

62

Stereotactic Body Radiation Therapy . . . . . . . . . . . . . . . . . . . . . . . . . . . . Laura A. Dawson

611

63

Stereotactic Body Radiation Therapy: Fractionated Radiation Therapy Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gordon W. Wong, Rafael R. Mañon, Wolfgang Tomé, and Minesh Mehta

64

Stereotactic Body Radiation Therapy: Brachytherapy Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Caroline L. Holloway, Desmond O’Farrell, and Phillip M. Devlin

PART V

635

643

Patient Care and Socioeconomic Issues

65

Complications and Management in Radiosurgery . . . . . . . . . . . . . . . . . . Isaac Yang, Penny K. Sneed, David A. Larson, and Michael W. McDermott

649

66

Cost-Effectiveness and Quality of Life . . . . . . . . . . . . . . . . . . . . . . . . . . . Minesh Mehta and May N. Tsao

663

67

Regulatory and Reimbursement Aspects of Radiosurgery . . . . . . . . . . Rebecca Emerick

673

68

Medicolegal Issues in Stereotactic Radiosurgery . . . . . . . . . . . . . . . . . . . April Strang-Kutay

681

69

The Semantics of Stereotactic Radiation Therapy . . . . . . . . . . . . . . . . . . Louis Potters

687

70

Building a Radiosurgery Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . N. Scott Litofsky and Andrea D’Agostino-Demers

691

71

Patient Care in Stereotactic Radiosurgery . . . . . . . . . . . . . . . . . . . . . . . . Terri F. Biggins

699

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

709

Contributors John R. Adler Jr., MD Professor of Neurosurgery, Stanford University Medical Center, and Attending Neurosurgeon, Stanford University Medical Center, Stanford, CA, USA Edward Ahn, MD Fellow in Neurosurgery, Department of Neurosurgery, Children’s Hospital of Boston, Boston, MA, USA Felipe C. Albuquerque, MD Assistant Director of Endovascular Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, AZ, USA Robert J. Amdur, MD Professor of Radiation Oncology, University of Florida College of Medicine, Gainesville, FL, USA Sepideh Amin-Hanjani, MD Assistant Professor, Neurosurgery, University of Illinois at Chicago, Chicago, IL, USA David W. Andrews, MD Professor and Vice Chairman, Chief, Division of Neuro-Oncologic Neurosurgery & Stereotactic Radiosurgery, Thomas Jefferson University, Philadelphia, PA, USA Michael L.J. Apuzzo, MD Edwin M. Todd and Trent H. Wells Professor of Neurosurgery, Radiation, Oncology, Biology and Physics, Department of Neurological Surgery, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA Frederick G. Barker II, MD Associate Professor, Department of Neurosurgery, Harvard Medical School; Associate Visiting Neurosurgeon, Brain Tumor Center, Massachusetts General Hospital, Boston, MA, USA

Gene H. Barnett, MD, FACS Professor of Surgery, Cleveland Clinic Lerner College of Medicine; Director, Brain Tumor Institute, Cleveland Clinic, Cleveland, OH, USA Fabrice Bartolomei, MD, PhD Service de Neurophysiologie Clinique, Université de la Méditerranée, Marseille, France Greg Bednarz, PhD Medical Physicist, Department of Radiation Oncology, Thomas Jefferson University, Jefferson Hospital for Neuroscience, Philadelphia, PA, USA Terri F. Biggins, RN, BSN Patient Care Coordinator, University of Maryland, Gamma Knife Center, Baltimore, MD, USA Frank J. Bova, PhD Professor of Neurosurgery, University of Florida, Gainesville, FL, USA David J. Brenner, PhD, DSc Professor of Radiation Oncology and Public Health, Center for Radiological Research, Department of Radiation Oncology, Columbia University Medical Center, New York, NY, USA Jeffrey N. Bruce, MD Professor of Neurological Surgery, Department of Neurosurgery, Columbia University—College of Physicians and Surgeons, New York, NY, USA Marc R. Bussière, MSc, DABR Medical Radiation Physicist, Department of Radiation Oncology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA J. Gregory Cairncross, MD, FRCPC Department of Clinical Neurosciences, University of Calgary, Foothills Hospital, Alberta, Canada

xv

xvi

contributors

John Caridi, MD Resident, Department of Neurosurgery, University of Maryland, Baltimore, MD, USA

E. Sander Connolly Jr., MD Associate Professor, Department of Neurological Surgery, Columbia University, New York, NY, USA

Indro Chakrabarti, MD, MPH Neurosurgery Chief Resident, University of Southern California, Los Angeles, CA, USA

Antonio Cossu, MTE 3DLine Medical Systems, Milano, Italy

Eric L. Chang, MD Associate Professor, Department of Radiation Oncology, University of Texas M.D. Anderson Cancer Center, Houston, TX, USA Steven D. Chang, MD Assistant Professor of Neurosurgery, Department of Neurosurgery, Stanford University School of Medicine, Stanford, CA, USA Paul H. Chapman, MD Professor, Department of Neurosurgery, Massachusetts General Hospital, Boston, MA, USA Patrick Chauvel, MD Service de Neurophysiologie Clinique, Université de la Méditerranée, Marseille, France Clark C. Chen, MD, PhD Fellow, Radiosurgery, Department of Neurosurgery, Massachusetts General Hospital, Boston, MA, USA Gregory Y. Chin, MD Attending Physician, Department of Head and Neck Surgery, Kaiser Permanente Walnut Creek Medical Center, Walnut Creek, CA, USA Lawrence S. Chin, MD Professor and Chairman, Department of Neurosurgery, Boston University School of Medicine, Boston, MA, USA Daniel T.T. Chua, FRCR Associate Professor, Department of Clinical Oncology, Queen Mary Hospital, The University of Hong Kong, Pokfulam, Hong Kong David B. Cohen, MD Functional Neurosurgery Fellow, Department of Neurosurgery, Allegheny General Hospital, Pittsburgh, PA, USA

William T. Couldwell, MD, PhD Professor and Joseph J. Yager Chair, Department of Neurosurgery, University of Utah, Salt Lake City, UT, USA Kevin J. Cullen, MD Director, University of Maryland Greenebaum Cancer Center, Professor of Medicine, University of Maryland Medical Center, Baltimore, MD, USA Andrea D’Agostino-Demers, MSN, EdD, CS, APRN, BC, NP Clinical Coordinator, Stereotactic Radiosurgery, Image-Guidance, and Functional Neurosurgery Programs, Division of Neurosurgery, UMASS Memorial Healthcare, Worcester, MA, USA Keith G. Davies, MD, FRCS Associate Professor, Department of Neurosurgery, Boston University School of Medicine, Boston, MA, USA Laura A. Dawson, MD Associate Professor, Department of Radiation Oncology, Princess Margaret Hospital, University of Toronto, Toronto, Ontario, Canada Jürgen Debus, MD, PhD Department of Radiation Oncology and Radiation Therapy, University of Heidelberg, Heidelberg, Germany Antonio A.F. de Salles, MD, PhD Professor, Department of Surgery, Division of Neurosurgery, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA Phillip M. Devlin, MD Assistant Professor, Department of Radiation Oncology, Harvard Medical School; and Chief, Division of Brachytherapy, Department of Radiation Oncology, Dana Farber/Brigham & Women’s Cancer Center, Boston, MA, USA

xvii

contributors

Robert L. Dodd, MD, PhD Endovascular Fellow, Department of Neurosurgery, Stanford University, Stanford, CA, USA

Cesare Giorgi, MD Neurosurgeon, Department of Computer-assisted Neuro and Radiosurgery, Ospedale S. Maria, Terni, Italy

Beverly Downes, MS Chief Medical Physicist, Stereotactic Radiosurgery Units, Jefferson Hospital for Neuroscience, Philadelphia, PA, USA

Oren N. Gottfried, MD Resident, Department of Neurosurgery, University of Utah, Salt Lake City, UT, USA

Rebecca Emerick, MS, MBA, CPA Executive Director, International RadioSurgery Association (IRSA), Harrisburg, PA, USA

Steven Grunberg, MD Professor of Medicine, Department of Medical Oncology, University of Vermont, Burlington, VT, USA

Loïc Feuvret Centre de protonthérapie d’Orsay-Institut Curie, Campus universitaire, Orsay, France

Jean Louis Habrand CPO-Institut Curie, Orsay, France

David Fiorella, MD, PhD Staff Neuroradiology, Department of Neuroradiology and Neurosurgery, Cleveland Clinic Foundation, Cleveland, OH, USA Markus Fitzek, MD Radiation Oncology Center, Tufts—New England Center, Tufts University School of Medicine, Boston, MA, USA John C. Flickinger, MD Professor, Department of Neurological Surgery, University of Pittsburgh, Pittsburgh, PA, USA William A. Friedman, MD Professor and Chair, Department of Neurosurgery, University of Florida College of Medicine, Gainesville, FL, USA Amar Gajjar, MD Professor of Pediatrics, University of Tennessee, Director, Division of Neuro Oncology; Co-leader Neurobiology and Brain Tumor Program, Member and Co-Chair Department of Oncology, St. Jude Children’s Research Hospital, Memphis, TN, USA

Motohiro Hayashi, MD, PhD Lecturer of the Department of Neurosurgery, Chief of Gamma Knife Center, Tokyo Women’s Medical University, Shinjuku, Tokyo, Japan Chris Heller, MD Department of Neurological Surgery, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA Caroline L. Holloway, MD, FRCPC Radiation Oncologist, Department of Radiation Oncology, BCCA—Centre for the Southern Interior, Kelowna, BC, Canada Kwan-Ngai Hung, FRCS Consultant Neurosurgeon, Department of Surgery, The University of Hong Kong, Queen Mary Hospital, Pokfulam, Hong Kong Masahiro Izawa, MD, PhD Assistant Professor, Department of Neurosurgery, Tokyo Women’s Medical University, Shinjuku, Tokyo, Japan

Steven L. Giannotta, MD Chairman, Department of Neurosurgery, University of Southern California, Los Angeles, CA, USA

Peter J. Jannetta, MD Professor, Department of Neurosurgery, Drexel University School of Medicine; Vice-Chairman, Department of Neurosurgery, Jannetta Center for Cranial Nerve Disorders, Allegheny General Hospital, Pittsburgh, PA, USA

Iris Gibbs, MD Assistant Professor of Radiation Oncology, Stanford University, Stanford, CA, USA

Siyong Kim, PhD Department of Radiation Oncology, Mayo Clinic, Jacksonville, FL, USA

xviii Jonathan P.S. Knisely, MD, FRCPC Associate Professor, Department of Therapeutic Radiology, Yale University School of Medicine; and Yale Cancer Center, Yale–New Haven Hospital, New Haven, CT, USA Ricardo J. Komotar, MD Resident, Neurosurgery, Department of Neurological Surgery, Columbia University, New York, NY, USA Douglas Kondziolka, MD, FRCS, FACS Professor of Neurological Surgery, University of Pittsburgh, Pittsburgh, PA, USA Hanne Kooy, PhD Research Associate, Department of Radiation Oncology, Massachusetts General Hospital, Boston, MA, USA Young Kwok, MD Department of Radiation Oncology, University of Maryland Medical Center, Baltimore, MD, USA David A. Larson, PhD, MD, FACR Professor of Radiation Oncology and Neurological Surgery, Director, CyberKnife Radiosurgery Program, Co-Director, Gamma Knife Radiosurgery Program, Department of Neurological Surgery and Radiation Oncology, University of California San Francisco, San Francisco, CA, USA John Y.K. Lee, MD Assistant Professor, Department of Neurosurgery, University of Pennsylvania; Medical Director, Penn Gamma Knife at Pennsylvania Hospital, University of Pennsylvania, Philadelphia, PA, USA

contributors

Roman Liscˇák, MD 3rd Faculty of Medicine, Clinical Department of Neurosurgery, Charles University; Department of Stereotactic and Radiation Neurosurgery, Na Homolce Hospital, Prague, Czech Republic N. Scott Litofsky, MD, FACS Associate Professor, Director of Neuro-Oncology, Director of Radiosurgery, Division of Neurological Surgery, University of Missouri-Columbia School of Medicine, Columbia, MO, USA James K. Liu, MD Resident, Department of Neurosurgery, University of Utah, Salt Lake City, UT, USA Jay S. Loefﬂer, MD Chief Radiation Oncology, Department of Radiation Oncology, Massachusetts General Hospital, Boston, MA, USA L. Dade Lunsford, MD, FACS Professor and Chairman, Department of Neurological Surgery, University of Pittsburgh, Pittsburgh, PA, USA Lijun Ma, PhD Associate Professor, Department of Radiation Oncology, University of California San Francisco, San Francisco, CA, USA Ann Maitz, MSc Assistant Professor, Department of Neurological Surgery, University of Pittsburgh, Pittsburgh, PA, USA Robert S. Malyapa, MD, PhD Assistant Professor, Department of Radiation Oncology, University of Florida College of Medicine, Jacksonville, FL, USA

Gregory P. Lekovic, MD, PhD, JD Resident Neurological Surgery, Division of Neurological Surgery, Barrow Neurological Institute, Phoenix, AZ, USA

Rafael R. Mañon, MD Department of Human Oncology, University of Wisconsin Hospital, Madison, WI, USA

Lucullus Leung, PhD Physicist, Department of Clinical Oncology, Queen Mary Hospital, Pokfulam, Hong Kong

David Mathieu, MD, FRCS(C) Visiting Assistant Professor, Department of Neurological Surgery, University of Pittsburgh, Pittsburgh, PA, USA

xix

contributors

Thomas Mattingly, MD Resident, Department of Neurosurgery, University of Maryland, Baltimore, MD, USA Carlos A. Mattozo, MD Professor, Department of Surgery, Division of Neurosurgery, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA Michael W. McDermott, MD, FRCSC Professor in Residence of Neurological Surgery, Halperin Endowed Chair, Vice Chairman, Department of Neurological Surgery, University of California San Francisco, San Francisco, CA, USA Cameron G. McDougall, MD Chief of Endovascular Neurosurgery, Barrow Neurological Institute— Neurosurgery, St. Joseph’s Hospital and Medical Center, Phoenix, AZ, USA Minesh Mehta, MD Professor and Chairman, Department of Human Oncology, University of Wisconsin Hospital, Madison, WI, USA Barry Meisenberg, MD Professor of Medicine, Chief Division of Hematology and Oncology, Department of Medicine, University of Maryland School of Medicine, Baltimore, MD, USA William M. Mendenhall, MD Professor, Department of Radiation Oncology, University of Florida College of Medicine, Gainesville, FL, USA

Alexander Muacevic, MD CyberKnife Center Munich, Munich, Germany Martin Murphy, PhD Associate Professor, Department of Radiation Oncology, Virginia Commonwealth University, Richmond, VA, USA Ajay Niranjan, MBBS, MS, MCh Assistant Professor of Neurological Surgery, University of Pittsburgh, Pittsburgh, PA, USA Georges Noel, MD Centre de lutte contre le Paul Strauss, Department of Radiotherapy, Strasbourg, France Josef Novotný Jr., MSc, PhD Assistant Professor, Department of Radiation Oncology, University of Texas Southwestern Medical Center, Dallas, TX, USA Desmond O’Farrell, CMD Senior Dosimetrist, Division of Brachytherapy, Department of Radiation Oncology, Dana Farber/Brigham & Women’s Hospital, Boston, MA, USA Alfred T. Ogden, MD Resident, Department of Neurological Surgery, Columbia University, New York, NY, USA

Thomas E. Merchant, DO, PhD Member and Chief, Division of Radiation Oncology, St. Jude Children’s Research Hospital, Memphis, TN, USA

Michael Y. Oh, MD Assistant Professor, Department of Neurosurgery, Drexel University School of Medicine; Co-Director, Stereotactic and Functional Neurosurgery, Department of Neurosurgery, Allegheny General Hospital, Pittsburgh, PA, USA

Stefanie Milker-Zabel, MD Departments of Radiation Oncology and Radiation Therapy, Hospital of Heidelberg, Heidelberg, Germany

Sangjin Oh, MD Fellow, Department of Neurology, University of Maryland School of Medicine, Baltimore, MD, USA

René-Olivier Mirimanoff, MD Professor, Department of Radiation Oncology, University Hospital of Lausanne (CHUV), Lausanne, Switzerland

Jatinder Palta, PhD Professor and Chief of Physics, Department of Radiation Oncology, University of Florida, Gainesville, FL, USA

J. Mocco, MD Resident, Neurosurgery, Department of Neurological Surgery, Columbia University, New York, NY, USA

Roy A. Patchell, MD Chief of Neuro-oncology, Professor of Neurology and Neurosurgery, University of Kentucky Medical Center, Lexington, KY, USA

xx

contributors

Shilpen Patel, MD Assistant Professor, Department of Radiation Oncology, University of Washington Medical Center, Seattle, WA, USA

Ali R. Rezai, MD Director, Brain Neuromodulation Center, Jane and Lee Seidman Chair in Functional Neurosurgery, The Cleveland Clinic Foundation, Cleveland, OH, USA

Vaibhav Patil, BA Department of Neurosurgery, New Jersey Medical School, Newark, NJ, USA

David Roberge, MD Assistant Professor, Department of Oncology, Division of Radiation Oncology, McGill University, Montreal, Quebec, Canada

Arie Perry, MD Washington University, Division of Neuropathology, St. Louis, MO, USA Alessia Pica, MD Doctor, Department of Radiation Oncology, University Hospital of Lausanne (CHUV), Lausanne, Switzerland Bruce E. Pollock, MD Professor, Department of Neurological Surgery and Radiation Oncology, Mayo Clinic College of Medicine, Rochester, MN, USA Neil C. Porter, MD Assistant Professor, Department of Neurology, University of Maryland School of Medicine, Baltimore, MD, USA

Leland Rogers, MD Radiation Oncologist, GammaWest Radiation Therapy, Salt Lake City, UT, USA Pantaleo Romanelli, MD Clinical Assistant Professor, Department of Neurology, State University of New York, Stony Brook, NY, USA; Consulting Assistant Professor, Department of Neurosurgery, Stanford University, Stanford, CA, USA; Director, Functional Neurosurgery, Department of Neurosurgery, IRCCS Neuromed, Pozzilli, Italy

Nader Pouratian, MD, PhD Resident Physician, Department of Neurological Surgery, University of Virginia, Charlottesville, VA, USA

Joshua M. Rosenow, MD Assistant Professor of Neurosurgery, Director of Stereotactic and Functional Neurosurgery, Department of Neurosurgery, Feinberg School of Medicine, Northwestern University; Assistant Professor of Neurosurgery, Director of Stereotactic and Functional Neurosurgery, Northwestern Memorial Hospital, Chicago, IL, USA

Pulak Ray, MD Resident, Department of Neurosurgery, Temple University, Philadelphia, PA, USA

Charles Sansur, MD, MHSc Resident, Department of Neurosurgery, Hospital of the University of Virginia, Charlottesville, VA, USA

William F. Regine, MD Professor and Chairman, Department of Radiation Oncology, University of Maryland Medical School, Baltimore, MD, USA

Raymond Sawaya, MD Professor and Chairman, Department of Neurosurgery, University of Texas M.D. Anderson Cancer Center, Houston, TX, USA

Jean Régis Professor, Departement de Neurochirurgie Centre Hospitalier, Er Universitaire La Timone, Marseille, France

Steven E. Schild, MD Professor, Department of Radiation Oncology, Mayo Clinic, Scottsdale, AZ, USA

Andrew Reisner, MD, FACS, FAAP Neurosurgeon, Department of Pediatric Neurosurgery, Children’s Healthcare of Atlanta, Atlanta, GA, USA

Michael Schulder, MD Professor and Vice-Chairman, Department of Neurosurgery, New Jersey Medical School, Newark, NJ, USA

Louis Potters, MD, FACR South Nassau Communities Hospital, Oceanside, NY, USA

xxi

contributors

Kathleen Settle, MD Chief Resident, Department of Radiation Oncology, University of Maryland Medical Systems, Baltimore, MD, USA Jonathan Sham, MD Professor, Department of Clinical Oncology, Queen Mary Hospital, The University of Hong Kong, Pokfulam, Hong Kong Jason Sheehan, MD, PhD Assistant Professor of Neurological Surgery and Neuroscience, Department of Neurological Surgery and Neuroscience, University of Virginia, Charlottesville, VA, USA David M. Shepard, PhD Director of Medical Physics, Swedish Cancer Institute, Seattle, WA, USA Andrew G. Shetter, MD, FACS Chairman of Functional Stereotactic Neurosurgery, Division of Neurological Surgery, Director of Pain Research Laboratory, Barrow Neurological Institute, Phoenix, AZ, USA Almon S. Shiu, PhD Professor, Department of Radiation Physics, University of Texas M.D. Anderson Cancer Center; Director Stereotactic Services, Department of Radiation Physics, M.D. Anderson Cancer Center, Houston, TX, USA Mansur E. Shomali, MD, CM Clinical Assistant Professor of Medicine, University of Maryland School of Medicine, Division of Endocrinology, Union Memorial Hospital, Baltimore, MD, USA Dennis Shrieve, MD, PhD Department of Radiation Oncology, University of Utah Medical Center, Salt Lake City, UT, USA Gabriela Šimonová, MD, PhD Department of Stereotactic Radioneurosurgery, Hospital Na Homolce, Prague, Czech Republic

Uttam K. Sinha, MD Associate Professor, Chief and Program Director, Department of Otolaryngology—Head and Neck Surgery, University of Southern California, Keck School of Medicine, Los Angeles, CA, USA Sait Sirin, MD Visiting Assistant Professor, Department of Neurological Surgery, University of Pittsburgh, Pittsburgh, PA, USA Penny K. Sneed, MD, FACR Professor in Residence, Department of Radiation Oncology, University of California San Francisco, San Francisco, CA, USA Luis Souhami, MD Professor and Associate Director, Department of Oncology, Division of Radiation Oncology, McGill University, Montreal, Quebec, Canada Paul W. Sperduto, MD, MAPP Co-Director, Gamma Knife Center, University of Minnesota Medical Center, Minneapolis, MN, USA Gary K. Steinberg, MD, PhD Bernard and Ronni Lacroute–William Randolph Hearst Professor of Neurosurgery and the Neurosciences; Chairman, Department of Neurosurgery, Stanford University School of Medicine, Stanford, CA, USA April Strang-Kutay, JD Attorney, Goldberg Katzman, P.C., East Petersburg, PA, USA Roger Stupp, MD Multidisciplinary Oncology Center, University of Lausanne Hospitals (CHUV), Lausanne, Switzerland John H. Suh, MD Chairman, Department of Radiation Oncology, Cleveland Clinic, Cleveland, OH, USA Mohan Suntharalingam, MD Professor and Vice Chairman, Department of Radiation Oncology, University of Maryland School of Medicine, Baltimore, MD, USA

xxii Nicholas J. Szerlip, MD Resident, Department of Neurosurgery University of Maryland School of Medicine, Baltimore, MD, USA Kintomo Takakura, MD, PhD President, Tokyo Women’s Medical University, Shinjuku, Tokyo, Japan Wolfgang Tomé, PhD Department of Human Oncology, University of Wisconsin Hospital, Madison, WI, USA May N. Tsao, MD, FRCP(C) Assistant Professor, Department of Radiation Oncology, University of Toronto, Toronto-Sunnybrook Regional Cancer Centre, Toronto, Ontario, Canada Elena Vera, BS Department of Neurological Surgery, Columbia University, New York, NY, USA Michael Y. Wang, MD Assistant Professor, Department of Neurological Surgery, University of Southern California, Los Angeles, CA, USA William J. Weiner, MD Professor and Chairman, Department of Neurology, University of Maryland School of Medicine; Professor and Chairman, Department of Neurology, University of Maryland Medical Center, Baltimore, MD, USA Martin H. Weiss, MD Professor of Neurological Surgery, Department of Neurological Surgery, USC; Attending Physician, Department of Neurosurgery, USC University Hospital, Los Angeles, CA, USA

contributors

Maria Werner-Wasik, MD Associate Professor, Department of Radiation Oncology, Jefferson Hospital for Neuroscience, Philadelphia, PA, USA David M. Wildrick, PhD Surgery Publications Coordinator, Department of Neurosurgery, University of Texas M.D. Anderson Cancer Center, Houston, TX, USA Gordon W. Wong, MD Department of Human Oncology, University of Wisconsin Hospital, Madison, WI, USA Isaac Yang, MD Resident, Department of Neurological Surgery, University of California San Francisco, San Francisco, CA, USA Lavanya Yarlagadda, MD Department of Medicine, University of Maryland, Baltimore, MD, USA Cedric Yu, PhD Department of Radiation Oncology, University of Maryland School of Medicine, Baltimore, MD, USA Cheng Yu, PhD Professor and Director of Radiation Oncology Physics, Department of Radiation Oncology, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA Gabriel Zada, MD Resident Physician, Department of Neurosurgery, University of Southern California, Los Angeles, CA, USA

PA R T I

The Fundamentals

1

1

The History of Stereotactic Radiosurgery Michael Schulder and Vaibhav Patil

The Early Years The history of radiosurgery can be said to begin with the discovery of X-rays by Wilhelm Konrad Roentgen on November 26, 1895. His report, “Uber eine neue art von strahlen” (“On a new kind of ray”), appeared 6 weeks later [1]. By January 1896, X-rays were being used to treat skin cancers. The discovery of radioactivity by Becquerel in 1896, and of radium by the Curies soon after, provided another means for the use of therapeutic ionizing radiation. Neurosurgical applications were not long in following. X-rays were used to treat patients with pituitary tumors as early as 1906, and radium brachytherapy was applied to treat similar conditions at about the same time [2]. Harvey Cushing, the father of American neurosurgery, had extensive experience with both X-ray and brachytherapy treatments, although he remained skeptical of the utility of either [3]. Other neurosurgeons continued to explore the uses of ionizing radiation throughout the ﬁrst half of the 20th century [4]. In 1951, Lars Leksell coined the term stereotactic radiosurgery (SRS) [5]. A ceaseless innovator, his goal was to develop a method for “the non-invasive destruction of intracranial . . . lesions that may be inaccessible or unsuitable for open surgery.” The ﬁrst procedures were done using an orthovoltage X-ray tube, mounted on an early model of what is now known as the Leksell stereotactic frame, for the treatment of several patients with trigeminal neuralgia. After experimenting with particle beams and linear accelerators, Leksell and his colleagues ultimately designed the Gamma Knife (GK), containing 179 cobalt sources in a hemispheric array (Fig. 1-1). The ﬁrst unit was operational in 1968. The potential of the GK to treat lesions was recognized by Leksell and colleagues early on. In the era before computed tomography (CT), these treatments were limited to patients with arteriovenous malformations (AVMs) [6] and acoustic neuromas, which could be imaged either on angiography or by polytomography, respectively [7]. At the same time, work was continuing elsewhere with focused heavy particle irradiation. Ernest Lawrence, one of the great ﬁgures of 20th century physics and a professor at the University of California Berkeley, invented the cyclotron in

1929, winning the Nobel Prize in 1939 (Fig. 1-2). In the 1950s, his brother John began a decades-long investigation of the use of heavy particles (proton beams, then helium ion beams) for the treatment of patients with pituitary and other intracranial disorders (Fig. 1-3) [8, 9]. Raymond Kjellberg, a neurosurgeon at the Harvard/Massachusetts General Hospital facility, spearheaded the use of proton beam treatments (Fig. 1-4) [10]. A large series of patients with arteriovenous malformations and pituitary tumors was amassed. Similar efforts were carried out in California with helium ions [11]. Particle beams have the advantage of depositing their energy at a distinct point known as the Bragg peak, with minimal exit dose. In practice, the beams must be carefully shaped and spread in order to treat patients with intracranial lesions. The expense of building and maintaining a cyclotron has limited the use of heavy-particle SRS to a few centers.

Acceptance The advent of CT in the mid-1970s, and magnetic resonance imaging some 10 years later, opened up the possibility of direct targeting of tumors and other “soft tissue” targets inside the skull. The 1980s saw the evolution of SRS from an esoteric technique, available at the original GK in Stockholm (and as fractionated treatments at a few heavy-particle accelerators around the world), to an emerging technology of increasing utility. As the potential horizons of SRS broadened, other investigators were able to adapt linear accelerators (“linacs”) for SRS. These devices were more available (and less expensive) than GKs or heavy-particle accelerators [12]. Working independently, in Buenos Aires, Argentina, and in Vicenza, Italy, respectively, Betti and Colombo reported the successful adaptation of linacs for SRS [13, 14]. Their systems allowed for the rotation of the linac gantry in a single plane. After several years of hacking through mounds of red tape, Lunsford and colleagues completed the installation of the ﬁrst American GK at the University of Pittsburgh [15]. This group was instrumental, via an ongoing series of peer-reviewed

3

4

m. schulder and v. patil

FIGURE 1-3. Particle beam accelerator, 1947. (Photo courtesy of the Lawrence Berkeley National Laboratory.)

FIGURE 1-1. Lars Leksell and his physicist colleague, Borje Larsson, preparing a patient for SRS with a particle beam accelerator in 1958. (Photo courtesy of L. Dade Lundsford, MD.)

publications, in placing the technique and clinical indications for SRS on a sound scientiﬁc basis. At about the same time, Winston and Lutz described the use of a commercially available stereotactic frame for linac radiosurgery [16]. Following in their footsteps, Loefﬂer and Alexander demonstrated how a linac dedicated to SRS could be a practical alternative to a GK [17]. In the late 1980s, Friedman and Bova elected not to install the second American GK unit, preferring to develop a new linac SRS system [18]. Other advantages of these linac systems, besides ubiquity and

FIGURE 1-2. Ernest Lawrence at the controls of a cyclotron. (Photo courtesy of the Lawrence Berkeley National Laboratory.)

lower cost, included the availability of collimators in a much greater variety of diameters than provided with the GK. This allowed for the use of single isocenters when treating patients whose targets were more than 18 mm in diameter, the width of the largest GK collimator. However, at around the same time, several GKs were installed in several sites around the world. As clinical experience increased, publications appeared, indications broadened, and vendors became increasingly interested, a debate emerged regarding the merits of the GK versus linac-based SRS. By now, clinical and physics studies seem to have settled the issue in that SRS can be delivered effectively

FIGURE 1-4. Raymond Kjellberg with a frame for proton beam therapy of a patient with an AVM. (Photo courtesy of Richard Wilson, Mallinckrodt Research Professor of Physics, Harvard University.)

1.

the history of stereotactic radiosurgery

5

and accurately with either method [12, 19]. Numerous reports demonstrating the efﬁcacy of SRS with few if any short-term complications and lower costs led to the proliferation of GK and linac units around the world.

Fractionation Linac-based systems also opened up the possibility of SRS without an invasive frame. In 1992, the relocatable Gill-ThomasCosman (GTC) frame was introduced. This device relied on an attached bite block, custom molded for each patient, and was shown to have a stereotactic accuracy of just over 2 mm [20]. Although not sufﬁciently accurate and precise for single-session SRS, the GTC frame opened up the era of fractionated stereotactic irradiation [21, 22]. This in turn began a debate that has not been settled: what to call this new method? fractionated SRS or rather stereotactic radiation therapy (SRT)? This semantic question reﬂects two different underlying views of SRS. The neurosurgeon views it as a type of minimally invasive surgery, whereas the radiation oncologist sees SRS as a technique of small-volume irradiation. Advocating for “FSRS” were neurosurgeons who attached importance to the stereotactic concept, which they viewed as being “neurosurgical.” On the other hand, radiation oncologists claimed that patients were being treated with the standard fractionation schemes that practitioners knew and had been employing for decades. The GTC or similar devices were merely another means of achieving three-dimensional conformality. Confounding this controversy was the introduction of new fractionation schemes. For instance, patients with vestibular schwannomas were treated with 2500 cGy in ﬁve fractions. Other regimens have been used, including frame-based GK to treat hospitalized patients over a 5-day period [23]. Whereas SRT generally was accepted as referring to a stereotactically focused treatment using a conventional fractionation scheme, some neurosurgeons and radiation oncologists insisted that there was nothing sacrosanct about the single-fraction treatment. Who was to say that 3 or 5 doses (i.e., far fewer than usual for radiation therapy, and potentially risky to the patient if not planned and delivered with great precision) were not SRS? Different new technologies made all these options possible, but the argument was honed most precisely by the introduction of a new, robotic device. John Adler, a neurosurgeon who trained at the Brigham and Women’s Hospital in Boston, spent a fellowship year with Lars Leksell in 1985 (Adler JR, personal communication). Excited by his exposure to the GK, Adler saw the potential of SRS being extended to other areas of the body. This required a method of delivering focused radiation without a stereotactic frame. Partnering with engineers at Stanford University and with private ﬁnancial backing, the CyberKnife ultimately came into being in 1994 (Fig. 1-5). The CyberKnife delivers SRS via an X-band linac with an output of 6 MV. It is nonetheless small enough to be mounted on an industrial robot, allowing for a theoretically inﬁnite number of beams to be aimed at the target. Treatments are fashioned using an inverse planning method; to allow for practical computation times, the number of beam origins (“nodes”) and robot angles are limited. Peer-reviewed publications have

FIGURE 1-5. The ﬁrst CyberKnife treatment, 1994. (Photo courtesy of John R. Adler, MD.)

demonstrated the acceptance of the CyberKnife [24–26]. These and other articles have fostered a useful debate regarding the concept of hypofractionation in SRS and indeed if such treatments are still “radiosurgical” [27, 28].

Extracranial Radiosurgery SRS was invented as a means of minimally invasive brain surgery and was expanded with the aid of digital imaging to include extracerebral, intracranial targets. Still, the concept of a highly focused, single- or several-session radiation treatment had obvious appeal for extracranial targets. The ﬁrst radiosurgical moves out of the intracranial compartment were in the logical direction of the skull base and past that into the paranasal sinuses, using either GK [29, 30] or linac units [31]. Creative modiﬁcations of standard stereotactic frames were described to allow for treatment of “lower” targets [32]. The adaptation of available equipment for SRS could go only so far. Hamilton and colleagues described the ﬁrst truly extracranial radiosurgical unit. This prototypical system did not rely on rigid frame ﬁxation to the skull and was designed to provide spinal SRS [33]. The need to surgically place a clamp on a spinous process, and to treat the patient in a prone position, limited the appeal of this groundbreaking concept. With the advent of newer technologies, spinal SRS has become a reality. Reports to date have employed the CyberKnife [34] or other linac-based systems [35]. More recently still, the inevitable and logical extension of SRS to non-CNS targets has begun. Work on CyberKnife treatment of tumors of the lung [36] and prostate [37] has been published. Despite the neurosurgical origins of SRS, all advocates of this concept, in its various forms, can only welcome its spread to other specialties in which neurosurgeons will have little role to play. Table 1-1 summarizes the historical landmarks in the development of SRS.

6

m. schulder and v. patil

TABLE 1-1. Historical landmarks in the development of SRS. Year

Author

Device/Event

1951

Leksell

1954

Lawrence

1962

Kjellberg

1967 1970 1980 1982 1984 1986

Leksell Steiner Fabrikant Betti/Columbo Bunge Winston/Lutz

1991 1992 1994 1997

Friedman Loefﬂer/Alexander Adler Krispel

Invention of SRS with rotating orthovoltage unit Heavy-particle treatment of pituitary for cancer pain Proton beam therapy of intracranial lesions Invention of GK GK SRS of AVMs Helium ion treatment of AVMs Linacs adapted for SRS Installation of commercial GK Linac SRS based on common stereotactic frame Linac system for highly conformal SRS Dedicated linac for SRS developed First CyberKnife treatment Rotating cobalt unit

Other Linac Systems and the Role of Industry The convergence of image-guidance technology and radiation delivery devices has encouraged the entry of multiple vendors into the SRS marketplace. This has reﬂected the undeniable logic of stereotactic localization and the resulting ability to focus radiation treatments on the smallest possible volume. Initially, in addition to the GK, there were a variety of framebased systems designed to provide single-fraction SRS. Vendors included Radionics (X-Knife), Zmed (the University of Florida system), BrainLAB, and Fischer-Leibinger. The acceptance of stereotactic fractionation by radiation oncologists, their reluctance to apply stereotactic frames, and to some extent patients’ preference for avoiding frame use have shifted the focus toward frameless systems. Long-established purveyors of linacs have begun to market stereotactic devices aimed primarily at radiation oncologists but usually with a nod toward neurosurgeons who often will prefer to treat patients with a single fraction, or at most several. Thus, Varian and Phillips (now a division of Elekta) have developed systems with integrated stereotactic localization (Trilogy and Synergy). At the same time, Radionics and BrainLAB have adapted their linac-based SRS devices for frameless use and have marketed directly to radiation oncologists. And to square the circle, American Radiosurgical, Inc., has as its sole product a modiﬁcation of the GK, using a limited number of cobalt-60 sources in a rotating helmet. This industrial involvement in the advancement of SRS and related techniques results from the expense of the equipment and the need for support personnel to ensure their proper functioning. From the days of the ﬁrst GK and on up to the emerging era of frameless, fractionated SRS, companies have played an invaluable role. Without them, SRS would never have come to deﬁne a new standard in patient care, as it so clearly has.

Organized Radiosurgery Neurosurgeons’ interest in SRS was slow to develop but has increased exponentially over time. In 1987, the year that the ﬁrst American GK was installed at the University of Pittsburgh and early work on linac SRS had been published, there were no SRS-related presentations at the meeting of the American Association of Neurological Surgeons (AANS). By 1998, there were 31 such abstracts in addition to practical courses and seminars devoted to the topic. SRS has remained a key item of interest at the major annual meetings of the AANS and of the Congress of Neurological Surgeons. In addition, the meetings of the American and World Societies for Stereotactic and Functional Neurosurgery feature SRS as one of the main topics. The International Stereotactic Radiosurgery Society (ISRS) was founded in 1993 and held its ﬁrst biannual meeting that year in Stockholm. At ﬁrst, the papers presented dealt entirely with the treatment of intracranial conditions. As SRS has moved below the skull base, studies regarding patients with such conditions as tumors of the spine, lung, pancreas, and prostate have been included in the ISRS program. Thus, the expertise of clinicians in ﬁelds completely unrelated to neurosurgery is being applied to the study of SRS. Neurosurgeons comprise the single biggest specialty group in the organization, followed by radiation oncologists and medical physicists. As interest in extracranial and indeed nonneurosurgical SRS inevitably increases, the membership of ISRS no doubt will evolve to reﬂect this broadening of interest. The ISRS publishes a peer-reviewed collection of selected manuscripts from each meeting, entitled Radiosurgery.

Conclusion Acceptance by neurosurgeons, surgical specialists, and radiation oncologists means that as SRS evolves, it will not be a technique for “radiosurgeons” but one of the methods available to treat patients with a wide variety of disorders. At the same time, the historical role of neurosurgeons in the development of SRS, their leadership in its reﬁnement and expansion over the last half century, their knowledge of neuroanatomy, and their understanding of central nervous system pathology and its treatment will ensure the continued active role of neurosurgeons in the ongoing growth of stereotactic radiosurgery.

References 1. Mould R. A Century of X Rays and Radioactivity in Medicine. Philadelphia: Institute of Physics Publishing, 1993. 2. Hirsch O. Uber methoden der operativen behandlung von hypophysistumoren auf endonasalem Wege. Arch Laryngol Rhinol 1910; 24. 3. Schulder M, Loefﬂer J, Howes A, et al. The radium bomb: Harvey Cushing and the interstitial irradiation of gliomas. J Neurosurg 1996; 84:530–532. 4. Schulder M, Rosen J. Therapeutic radiation and the neurosurgeon. Neurosurg Clin N Am 2001; 12(1):91–100, viii. 5. Leksell L. The stereotaxic method and radiosurgery of the brain. Acta Chir Scand 1951; 102:316–319. 6. Steiner L, Leksell L, Greitz T. Stereotaxic radiosurgery for cerebral arteriovenous malformations. Acta Chir Scand 1972; 138: 459–464.

1.

the history of stereotactic radiosurgery

7. Leksell L. Stereotactic radiosurgery. J Neurol Neurosurg Psychiatry 1983; 46:797–803. 8. Kirn TF. Proton radiotherapy: some perspectives. JAMA 1988; 259:787–788. 9. Skarsgard LD. Radiobiology with heavy charged particles: a historical review. Phys Med 1998; 14(Suppl 1):1–19. 10. Kjellberg RN, Abe M. Stereotactic Bragg Peak proton beam therapy. In: Lunsford LD, ed. Modern Stereotactic Neurosurgery. Boston: Martinus Nijhoff, 1988:463–470. 11. Fabrikant J, Lyman J, Frankel K. Heavy charged particle Bragg peak radiosurgery for intracranial vascular disorders. Radiat Res Suppl 1985; 8:S244–258. 12. Podgorsak E, Pike G, Olivier A, et al. Radiosurgery with high energy photon beams: a comparison among techniques. Int J Radiat Oncol Biol Phys 1989; 16:857–865. 13. Betti O, Derechinsky V. Hyperselective encephalic irradiation with a linear accelerator. Acta Neurochir 1984; Suppl 33:385– 390. 14. Columbo F, Benedetti A, Pozza F, et al. External stereotactic irradiation by linear accelerator. Neurosurgery 1985; 16:154–160. 15. Lunsford LD, Flickinger JC, Linder G, et al. Stereotactic radiosurgery of the brain using the ﬁrst United States 210 cobalt-60 source gamma knife. Neurosurgery 1989; 24:151–159. 16. Winston KR, Lutz W. Linear accelerator as a neurosurgical tool for stereotactic radiosurgery. Neurosurgery 1988; 22:454–464. 17. Loefﬂer J, Shrieve D, Wen P, et al. Radiosurgery for intracranial malignancies. Semin Radiat Oncol 1995; 5:225–234. 18. Friedman W, Bova F. The University of Florida radiosurgery system. Surg Neurol 1989; 32:334–342. 19. Luxton G, Petrovich Z, Joszef G, et al. Stereotactic radiosurgery: principles and comparison of treatment methods. Neurosurgery 1993; 32:241–259. 20. Gill SS, Thomas DG, Warrington AP, et al. Relocatable frame for stereotactic external beam radiotherapy. Int J Radiat Oncol Biol Phys 1991; 20:599–603. 21. Andrews DW, Silverman CL, Glass J, et al. Preservation of cranial nerve function after treatment of acoustic neurinomas with fractionated stereotactic radiotherapy. Preliminary observations in 26 patients. Stereotact Funct Neurosurg 1995; 64:165–182. 22. Combs SE, Volk S, Schulz-Ertner D, et al. Management of acoustic neuromas with fractionated stereotactic radiotherapy (FSRT): long-term results in 106 patients treated in a single institution. Int J Radiat Oncol Biol Phys 2005; 63:75–81.

7

23. Noren G. Gamma knife radiosurgery of acoustic neurinomas. A historic perspective. Neurochirurgie 2004; 50:253–256. 24. Chang SD, Murphy M, Geis P, et al. Clinical experience with image-guided robotic radiosurgery (the CyberKnife) in the treatment of brain and spinal cord tumors. Neurol Med Chir (Tokyo) 1998; 38:780–783. 25. Ishihara H, Saito K, Nishizaki T, et al. CyberKnife radiosurgery for vestibular schwannoma. Minim Invasive Neurosurg 2004; 47: 290–293. 26. Mehta VK, Lee QT, Chang SD, et al. Image guided stereotactic radiosurgery for lesions in proximity to the anterior visual pathways: a preliminary report. Technol Cancer Res Treat 2002; 1:173– 180. 27. Adler JR Jr, Colombo F, Heilbrun MP, et al. Toward an expanded view of radiosurgery. Neurosurgery 2004; 55:1374–1376. 28. Pollock BE, Lunsford LD. A call to deﬁne stereotactic radiosurgery. Neurosurgery 2004; 55:1371–1373. 29. Firlik KS, Kondziolka D, Lunsford LD, et al. Radiosurgery for recurrent cranial base cancer arising from the head and neck. Head Neck 1996; 18:160–165; discussion 166. 30. Kondziolka D, Lunsford LD. Stereotactic radiosurgery for squamous cell carcinoma of the nasopharynx. Laryngoscope 1991; 101:519–522. 31. Kaplan ID, Adler JR, Hicks WL Jr, et al. Radiosurgery for palliation of base of skull recurrences from head and neck cancers. Cancer 1992; 70:1980–1984. 32. Samblas JM, Bustos JC, Gutierrez-Diaz JA, et al. Stereotactic radiosurgery of the foramen magnum region and upper neck lesions: technique modiﬁcation. Neurol Res 1994; 16:81–82. 33. Hamilton A, Lulu B, Fosmire H, et al. Preliminary clinical experience with linear accelerator-based spinal stereotactic radiosurgery. Neurosurgery 1995; 36:311–319. 34. Gerszten PC, Welch WC. CyberKnife radiosurgery for metastatic spine tumors. Neurosurg Clin N Am 2004; 15:491–501. 35. De Salles AA, Pedroso AG, Medin P, et al. Spinal lesions treated with Novalis shaped beam intensity-modulated radiosurgery and stereotactic radiotherapy. J Neurosurg 2004; 101(Suppl 3):435–440. 36. Whyte RI, Crownover R, Murphy MJ, et al. Stereotactic radiosurgery for lung tumors: preliminary report of a phase I trial. Ann Thorac Surg 2003; 75:1097–1101. 37. King CR, Lehmann J, Adler JR, et al. CyberKnife radiotherapy for localized prostate cancer: rationale and technical feasibility. Technol Cancer Res Treat 2003; 2:25–30.

2

Neuroimaging in Radiosurgery Treatment Planning and Follow-up Evaluation Clark C. Chen, Paul H. Chapman, Hanne Kooy, and Jay S. Loefﬂer

Introduction Radiosurgery refers to the precise delivery of a large, single dose of radiation to a focal target. The focal dose distribution allows for a positive therapeutic gain. Because of the opacity of the cranial vault, target volume deﬁnition relies entirely on the anatomic accuracy of the available imaging modalities. The need for accurate anatomic visualization is magniﬁed by the use of higher radiation dose delivered in radiosurgery as compared with radiotherapy. To achieve maximal spatial accuracy in radiosurgical planning, an understanding of the basic principles underlying neuroimaging as well as the limitations associated with each imaging modality is mandatory. Additionally, the optimal management of patients after radiosurgery requires knowledge of the expected neuroimaging changes as they relate to clinical outcome. These issues will be reviewed in this chapter.

Imaging Modalities Since its inception with the discovery of X-rays in 1895, radiology has played a pivotal role in the diagnosis and treatment of various neurosurgical lesions. The advent of computed tomography (CT) imaging in the 1970s marked a major step forward in the application of imaging in radiotherapeutic planning by allowing improved anatomic resolution as well as calculation of electron density maps. Improved soft tissue resolution was achieved with the introduction of magnetic resonance imaging (MRI), a technique based on differential nuclear interaction rather than differential density. Advances made in computational technology in the past decade have enabled the superposition of CT and magnetic resonance (MR) images in order to maximize anatomic delineation. More recently, signiﬁcant strides in functional imaging have further reﬁned target deﬁni-

tion in radiosurgical planning (Fig. 2-1). The following section will review the basic principles underlying the various neuroimaging modalities as well as limitations associated with each modality.

Computed Tomography Imaging Computed tomography provides cross-sectional images of the body using mathematical reconstructions based on X-ray images taken circumferentially around the subject. In practice, X-ray transmissions through the subject from a rotating emitter are detected and digitally converted into a grayscale image. Because CT images are ultimately a compilation of X-ray transmissions, the physical principles underlying the two modalities are identical; that is, structural discrimination is made based on the relative atomic composition, and therefore the electron density, of the tissue imaged. CT images, however, offer improved anatomic resolution because each image represents the synthesis of information from multiple X-ray images (Fig. 2-1a). Besides improved anatomic delineation, CT imaging aids radiosurgical planning in another way. Because the pixel intensity on a CT image reﬂects the electron density of the tissues imaged, the pixel intensity can be mathematically converted into electron density maps (electrons per cm3). This information can be used to deﬁne isodose lines in radiosurgical planning. Without this information, actual radiation dose delivered can deviate from the desired dose by as much as 20% as a result of tissue inhomogeneity [1]. Despite yielding improved anatomic resolution as well as electron density information, delineation of soft tissue structures by CT imaging is suboptimal, even with the aid of intravenous contrast agents. For the most part, delineation of soft tissue structures is achieved by the use of MRI, especially for targets in the cranial base.

9

10

c.c. chen et al.

FIGURE 2-1. CT, MR, and MRS images from a patient with a left cerebellar tumor. (a) CT imaging without intravenous contrast shows a poorly deﬁned left cerebellar mass with effacement of the fourth ventricle and displacement of the brain stem. (b) Intravenous contrast administration improves the anatomic resolution of the left cerebellar mass, revealing a densely enhancing mass with surrounding edema. (c) The same lesion is visualized using T1-weighted MRI. (d) MRI after gadolinium administration reveals a heterogeneously enhancing mass. The homogeneously enhancing tissue on CT is further resolved into tissues of varying intensity on MRI, demonstrating the superiority of MRI over CT in soft tissue resolution. The numbered grid corresponds with the MR spectral arrays shown in (e). The grid is placed over

normal-appearing tissue. (e) The various chemical peaks are as indicated in box 9. The thick arrow indicates the choline peak. The arrowhead represents the creatine peak. The thin arrow designates the N-acetylaspartate (NAA) peak. The MRS in box 9 is typical of normal tissue, with comparable choline and creatine peaks and a notable NAA peak. (f) The numbered MRS grid is placed over the diseased tissue. The MRS is shown in (g). (g) The various chemical peaks are labeled in box 1. The diseased tissue shows an elevated choline peak (thick arrow) relative to a diminished creatine peak (arrowhead). The NAA peak is also decreased (thin arrow) relative to normal tissue. The accumulation of lactate (double arrow) is another signature of diseased tissue.

Magnetic Resonance Imaging

source of error involves the imperfection of the input magnetic ﬁeld. The input magnetic ﬁeld in MRI is produced by electric currents passing through sets of mutually orthogonal coils. Ideally, the magnetic ﬁeld generated should be uniform such that a linear relationship between space and resonance frequency can be established [3]. However, such uniform ﬁelds cannot be easily achieved in practice. This phenomenon is referred to as gradient ﬁeld nonlinearity and tends to escalate with increasing distance from the central axis of the main magnet. For the most part, gradient ﬁeld nonlinearity can be corrected computationally. Prior to correction, gradient ﬁeld nonlinearity can induce spatial distortions as large as 4 mm. After computational correction, the distortion is minimized to <1 mm [4]. A more complex MR distortion that is more difﬁcult to correct computationally involves electromagnetic interactions between the imaged tissue and the input magnetic ﬁeld. This distortion is often referred to as resonance offset. Resonance offset occurs because hydrogen atoms carry with them an inherent magnetic ﬁeld. Thus, placement of hydrogen-bearing tissues in a magnetic ﬁeld necessarily induces a perturbation in the input magnetic ﬁeld. This perturbation disrupts the linear relationship between space and resonance frequency as to produce

The human body consists primarily of fat and water, both having a high content of hydrogen atoms. MRI exploits the nuclear spin property of these hydrogen atoms as a means to attain soft tissue resolution. In MRI, a radiofrequency pulse is applied to the imaged subject. As a result, the nuclear spin states of these atoms shift from that of equilibrium to that of excitation. To return to their equilibrium state, the law of energy conservation dictates that an energy equal to that absorbed must be emitted. The energy release between nuclear spin state transitions can be measured and analyzed. Because the process of energy absorption and emission is affected by the local chemical environment, hydrogen atoms in soft tissues of varying chemical composition will absorb and emit differential energy. Mathematical transformation of this information yields ﬁne-resolution maps of soft tissue structures (Fig. 2-1c, d). Because tumor and normal tissues often differ in chemical composition [2], the same principle allows delineation of these tissue types. Because of the complexity of the nuclear interactions involved in MRI, the modality is subject to many sources of error, resulting in distortion of the image obtained. One such

2.

neuroimaging in radiosurgery planning and evaluation

geometric distortions. The physics of this perturbation is complex because it depends on the inherent magnetic properties as well as the volume and shape of the imaged object. Resonance offset distortions tend to be largest at the interface of materials that differ in magnetic properties, such as at the airwater interface. In anatomic imaging, this translates into large distortions at the air-bone or air-tissue interfaces. Studies reveal that distortions at these interfaces can be as large as 2 mm [3, 5, 6]. Although the development of higher-strength magnets has allowed for improved resolution of soft tissue structures as well as minimization of geometric distortions related to gradient ﬁeld nonlinearity, higher-strength magnets do not address the issue of resonance offset. Because resonance offset is a product of the input magnetic ﬁeld and the local ﬁeld imposed by the imaged tissue, increasing the strength of the input magnetic ﬁeld will magnify the effects of resonance offset [7]. The accuracy of MR as a stand-alone imaging modality has been determined by a number of investigators [8–12]. Most investigators report a localization uncertainty of 2 to 3 mm [8– 11], but maximal absolute errors of 7 to 8 mm have also been reported [12]. These studies reveal that error in ﬁducial localization is ampliﬁed by subsequent mathematical transformation. Though the degree of localization uncertainty varies between studies, the reported uncertainty consistently remains greater than 1 mm, failing to achieve the current radiosurgical standard set forth by the American Society of Therapeutic Radiology and Oncology (ASTRO) [13–15]. Another downside of MRI as it pertains to radiosurgical planning is the absence of electron density information (see earlier “Computed Tomography Imaging” section). Contrary to CT imaging where the image is derived based on differential electron density, pixel intensities in MR images bear no correlation with electron density. For radiosurgical planning using MR as the only imaging modality, image processing and assignment of hypothetical electron density values are required. Such strategies have led to suboptimal radiosurgical plans [16]. Motion artifact is another consideration affecting spatial accuracy in MRI. The prolonged duration required for image acquisition increases the potential for patient movement. Even with a cooperative patient, motion artifact occurs with breathing and internal physiologic motions. The resultant motion compromises the accuracy of spatial resolution. Though MRI is inadequate as a stand-alone modality in radiosurgical planning, combining MR and CT images has led to radiosurgical plans that are superior to plans derived from each modality alone [17–20]. For example, Shuman et al. reported that the incorporation of MR information into CTbased radiotherapy plans resulted in better deﬁnition of tumor volume in 53% of the cases [18]. These observations have led to the development of algorithms for superimposing MR and CT images.

CT-MR Image Integration The differences between CT and MRI illustrate the conceptual distinction between geometric and diagnostic accuracy. Although CT imaging is geometrically accurate due to absence of spatial distortion effects, disease tissues are often missed by this modality. As such, CT imaging is diagnostically inaccurate.

11

On the other hand, due to enhanced soft tissue resolution, MRI affords enhanced diagnostic accuracy; however, the spatial accuracy is limited due to MR distortion effects. Algorithms have been developed to maximally utilize the different types of information afforded by CT and MRI (Fig. 2-2). Simple approaches to image integration involve manual superposition of equivalent views of MR and CT images, using bony landmarks as correlation points. Such approaches, however, are labor intensive and error-prone with uncertainties of up to 8 mm [21]. Advances in computational technology have allowed for the development of automated algorithms for superposition of CT and MR images in three-dimensional space. One way of integrating CT and MR images requires that the patient be placed in an immobilization device, such as the stereotactic frame. The immobilization device minimizes motion artifacts and ensures that the images are acquired in a predetermined manner. Fiduciary markers are used to establish the spatial relationship between the target and the head frame. Additionally, they serve as coregistration points between the MR and CT images. Because image acquisition and correlative points are ﬁxed in space in a predetermined way, this mode of image fusion is sometimes referred to as prospective image coregistration [14]. Alternatively, image coregistration can be done with images that are not acquired in a predetermined manner. This mode of image fusion is also known as retrospective coregistration. Retrospective image coregistration relies on matching corresponding anatomic landmarks instead of ﬁduciary markers. The CT and MR images are integrated on the basis of aligning these anatomic landmarks [22]. Various computational techniques, including point matching [23], line matching, and iterative matching [24], have been developed for retrospective image superposition. Whether one method is superior to another remains an area of research. In general, with proper training and quality control, most current algorithms will coregister MR and CT images to an uncertainty of 1 to 2 mm using prospective registration and of 2 to 3 mm using retrospective registration [14].

Contrast Administration Contrast administration takes advantage of the observation that disease processes, such as tumor growth, often result in vascular encroachment or faulty angiogenesis [2]. These processes allow contrast material to escape the vasculature and preferentially accumulate in the diseased tissue. The accumulation of contrast material can be easily visualized on CT or MRI (Fig. 2-1b, d). In malignant gliomas for instance, contrast enhancement correlates with diseased tissue. Kelly et al. evaluated 195 brain tumor biopsies acquired from various locations relative to the contrast-enhancing regions of CT or MRI scans and showed that the regions of contrast enhancement best correlated with regions of tumor burden [25]. Because contrast-enhancing volumes are used for radiosurgery target deﬁnition, diseased tissues without contrast enhancement often escape therapy. Investigators have used various functional imaging modalities to address this issue. Although these modalities hold tremendous promise, they are limited by poor anatomic resolution. As such, functional imaging

12

c.c. chen et al.

FIGURE 2-2. Fusion of MR and CT images in radiosurgical planning. (a) CT image of a patient with left frontal metastatic lesion. The image is selected to illustrate the continuity of the ventricular contour and the cranial vault as landmarks to gauge the spatial discrepancy when comparing MR and CT images. The lesion is not shown in (a). (b) Equivalent T1-weighted MR image of the CT image shown in (a). Again, note the continuity of the ventricular contour and the cranial vault. (c) Superposition of (a) and (b) without correction of MR distortion shows spatial discrepancy as evidenced by the discontinuity of ventricular contour and the cranial vault at the transition point. The CT-derived image is shown on the top-half panel. The MR image is shown on the

bottom-half panel. (d) After computational correction of MR distortion, continuity of the transition point is restored and various anatomic landmarks are coregistered. (e) Three-dimensional view of the metastatic lesion in relation to the stereotactic frame and a surface rendering of the patients head. The lines interconnecting the small red, green, and yellow spheres indicate the planes through which radiation is delivered. (f) Axial, (g) sagittal, and (h) coronal views of the lesion on MRI after correcting for MR distortion effects. The colored lines represent the various isodose contours. The magnitude of the radiation delivered is shown in the right lower corner of each panel.

is most useful in conjunction with traditional anatomic imaging modalities. In many instances, the clinical applications of functional imaging remain investigational.

intense upregulation of glucose metabolism in tumor cells. Once inside the cell, 18F-FDG undergoes phosphorylation to yield an intermediate that cannot undergo further metabolic processing or cellular export. The phosphorylated intermediate is, therefore, preferentially transported into tumor cells and trapped there [26]. Studies investigating the use of 18F-FDG in guiding radiosurgery for treatment of gliomas yielded mixed results. Tralins et al. reported a series of 27 patients who underwent conventional MR or CT scanning as well as 18F-FDG PET. In this study, a multivariate analysis revealed 18F-FDG PET ﬁndings as the only variable that retained statistical signiﬁcance in predicting time to tumor progression and overall survival. Moreover, the 18F-FDG PET deﬁned target volumes differing from those deﬁned by MR or CT imaging by at least 25% in all patients [27]. Gross et al., on the other hand, reported that regions of 18F-FDG abnormal uptake closely correlated with regions of contrast enhancement in their 18 patients. In a minority of patients, 18F-FDG PET did affect target volume deﬁnition. These changes, however, were not associated with improved survival when compared with historical controls [28]. Likewise, Prado et al. reported that the inclusion of PET scan data minimally altered radiation planning in most patients [29]. These conﬂicting data can, in part, be attributed to the variability and subjectivity involved in PET image interpretation.

Positron Emission Tomography and Single-Photon-Emission Computed Tomography One type of functional imaging relies on visualizing tracer molecules that preferentially accumulate in diseased tissues. This type of imaging includes positron emission tomography (PET) and single-photon-emission computed tomography (SPECT). PET is designed to detect the preferential accumulation of positron-emitting radioactive tracer compounds in the diseased tissue. The emitted positron collides with an electron to yield opposing gamma rays. These emissions are detected by a gamma-ray camera, thereby generating images of regional radioactivity (Fig. 2-3). Similarly, SPECT is designed to detect the preferential accumulation of tracer compounds bearing photon-emitting isotopes. Photon emission is detected by a rotating gamma camera detection system and reconstructed into three-dimensional tomographic images. In tumor neuroimaging, the enhanced metabolic state of the tumor cells is often exploited to achieve preferential tracer accumulation in these tissues. For instance, 18-ﬂuorodeoxyglucose (18F-FDG), a commonly used PET tracer, is preferentially transported into tumor cells relative to normal cells due to an

2.

neuroimaging in radiosurgery planning and evaluation

13

FIGURE 2-3. Use of 18F-FDG PET in neuroimaging. (a) A left parietal-occipital mass (arrow) that shows gadolinium enhancement on T1-weighted MR imaging. (b) The same lesion shows increased 18FFDG accumulation on PET imaging (arrow). The signal intensity of the lesion on 18F-FDG PET imaging is comparable with those of the gray matter. (c) Bilateral gadolinium enhancement in a patient undergoing treatment for malignant glioma on a T1-weighted MR sequence. (d)

PET imaging showed preferential 18F-FDG accumulation in the right hemispheric lesion (arrow). Biopsy specimen of this lesion reveals recurrent glioma. (From Davis WK, Boyko OB, Hoffman JM, et al. [18F] 2-ﬂuoro-2deoxyglucose-positron emission tomography correlation of gadolinium-enhanced MR imaging of central nervous system neoplasia. AJNR 1993; 14:515–523. Copyright by American Society of Neuroradiology.)

Because normal gray matter exhibits high physiologic uptake of 18F-FDG, the distinction between normal gray matter and tumor is sometimes subjective, especially in cases of signiﬁcant anatomic distortion secondary to mass effect. Additionally, regions known to be at high risk for tumor inﬁltration, such as regions of edema, often display increased 18F-FDG uptake. In many instances, different thresholds are used for deﬁning abnormal 18F-FDG uptake [27–30]. Despite their differences, the various studies suggest that, in selected patients, the inclusion of PET or SPECT can aid in the deﬁnition of target volumes in radiosurgery for gliomas. The extent of clinical beneﬁt and the criteria for patient selection await future investigations. The likelihood of routine PET or SPECT for radiosurgical planning should increase with the development of tracer compounds that exhibit high speciﬁcity to diseased tissues.

of 46 patients with malignant gliomas, patients with MRS abnormality outside of the MR-deﬁned target volume showed decreased median survival relative to those with MRS abnormality inside the MR-deﬁned tumor volume (10.7 months vs. 17.4 months, p = 0.002) [32]. Other studies have conﬁrmed the correlation between untreated MRS abnormality and worse prognosis [33–37]. These studies suggest that MRS data should be taken into consideration in target volume determination for the treatment of gliomas.

Magnetic Resonance Spectroscopy Another type of functional imaging capitalizes on the ability of MRI to measure the levels of biochemical metabolites. Three metabolites commonly used to distinguish tumor and healthy tissue include choline, creatine, and N-acetylaspartate (NAA) (Fig. 2-1d–g; Fig. 2-4). Choline is an essential component of the cell membrane. The level of choline reﬂects the rapidity of membrane turnover and is increased in rapidly proliferating tumors. Creatine is a metabolic intermediate for the synthesis of phosphocreatine, an energy source for cellular metabolism. The level of creatine corresponds with the level of cellular energy reserve, which is decreased in tumor tissues. NAA is a marker for neuronal differentiation and is decreased in tumors [2]. Using elevated choline and decreased NAA as criteria, Pirzkall et al. compared the magnetic resonance spectroscopy (MRS)-deﬁned tumor volume to that deﬁned by contrastenhanced MRI for malignant gliomas. The authors report that the MRS-deﬁned volume extended outside of the MRI-deﬁned volume by <2 cm in 88% of the patients [31]. In another study

MR Perfusion Imaging Like levels of biochemical metabolites, perfusion parameters such as cerebral blood volume (CBV) can be measured using MRI techniques (Fig. 2-5). CBV is measured by monitoring the transit of a rapid bolus of contrast with respect to time. This parameter is an indirect measure of tissue vascularity, a property often associated with tumor burden. It is, therefore, not surprising that MR-derived measurements of cerebral blood volume correlate with tumor grading and clinical outcome. In a series of 28 patients with gliomas, pretreatment high CBV intensity was associated with shorter median survival [2, 38, 39]. The use of CBV in radiosurgical planning is limited by several factors. CBV values in tumor volumes are often greater than CBVs of normal white matter but comparable with CBVs of normal gray matter. Thus, distinguishing tumor and cortex is problematic, especially in the context of anatomic distortion caused by large tumors. Additionally, regions of increased CBV correlate well with regions of contrast enhancement. As such, incorporation of CBV information will only alter radiosurgical plans in a subpopulation of patients.

MR Diffusion Weighted Imaging The white matter in a normal cerebrum is organized into tracts that allow communication between cortical neurons. As a result of this high degree of organization, water molecules in the

14

c.c. chen et al.

FIGURE 2-4. MRS images from a patient with high-grade glioma. (a) T1-weighted MRI reveals an area of hypointensity in the basal frontal lobe involving the anterior limb of the internal capsule. (b) The lesion does not enhance with gadolinium administration but shows increased signal intensity on a (c) FLAIR sequence. (d) The numbered MRS grid is placed over normal-appearing tissue. The MRS is shown in (e). (e) The various chemical peaks are as indicated in box 1. The

MRS shown is typical of normal tissue, with comparable choline (thick arrow) and creatine peaks (arrowhead) and a notable NAA peak (thin arrow). (f) The numbered MRS grid is placed over the diseased tissue. The MRS is shown in (g). (g) The various chemical peaks are labeled in box 10. The diseased tissue shows an elevated choline peak (thick arrow) relative to a diminished creatine peak (arrowhead). The NAA peak is also decreased (thin arrow) relative to normal tissue.

cerebral cortex diffuse in a highly directional manner. The extent of this directional diffusion can be estimated using specialized MR techniques and is referred to as apparent diffusion coefﬁcient (ADC) (Fig. 2-6). Because gliomas often distort cerebral architecture, regions with altered ADC are expected to correlate with tumor burden. This expectation was demonstrated in several studies [40–42]. These studies revealed that

patients with lower ADC values in the tumor volume showed shorter median survival than patients with normal or nearnormal ADC values (12 months vs. 21.7 months). These studies suggest that ADC maps may be helpful in guiding radiosurgical planning, especially in cases where conventional MRI and ADC maps yield discordant information with regard to tumor volume.

FIGURE 2-5. Application of MR perfusion imaging in tumor diagnosis. (a) T1-weighted MRI shows a heterogeneous right temporal lesion. (b) Gadolinium administration reveals peripheral enhancement and septation of the lesion. A nodular enhancing region is seen in the

right lower corner of the lesion. (c) MR perfusion shows increased cerebral blood volume (CBV) correlating with the contrast-enhancing rim, septation, and nodule. Biopsy of the lesion reveals a grade IV glioma.

2.

neuroimaging in radiosurgery planning and evaluation

15

FIGURE 2-6. MR diffusion imaging in tumor assessment. (a) T1weighted MRI shows a heterogeneous right temporal lesion. (b) Gadolinium administration reveals a multicystic lesion with a central region

of enhancement. (c) MR diffusion imaging shows decreased apparent diffusion coefﬁcient (ADC) signal in the region of central enhancement. Biopsy of the lesion reveals a grade IV glioma.

Combined Imaging Modality

demonstrated by other studies, AVM obliteration occurred 2 to 3 years after radiosurgical treatment. The risk of hemorrhage during this latency interval was the same as that seen in untreated AVMs [45–53]. Radiosurgical treatment of AVMs requires a precise deﬁnition of the nidus in three-dimensional space. This precise deﬁnition can be achieved only by the combination of MRI, CT imaging, and cerebral angiography. Whereas MR and CT images afford anatomic resolution, they cannot discriminate between the AVM nidus and the feeding arteries and draining veins [54]. This distinction is crucial because the goal of AVM treatment lies in the obliteration of the former while sparing the latter. On the other hand, cerebral angiography offers a limited deﬁnition of the nidus margin without the corresponding MR and CT imaging, especially in cases of irregularly shaped AVMs (Fig. 2-7). Conventional angiographic studies yield two-dimensional projections of the AVM nidus, a three-dimensional lesion. The spatial information lost as a result of dimensional reduction represents a source of inaccuracy in nidus deﬁnition [48, 55–57]. For instance, angiographic projections may outline different AVM nidus margins depending on the angle of projection and the nidus geometry [54, 56]. Not surprisingly, studies examining the value of incorporating CT and MR information into conventional angiography–based radiosurgical plans yielded data supporting the superiority of the combined approach. In one study, inclusion of CT imaging in angiography-based plans resulted in a mean isocenter shift of 3.6 mm in 44 of 81 (54%) patients and changes in the diameter of collimator beam in an equal number of patients [55]. Other studies have reported similar ﬁndings [57]. CT- and MR-based angiograms are proposed alternatives to conventional angiogram in radiosurgical planning (Fig. 2-8); however, the spatial accuracy and the resolution of vessel architecture afforded by CT and MR angiograms are insufﬁcient as stand-alone modalities [58, 59]. Tanaka et al. compared AVM resolution by MR angiography (MRA), CT angiography (CTA), and conventional angiogram in terms of feeding vessel and draining vein visualization [59]. In this study, only 20% to 30% of feeding vessels and draining veins detected by conventional angiogram were identiﬁed by MRA or CTA. Combined use of CTA and MRA did not further improve AVM resolution. The work by Aoyama et al. further illustrated the inadequacy of CT and MR angiograms as stand-alone modalities in radiosurgery

Given the complexity of physiology and pathology underlying tumor biology, it is unlikely that any single imaging modality will allow perfect deﬁnition of the diseased volume [43]. The failure to precisely deﬁne tumor volume will result in inadequate or excessive radiation treatment and suboptimal clinical outcomes. Precise tumor volume deﬁnition likely requires a synthesis of information obtained from contrast-enhanced CT or MRI, PET, SPECT, MRS, CBV, and ADC in a meaningful way. The optimal algorithm for the synthesis of this information remains an area of active research.

Cerebral Angiography The primary application of cerebral angiography in radiosurgical planning lies in the treatment of cerebral arteriovenous malformations (AVMs). AVMs represent abnormal communications between vessels of disproportionately unbalanced hydrodynamic stress. The region of abnormal communication is referred to as the nidus. Due to increased hydrodynamic stress, the nidus in an AVM is at high risk for rupture, causing intracranial hemorrhage. The goal of AVM treatment is to eliminate this risk by obliterating the AVM nidus. Estimates of the annual risk for hemorrhage secondary to AVM rupture lie in the range 2.2% to 4.0%, with an associated fatality of roughly 10%. Whereas surgical resection is the treatment of choice for AVMs, radiosurgical treatment is often performed in cases of surgical inaccessibility, patient preference, or severe preoperative morbidity. The risks associated with surgical resection of AVMs are graded by the Spetzler-Martin scale, which reﬂects the importance of AVM size, location, and venous drainage. Higher grades are associated with increased postoperative morbidity and mortality. In most series, complete surgical resection of grade I or II AVMs is associated with minimal surgical complications (0 to 10%). Resection of grade IV or V AVMs is associated with complication rates exceeding 40% [44]. The efﬁcacy of radiosurgery in AVM treatment has been demonstrated in a number of studies [45–53]. For instance, Pollock et al. reported the results of stereotactic Gamma Knife radiosurgery for 65 patients with Spetzler-Martin grade I or II AVMs who opted not to undergo surgery. The series reported a cure rate of 84% and a complication rate of 7.7% [52]. As was

16

c.c. chen et al. FIGURE 2-7. MRI as an adjunct to conventional angiography in deﬁning AVM nidus. (a) Anterior-posterior and (b) lateral views of an AVM during mid–arterial phase (right internal carotid artery injection) revealed a large AVM in the right parietal temporal region supplied by distal right middle cerebral artery, posterior cerebral artery, and external carotid artery branches. (c) Coronal T2-weighted and (d) axial T1-weighted images further reﬁne the anatomic geometry of this complex AVM.

FIGURE 2-8. Comparison of CTA, MRA, and conventional angiography in AVM resolution. (a) Axial CTA image of a patient who presented with a left parietal-occipital intracranial hemorrhage. An AVM nidus is visualized on the superoposterior aspect of the hematoma. (b) Axial image reconstruction affords improved anatomic resolution of the AVM nidus, demonstrating feeder vessels from the middle cerebral artery and the posterior cerebral artery. (c) Sagittal reconstruction of the CTA images shows venous drainage of the AVM into the superior sagittal sinus. (d) Three-dimensional reconstruction of the CTA images

allows visualization of the spatial geometry of the AVM nidus. (e) Cranial-caudal view of an MRA demonstrating the left parietaloccipital hematoma as well as an enlarged, left middle cerebral artery branch. The AVM nidus is poorly visualized. (f) Anterior-posterior and (g) lateral views of the AVM during the mid–arterial phase of a conventional angiogram (left internal carotid artery injection) show detailed view of the arterial feeders and venous drainage of the AVM. The resolution afforded by conventional angiogram in this regard is superior to that of CTA or MRA.

2.

neuroimaging in radiosurgery planning and evaluation

planning for AVM treatment [58]. These authors investigated the spatial discrepancy between AVM targets as deﬁned by MRA, CTA, and conventional stereotactic angiogram. The authors reported a mean discrepancy of 2 to 3 mm in the center of the target volume when MRA and CTA targets were compared with conventional angiography–deﬁned targets. Discrepancies were noted in the left-right, anterior-posterior, and cranio-caudal directions. Independent of the size of the AVM, the discrepancy was greater than 5 mm in one-third of the cases. Despite their limitations, CT and MR angiograms allow for a more precise deﬁnition of the AVM nidus when used in conjunction with conventional angiography [48, 55–57]. In a series of 28 patients, Kondziolka et al. reported that MR angiography provided information on irregularly shaped AVMs that was not visualized by conventional angiography alone. The utility of MR angiography was especially evident in situations where conventional angiography showed different superior and inferior nidus margins on different projections. The improved nidus deﬁnition resulted in modiﬁcation of treatment plans in 16 of 28 (55%) cases [56]. Others have reported similar results [48, 55, 57]. Conventional angiography for the purpose of radiosurgical planning is generally done by placing the patient in an immobilizing stereotactic head frame in order to ensure maximal spatial accuracy. Recent advances in image acquisition and retrospective image coregistration techniques have raised questions about the necessity of such practice. Three-dimensional rotatory angiography is an imaging technique that combines the principles of CT with those of conventional angiography. As the contrast material is injected through the cerebral vasculature, X-ray transmissions from a rotating emitter are detected and digitally converted into a high-resolution view of the cerebral architecture in three dimensions. Radiosurgical planes derived using nonstereotactic three-dimensional angiography have been compared with those derived from conventional stereotactic angiography. In one study, this comparison revealed target coordinate discordance in 5 of the 20 patients. Coordinate discordance ranged from 0.3 to 1 mm with a mean of 0.7 mm [54]. Today, the gold standard for radiosurgical treatment of AVMs remains a combination of stereotactic cerebral angiography, CT-based or MR-based imaging, (including CTA and MRA), and three-dimensional angiography. The resultant information enables a detailed geometric reconstruction of nidus anatomy that is required in complex treatment strategies [60]. With improvement in algorithms for image acquisition and coregistration, nonstereotactic three-dimensional angiography may supplant the need for stereotactic angiography in radiosurgical planning for selected patients.

Radiologic Considerations in Radiosurgical Planning The following section will review pertinent radiologic features that affect radiosurgical planning, including size of lesion and proximity to critical neuroanatomic structures. Nonradiologic factors that affect radiosurgical planning are reviewed elsewhere [61, 62].

17

Size of the Lesion The size of the lesion treated remains an important criterion for determining the appropriateness of radiosurgery versus radiotherapy. As the size of the irradiated target volume increases, undesired radiation of the surrounding nontarget tissue increases in an exponential manner. The clinical impact of this geometric inevitability is magniﬁed by the higher doses of radiation delivered in radiosurgery. Generally, single-dose irradiation of normal cerebral parenchyma should be restricted to <12 to 15 Gy, because doses exceeding this range are associated with increased risks of neurologic deﬁcits [63]. As a result of this dose restriction, lesions with sizes >4 cm are usually not treated radiosurgically.

Critical Neuroanatomic Structures Radiosurgery may be contraindicated for the treatment of lesions in the immediacy of highly radiosensitive neuroanatomic structures. For instance, because of the radiosensitivity of cranial nerve II, radiosurgery is contraindicated in the treatment of lesions in the proximity of or intrinsic to the optic nerve, chiasm, or tracts. The distance between the tumor margin and the optic apparatus should be at least 4 mm before radiosurgery is considered [64, 65]. The dose delivered to the optic apparatus should be restricted to less than 10 Gy to minimize the risk of optic neuropathy. Like cranial nerve II, cranial nerves VII and VIII are more radiosensitive than other neuroanatomic structures [66]. Detailed delineation of these structures on neuroimaging is required for radiosurgical planning. Lesion localization relative to regional cerebral anatomy is another consideration in radiosurgical planning because this spatial relationship is a major predictor for posttreatment complications. Flickinger et al. reviewed 332 patients with AVMs treated with radiosurgery and correlated the risk of posttreatment neurologic injury to the location of the lesion. The risk for neurologic deﬁcit is maximal when the lesions are located in the deep gray matters (thalamus, basal ganglia) and brain stem (pons/midbrain). Minimal risk for deﬁcit was seen in the lesions located in the frontal and temporal lobe [63]. Thus, depending on lesion location, radiation doses should be adjusted to minimize the risk of posttreatment neurologic deﬁcit.

Evaluation of Treatment Efﬁcacy Radiation can induce imaging changes that are unrelated to the underlying disease process. Misinterpretation of these imaging results can lead to inappropriate treatment. For instance, radiation induces cytotoxicity and also disrupts cerebral vascular architecture. These changes, often referred to as radiation necrosis, can lead to imaging ﬁndings that are indistinguishable from tumor recurrence on contrast-enhanced MRI. As another example, radiation can induce inﬂammatory changes, causing a temporary increase in the volume of contrast enhancement that is unrelated to tumor regrowth [67, 68]. Thus, optimal patient management requires an understanding of the imaging ﬁndings as they relate to clinical outcome. This issue will be addressed

18

c.c. chen et al.

in the following section. Means of distinguishing radiation necrosis from tumor recurrence will be discussed in the context of malignant gliomas and intracranial metastasis. Temporary increases in the volume of contrast enhancement will be reviewed in the context of vestibular neuroma. The utility of serial MR in the management of trigeminal neuralgia will also be discussed. Finally, the complexity of posttreatment management after AVM radiosurgery will be reviewed.

Malignant Gliomas and Metastatic Disease For malignant gliomas, radiosurgery represents a second-line treatment option, usually offered in the context of recurrence after primary therapy or as adjuvant therapy. In contrast, radiosurgery is a ﬁrst-line treatment option for intracranial metastasis without signiﬁcant mass effect. In both diseases, typical follow-up regimens include neurologic examination and contrast-enhanced MRI 6 to 8 weeks after radiosurgery and then at roughly 3-month intervals. More frequent imaging is performed with neurologic deterioration or with interval enlargements of the treated lesion [67, 68]. For both malignant gliomas and metastatic diseases, response to radiosurgery is radiologically deﬁned by a decrease in the size of the contrast-enhancing volume on conventional MRI [67, 68]. Complete and partial responses are generally deﬁned as the complete disappearance or a partial decrease in the size of the enhancing lesion, respectively. No change in the contrast-enhancing volume is usually referred to as stable disease. Using these deﬁnitions, Ross et al. described the imaging ﬁndings after radiosurgery for malignant gliomas and intracranial metastasis. Based on imaging obtained roughly 3 months after radiosurgery, 17% of the patients with malignant glioma showed partial or complete response; 10% showed stable disease, and 73% showed an increase in the size of contrast enhancement. In patients with intracranial metastasis, images obtained in the same time frame revealed 51% complete or partial response, 27% stable disease, and 22% contrastenhancement size increase [67]. Comparable results are reported by other studies [69–72]. The use of corticosteroids is common in the treatment of cerebral edema in both malignant gliomas and intracranial metastasis. Questions are often raised as to whether corticosteroid administration affects MRI ﬁndings after radiosurgery. Case series addressing this issue revealed that whereas corticosteroids reduced the extent of peritumoral edema on T2weighted MRI, their administration did not alter the size of the contrast-enhancing volume [67]. The physiology underlying increased contrast uptake is complex and cannot be equated with treatment failure in all cases. Two independent series that detailed the imaging changes after radiosurgical treatment of intracranial metastatic disease revealed that 6% to 12% of all radiosurgically treated lesions displayed a transient increase in contrast uptake [67, 68]. This transient increase occurred at 3 months after treatment (range, 2 to 10 months) and resolved after an additional 6 months (range, 2 to 6 months). Microsurgical resection of these contrast-enhancing volumes revealed hyalinized thrombosis, tumor necrosis, and granulomatous changes [73]. Genuine treatment failures, documented by surgical biopsy and enlargement of

contrast-enhancing volume, on the other hand, were more likely to occur at 6 months after treatment (range, 3 to 24 months). Thus, although contrast-enhancing volume represents an important predictor for therapeutic efﬁcacy [67], it must be interpreted with caution. Another caveat in equating increased contrast uptake with radiosurgical failure involves the phenomenon of radiation necrosis. Radiation necrosis is a term used to describe radiologic changes (primarily visualized in the form of increasing size of contrast-enhancing volume) resulting from radiation-induced cytotoxicity that is unrelated to the underlying disease process. Some reports suggest that radiation necrosis is more likely to produce a fuzzy, indiscrete pattern of enhancement in contrast with the discrete pattern of enhancement seen in tumor recurrence [68]; however, 10% to 20% of post-radiosurgery patients develop radiologic ﬁndings indistinguishable from tumor recurrence [74]. The functional imaging modalities described earlier have been employed as a means to better distinguish radiation necrosis from tumor recurrence. Early results are promising in this regard. In general, whereas functional imaging modalities exhibit high degrees of sensitivity for detecting tumor recurrence, the speciﬁcity remains somewhat poor. For instance, Tusyuguchi et al. compared the methionine PET ﬁndings in eight cases of biopsy-proven glioma recurrence with six cases of biopsy-proven radiation necrosis. They calculated the ratio of methionine accumulation in the regions of contrast enhancement relative to regions of normal gray matter. They found this ratio elevated in cases of tumor recurrence when compared with cases of radiation necrosis. In this study, the sensitivity and speciﬁcity of methionine PET for tumor recurrence detection were 100% and 60%, respectively [75]. Comparable results are reported for thallium 201 SPECT [76]. Measurement of cerebral blood volume (CBV) on MRI represents another proxy for tumor recurrence [38, 39, 77]. Essig et al. described their experience with 18 patients imaged at 6 weeks and 3 months after radiosurgery for solitary metastasis. In this study, a decrease in the CBV of the radiosurgically treated volume at 6 weeks predicted treatment outcome with a sensitivity of 97% and a speciﬁcity of 71% [77]. Perhaps the most promising imaging modality for determining radiation necrosis versus tumor recurrence involves the use of MRS. As previously described, choline is a cell membrane component that reﬂects the extent of membrane turnover. Elevated levels of choline are associated with rapidly proliferating tumors and poor clinical outcomes [2]. The level of creatine is a proxy for cellular metabolism and is decreased in tumor cells. Using measurements of choline and creatine as guides, Rabinov et al. were able to distinguish recurrent tumor from radiation necrosis in 13 of 14 cases [74]. Others have reported similar ﬁndings [35, 78]. Future imaging evaluation after radiosurgery likely will involve the synthesis of information obtained from different imaging modalities. The development of algorithms for data integration requires clinical-imaging correlation. To date, most studies carried out for such purpose are retrospective in design and involve a small number of patients. Ultimately, prospective and randomized studies will be required for the purpose of correlating imaging ﬁndings with clinical outcomes.

2.

neuroimaging in radiosurgery planning and evaluation

Benign Lesions: Acoustic Neuroma Imaging ﬁndings after radiosurgical treatment for acoustic neuromas are discussed as a review of the basic principles underlying the management of benign intracranial lesions after radiosurgery. Acoustic neuromas, also known as vestibular schwannomas, are rare, benign tumors that arise from the Schwann cells associated with the eighth cranial nerve. Their incidence is approximately 1 in 100,000 in the general population. Historically, surgical resection has been the treatment of choice. Two important goals of acoustic neuroma surgery are facial nerve and hearing preservation. The likelihood of achieving these goals is largely a function of the tumor size. In recent years, radiosurgery has emerged as an alternative to microsurgical treatment for small acoustic neuromas (<2 cm). In the largest series reported to date (827 patients over a span of 10 years at the University of Pittsburgh), radiosurgical administration of 12 to 20 Gy to the schwannoma achieved local control in 97% of the cases after 10 years [79]. Other series have reported similar ﬁndings [80–84]. After radiosurgery, the rate of House-Brackman grade I/II facial nerve function preservation ranged from 70% to 95%; Robertson-Garner serviceable hearing preservation ranged from 13% to 40% [80–84]. In multiple retrospective comparisons of radiosurgery and microsurgical resections, no statistically signiﬁcant difference was observed in the rate of tumor recurrence, facial nerve preservation, or hearing preservation [82, 85]. Currently, MRI with contrast enhancement is the gold standard for the evaluation of treatment response [86, 87]. Typically, follow-up neuroimaging and clinical examinations are performed at 6 and 12 months during the ﬁrst year and every 12 months thereafter. In the largest clinical series to date, Nakamura et al. reported their experiences with serial MRI after Gamma Knife radiosurgery for vestibular schwannoma. They classiﬁed the changes in the posttreatment contrast-enhancing volume into four categories. The ﬁrst category consisted of schwannomas that showed initial enlargement followed by sustained regression (25/78, or 32%). This temporary enlargement peaked in roughly 1 year and regressed within an additional 2 years. In many cases, the schwannoma doubled in size before regression. The tumor may not have regressed to the size of the initial lesion. Instead, many tumors regressed to an intermediate size and remained stable at that size. The second category consisted of tumors that showed repeated enlargement followed by regression (8/78, or 9%). Many of these tumors were cystic schwannomas (5/8), with size ﬂuctuations resulting from enlargement or collapse of the cystic component. Size ﬂuctuations in noncystic schwannomas also occurred (3/8). Again, the size enlargement could reach a doubling of the initial lesion size before regression. The third category consisted of schwannomas that remained stable in size or regressed in size (21/78, or 27%). The fourth category consisted of continual tumor enlargement (7/78, or 9%) [87]. Microsurgical resection of tumors that showed radiologic “progression” often revealed hyalinized thrombosis, thickened vascular wall, and granulomatous changes [88, 89]. Thus, some of the cases of radiologic “progression” may have represented inﬂammatory changes rather than genuine treatment failure. Because up to 41% (combined category 1 and 2) of schwanno-

19

mas treated with radiosurgery displayed a temporary increase in the volume of contrast enhancement (sometimes doubling in size), cautious observation may be warranted for at least 2 years after treatment, provided the imaging changes are not associated with clinical deterioration or signiﬁcant compression of the brain stem. Similar temporary enlargement of contrast-enhancing volumes is seen after radiosurgery for pituitary adenomas and other benign diseases [90]. As such, in the management of patients with benign intracranial lesions after radiosurgery, surgical intervention or a second round of treatment should be reserved for cases with continual disease progression after or for cases with neurologic deterioration. Another radiologic ﬁnding pertinent to the management of vestibular schwannoma after radiosurgery is that of hydrocephalus. It is estimated that roughly 10% of patients with vestibular schwannoma develop communicating hydrocephalus after radiosurgical or radiation treatment [91]. In most cases, the hydrocephalus is not associated with tumor enlargement or cerebrospinal ﬂuid (CSF) ﬂow obstruction. The etiology of the hydrocephalus is unclear though many investigators attribute the phenomenon to CSF malabsorption secondary to tumor necrosis. It is also unclear whether radiation contributes to the process, as a comparable percentage of vestibular schwannoma patients without radiation treatment develops communicating hydrocephalus [92]. Regardless of the etiology, prompt identiﬁcation of ventricular enlargement on imaging and clinical ﬁndings associated with hydrocephalus are needed in order to ensure timely neurosurgical intervention.

Trigeminal Neuralgia Imaging ﬁndings after radiosurgical treatment for trigeminal neuralgia are reviewed to illustrate a clinical scenario where routine, serial MRI is not warranted. Trigeminal neuralgia is a facial pain syndrome consisting of paroxysmal, lancinating pain occurring in the distribution of cranial nerve V. Most patients with trigeminal neuralgia are successfully treated with anticonvulsives, antidepressants, neuroleptics, or opioids. Options available for the treatment of medically resistant trigeminal neuralgia include microvascular decompression, thermal, chemical, or radiofrequency ablative procedures, and radiosurgery [93]. In most studies, excellent responses to radiosurgery are reported in 70% to 90% of patients after treatment [94, 95]. Many institutions obtain a MR contrast-enhanced scan approximately 6 months after radiosurgical treatment. The scan is performed primarily for the purpose of target site veriﬁcation. In one report, all patients treated with 45 Gy at the 50% isodose line developed contrast enhancement of the target zone within 6 months of treatment [94]. The detection of contrast enhancement on cranial nerve V after radiosurgery, therefore, served as a conﬁrmation of accurate targeting. Studies correlating clinical responses to contrast enhancement of cranial nerve V have yielded mixed results. Some studies suggest that the exact region of enhancement relative to the pontine edge and along the retrogasserian portion of cranial nerve V correlate well with treatment outcome [96–98]. Others report poor correlation between contrast enhancement and clinical response [99]. In many instances, beneﬁcial clinical

20

c.c. chen et al.

responses or treatment failures are apparent before the onset of contrast enhancement [99]. Despite the lack of consistent correlation between radiologic ﬁndings and clinical response, a case can be made for obtaining a MR contrast-enhanced scan at the 6-month followup evaluation, because it allows target conﬁrmation as well as an assessment of potential brain-stem injuries. Serial MRI in patients without neurologic or radiologic changes at the initial posttreatment scan, however, does not appear warranted. In patients who exhibit good clinical response to radiosurgery without incurring neurologic deﬁcits, it is also reasonable to forgo the MRI at the 6-month follow-up evaluation, because the imaging ﬁndings are unlikely to alter patient management.

Arteriovenous Malformations Therapeutic efﬁcacy of radiosurgery for AVMs depends on radiation-induced endothelial cell and mesenchymal cell proliferation, causing progressive vasoocclusion [100]. Although it is generally agreed that the peak effect of this process occurs between 2 and 3 years after radiation treatment, the actual time course of AVM obliteration varies widely [45–53]. Complete angiographic obliteration, as determined by conventional angiography, remains the gold standard for deﬁning treatment success and can occur as early as 4 months or as late as 5 years after treatment [101, 102]. Because of the variable time course in therapeutic response, the frequency and timing of neuroimaging follow-up evaluation vary widely between institutions [103–105]. Because of the risks associated with conventional cerebral angiography and because most AVMs obliterate between 2 and 3 years after radiosurgery, there is a tendency to delay cerebral angiography until the presumed time of obliteration [103, 105]. Some investigators, however, advocate early cerebral angiography on grounds that ﬁndings on these studies are highly predictive of the ﬁnal treatment outcome. Oppenheim et al. reported their experience with 138 patients radiosurgically treated for AVMs. These patients underwent early angiography 6 to 18 months after treatment. Eighty-four percent of the radiosurgically treated AVMs with evidence of early regression (deﬁned as >75% reduction in size at the time of early angiography) eventually developed complete obliteration on subsequent angiograms. On the other hand, only 10% of the AVMs without evidence of early regression (<50% reduction in size on the early angiogram) developed complete obliteration. The authors concluded that the identiﬁcation of AVMs unresponsive to radiosurgery at a early stage will facilitate planning for subsequent treatment strategies [104]. To avoid the risks associated with conventional cerebral angiograms, investigators have identiﬁed CT and MR imaging ﬁndings that correlate well with treatment response. One such ﬁnding involves contrast enhancement in the region of the AVM nidus after radiosurgery. Contrast enhancement of the AVM nidus on CT imaging was ﬁrst described in two patients with eventual AVM obliteration after radiosurgery [106]. Subsequent studies reported good correlation between nidus enhancement on CT or MR imaging after radiosurgery and therapeutic efﬁcacy [102, 107]. These studies revealed that the volume of contrast enhancement corresponds with the radiosurgical target volume. The degree of enhancement increases

with time and is correlated with a reduction in the nidus size on conventional angiography. The onset for contrast enhancement is typically 6 to 24 months after radiosurgery. Contrast enhancement on CT tends to resolve 1 to 2 years after angiographic demonstration of complete AVM obliteration, whereas enhancement on MRI tends to persist even after disappearance of contrast enhancement on CT [102]. MR and CT angiograms are also used to evaluate cerebral AVMs after radiosurgery. Though the spatial resolution of CT and MR imaging remains poor, AVM changes visualized using these modalities correlate well with those detected using conventional angiography [102, 108–110]. As such, serial MR or CT angiograms are often performed for routine monitoring while the deﬁnitive conventional angiography is postponed until 2 to 3 years after radiosurgery. Radiosurgery of AVMs inevitably results in irradiation of the surrounding nontarget tissues. It is, therefore, not surprising that an estimated 28% to 50% of the patients undergoing treatment develop T2 signal abnormalities in these regions [100, 111]. The onset of these ﬁndings occurs between 1 day and 44 months after treatment. Regression of the signal abnormality is seen in 80% to 90% of the cases and tends to occur 5 to 8 months after the initial onset [112]. As previously discussed, the risk of neurologic deﬁcit expected due to these abnormalities depends on their location. Maximal risk for deﬁcit is expected in cases of signal abnormalities in the deep gray matters (thalamus, basal ganglia) and brain stem (pons/midbrain). Minimal risk of deﬁcit is expected with frontal and temporal lobe abnormalities [63]. Whereas conventional angiography represents the gold standard for deﬁning therapeutic efﬁcacy for AVM treatment, disappearance of the AVM on angiography after radiosurgery does not always indicate disease eradication. Shin et al. followed 236 radiosurgery-treated AVMs between 1 and 133 months after angiographic conﬁrmation of obliteration. The authors identiﬁed four patients who developed intracranial hemorrhage between 16 and 51 months after angiographic conﬁrmation. No evidence of residual AVM was found on retrospective review of the conﬁrmation angiograms. Two of the patients underwent surgical resection. Histologic analysis of the resection specimen revealed evidence of vasoocclusion as well as small residual AVM vessels. The only radiologic ﬁndings associated with these hemorrhages were the persistence of contrast enhancement on CT and MRI after angiographic evidence of AVM obliteration [113]. Given these ﬁndings, yearly followup evaluation after angiographic evidence of AVM obliteration may be warranted. Treatment of patients with AVMs remains one of the most complex and challenging in the ﬁeld of radiosurgery. Optimal patient management requires an understanding of the pathophysiology, neuroanatomy, as well as the radiologic manifestations after treatment. As such, treatment efforts should involve collaborative inputs from experienced radiation oncologists, neurosurgeons, neuroradiologists, and the patient.

Neuroimaging for Radiation-Associated Secondary Tumors The probability of secondary tumors arising from radiosurgery is quite low [114]. Thus, once a patient has achieved a positive

2.

neuroimaging in radiosurgery planning and evaluation

result from treatment (e.g., complete obliteration of an AVM documented by angiography), we do not recommend further imaging to look for secondary tumor formation.

Conclusion There is no doubt that advances in neuroimaging will help to reﬁne all aspects of radiosurgery and improve treatment efﬁcacy. For many modalities, clinical applications remain poorly deﬁned and await further investigation. For radio surgeons and therapists, the challenge lies in understanding the basis and the limitations associated with the various imaging modalities. Ultimately, prospective and randomized studies correlating imaging ﬁndings and clinical outcomes are required for developing guidelines for optimal patient care.

References 1. Battista J, Rider W, Van DJ. Computed tomography for radiotherapy planning. Int J Radiat Oncol Biol Phys 1980; 6:99–107. 2. Nelson SJ, Cha S. Imaging glioblastoma multiforme. Cancer J 2003; 9:134–145. 3. Sumanaweera TS et al. Characterization of spatial distortion in magnetic resonance imaging and its implications for stereotactic surgery. Neurosurgery 1994; 35:696–703; discussion 703–704. 4. Chang H, Fitzpatrick JM. A technique for accurate magnetic resonance imaging in the presence of ﬁeld inhomogeneities. IEEE Trans Med Imaging 1992; 11:319–329. 5. Sumanaweera T et al. MR susceptibility misregistration correction. IEEE Trans Med Imaging 1993; 12:251–259. 6. Sumanaweera T et al. Quantifying MRI geometric distortion in tissue. Magn Reson Med 1994; 31:40–47. 7. Mack A et al. Analyzing 3-tesla magnetic resonance imaging units for implementation in radiosurgery. J Neurosurg 2005; 102(Suppl): 158–164. 8. Bednarz G et al. Evaluation of the spatial accuracy of magnetic resonance imaging-based stereotactic target localization for gamma knife radiosurgery of functional disorders. Neurosurgery 1999; 45:1156–1161; discussion 1161–113. 9. Kondziolka D et al. A comparison between magnetic resonance imaging and computed tomography for stereotactic coordinate determination. Neurosurgery 1992; 30:402–406; discussion 406– 407. 10. Orth RC et al. Development of a unique phantom to assess the geometric accuracy of magnetic resonance imaging for stereotactic localization. Neurosurgery 1999; 45:1423–1429; discussion 1429– 1431. 11. Walton L et al. A phantom study to assess the accuracy of stereotactic localization, using T1-weighted magnetic resonance imaging with the Leksell stereotactic system. Neurosurgery 1996; 38:170– 176; discussion 176–178. 12. Walton L et al. Stereotactic localization with magnetic resonance imaging: a phantom study to compare the accuracy obtained using two-dimensional and three-dimensional data acquisitions. Neurosurgery 1997; 41:131–137; discussion 137–139. 13. Task Force on Stereotactic Radiosurgery. Consensus statement on stereotactic radiosurgery quality improvement. The American Society for Therapeutic Radiology and Oncology, Task Force on Stereotactic Radiosurgery and the American Association of Neurological Surgeons. Int J Radiat Oncol Biol Phys 1994; 28:527–530. 14. Khoo VS et al. Magnetic resonance imaging (MRI): considerations and applications in radiotherapy treatment planning. Radiother Oncol 1997; 42:1–15.

21

15. Shaw E et al. Radiation Therapy Oncology Group: radiosurgery quality assurance guidelines. Int J Radiat Oncol Biol Phys 1993; 27:1231–1239. 16. Schad LR et al. Radiotherapy treatment planning of basal meningiomas: improved tumor localization by correlation of CT and MR imaging data. Radiother Oncol 1992; 25:56–62. 17. Just M et al. MRI-assisted radiation therapy planning of brain tumors—clinical experiences in 17 patients. Magn Reson Imaging 1991; 9:173–177. 18. Shuman WP et al. MR imaging in radiation therapy planning. Work in progress. Radiology 1985; 156:143–147. 19. Shuman WP et al. The utility of MR in planning the radiation therapy of oligodendroglioma. AJR Am J Roentgenol 1987; 148: 595–600. 20. Ten Haken RK et al. A quantitative assessment of the addition of MRI to CT-based, 3-D treatment planning of brain tumors. Radiother Oncol 1992; 25:121–133. 21. De Salles AA et al. Transposition of target information from the magnetic resonance and computed tomography scan images to conventional X-ray stereotactic space. Appl Neurophysiol 1987; 50:23–32. 22. Pelizzari CA et al. Accurate three-dimensional registration of CT, PET, and/or MR images of the brain. J Comput Assist Tomogr 1989; 13:20–26. 23. Judnick JW et al. Radiotherapy technique integrates MRI into CT. Radiol Technol 1992; 64:82–89. 24. Evans PM et al. Image comparison techniques for use with megavoltage imaging systems. Br J Radiol 1992; 65:701–709. 25. Kelly PJ et al. Stereotactic histologic correlations of computed tomography- and magnetic resonance imaging-deﬁned abnormalities in patients with glial neoplasms. Mayo Clin Proc 1987; 62: 450–459. 26. Apisarnthanarax S, Chao KS. Current imaging paradigms in radiation oncology. Radiat Res 2005; 163:1–25. 27. Tralins KS et al. Volumetric analysis of 18F-FDG PET in glioblastoma multiforme: prognostic information and possible role in deﬁnition of target volumes in radiation dose escalation. J Nucl Med 2002; 43:1667–1673. 28. Gross MW et al. The value of F-18-ﬂuorodeoxyglucose PET for the 3-D radiation treatment planning of malignant gliomas. Int J Radiat Oncol Biol Phys 1998; 41:989–995. 29. Pardo FS et al. Functional cerebral imaging in the evaluation and radiotherapeutic treatment planning of patients with malignant glioma. Int J Radiat Oncol Biol Phys 1994; 30:663–669. 30. Nuutinen J et al. Radiotherapy treatment planning and long-term follow-up with [(11)C]methionine PET in patients with low-grade astrocytoma. Int J Radiat Oncol Biol Phys 2000; 48:43–52. 31. Pirzkall A et al. MR-spectroscopy guided target delineation for high-grade gliomas. Int J Radiat Oncol Biol Phys 2001; 50:915–928. 32. Chan A, Nelson S. Use of 1H magnetic resonance spectroscopic imaging in managing recurrent glioma patients undergoing radiosurgery. ISMRM Workshop Proceedings on MR of Cancer 2002; 227. 33. Chan AA et al. Proton magnetic resonance spectroscopy imaging in the evaluation of patients undergoing gamma knife surgery for Grade IV glioma. J Neurosurg 2004; 101:467–475. 34. Graves EE et al. A preliminary study of the prognostic value of proton magnetic resonance spectroscopic imaging in gamma knife radiosurgery of recurrent malignant gliomas. Neurosurgery 2000; 46:319–326; discussion 26–28. 35. Graves EE et al. Serial proton MR spectroscopic imaging of recurrent malignant gliomas after gamma knife radiosurgery. AJNR Am J Neuroradiol 2001; 22:613–624. 36. McKnight TR et al. Histopathological validation of a three-dimensional magnetic resonance spectroscopy index as a predictor of tumor presence. J Neurosurg 2002; 97:794–802.

22

c.c. chen et al.

37. Nelson SJ et al. In vivo molecular imaging for planning radiation therapy of gliomas: an application of 1H MRSI. J Magn Reson Imaging 2002; 16:464–476. 38. Cha S. Perfusion MR imaging: basic principles and clinical applications. Magn Reson Imaging Clin N Am 2003; 11:403–413. 39. Cha S. Perfusion MR imaging of brain tumors. Top Magn Reson Imaging 2004; 15:279–289. 40. Castillo M et al. Apparent diffusion coefﬁcients in the evaluation of high-grade cerebral gliomas. AJNR Am J Neuroradiol 2001; 22:60–64. 41. Moffat BA et al. Functional diffusion map: a noninvasive MRI biomarker for early stratiﬁcation of clinical brain tumor response. Proc Natl Acad Sci USA 2005; 102:5524–5529. 42. Nelson S. Imaging of brain tumors after therapy. Neuroimaging Clin N Am 1999; 9:801–819. 43. Benjamin R, Capparella J, Brown A. Classiﬁcation of glioblastoma multiforme in adults by molecular genetics. Cancer J 2003; 9:982–990. 44. Spetzler RF, Martin NA. A proposed grading system for arteriovenous malformations. J Neurosurg 1986; 65:476–483. 45. Flickinger JC, Kondziolka D, Lunsford LD. Clinical applications of stereotactic radiosurgery. Cancer Treat Res 1998; 93:283–297. 46. Friedman WA. LINAC radiosurgery. Neurosurg Clin N Am 1990; 1:991–1008. 47. Friedman WA, Bova FJ. Linear accelerator radiosurgery for arteriovenous malformations. J Neurosurg 1992; 77:832–841. 48. Friedman WA, Bova FJ, Mendenhall WM. Linear accelerator radiosurgery for arteriovenous malformations: the relationship of size to outcome. J Neurosurg 1995; 82:180–189. 49. Friedman WA, Bova FJ, Spiegelmann R. Linear accelerator radiosurgery at the University of Florida. Neurosurg Clin N Am 1992; 3:141–166. 50. Kondziolka D, Lunsford LD. The case for and against AVM radiosurgery. Clin Neurosurg 2001; 48:96–110. 51. Lunsford LD et al. Stereotactic radiosurgery of brain vascular malformations. Neurosurg Clin N Am 1992; 3:79–98. 52. Pollock BE et al. Patient outcomes after stereotactic radiosurgery for “operable” arteriovenous malformations. Neurosurgery 1994; 35:1–7; discussion 7–8. 53. Steiner L et al. Clinical outcome of radiosurgery for cerebral arteriovenous malformations. J Neurosurg 1992; 77:1–8. 54. Colombo F et al. Three-dimensional angiography for radiosurgical treatment planning for arteriovenous malformations. J Neurosurg 2003; 98:536–543. 55. Blatt DR, Friedman WA, Bova FJ. Modiﬁcations based on computed tomographic imaging in planning the radiosurgical treatment of arteriovenous malformations. Neurosurgery 1993; 33:588–595; discussion 595–596. 56. Kondziolka D et al. Stereotactic magnetic resonance angiography for targeting in arteriovenous malformation radiosurgery. Neurosurgery 1994; 35:585–590; discussion 590–591. 57. Levy RP et al. Computed tomography slice-by-slice target-volume delineation for stereotactic proton irradiation of large intracranial arteriovenous malformations: an iterative approach using angiography, computed tomography, and magnetic resonance imaging. Int J Radiat Oncol Biol Phys 1996; 35:555–564. 58. Aoyama H et al. Comparison of imaging modalities for the accurate delineation of arteriovenous malformation, with reference to stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 2005; 62:1232–1238. 59. Tanaka H et al. Initial experience with helical CT and 3D reconstruction in therapeutic planning of cerebral AVMs: comparison with 3D time-of-ﬂight MRA and digital subtraction angiography. J Comput Assist Tomogr 1997; 21:811–817. 60. Kakizawa Y et al. Compartments in arteriovenous malformation nidi demonstrated with rotational three-dimensional digital sub-

61.

62.

63.

64.

65.

66.

67.

68.

69.

70.

71. 72.

73.

74.

75.

76.

77.

78.

79.

80.

traction angiography by using selective microcatheterization. Report of three cases. J Neurosurg 2002; 96:770–774. Debus J et al. Brainstem tolerance to conformal radiotherapy of skull base tumors. Int J Radiat Oncol Biol Phys 1997; 39: 967–975. Debus J et al. Dose-volume tolerance of the brainstem after high-dose radiotherapy. Front Radiat Ther Oncol 1999; 33:305– 314. Flickinger JC et al. Development of a model to predict permanent symptomatic postradiosurgery injury for arteriovenous malformation patients. Arteriovenous Malformation Radiosurgery Study Group. Int J Radiat Oncol Biol Phys 2000; 46:1143–1148. Leber KA, Bergloff J, Pendl G. Dose-response tolerance of the visual pathways and cranial nerves of the cavernous sinus to stereotactic radiosurgery. J Neurosurg 1998; 88:43–50. Tishler RB et al. Tolerance of cranial nerves of the cavernous sinus to radiosurgery. Int J Radiat Oncol Biol Phys 1993; 27:215– 221. Foote KD et al. Analysis of risk factors associated with radiosurgery for vestibular schwannoma. J Neurosurg 2001; 95:440– 449. Huber PE et al. Transient enlargement of contrast uptake on MRI after linear accelerator (linac) stereotactic radiosurgery for brain metastases. Int J Radiat Oncol Biol Phys 2001; 49:1339– 1349. Ross DA et al. Imaging changes after stereotactic radiosurgery of primary and secondary malignant brain tumors. J Neurooncol 2002; 56:175–181. Auchter RM et al. A multiinstitutional outcome and prognostic factor analysis of radiosurgery for resectable single brain metastasis. Int J Radiat Oncol Biol Phys 1996; 35:27–35. Engenhart R et al. Long-term follow-up for brain metastases treated by percutaneous stereotactic single high-dose irradiation. Cancer 1993; 71:1353–1361. Flickinger JC et al. Radiosurgery: its role in brain metastasis management. Neurosurg Clin N Am 1996; 7:497–504. Mehta MP et al. Stereotactic radiosurgery for glioblastoma multiforme: report of a prospective study evaluating prognostic factors and analyzing long-term survival advantage. Int J Radiat Oncol Biol Phys 1994; 30:541–549. Shrieve DC et al. Comparison of stereotactic radiosurgery and brachytherapy in the treatment of recurrent glioblastoma multiforme. Neurosurgery 1995; 36:275–282; discussion 82–84. Rabinov JD et al. In vivo 3-T MR spectroscopy in the distinction of recurrent glioma versus radiation effects: initial experience. Radiology 2002; 225:871–879. Tsuyuguchi N et al. Methionine positron emission tomography for differentiation of recurrent brain tumor and radiation necrosis after stereotactic radiosurgery—in malignant glioma. Ann Nucl Med 2004; 18:291–296. Kimura T et al. Evaluation of the response of metastatic brain tumors to stereotactic radiosurgery by proton magnetic resonance spectroscopy, 201TlCl single-photon emission computerized tomography, and gadolinium-enhanced magnetic resonance imaging. J Neurosurg 2004; 100:835–841. Essig M et al. Assessment of brain metastases with dynamic susceptibility-weighted contrast-enhanced MR imaging: initial results. Radiology 2003; 228:193–199. Wald LL et al. Serial proton magnetic resonance spectroscopy imaging of glioblastoma multiforme after brachytherapy. J Neurosurg 1997; 87:525–534. Lunsford LD et al. Radiosurgery of vestibular schwannomas: summary of experience in 829 cases. J Neurosurg 2005; 102(Suppl): 195–199. Harsh G et al. Proton beam stereotactic radiosurgery of vestibular schwannomas. Int J Radiat Oncol Biol Phys 2002; 54:35–44.

2.

neuroimaging in radiosurgery planning and evaluation

81. Karpinos M et al. Treatment of acoustic neuroma: stereotactic radiosurgery vs. microsurgery. Int J Radiat Oncol Biol Phys 2002; 54:1410–1421. 82. Pollock B et al. Outcome analysis of acoustic neuroma management: a comparison of microsurgery and stereotactic radiosurgery. Neurosurgery 1995; 36:215–224. 83. Prasad D, Steiner M, Steiner L. Gamma surgery for vestibular schwannoma. Neurosugery 2000; 92:745–759. 84. Petit J et al. Reduced-dose radiosurgery for vestibular schwannomas. Neurosurgery 2001; 49:1299–1306. 85. Karpinos M et al. Treatment of acoustic neuroma: stereotactic radiosurgery vs. microsurgery. Int J Radiat Oncol Biol Phys 2002; 54:1410–1421. 86. Linskey ME, Lunsford LD, Flickinger JC. Neuroimaging of acoustic nerve sheath tumors after stereotaxic radiosurgery. AJNR Am J Neuroradiol 1991; 12:1165–1175. 87. Nakamura H et al. Serial follow-up MR imaging after gamma knife radiosurgery for vestibular schwannoma. AJNR Am J Neuroradiol 2000; 21:1540–1546. 88. Hirato M et al. Gamma knife radiosurgery for acoustic schwannoma: early effects and preservation of hearing. Neurol Med Chir (Tokyo) 1995; 35:737–741. 89. Kobayashi T, Tanaka T, Kida Y. The early effects of gamma knife on 40 cases of acoustic neurinoma. Acta Neurochir Suppl 1994; 62:93–97. 90. Tung GA et al. MR imaging of pituitary adenomas after gamma knife stereotactic radiosurgery. AJR Am J Roentgenol 2001; 177:919–924. 91. Sawamura Y et al. Management of vestibular schwannoma by fractionated stereotactic radiotherapy and associated cerebrospinal ﬂuid malabsorption. J Neurosurg 2003; 99:685–692. 92. Pirouzmand F, Tator CH, Rutka J. Management of hydrocephalus associated with vestibular schwannoma and other cerebellopontine angle tumors. Neurosurgery 2001; 48:1246–1253; discussion 53–54. 93. Cheshire WP. Trigeminal neuralgia: diagnosis and treatment. Curr Neurol Neurosci Rep 2005; 5:79–85. 94. Alberico RA, Fenstermaker RA, Lobel J. Focal enhancement of cranial nerve V after radiosurgery with the Leksell gamma knife: experience in 15 patients with medically refractory trigeminal neuralgia. AJNR Am J Neuroradiol 2001; 22:1944–1948. 95. Cheuk AV et al. Gamma knife surgery for trigeminal neuralgia: outcome, imaging, and brainstem correlates. Int J Radiat Oncol Biol Phys 2004; 60:537–541. 96. Kondziolka D, Lunsford LD, Flickinger JC. Gamma knife radiosurgery as the ﬁrst surgery for trigeminal neuralgia. Stereotact Funct Neurosurg 1998; 70(Suppl 1):187–191. 97. Urgosik D et al. Gamma knife treatment of trigeminal neuralgia: clinical and electrophysiological study. Stereotact Funct Neurosurg 1998; 70(Suppl 1):200–209. 98. Young RF et al. Gamma Knife radiosurgery for treatment of trigeminal neuralgia: idiopathic and tumor related. Neurology 1997; 48:608–614.

23

99. Friedman DP, Morales RE, Goldman HW. Role of enhanced MRI in the follow-up of patients with medically refractory trigeminal neuralgia undergoing stereotactic radiosurgery using the gamma knife: initial experience. J Comput Assist Tomogr 2001; 25:727–732. 100. Morikawa M et al. Radiosurgery for cerebral arteriovenous malformations: assessment of early phase magnetic resonance imaging and signiﬁcance of gadolinium-DTPA enhancement. Int J Radiat Oncol Biol Phys 1996; 34:663–675. 101. Lunsford LD et al. Stereotactic radiosurgery for arteriovenous malformations of the brain. J Neurosurg 1991; 75:512–524. 102. Yamamoto M et al. Long-term results of radiosurgery for arteriovenous malformation: neurodiagnostic imaging and histological studies of angiographically conﬁrmed nidus obliteration. Surg Neurol 1992; 37:219–230. 103. Kauczor HU et al. 3D TOF MR angiography of cerebral arteriovenous malformations after radiosurgery. J Comput Assist Tomogr 1993; 17:184–190. 104. Oppenheim C et al. Radiosurgery of cerebral arteriovenous malformations: is an early angiogram needed? AJNR Am J Neuroradiol 1999; 20:475–481. 105. Young C et al. Radiosurgery for arteriovenous malformations: the University of Toronto experience. Can J Neurol Sci 1997; 24:99–105. 106. Loefﬂer JS et al. Stereotactic radiosurgery of the brain using a standard linear accelerator: a study of early and late effects. Radiother Oncol 1990; 17:311–321. 107. Marks MP et al. Intracranial vascular malformations: imaging of charged-particle radiosurgery. Part II. Complications. Radiology 1988; 168:457–462. 108. Abe T et al. Magnetic resonance angiography of cerebral arteriovenous malformations. Neurol Med Chir (Tokyo) 1995; 35:580– 583. 109. Noorbehesht B, Fabrikant JI, Enzmann DR. Size determination of supratentorial arteriovenous malformations by MR, CT and angio. Neuroradiology 1987; 29:512–518. 110. Quisling RG et al. Persistent nidus blood ﬂow in cerebral arteriovenous malformation after stereotactic radiosurgery: MR imaging assessment. Radiology 1991; 180:785–791. 111. Flickinger JC et al. Radiosurgery and brain tolerance: an analysis of neurodiagnostic imaging changes after gamma knife radiosurgery for arteriovenous malformations. Int J Radiat Oncol Biol Phys 1992; 23:19–26. 112. Guo WY et al. Gamma knife surgery of cerebral arteriovenous malformations: serial MR imaging studies after radiosurgery. Int J Radiat Oncol Biol Phys 1993; 25:315–323. 113. Shin M et al. Risk of hemorrhage from an arteriovenous malformation conﬁrmed to have been obliterated on angiography after stereotactic radiosurgery. J Neurosurg 2005; 102:842– 846. 114. Loefﬂer JS, Niemierko A, Chapman PH. Second tumors after radiosurgery: tip of the iceberg or a bump in the road? Neurosurgery 2003; 52:1436–1440.

3

Techniques of Stereotactic Radiosurgery Chris Heller, Cheng Yu, and Michael L.J. Apuzzo

Introduction Stereotactic navigation and radiosurgery share a rich history highlighted by innovative minds, technological advances, and clinical success. Although the two concepts are in many ways intimately joined, radiosurgery could not exist without a thorough understanding of the fundamental principles of stereotaxis. Just as the original concept of radiosurgery was a natural, albeit inspired, application of the ideas of stereotactic navigation, advances in radiosurgery over the years have largely been preceded by innovations in stereotaxy.

History The ﬁrst device intended for experimental and surgical intervention in deep-seated structures of the brain was developed for animal studies by Sir Victor A.H. Horsley and Robert H. Clarke in 1908. Using an early atlas consisting of serial transverse sections of the animal brain, each structure of interest could be assigned three-dimensional coordinates and thus accurately located using the stereotactic device [1]. Almost 40 years passed, however, until Ernest A. Spiegel and Henry T. Wycis introduced a similar device intended for targeting locations within the human cerebrum [2]. Their device was similar to that of Horsley and Clarke in that it was a rectangular design using Cartesian coordinates in an orthogonal frame of reference, but it represented a considerable advancement in target localization. Spiegel and Wycis pioneered the use of intracranial structures rather than skull landmarks for navigational reference points [3, 4]. Plain X-ray and air and contrast ventriculography were used to visualize structures such as the foramen of Monroe, the pineal gland, and the habenular calciﬁcation. In 1952, they published their own brain atlas entitled Stereoencephalotomy, which provided the relative positions of these radiographic landmarks to structures of interest within the brain [3–5]. The ﬁrst clinical use of this device was for making functional lesions within the medial globus pallidus for the treatment of Huntington’s chorea. Early success would pave the way for future treatments of other movement disorders as well as epilepsy, psychiatric disorders, and pain [6] (see Fig. 3-1).

From Stereotaxis to Radiosurgery In 1949, Swedish neurosurgeon Lars Leksell (Fig. 3-2) introduced a stereotactic apparatus for application to intracerebral surgery [4]. Though the instrument developed by Spiegel and Wycis preceded it, Leksell’s design proved more versatile for a number of reasons. His innovative design placed the target at the center of a semicircular arc rather than a rectangular box, thus eliminating the need for trigonometric calculations for angled trajectories. It also allowed for alteration of the trajectory along the arc without the need to recalculate the coordinates of the target. Additionally, he introduced the use of skull pins for rigid ﬁxation of the base frame, which increased the overall precision of the system [7]. Two years after the introduction of his stereotactic frame, Dr. Leksell developed the concept of radiosurgery. With his ability to accurately localize a target in three-dimensional space, he postulated that narrow beams of radiation intersecting at a common point could be used to deliver high doses of energy to a chosen volume in space. Using this approach, lesions deep within the brain or at the base of the skull could be treated with minimal disruption of surrounding normal tissue [7]. Leksell’s original concept used multiple 300-kV stationary collimated photon beams focused on a common point [8]. Further reﬁnement of the idea led to the use of multiple ﬁxed gamma-emitting cobalt-60 (60Co) sources. This arrangement was called the Gamma Knife and was ﬁrst installed for clinical use at Sophiahemmet Hospital in Sweden in 1968 [9]. Designed as a means to ablate epileptogenic foci and to create functional lesions within deep ﬁber tracts and nuclei, the Gamma Knife was eventually applied to small tumors and arteriovenous malformations. The second Gamma Knife was installed at the Karolinska Hospital in Stockholm in 1974. It was capable of creating more spherical isodose distributions compared with the disk-shaped lesions seen with the ﬁrst unit, but with lesion localization still accomplished via diagnostic X-ray, angiography, and air ventriculography, the capabilities of radiosurgery were limited greatly by the imaging technology of the day.

Notable Contributors Although Leksell is viewed as the pioneer of stereotactic radiosurgery, other notable ﬁgures have contributed greatly to the

25

26

c. heller et al.

FIGURE 3-1. Examples of early stereotactic frames: (a) Horsley-Clarke device, (b) Leksell stereotactic device, (c) Todd-Wells stereotactic frame.

principles of target localization and thus the feasibility of radiosurgery. In 1957, Jean Talairach of the Centre Hospitalier Ste. Anne in Paris introduced his own stereotactic instrument, but more important was his publication of the Atlas d’Anatomie Stereotaxique [10]. Using this highly detailed atlas, he proposed the use of the line connecting the anterior and posterior commissurae as a common reference for localizing neural structures [6]. Among the most notable of these contributors were neurosurgeon Edwin M. Todd and engineer Trent H. Wells, who in 1965 introduced the Todd-Wells Stereotactic unit. Using an arc-based design similar to Leksell’s, it became the ﬁrst widely produced, distributed, and utilized stereotactic device of its era and led to the establishment of stereotaxy in mainstream neurosurgery [4].

In the late 1970s, Wells collaborated with Russell Brown and Theodore Roberts of the University of Utah to develop the Brown-Roberts-Wells (BRW) stereotactic device. The system’s innovative “N-bar” ﬁducial system allowed for precise localization of targets using the newly developed computed tomography (CT) imaging technology [4]. Using a handheld computer, the user was able to convert two-dimensional reference points from CT images into three-dimensional target coordinates. The Cosman-Roberts-Wells (CRW) frame was the followup to the BRW with design improvements including a movable arc, which allowed for an inﬁnite number of non-preselected entry points, and a phantom simulator for preoperative conﬁrmation of target coordinates [11]. Early experience with these devices at the University of Utah and the Los Angeles County/ University of Southern California Hospitals conﬁrmed a great leap forward in target localization utilizing modern imaging techniques and ushered in a new era in stereotactic radiosurgery [12]. The ability to precisely localize a target in three-dimensional space using modern imaging techniques expanded the therapeutic abilities of the Gamma Knife and made it possible for conventional radiotherapy devices such as the linear accelerator to be used for radiosurgery. Subsequent evolution of these techniques and the integration of magnetic resonance imaging (MRI), positron emission tomography (PET), and three-dimensional CT have seen the establishment of four widely utilized modalities of stereotactic radiosurgery: Gamma Knife, conventional and robot-assisted linear accelerator–based systems, and charged particle beams.

Gamma Knife

FIGURE 3-2. The original concept of radiosurgery is credited to Swedish Neurosurgeon Lars Leksell.

There are currently two radiosurgical devices that utilize 60Co sources, but the Leksell Gamma Knife is by far the most widely used (Fig. 3-3). The Leksell Gamma Knife differs from all other forms of radiosurgery in that the radiation energy is delivered to a ﬁxed point and the position of the target is manipulated to create the desired volume of treatment.

3.

techniques of stereotactic radiosurgery

FIGURE 3-3. The Gamma Knife treatment area at the USC University Hospital, Los Angeles, California.

Target Localization Target localization begins with the application of the Leksell head frame, which is ﬁxed to the skull using four percutaneous pins. Conscious sedation and local anesthetic minimize the discomfort of this procedure, which is generally well tolerated. Tumor location must be taken into account when placing the head frame. Interactions between the patient’s head, the frame, and the treatment helmet make it imperative that the target is as close to the center of the frame as possible. A variation of the “N-bar” ﬁducial device is then attached to the frame, and magnetic resonance (MR) and/or CT images are obtained and transferred to the Leksell Gamma Plan workstation where the treatment plan is created. With the ﬁrmly afﬁxed frame as a reference point, each pixel of imaging data is assigned an x, y, and z coordinate value. This allows the dose planning software to know the precise relative locations of the target, critical structures, and the focal point of the treatment beams.

patient is set either manually or by an automatic positioning system (APS) so that the x, y, and z coordinates of the target match up with the focal point of the radiation beams. The patient is repositioned throughout the treatment as needed to accommodate all of the shots from the treatment plan and create the ideal treatment volume and dose. Treatment time depends on the size and number of targeted lesions, the prescribed dose of radiation, and the strength of the cobalt source. With a half-life of 5.26 years, treatment times become prolonged as the cobalt source ages. Once the treatment is concluded, the head frame is removed, and the patient is discharged home to resume normal activities. The rigid ﬁxation of the Gamma Knife provides accuracy in the delivery of radiation and allows for high-energy treatment to a discrete volume with minimal damage to adjacent normal tissue. The need for a head frame makes hypofractionation impractical and, along with the limited amount of space inside the treatment helmet, limits the ability of the Gamma Knife to treat certain lesions at the extreme periphery of the intracranial space and precludes treatment of lesions elsewhere in the body.

Linear Accelerator–Based Systems The linear accelerator (Fig. 3-4) uses microwave technology to accelerate electrons in a part of the accelerator called the “wave guide” and then allows these electrons to collide with a heavy metal target. As a result of the collisions, high-energy photons are scattered from the target. A portion of these X-rays passes through a collimator to form a beam that is directed to the target. The linear accelerator (linac) is the most common instrument used for external beam radiotherapy, but with the integration of stereotactic localization, it can be used for radiosurgery as well [8]. There have been a number of recent advancements to linear accelerator radiosurgery such as frameless localization, which will be discussed later, and multileaf collimation. As the highenergy photon source travels along its treatment arc, the shape

Dose Planning When a target is placed at the focal point of the Gamma Knife, the volume of energy delivered is approximately spherical and termed a “shot.” A treatment plan consists of one or more “shots” positioned in such a way to conform to the often irregular volume of the target. Shots can differ from one another by their x, y, z coordinates, volume, and radiation dose. The volume of a shot depends on the size of the collimator used. The Gamma Knife includes four interchangeable treatment helmets with collimators measuring 4, 8, 14, and 18 mm in diameter. The radiation dose delivered to each volume depends on the length of time the target is left at the focal point of the beams. The plan is designed so that the entire lesion receives a high percentage (e.g., 50%) of the maximum treatment dose, and surrounding structures are relatively spared.

Treatment Once the dose planning is complete, the head frame is ﬁxed to the cast-steel helmet of the Gamma Knife. The position of the

27

FIGURE 3-4. A linac treatment facility.

28

c. heller et al.

FIGURE 3-5. The linac treatment gantry and couch rotate on axes that are perpendicular with respect to each other and intersect at the target point.

of the lesion from the beam’s point of view changes. This primarily affects beam conformity and dose distribution within the target. Multileaf collimators modify the cross-sectional shape of the treatment beam to match that of the lesion. Using a single isocenter, a multileaf collimated photon beam can achieve good beam conformity and homogeneous dose distribution. Accuracy of multileaf collimation, however, is a function of the size of each individual leaf and may diminish when applied to smaller lesions. Photon beams passing through circular collimators can likewise achieve good beam conformity through the use of multiple isocenters. This does, however, result in a more heterogeneous dose distribution throughout the target.

Target Localization Linac-based systems, when used for radiosurgery, have traditionally obtained target coordinates using CT and/or MR imaging with a frame afﬁxed to the head. Recent advances in hardware and software technology, however, have introduced frameless target localization and patient position tracking to linac-based radiosurgery. There are several methods for tracking patient and target position without the use of a head frame. TomoTherapy, which delivers high-energy photons from a device similar in design to a diagnostic CT scanner, has the ability to use “step down energy” for position-tracking purposes. During treatment, lower-energy photons can be used to periodically obtain images similar to conventional diagnostic CT imaging to conﬁrm proper position. Some systems utilize reformatted CT data in comparison with real-time diagnostic X-ray images to allow for patient positioning and tracking. Skeletal landmarks and/or radiofrequency or infrared ﬁducials can be used as reference points precluding the need for a rigid skull frame. Other linac-based systems include a lightweight and robotic linear accelerator (CyberKnife; Accuracy, Inc., Sunnyvale, CA) and modiﬁed linear accelerators to allow image guidance, such as Novalis (BrainLAB, Inc., Westchester, IL), Synergy (Eletka Oncology, Stockholm, Sweden), Trilogy (Varian Medical Systems, Palo Alto, CA), and Artiste (Siemens, Concord, CA).

Dose Planning As opposed to the ﬁxed-beam design of the Gamma Knife, linac-based systems produce a single beam that can be positioned in a number of different ways to create an ideal treatment plan. As discussed below, the treatment gantry rotates in a single plane about a horizontal axis. The use of treatment arcs rather than a stationary beam prevents surrounding tissues from receiving unwanted doses of radiation. As the couch rotates about a perpendicular axis with respect to the gantry, multiple non-coplanar treatment arcs are possible. The main goals with any treatment plan are to maximize tumor coverage with a sharp dose fall-off to limit damage to normal tissues. These goals are inﬂuenced primarily by collimator size, the number of isocenters used, and the number of treatment arcs per isocenter. Whereas a larger collimator with fewer isocenters may shorten treatment time, the opposite approach may be necessary to achieve ideal conformity and dose fall-off.

Treatment Once the treatment plan is ﬁnalized, the patient is placed on the treatment couch and the target location is conﬁrmed. In the case of traditional linac-based radiosurgery, the head frame is secured to either a ﬂoor stand or a couch-mounted holder, though frameless localization is becoming more prominent as previously discussed. The x, y, and z coordinates of the target are centered at a point intersected by the vertical axis around which the couch rotates. The linear accelerator rotates in a single plane around a horizontal axis that intersects the vertical axis of couch rotation at the target site [8] (Fig. 3-5). Treatment is typically done as an outpatient. When frameless localization is used, linac-based systems can be used for hypofractionated radiosurgery as well.

Charged Particle Beam Charged particle beam devices (Fig. 3-6) such as proton beam systems at Harvard Medical School and at Loma Linda University Medical Center in California function on the principle of

3.

techniques of stereotactic radiosurgery

29

Treatment The patient lies on the treatment couch and the head frame or body immobilization device is locked into place in preparation for treatment. There are many strategies as discussed previously that can be used to track and conﬁrm appropriate positioning throughout the treatment. In general, the treatment is an outpatient procedure.

Robot-Assisted Linac Radiosurgery

FIGURE 3-6. The Proton Beam at Loma Linda University Medical Center, Loma Linda, California.

The principles of radiosurgery and frameless stereotaxy were brought together in 1994 when the ﬁrst prototype of the CyberKnife stereotactic radiosurgery system (Fig. 3-7) was installed for clinical use at Stanford University. Developed in the early 1990s by neurosurgeon John Adler Jr., the CyberKnife consists of a compact 6-MV linac guided by a robotic arm with 6 degrees of freedom eliminating physical limitations on beam position [13]. It was designed to overcome the limitations that framed stereotactic localization places on radiosurgery.

Target Localization the Bragg peak [8]. When charged particles such as protons or helium ions are delivered to a lesion, the energy they release is related to their velocity. When they slow down and ﬁnally stop at a depth determined by the initial energy of the beam and type of tissue traversed, they release the majority of their energy in what is called the Bragg peak. The release of energy falls off precipitously thereafter. This allows for theoretically precise delivery of large amounts of energy to a well-deﬁned volume but makes it very important to precisely deﬁne the target location and monitor patient movement to avoid unwanted radiation dosage to normal tissues.

Target localization for CyberKnife treatment is both innovative and traditional combining the latest in computer technology with diagnostic X-ray imaging. Treatment planning consists of a ﬁne-cut, contrast-enhanced CT scan of the target area, which is reformatted by the computer software into multiple digital reconstructed radiographs (DRRs) to resemble diagnostic Xray images from a large number of viewing angles. Once the patient is lying on the treatment couch, a customﬁt immobilization device is placed over the treatment area to limit movement. The CyberKnife includes two ceiling-mounted X-ray imaging devices that, by projecting onto amorphous silicon detector plates rather than conventional X-ray ﬁlm,

Target Localization When employed as a tool for radiosurgery, target localization for charged particle beam treatment can be accomplished using a variety of patient immobilization and tracking strategies. Fiducial markers placed either on or beneath the skin surface can be used as reference points for imaging data as well as for patient tracking purposes during treatment. When intracranial lesions are targeted, a skull frame is used and is secured by either percutaneous pins or a custom-molded mouthpiece.

Dose Planning Charged particle beam systems need not utilize treatment arcs because of their unique properties as discussed above. Although a treatment plan may include more than one beam entry site to achieve desired tumor coverage and dose distribution, the beam is stationary while ﬁring. Beam conformity is achieved through the use of custom-designed apertures that are shaped to match the proﬁle of the lesion from the point of view of each beam used. An additional shielding device known as a compensator is custom-designed to match the depth proﬁle of each lesion. This ensures that the charged particles will release their energy at the desired depth throughout the lesion.

FIGURE 3-7. The CyberKnife treatment facility at the USC/Norris Cancer Hospital.

30

c. heller et al.

provide real-time digital images of the target area. Prior to commencing treatment, the baseline X-ray image is compared with the reformatted DRR images and the best match is determined based on skeletal landmarks and/or ﬁducial markers. Any translational or rotational errors are compensated for by the movable treatment couch during the initial set-up. Real-time X-ray images are obtained during treatment, and any movement by the patient is calculated in terms of deviation from the baseline. This information is directed to the robot arm that is guiding the linac, and appropriate corrections are made. The robot is able to correct for patient movements of up to 1 cm in three translational dimensions and 5 degrees in three rotational dimensions.

Dose Planning CT images are primarily used to outline the intended target when planning for treatment with the CyberKnife. When a lesion is not well visualized on CT, however, the CyberKnife software has the ability to merge MR and PET images with the existing CT images to enhance targeting. The target lesion is carefully outlined on the computer workstation along with all adjacent critical structures. The physicist is able to input a number of variables into the computer such as the desired radiation dose to the target lesion and maximum allowable dosage to critical structures. Using this information, the planning software creates a treatment plan consisting of multiple stationary beam positions to satisfy the desired dosage parameters. If a solution cannot be found to satisfy these parameters, the physicist must modify one or more variables such as tumor coverage, dose to critical structures, or collimator size. The CyberKnife’s robotic arm could theoretically position the beam at an inﬁnite number of points. For ease of calculation, however, there are 101 predetermined stopping points termed nodes. At each node, the beam can assume one of 12 positions resulting in a total of 1212 different possible beam locations at the planning software’s disposal [13].

Treatment The patient is placed supine on the treatment couch and ﬁtted with the custom-designed immobilization device to limit excessive movement. Once the baseline position is achieved, the treatment begins. The unique feature of the CyberKnife is its ability to reposition the treatment beam in real time to compensate for patient movement. Because of its frameless targeting capabilities, the CyberKnife is ideal for hypofractionated treatments.

Conclusion The development of radiosurgery from the early 20th century to the present time has been one of the most signiﬁcant instances of interdisciplinary collaboration in the history of medicine. The free exchange of ideas, technology, and innovation among physicians, physicists, and engineers has revolutionized the treatment of some of the most challenging medical disorders. Target localization has always been and continues to be central to the advancement of radiosurgery, which has become the standard of care in the primary and adjuvant treatment of a wide range of neoplastic diseases. Additionally, as the sophistication and clarity of functional brain mapping begins to approach that of anatomic imaging, radiosurgery is poised to realize its originally intended role as a noninvasive means of treatment for epilepsy, movement disorders, and psychiatric illnesses.

References 1. Horsley V, Clarke RH. The structure and function of the cerebellum examined by a new method. Brain 1908; 31:45–124. 2. Spiegel EA, Wycis HT, Marks M, Lee A. Stereotaxic apparatus for operations on the human brain. Science 1947; 106:349– 350. 3. Apuzzo MLJ, Chen JC. Stereotaxy, navigation and the temporal concatenation. Stereotact Funct Neurosurg 1999; 72:82–88. 4. Chen JC, Apuzzo MLJ. Localizing the point: evolving principles of surgical navigation. Clin Neurosurg 2000; 46:44–69. 5. Spiegel EA, Wycis HT. Stereoencephalotomy, Part 1. New York: Grune & Stratton, 1952. 6. Nashold B. The history of stereotactic neurosurgery. Stereotact Funct Neurosurg 1994; 62:29–40. 7. Leksell L. The stereotaxic method and radiosurgery of the brain. Acta Chir Scand 1951; 102:316–319. 8. Luxton G, Petrovich Z, Jozsef G, et al. Stereotactic radiosurgery: principles and comparison of treatment methods. Neurosurgery 1993; 32(2):241–258. 9. Leksell L. Stereotactic radiosurgery. J Neurol Neurosurg Psychiatry 1983; 46:797–803. 10. Talairach J, David M, Tournoux P, et al. Atlas D’Anatomie stereotaxique. Paris: Masson, 1957. 11. Couldwell WT, Apuzzo MLJ. Initial experience related to the use of the Cosman-Roberts-Wells stereotactic instrument. J Neurosurg 1990; 72:145–148. 12. Heilbrun MP, Roberts T, Apuzzo MLJ, et al. Preliminary experience with Brown-Roberts-Wells (BRW) Computerized Tomography Stereotaxic Guidance System. J Neurosurg 1983; 59:217–222. 13. Kuo J, Yu C, Petrovich Z, Apuzzo MLJ. The CyberKnife Stereotactic Radiosurgery System: description, installation, and an initial evaluation of use and functionality. Neurosurgery 2003; 53(5): 1235–1239.

PA R T II

Radiation Biology and Physics

31

4

The Physics of Stereotactic Radiosurgery Siyong Kim and Jatinder Palta

Introduction In radiosurgery, instead of using a surgical knife when treating a patient, high-energy ionizing radiation is the tool of choice. To understand the tumoricidal effects of ionizing radiation, it is important to know how radiation interacts with matter. This chapter describes general concepts and principles in radiation physics, including basic physics that are applicable to stereotactic radiosurgery. Commonly used delivery systems are also brieﬂy reviewed. This chapter is written for non–physics professionals, especially neurosurgeons and radiation oncologists.

The Radiation Source for Radiosurgery Ionizing radiation is any electromagnetic or particulate radiation capable of producing ions either directly or indirectly in its passage through matter [1, 2]. Among the many radiations we have, the photon, deﬁned as a discrete packet of electromagnetic energy, is the most commonly used for patient treatment [3, 4]. A photon is called by several different names depending on its energy level, such as radio waves, infrared, visible light, ultraviolet, and X-rays in the form of energy as illustrated in Figure 4-1. Only the X-ray range of photons is a form of ionizing radiation, and, though the energy range of X-rays is fairly wide, only a high-energy range is used for treatment. Photon treatment beams can be obtained either from radioisotope sources or from X-ray–generating machines. Currently, cobalt-60 (60Co) is the most widely used radioisotope [5–9] for radiosurgery, and the medical linear accelerator is the most popular X-ray– generating machine [10–14]. Protons, deﬁned as positively charged particles found in the nucleus of an atom, are also used for treatment [15–19]. Proton beams can be obtained from particle accelerators such as cyclotrons and synchrotrons [20], which cost a tremendous amount of money; thus, very few proton facilities exist worldwide, and there are currently only ﬁve in the United States. Because the majority of radiosurgery treatments are performed with photon beams, emphasis is placed on the photon beam.

Cobalt-60 Cobalt-60 is the radioactive isotope used as the radiation source in the Leksell Gamma Knife treatment machine (Elekta, Nor-

cross, GA), which is described later. A radioactive isotope is an atom with an unstable nucleus that tries to stabilize itself through a process called radioactive decay by emitting ionizing radiation such as alpha, beta, and/or gamma particles. When 60 Co undergoes radioactive decay, it emits beta particles and two strong gamma radiations, one with 1.17 MeV of energy and the other with 1.33 MeV of energy (Fig. 4-2). MeV, a commonly used unit of energy for particles in the treatment energy range, means 1 million electron-volts, where 1 electron-volt is the energy gained by an electron that accelerates through a potential difference of 1 volt. A real treatment beam of Gamma Knife can include photons of energy different from 1.17 and 1.33 MeV because some photons experience interactions with 60Co itself and/or its capsulation material and lose a part of that energy. Therefore, the effective energy of Gamma Knife is slightly lower than 1.25 MeV (the average of 1.17 and 1.33 MeV). Cobalt-60 ultimately decays to nonradioactive nickel. The half-life of 60Co, that is, the length of time needed for 60Co to lose half of its radioactivity from decay, is 5.26 years. At the end of 1 half-life, only 50% of the original radioactive material remains. This means that treatment time after 5 years is double that at the time of initial installation, and that the source of Gamma Knife needs to be replaced after a certain period of time to avoid long treatment times.

X-Ray Production in a Linear Accelerator One of the interesting characteristics of a fast-moving electron is that when it interacts with a material, it can produce an X-ray. A part or whole of the kinetic energy of the electron is transformed into electromagnetic energy. This principle is used to obtain X-rays in most X-ray production machines in which electrons are accelerated and induced to hit a target and then generate an X-ray. The amount of X-ray particles produced increases as the kinetic energy of the incident electrons increases. Electrons are accelerated by potential differences in diagnostic X-ray tubes up to several hundred kilo-electron-volts (keV). The acceleration mechanism in a linear accelerator, however, is somewhat different because it uses microwave technology similar to that used for radar to accelerate electrons in a linear tube, often called an accelerator tube [21]. In linear accelerators, electrons are usually accelerated to the energy range of 4 to 25 MeV. A 6-MeV electron is most often used to create an X-ray for radiosurgery. When an electron creates a

33

34

s. kim and j. palta

FIGURE 4-1. The electromagnetic spectrum is illustrated. Therapeutic X-ray or ionizing radiation is in the energy range of 105 eV and higher. The energy of photon beams commonly used in radiosurgery are higher than 106 eV.

photon, theoretically, the photon can take any energy from zero to the same as the energy of the incident electron. In other words, the photon beam produced by a 6-MeV electron beam can have an energy spectrum of 0 to 6 MeV. The energy of an X-ray produced in a linear accelerator is denoted MV instead of MeV. The major components and auxiliary systems of typical medical linear accelerators are shown in a block diagram (Fig. 4-3). The role of each part is brieﬂy explained below.

Power Supply The power supply provides direct current (DC) power to the modulator.

Electron Gun The electron gun produces electrons by thermionic emission. In a conducting material, the electrons all exist at or below the baseline electron energy at low temperatures. As the temperature is increased, some of these electrons have sufﬁcient energy to pass over the surface-potential barrier between the material and the vacuum. This process of increasing the temperature of a bulk material to increase the number of electrons that can leave the material is called thermionic emission. These electrons are pulse-injected into the accelerator tube according to the signal from the modulator.

Radiofrequency Source Modulator The modulator generates high-voltage pulses lasting a few microseconds that are simultaneously introduced to the radiofrequency source and the electron gun.

Radiofrequency (RF) is a range of electromagnetic frequencies above sound and below visible light, generally in the 30-kHz to 300-GHz range. RF is used for all broadcast transmissions including AM and FM radio, television, shortwave, microwave,

Bending Magnet 60 27 Co

5.26 y Electron Gun

Accelerator Tube/ Wave-guide System

Modulator

RF Generator (Magnetron)/ Amplifier (Klystron)

b 1– (99.8%) 0.313 MeV

g 1 (99.8%) 1.173 MeV

Power Supply

g 2 (100%) 1.332 MeV

60 28 Ni 60

FIGURE 4-2. For Co, most of the disintegrations (99.8%) are the emission of a b 1− with a maximum energy of 0.313 MeV. This leads to an excitation of 60Ni, which releases its energy quickly by emitting two gamma rays (1.173 MeV and 1.332 MeV) in cascade. In some disintegration (0.12%), a b 2− particle with a maximum energy of 1.486 MeV is released and leads to the lowest excited state of 60Ni.

RF Generator for Klystron

Target Treatment Head

b 2– (0.12%) 1.489 MeV

FIGURE 4-3. The power supply provides DC to the modulator, which provides a high-voltage pulse to the RF generator and electron gun simultaneously. The electron gun introduces the electron in pulse to the accelerator tube. Electron production occurs through the thermionic process. The RF generator introduces a microwave to the accelerator tube. The electron and the microwave are timed exactly to be met in the accelerator tube by the modulator. The accelerator tube accelerates the electrons via the resonance cavity. As the accelerated electron comes out of the accelerator tube, the path of the electron is bent through a bending magnet to hit the target and generates an X-ray.

4.

the physics of stereotactic radiosurgery

35

and satellite transmissions. In a linear accelerator used for radiosurgery, a magnetron is often the RF source. The magnetron is a diode-type electron tube that functions as a high-power oscillator to generate microwave pulses. Pulse duration is several microseconds long, and the repetition rate is several hundred pulses per second. Following the pulses from the modulator, microwave pulses are introduced into the accelerator tube through the wave-guide system. The typical frequency of microwaves used in a linear accelerator is about 3 GHz. In some designs, a klystron is used instead of a magnetron. The klystron is a microwave ampliﬁer rather than a source, therefore, it requires a low-power microwave oscillator called an RF driver as an RF source. In general, a klystron produces microwaves of higher power than the magnetron and is often used with high-energy linear accelerators (>10 MV).

Wave-Guide System The wave-guide system is made up of a pressurized rectangular tube through which microwave pulses generated by either a magnetron or klystron are transmitted into the accelerator tube.

Accelerator Tube An accelerator tube consists of many RF resonant cavities. An RF cavity allows power to be coupled into the particle beam. In an electron linear accelerator, RF accelerating cavities are basically microwave resonators. Resonance is a very common physical phenomenon that can be observed in many types of systems. Imagine a child on a swing, for example. If the child pushes the swing at just the right time, the push can make the swing go higher and higher with very little effort. The swing has a natural frequency of oscillation. Pushing “at just the right time” means that energy is put into the system at a frequency known as the fundamental frequency. Now the system is said to be in resonance condition. The amplitude of the motion can increase rapidly with resonance. An analogous phenomenon can occur in a linear accelerator with RF resonance, which is generally derived from an RF cavity that is typically in a right circular cylindrical shape with connecting holes to allow a charged particle’s beam to pass through for acceleration. Figure 4-4 shows a typical cylindrically symmetric RF cavity. In a fundamental RF mode, the electric ﬁeld is roughly parallel to the beam axis and radically decays to

FIGURE 4-4. A typical cylindrically symmetric cavity.

FIGURE 4-5. As the electron travels into a uniform magnetic ﬁeld, it experiences a downward force. This force causes the electron to travel in a circular path. The crosses in the ﬁgure indicate the uniform magnetic ﬁeld pointing into the page.

zero upon approach to the cavity walls. The pulsed electron particle beam traverses the cavity through the centered hole, creating an accelerating force along the axis of the cavity due to the electric ﬁeld.

Bending Magnet When an electron moves in a magnetic ﬁeld, it experiences force. The direction of the force is at a right angle to both the direction of the magnetic ﬁeld and the direction of the electron motion (Fig. 4-5). Using this principle, the path of accelerated electrons is controlled so that they can hit the target normally (note the electron path, bending magnet, and target in Fig. 4-3). Magnets used for this purpose are called bending magnets. In certain cases, straight-ahead beam designs can be applied without the bending magnet if the acceleration structure is short enough to be placed in a vertical direction. A linear accelerator in straight-ahead beam design is used in the CyberKnife unit, which will be better described later.

Treatment Head The major parts of the treatment head are the target (often called the X-ray target), primary collimator, ﬂattening ﬁlter, ion chamber, and secondary collimators (Fig. 4-6). Accelerated electrons collide with the target to generate an X-ray beam. For a given energy of electrons, more X-rays can be produced if a target material with a higher atomic number (Z number) is used. Tungsten and gold are commonly used as target materials. The production of an X-ray is based on a process called bremsstrahlung (meaning “braking radiation”), which is the result of radiative collisions between a fast-moving electron and a nucleus. When the electron passes near a nucleus, it can be deﬂected from its path by Coulomb forces of attraction and lose its energy as bremsstrahlung radiation (i.e., an X-ray). The Xrays are produced in all directions, but high-energy X-rays are produced in the forward direction (Fig. 4-7). Therefore, in a linear accelerator, higher photon intensity is observed in the area close to the central line (i.e., the extended line following the direction of the incident electron) below the X-ray target.

36

s. kim and j. palta

The amount of radiation produced by the linear accelerator is monitored by an ionization chamber, often called a monitor chamber. The monitor chamber not only controls the amount of radiation delivered to the patient, but it also checks and controls the uniformity of the delivered radiation. Whereas the primary collimator deﬁnes the maximum aperture of the radiation ﬁeld, a real treatment ﬁeld is determined by the secondary collimator, which deﬁnes a rectangular ﬁeld that ﬁts to each patient treatment in routine radiation therapy. In radiosurgery, smaller ﬁelds are needed. Thus, an additional collimator system, either a circular collimator or micro-multileaf collimator [22], is used.

A Photon’s Interaction with Matter FIGURE 4-6. A schematic of the treatment head of a medical linear accelerator is illustrated. Initially, as the accelerated electrons hit an X-ray target, an X-ray beam is generated via the bremsstrahlung process. This X-ray beam is collimated by the primary collimator. As it exits the primary collimator, the X-ray beam proﬁle is ﬂattened out by the ﬂattening ﬁlter. Then, it goes through the ion chamber, and, ﬁnally, it gets collimated by the secondary collimator. In radiosurgery, generally one more collimator is added to make a very small ﬁeld (e.g., circular collimator or micro-multileaf collimator).

The X-ray beam is ﬁrst collimated by the primary collimator, which has a circular aperture through which a photon beam can pass to make a radiation ﬁeld. The primary collimator is typically made of tungsten. For patient treatment, a radiation beam with uniform intensity across the ﬁeld is desirable, and to create a uniform photon beam, a conical ﬁlter called a ﬂattening ﬁlter is used. The central area of the ﬂattening ﬁlter is thicker to attenuate more photons in the central area. In general, a high-Z-number material is used for the ﬂattening ﬁlter to effectively attenuate photons.

FIGURE 4-7. Schematic illustration of the angular distribution of emission of bremsstrahlung X-rays around a target. At the kinetic energy of an electron beam less than 100 keV, X-rays are emitted relatively equally in all directions. As the kinetic energy of the electron beam increases, the direction of the bremsstrahlung X-ray emission becomes signiﬁcantly forward. In megavoltage X-ray machines, the transmission X-rays are therefore used to treat a patient.

The effectiveness of radiation treatment is based on how much radiation energy is deposited in the tumor compared with the energy deposited in the surrounding normal tissues. Energy deposition occurs through interactions between a radiation beam and the human body at either the atomic or nuclear level. When photons traverse a material, they reduce the intensity of the incident photons. The three most common ways for a photon to interact with matter for energies above a few keV are photoelectric absorption, Compton scattering, and pair production. These three mechanisms account for more than 99% of the interactions between photons and matter, and the probability of each depends on the energy of the photon and the material with which it interacts. Because these interactions are quantum in nature, a speciﬁc interaction is never guaranteed. It is simply more or less probable than the others (depending on the energy and the material).

Photoelectric Absorption In photoelectric absorption, an X-ray photon is absorbed by the interacting material and an electron from a shell of the atom is ejected, leaving the atom ionized, as illustrated in Figure 4-8. The ionized atom returns to the neutral state with the emission

FIGURE 4-8. In a photoelectric effect, an incident photon striking a bounded electron is absorbed, and the electron is ejected from the atom with a kinetic energy equal to the difference between the incident photon energy and the binding energy of the electron. The probability of the photoelectric effect increases as the atomic number of the target material increases; however, it decreases as the energy of the incident photon increases. (E is the energy of the incident photon, T is the kinetic energy of the ejected electron, and BE is the binding energy. Photon energy is expressed as hν, where h is Plank’s constant and ν is frequency.)

4.

the physics of stereotactic radiosurgery

of an X-ray, called the characteristic X-ray of the atom. The kinetic energy transferred to the ejected electron during the interaction is the same as the incident photon energy minus the ejected electron’s binding energy to the atomic nucleus. Photoelectric absorption occurs predominately at low energies of photons and for high-atomic-number materials. In water (one of the closest materials to tissue), the probability of this interaction becomes negligible once photon energy reaches above 100 keV. Because the effective energies of photon beams from the Gamma Knife (i.e., 60Co source) and linear accelerator are of the order MeV, the effect of photoelectric absorption is not signiﬁcant in radiosurgery.

Compton Scattering In Compton scattering, also known as an incoherent scattering, the incident X-ray photon ejects an electron from an atom, loses some of its energy to the ejected electron, and continues to move in a direction different from the initial direction. The resulting incident photon is called a scattered photon, whereas it is called a primary photon before the interaction. The energy and momentum are conserved in this process. Kinetic energy transferred to the electron is the energy difference between the primary and scattered photons. The energy of the scattered photon depends on the direction (i.e., the angle with respect to the direction of the primary photon). The scattered photon has minimal energy when it is backscattered (180° from its direction of travel). If it is scattered in the same direction as the primary, the scattered photon energy is the same as the primary photon energy. Compton scattering is important for low-atomic-number materials and photon energies of 100 keV to 10 MeV. Therefore, Compton scattering is the most signiﬁcant interaction for photons used in radiosurgery. Figure 4-9 illustrates the Compton scattering interaction.

37

FIGURE 4-10. When a photon with energy equal to or higher than 1.02 MeV comes very close to the nucleus of an atom, it interacts with the nuclear Coulombic force ﬁeld and can create a pair of an electron and a positron. The total kinetic energy of the electron and position is the energy of the incident photon minus 1.02 MeV. The probability of pair production is proportional to the atomic number of the target material and the incident photon energy.

Pair Production Photons have a quantum of energy but zero mass. In pair production, the incident photon energy is completely absorbed in the medium and an electron and its antiparticle, a positive electron called a positron, are created. This interaction is illustrated in Figure 4-10. Pair production can only occur when the energy of the incident photon is greater than 1.02 MeV, which is the minimum energy needed to create an electron and a positron. Positrons are very short lived and disappear with the creation of two 0.51-MeV photons each, a process called positron annihilation. The total kinetic energy transferred to the electron and positron pair is equal to the incident photon energy minus 1.02 MeV. Pair production is of particular importance when high-energy photons pass through materials of a high atomic number but plays a small role in photon beams used for radiosurgery.

Dose and Dosimetry

FIGURE 4-9. In Compton scattering, the incident photon strikes an electron and ejects it out of the atom to which it was bound. During this process, the incident photon energy is transferred to the ejected electron. The least energy is transferred to the electron when the incident photon has no scattering (i.e., θ = 0°), and the most energy is transferred to the electron when the incident photon is scattered backward (i.e., θ = 180°). Compton scattering becomes most signiﬁcant at an energy range of 100 keV to 10 MeV, and it is almost independent of the atomic number (i.e., Z value) of the interacting material. Thus, for the radiosurgery energy range, Compton scattering is the dominating interaction. (E is the energy of the incident photon, T is the kinetic energy of the ejected electron, and E′ is the energy of the scattered photon.)

The physical quantity that is used in radiation therapy to kill neoplastic cells is the kinetic energy of the incident radiation beam. Once photons enter a medium, they start to interact with the medium with the probability of interactions described earlier. Any photon particle loses a part or all of its energy when it initially interacts with a medium and most of the lost energy is absorbed by the medium around the interaction point. Energy absorbed in a unit mass of the interaction medium during interactions with a radiation beam is called the dose or, more speciﬁcally, the radiation dose. The commonly used unit of dose is the gray (Gy), which is equivalent to J/kg (joules per kilogram), where joule is a unit of energy (1 J is the approximate energy required to lift 1 kg by 0.1 m assuming gravity accelerates at 10 m/s2). Depending on the amount of dose, cGy (centigray, or one-hundredth of a Gy) is also often used. Dosimetry is either the measuring or calculating of dose. As you may have noted, the energy absorption mechanism during photon interaction is a two-step process: (1) energy transfers from the photon to the ejected electron, and (2) energy is deposited from the ejected electron to the medium (Fig. 411). Thus, energy deposition occurs within a volume around the

38

s. kim and j. palta

There are two commonly used dose-calculation methods: measurement-based and model-based. Measurement-based approaches directly use dose distributions measured in phantoms. On the other hand, model-based approaches try to calculate the dose from the ﬁrst principles with limited use of measured data.

Measurement-Based Dose Calculation Dose calculation is mainly based on three parameters: percent depth dose, proﬁle, and output.

Percent Depth Dose FIGURE 4-11. A photon enters a volume of a medium and transfers a portion of its energy, thus giving the electron a kinetic energy. The electron will travel a short distance and deposit its energy in a medium. The original photon will continue to travel within the medium as it transfers more energy to another electron.

spot of interaction rather than at a point of interaction. If the photon still has kinetic energy after the ﬁrst interaction (e.g., Compton scattering), it can travel and interact at another point, then cause another energy deposition. This second interaction is independent of the ﬁrst interaction, thus, third, fourth, and multiple scatterings are possible. The energy absorbed from scattered photons is called the scattered dose, contrary to the primary dose, which is the energy absorbed in the ﬁrst interaction. The same thing happens when a patient is irradiated with a photon beam. Photons interact with human tissue and deposit energy into the tissue. Radiation therapy’s aim is to kill the target tissue (or cell) through enough impact from the energy deposition while minimizing the impact to the normal tissue. In other words, local control of treatment depends on how much dose is delivered to the target (and how accurately, too, of course), and normal tissue complication is dependent on how much unwanted dose is given to the normal structure.

When a photon beam is incident on a medium, it creates a dose distribution within the medium. Relative dose at any depth is called the depth dose (DD) and when it is expressed as a percentile it is called the percent depth dose (PDD). Let us assume a certain number of photons in a hypothetical narrow beam are incident on a thin medium (Fig. 4-12a). The number of photons decreases exponentially because some photons interact with the medium and disappear from the path of the narrow beam. Now think of energy deposition (i.e., dose) at depth. The absorbed dose is proportional to the amount of radiation interactions. Energy, however, is deposited by ejected electrons rather than by photons themselves. Because most of the ejected electrons lose energy continuously while traveling through the medium, there is a distance lag between the photon interactions and the dose. This mechanism also creates a so-called build-up region in the shallow depth of the medium. Although the number of interactions is the highest at the surface of the medium, dose is almost zero (because most ejected electrons move deeper into the medium). Thus, dose is close to zero at the surface, starts to build up with depth, and reaches the maximum. Then, the dose begins to decrease exponentially. Figure 4-13 shows a typical shape of DD (or PDD) for a narrow beam. In reality, the target volume is ﬁnite in size (Fig. 4-12b). In this scenario,

FIGURE 4-12. (a) In the thin line of beam and thin medium, the scattered photons are out of the medium and never return back to the medium. There is thus no contribution by the scattered photons to the dose. Percent depth dose (PDD) is mainly determined by the energy deposition of the primary photons. (b) On the other hand, in the thick medium with the thin line of photon beam, the scattered photons can further interact with the medium and return to the center line of PDD measurement. In general, however, this effect is insigniﬁcant. (c) In the

broad beam and thick medium, the single or multiple scattering of photons can occur at any line of beam. The broad beam is considered as a bundle of thin lines of beam. The scattered photons coming from each beam line to the center line of interest contribute to the total dose. PDD values beyond the dmax (depth of dose maximum) increase with ﬁeld size because there are more line beams for larger ﬁelds. Because most radiosurgery ﬁelds are small, radiosurgery beams are close to case (b).

4.

the physics of stereotactic radiosurgery

39

DD (log scale)

Max

FIGURE 4-15. An ideal dose proﬁle in a plane is illustrated.

Incident Photon Depth (cm) Build-up Region FIGURE 4-13. A typical percent depth dose with a narrow beam is illustrated. The dose is close to zero at the surface and starts to build up with depth until reaching the maximum. This region is called the build-up region. After the depth of dose maximum, it then begins to decrease exponentially. Note that depth dose (DD) is displayed in the “log” scale whereas depth is in the “linear” scale (i.e., log-linear plot). In a log-linear plot, an exponential decrease is expressed as a line.

some of the scattered photons can scatter back to the center of the ﬁeld and contribute to the dose at depth. This effect is minimal, however. The scatter effect is signiﬁcantly enhanced with a broad beam (Fig. 4-12c) because more scattered photons can contribute to the dose. The chance of multiple scattered photons coming to the central line increases with ﬁeld size because more scattered photons exist in a larger ﬁeld, so PDD increases with ﬁeld size (Fig. 4-14). In radiosurgery, however, most ﬁelds are relatively small, so the scatter effect is insigniﬁcant. Thus, most radiosurgery beams are similar to the case shown in Figure 412b, and PDD can be analytically expressed as a simple exponential function.

Proﬁle or Off-Axis Ratio Figure 4-15 shows an ideal dose proﬁle, often called the off-axis ratio (OAR), in a plane where a certain amount of dose is given

100

PDD(%)

5×5 cm2 10×10 cm2 20×20 cm2

uniformly inside the ﬁeld and no dose is deposited outside the ﬁeld. In reality, it is not possible to obtain such a proﬁle. A typical dose proﬁle in a plane at a certain depth for a clinical radiation beam is shown in Figure 4-16a. The dose is uniform in the central area but starts to gradually decrease as it nears the ﬁeld edge. It then drops rapidly at the geometric edge of the ﬁeld and slowly approaches zero dose (<1%) away from the ﬁeld edge. The region of rapid dose fall-off is called the penumbra. When a ﬁeld is large, the ﬂat plateau in the central area is clearly observed. As the ﬁeld size becomes smaller, the plateau disappears as illustrated in Figure 4-16b, which is a proﬁle for a typical radiosurgery beam. In radiosurgery, most radiation ﬁelds (<50 mm in diameter) show either no plateau region or a very narrow plateau region. Because of the large dose variation within a small region, a high-resolution measurement detector is required for dose proﬁle measurement [23–29]. The shape of the individual radiation ﬁeld is deﬁned by the collimator system that is speciﬁcally based on the radiosurgery delivery system.

Output The magnitude of radiation produced by the delivery system is called the output. In general, this is a relative value that is normalized to the output of a reference ﬁeld size. Output varies with ﬁeld size (or collimator size). Larger ﬁeld sizes have higher output. With the PDD, OAR, and output for a given geometry, relative dose at any point is simply a multiplication of these three parameters. Absolute dose is obtained when this value is multiplied with the machine calibration factor and monitor units (MU) in the linear accelerator (or calibration dose rate and treatment time in Gamma Knife).

Model-Based Dose Calculation Most model-based dose-calculation methods deﬁne a kernel, which is a function that describes how the dose is distributed within a medium (mostly water). Two commonly used kernels are point kernel and pencil-beam kernel.

50

RELATIVE DEPTH FIGURE 4-14. Comparison of PDD curves for various ﬁeld sizes. As the depth gets deeper, PDD increases with ﬁeld size because of the increased amount of scatter dose contribution. Note that it is intentionally shown in “linear-linear” scale to illustrate that PDD is often drawn in two ways, “linear-linear” and “log-linear” (see Fig. 4-13).

FIGURE 4-16. (a) A typical dose proﬁle is illustrated for a conventional beam. The region of rapid dose change is called the penumbra. (b) A typical dose proﬁle for a radiosurgery beam.

40

s. kim and j. palta

FIGURE 4-17. A point kernel is illustrated, which is a distribution of energy deposited at any point to total energy released at the primary interaction point.

Point Kernel When a photon particle interacts at a point within a medium, a certain amount of energy is released per mass, which is called the total energy released per unit mass (TERMA). A part of this energy is absorbed at the point of interaction (often called the primary interaction point), but the rest of it is dispersed around the point. The scattered photon can also travel to any other points, interact, and deposit energy around the area of interaction. Therefore, a photon interaction can contribute dose to the whole volume. The point kernel describes how dose is distributed with respect to the interaction point within the medium, as illustrated in Figure 4-17. In general, the distributed dose is expressed as a ratio between the energy deposited at any point and the total energy released during the interaction at the primary interaction point. The dose at any point is a summation of kernels multiplied with the TERMA from all of the primary photon interactions. TERMA is easily calculated using analytical expression of interaction mechanisms.

Pencil-Beam Kernel The pencil-beam kernel includes dose contributions from all of the primary interactions along a ray instead of at one interac-

FIGURE 4-18. A pencil-beam kernel is illustrated, which is a distribution of deposited energy from all primary interactions along a pencilbeam line.

tion point. Figure 4-18 illustrates a pencil-beam kernel. If photon energy ﬂuence (the same as the number of photons times the corresponding energy per area) is known, dose at any point is simply a summation of kernels multiplied with energy ﬂuence. In general, kernel values are preobtained by calculations, and sometimes the pencil-beam kernel can be extracted from measurement data.

Isodose Line and Dose-Volume Histogram Once the dose is calculated in the treatment planning system, each dose distribution is usually displayed in an isodose line, which is made up of all the points of the same dose connected in a line. In radiosurgery, isodose lines are expressed as a percentile of the maximum dose. Dose distribution within the target is much more nonuniform in radiosurgery than in conventional radiotherapy. The maximum dose is usually observed in the center of the target, and it is usual to have very stiff dose fall-offs near the target boundary. Prescription is assigned to isodose lines of 50% to 80% depending on the situation. A prescription isodose line is higher in the case of a single isocenter and lower with multiple isocenters. Figure 4-19 shows

FIGURE 4-19. An isodose line connects all points of the same dose. In this ﬁgure, absolute isodose lines are displayed in axial, sagittal, and coronal planes of the CT image.

4.

41

the physics of stereotactic radiosurgery

Dose Volume Histogram 1.0 0.9 0.8

Norm.Volume

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0

500

1000

1500

2000

2500

3000

Dose (cGy) FIGURE 4-20. In a dose-volume histogram (DVH), accumulated volume (i.e., histogram) is described according to dose for any particular volume of interest (e.g., the target). In this case, 95% of the volume of the target will get a dose equal to or more than 2000 cGy.

isodose lines displayed in absolute value on three (axial, sagittal, and coronal) planes of tomographic image. Dose-volume histogram (DVH) describes accumulated volume (i.e., histogram) according to dose for any particular volume of interest (e.g., the target) as illustrated in Figure 4-20. In Figure 4-20, 95% of the volume of the target will get a dose equal to or more than 2000 cGy. Although the DVH does not show spatial information, it is a useful tool for plan evaluation [30, 31]. In radiosurgery, the DVH for the target usually shows a gentle slope on the high-dose side because of a signiﬁcantly nonuniform dose proﬁle of small ﬁelds. With the lower isodose line selected for prescription, a gentler DVH slope on the highdose side is expected. In the case of multiple isocenters, ﬁeld overlap causes a signiﬁcant dose increase in the central volume of the target, resulting in a high dose tail in DVH.

a set of four collimator helmets, and a control unit. Figure 4-21 and Figure 4-22 show a photograph of a Gamma Knife unit and a simple schematic diagram, respectively. The 60Co radioactive source is in pellet form, and 20 pellets are encapsulated in a steel capsule. Each steel capsule is 1 mm in diameter and 20 mm in height. Sources are aligned with three collimator systems, the precollimator, the stationary collimator, and the ﬁnal collimator of a helmet. Both the precollimator and the ﬁnal collimator of the helmet are made of tungsten alloy whereas the stationary collimator is made of lead. The total

Photon Beam Delivery Systems Gamma Knife In the late 1960s, Leksell introduced a prototype of Gamma Knife, which is a dedicated radiosurgery unit [5]. Leksell’s prototype incorporated 179 60Co radioactive sources placed over a 60° × 160° sphere. The modern Gamma Knife models incorporate 201 60Co radioactive sources that converge and focus on the treatment target at a source-to-target distance of about 40 cm [6, 7]. Each source contributes a clinically insigniﬁcant dose following the beam line. However, when many beam lines are converged on the focus where the target is, the therapeutic dose is delivered to the target while the surrounding normal tissue dose remains under the limit. Gamma Knife mainly consists of the radiation unit, the operating table and sliding couch,

FIGURE 4-21. A photograph of a Gamma Knife unit.

42

s. kim and j. palta

Rear shield

Middle shield Front shield

Central body

Upper shielding door

Plug

Treatment table Helmet

Couch

Position indicator Trolley .

.

Lower shielding door

FIGURE 4-23. A photograph of a collimator helmet. A properly chosen helmet, depending on the target size, is connected to the sliding couch.

FIGURE 4-22. A simple schematic diagram of a Gamma Knife unit.

The current state-of-the-art Gamma Knife units are equipped with several auxiliary capabilities. They provide software that tracks the patient’s position 10 times per second. A robotic automatic positioning system repositions each isocenter with submillimeter precision without operator intervention. Radiosurgery using Gamma Knife has been performed for more than 250,000 patients worldwide during the past 30 years. Many clinical studies with longer than 15 years of follow-up have been reported.

activity of the Gamma Knife unit with new 201 sources loaded is of the order 6000 Ci (2.22 × 1014 Bq, where Bq is the same as disintegration per second, and Ci is 3.7 × 1010 Bq). The activity of each source is about 30 Ci and its variation is expected to be within ±5%. With 6000 Ci, a typical dose rate is about 300 cGy/ min at the center of the 16-cm-diameter, spherical water phantom. As described earlier, this value drops down to 150 cGy/min (50% of initial dose rate) 1 half-life (5.26 years) later. Four helmets of different collimator sizes, 4, 8, 14, and 18 mm in diameter, are provided. Figure 4-23 is a photograph of a collimator helmet. A proper helmet is chosen depending on the target size and is connected to the sliding couch. The Leksell stereotactic frame (Fig. 4-24) is ﬁxed to the patient’s skull and connected to the couch. Once the target position is aligned to the focus, the sliding couch is pushed into the mechanically set treatment position. Its mechanical spatial accuracy is claimed to be in submillimeters (about 0.3 mm). In treatment mode, the shielding doors are open, resulting in an increase of scatter dose inside the treatment room. When the sliding couch is withdrawn and the shielding doors are still open, the scatter dose becomes about 1.3 times higher. Because of the unit’s design, however, no primary beam directly escapes the unit even with the doors open. When the doors are closed, exposure rates by leakage radiation are signiﬁcantly reduced.

FIGURE 4-24. Picture of a Leksell stereotactic frame. The frame is ﬁxed to the patient’s skull and is connected to the couch.

.

..

..

.

4.

the physics of stereotactic radiosurgery

43

Isocentric Linear Accelerator Most clinical linear accelerators are isocentric, which means that the gantry, the collimator, and the table can be rotated with respect to one point in the space called the isocenter. In reality, the isocenter may not be a point because three rotation axes (gantry, collimator, and table) often do not intersect at a common point. Instead, they only near each other within a sphere. The center of this sphere is considered the nominal isocenter. The size of the sphere is the key factor in the linear accelerator’s spatial accuracy, and the smaller the better. Common consensus says that the sphere needs to be less than 1 mm in diameter for radiosurgery based on the fact that the resolution of modern imaging modalities is around 1 mm. Another important factor for spatial accuracy is how precisely positioned the patient is. In a linear accelerator, the patient’s positioning is usually determined using the treatment room laser system. Three lasers, one from the ceiling and two from side-walls, are aligned to point to the nominal isocenter. By matching the laser crosshairs and the desired marks on the frame attached to the patient’s skull, the target is aligned to the nominal isocenter. Therefore, laser system accuracy is signiﬁcantly important when used for patient positioning. As described earlier, linear accelerators are equipped with a so-called secondary collimator system that provides a rectangular ﬁeld for conventional radiotherapy, but rectangular ﬁelds are not considered appropriate for radiosurgery. Therefore, additional circular collimators are commonly used as tertiary collimators [10–14]. Circular beams are superior to rectangular beams for reasons like the sharper beam, easier dosimetry calculation, more precise beam delivery, and better ﬁeld deﬁnition for small ﬁelds. Because tertiary collimation happens closer to the patient, the beam is more accurately aligned and the penumbra is reduced. More rapid dose fall-off outside the target can therefore be obtained. Collimator sizes typically range from 5 to 40 mm to treat a variety of lesion sizes and shapes. Most circular collimators are made of tungsten alloy or lead and are 5 to 10 cm thick. Figure 4-25 shows several circular collimators with different diameters used at the University of Florida.

FIGURE 4-25. Several circular collimators in different sizes of diameter (5, 10, 12, 14,20, 24, and 30 mm) used at the University of Florida are shown.

FIGURE 4-26. A setup of the ﬂoor stand used at the University of Florida is shown. The ﬂoor stand is mechanically ﬁxed to the holes in the ﬂoor.

At the University of Florida, developed was a special device called a ﬂoor stand [32, 33] that improves the accuracy of the beam’s focus to the isocenter by adding a set of bearings to the stereotactic collimator system that accounts for imperfections in the gantry rotation. A set of bearings is also attached to the patient and the target area to bypass the inaccuracies of the table rotation. It is reported that the ﬂoor stand can provide mechanical spatial accuracy within 0.2 ± 0.1 mm to deﬁne the isocenter [32, 33]. Figure 4-26 shows how the ﬂoor stand is set up. With the recent introduction of a micro-multileaf collimator (mMLC), which has many collimator leaves that can be individually moved to conform to the shape of the target, it is possible to do two-dimensional conformal radiosurgery for irregularly shaped targets with a single isocenter (Fig. 4-27). In addition, dynamic conformal radiosurgery can be performed with the aid of the automatic ﬁeld shaping capability while the gantry rotates. It is also possible to do intensity modulated stereotactic radiosurgery (IMSR) if the treatment planning system supports the appropriate dose optimization [34]. The collimator leaf is usually made of tungsten. The width of each leaf is an important parameter related to the conformality and it ranges from 1.5 to 6 mm at the isocenter depending on the manufacturer. State-of-the-art linear accelerators incorporate a volumetric imaging system referred to as a cone beam computed tomography (CT) system. A diagnostic X-ray tube is added on a separate gantry at a 90° angle from the treatment gantry, and a ﬂat panel detector is installed on the opposite side as shown in

44

s. kim and j. palta

FIGURE 4-27. A photograph of a commercial mMLC. As can be seen, there are many collimator leaves, and each of them can be individually moved to conform to the shape of the target.

FIGURE 4-29. A CyberKnife unit uses a miniature linear accelerator mounted on a robotic arm.

Figure 4-28. Volumetric imaging can be obtained in a couple of minutes by recording multiple cone beam geometry planar images while the gantry rotates. Currently, it is not clear whether this capability needs to be used for radiosurgery. This advancement opened the door to exploring the possibilities with imageguided radiosurgery using isocentric linear accelerators.

beam. It is much smaller and lighter than a conventional S-band (2856 MHz) linear accelerator. CyberKnife incorporates an image-guided system consisting of two diagnostic Xray tubes mounted on the ceiling and two detectors placed next to the lateral sides of the table. With this X-ray imageguided system, CyberKnife continuously and automatically tracks a tumor’s precise location throughout the procedure, detects any tumor or patient movement, and makes corrections as needed. Because of its image-guidance capability, the radiosurgery procedure is performed without an invasive head frame. Even without the frame, the use of intelligent robotics technology enables treatments with submillimeter accuracy. The robot can position the linear accelerator at any point within a spherical shell of 60 to 100 cm from the target with a precision of ±0.5 mm. In principle, the frameless approach gives more choices in beam-angle selection compared with radiosurgery with a frame. The robotic arm must, however, avoid a collision with the patient, table, and imaging system as well as direct beam incidence into the imaging system. Another concern is the long treatment time compared with other delivery systems like the Gamma Knife and the isocentric linear accelerator. A total of 12 collimators from 5 to 60 mm in diameter are provided. Installing the unit would require a 10-ft ceiling over a 12 × 16 ft2 area.

CyberKnife CyberKnife, developed by Accuray Inc. (Santa Clara, CA) in collaboration with Stanford University (Stanford, CA), uses a miniature linear accelerator mounted on a robotic arm as shown in Figure 4-29 [35, 36]. The miniature linear accelerator operates in the X-band RF (9300 MHz) and provides a 6-MV

Proton Therapy

FIGURE 4-28. In an advanced linear accelerator, a diagnostic X-ray tube is added on a separate gantry at a 90° angle to the treatment gantry, and a ﬂat panel detector is installed on the opposite side. Using an X-ray tube and cone beam geometry reconstruction software, volumetric images can be obtained in the treatment room.

Clinical proton beams are produced by injecting hydrogen atoms with their electron stripped in an accelerating structure; electric ﬁelds accelerate the free protons to the desired clinical energy. The proton accelerators fall into two broad categories [20]. Cyclotrons are the classic proton accelerators in which protons are injected at the center of two halves (called “dees”) of an electrically powered circular dipole magnet with a constant magnetic ﬁeld (Fig. 4-30). The proton acceleration occurs within the narrow gap between the two half-magnets employing a correctly phased alternating electric ﬁeld. The protons gain energy, twice for each revolution, while moving in a constant circular path. Accelerated protons with a ﬁxed energy and a

4.

the physics of stereotactic radiosurgery

FIGURE 4-30. In cyclotrons, protons are injected at the center of two halves (“dees”) of an electrically powered circular dipole magnet with a constant magnetic ﬁeld. The proton acceleration occurs within the narrow gap between the two half-magnets employing a correctly phased alternating electric ﬁeld. The protons gain energy, twice for each revolution, while moving in a constant circular path. Accelerated protons with a ﬁxed energy and a continuous beam current can be extracted.

45

a constant radius (Fig. 4-31). As the proton energy increases, the magnetic ﬁeld strength also increases to keep them in a stable orbit. The acceleration cycle in synchrotrons is typically 2 seconds. Therefore, it is only possible to obtain proton beams in pulses with a repetition frequency of 30 cycles per minute and the energy of the protons can be varied from pulse-to-pulse. Operating the synchrotrons is complex because it requires rapid magnetic ﬁeld variations to operate. However, synchrotrons do not require energy degraders, thus minimizing secondary radiation production in the beam line. An elaborate beam transport system (BTS) carries the protons exiting the accelerator to the patient site. The beam line consists of focusing, bending, and switching magnets that steer a pencil beam of protons to different treatment rooms. The length of the beam line can easily exceed 30 m. A stable and precise BTS is required for reproducible dosimetry and patient treatments. The stability of the centroid of the beam position in the beam line must be better than 1.00 mm. Therefore, the bending and focusing magnets have to be very stable, mechanically and electrically, during the operation of the proton machine.

A Proton’s Interaction with Matter continuous beam current of up to 1 milliampere (mA) can be extracted and transported into an evacuated tube (beam line). Proton energy is changed by inserting an energy degrader in the path of the beam extracted from the cyclotron. The other type of proton accelerator is the synchrotron, which allows preaccelerated protons (up to a few MeV) to be injected into an accelerating chamber equipped with a number of electromagnets. The protons are accelerated in an orbit with

Protons are positively charged elementary particles that continuously lose energy and go through small-angle deﬂections as they travel through matter before they get absorbed. While the radiation dose deposited by photons drops off exponentially with penetration depth, the dose deposited by protons increases very slowly for about three-quarters of its range of travel in the medium before increasing sharply, reaching a maximum value before rapidly dropping off to zero. The depth at which the maximum energy is deposited by protons is called the Bragg peak. Figure 4-32 illustrates a depth dose of a proton beam showing its Bragg peak. The typical peak to entrance dose ratio of narrow proton beams is 4 or higher. The peak’s position is proportional to the energy of the proton beam. The range of

FIGURE 4-31. In synchrotrons, the protons are accelerated in an orbit with a constant radius. As the proton energy increases, the magnetic ﬁeld strength also increases to keep them in a stable orbit.

FIGURE 4-32. The depth dose of a proton beam is almost constant to a certain depth and starts to increase, then it jumps signiﬁcantly at the farthest end of the path. This peak is called the Bragg peak.

46

s. kim and j. palta

230-MeV protons is approximately 33.00 cm, which is sufﬁcient to penetrate any part of the body. The typical range of proton energies used for radiosurgery is 70 to 150 MeV (4 to 15 cm depth). The clinical target volumes are usually thicker than the width of the Bragg peaks, therefore, the range of the proton beam must be modulated to spread out the Bragg peak. By changing the energy of the incident proton beam with a variable range shifter, the Bragg peaks will “stack” at different depths. The variable range shifter can be a binary ﬁlter, a double-wedge variable absorber, or a circular wedge. Scattering foils or active scanning of a narrow beam with scanning magnets will laterally spread the proton beams to achieve uniform intensity across the target volume.

Dose Characteristics in Different Modalities In 1951, Leksell introduced the concept of stereotactic radiosurgery, and the ﬁrst patient was treated with the Gamma Knife prototype in 1967. Since then, Gamma Knife has only seen improvements with the establishment of many linear accelerator–based radiosurgery programs during the past 20 years. The use of proton beams for radiosurgery is also increasing. Some research groups have investigated and compared the dosimetric characteristics of each modality. The linear accelerator is thought to be able to provide dosimetric characteristics similar to Gamma Knife when multiple, non-coplanar arcs are used. In the case of more than ﬁve non-coplanar arcs, the full width at the half maximum (FWHM) of a 5-mm circular collimator ﬁeld is very close to that of Gamma Knife with a 4-mm helmet. The dose fall-off of the linear accelerator is also almost identical when more than ﬁve non-coplanar arcs are used. A recent study for trigeminal neuralgia treatment reports that compared with Gamma Knife, linear accelerator–based deliveries exhibit a ﬂatter top at the high isodose line level (>50%) and faster dose-off at the low isodose line levels (<50%) [37]. The penumbral widths (80% to 20%) were 1 mm for Gamma Knife with a 4-mm helmet and 2.1 mm for the 6-MV linear accelerator with a 5-mm collimator in another study [38]. Radiosurgery with a proton beam is different from both a linear accelerator and the Gamma Knife. Proton radiosurgery is performed with shaped and fully compensated ﬁelds, widely separated in the entrance angle, and with a single isocenter based on the characteristics of the proton beam, that is, the Bragg peak [39]. The proton beam, Gamma Knife, and linear accelerator have been compared in ﬁve clinical cases [3]. The optimal modality for stereotactic radiosurgery depends on target size, shape, and location; however, all modalities are equally good if the target is small and regular. In the case of trigeminal neuralgia, linear accelerator–based stereotactic radiosurgery is effective and can be comparable with Gamma Knife [38]. Gamma Knife is superior to the linear accelerator with regard to the conformality. In contrast, the linear accelerator is superior with regard to the dose fall-off outer region of the target volume [40]. The linear accelerator with mMLC can be better than Gamma Knife when hearing preservation is important during treatment [41].

Quality Assurance In most conventional radiation therapy treatments, prescribed dose is delivered in many fractions, and dose per each fraction is relatively small; however, a signiﬁcantly large dose is delivered in a single treatment during radiosurgery. Therefore, the impact of the inaccurate localization of the target or dose delivery in radiosurgery can be disastrous. To minimize such disasters, stereotactic radiosurgery quality assurance (QA) should be stringent. Both an international QA task group and Task Group 42 of the American Association of Physicists in Medicine (AAPM) have published general recommendations [42, 43]. The Radiation Therapy Oncology Group (RTOG) has also published guidelines [44]. In general, radiosurgery QA consists of common QA for the overall performance of all radiosurgery equipment, valid for a relatively long term, and speciﬁc QA for the calibration and preparation of equipment on each treatment day.

Common QA Common QA procedures need to be set to periodically verify equipment status and performance. It should include target localization QA, basic dosimetry QA, treatment planning QA, and output calibration and delivery QA. A detailed QA schedule must be set based on frequency and consequence: any item that has either high failure rate or severe consequence must frequently be checked. Overall target localization can be tested using a phantom containing a target located in the known geometry. Accuracy of within 1 mm is recommended. The treatment planning system’s dosimetric and nondosimetric conditions should both be checked. Speciﬁc items are well described in the recommendations of the AAPM Task Group 53 [45]. Dosimetry-related QA procedures are speciﬁc to the delivery system. In the Gamma Knife unit, the following are periodically veriﬁed: dose rate at the center of a 16-cm-diameter tissue-equivalent sphere at the focus; shutter error; connections of frame, collimator helmets, and sliding couch; and leak test on collimator helmets. For linear accelerators, output is checked daily. If the laser system is used for patient alignment, its accuracy needs to be veriﬁed rigorously. The most important QA item is the mechanical accuracy of the nominal isocenter. Less than a 1-mm diameter around the isocenter sphere is recommended [14, 46]. If an independent ﬂoor stand is used, like at the University of Florida, the accuracy is directly dependent on the ﬂoor stand. Thus, no additional QA tests on the laser and linear accelerator isocenter, other than those for routine conventional radiotherapy, are required.

Speciﬁc QA The goal of speciﬁc quality assurance is to make sure that the machine is running adequately, that all auxiliary devices are prepared and working properly, and that treatment parameters are correctly set just before and during patient treatments. Before placing the frame into the patient, the treatment streamline should be checked to minimize unnecessary treatment

4.

the physics of stereotactic radiosurgery

delay or cancellation after the frame is on. Figure 4-33 shows a sample of a patient-speciﬁc QA checklist used for linear accelerator–based radiosurgery at the University of Florida. As you can see, check items are listed chronologically. Once the secondary collimator is set as needed, the collimator drives are physically disconnected to prevent an accidental change of the

47

secondary collimator. To prevent the treatment table from accidentally dropping, a mechanical block is applied to the column of the table. Because the patient’s head is connected to the ﬂoor stand while his or her body is on the table, an abrupt drop of the table could cause serious problems as serious as a patient death. For the ﬁrst isocenter treatment, the accuracy of focusing

FIGURE 4-33. A sample of a patient-speciﬁc QA checklist used for linear accelerator–based radiosurgery at the University of Florida. In this sheet, checklist items are listed chronologically.

48

s. kim and j. palta

FIGURE 4-34. An example of veriﬁcation ﬁlm. It can be seen that the steel ball in the calibration jig and radiation beams are well aligned.

the radiation ﬁelds to the isocenter is checked by a WinstonLutz ﬁlming test. A staff member sets the ﬂoor stand to the treatment position. A steel ball in the calibration jig is set to the isocenter coordinates independently by another staff member and attached to the ﬂoor stand. Then, the veriﬁcation ﬁlm is irradiated in four different beam angles with a 24-mm-diameter circular collimator. If both the ﬂoor stand and calibration jig are set to the isocenter correctly, and beam alignment is appropriate in all beam angles, images of the ball must be centered within all the circular ﬁelds. Figure 4-34 shows an example of the veriﬁcation ﬁlm. Once the ﬁlm is conﬁrmed satisfactory, the patient is moved to the ﬂoor. Before the treatment beam is on, the size of the circular collimator and the treatment parameters, such as the table angle, gantry start angle, gantry end angle, and monitor unit, are veriﬁed. In subsequent isocenter treatments, the ﬂoor stand is set by a physicist and coordinates are independently double-checked by two other staff members. Speciﬁc QA in the Gamma Knife unit is relatively less intense. It is basically ready for treatment in most cases, and only a few safety checks need to be performed.

Conclusion Stereotactic radiosurgery is a special radiotherapy technique that holds promise not only radio-therapeutically but also as a neurosurgical tool. The underlying physics principles for radiosurgery are the same as those for conventional radiation therapy; however, the accuracy requirements in dosimetry and patient positioning for radiosurgery are at the millimeter level. There are several dosimetric parameters that must be measured in tissue-equivalent phantoms such as water including dose calibration, percentage depth doses, relative dose output factors, and cross-beam proﬁles. Radiosurgery treatment ﬁelds are often very small, therefore, the spatial resolution requirements for radiation detectors are much more stringent in radiosurgery.

References 1. Evans RD. The Atomic Nucleus. Malabar, FL: Krieger, 1955. 2. Attix FH. Introduction to Radiological Physics and Radiation Dosimetry. New York: Wiley-Interscience, 1986. 3. Khan FM. The Physics of Radiation Therapy, Second ed. Philadelphia: Lippincott Williams & Wilkins, 1992. 4. Johns HE, Cunningham JR. The Physics of Radiology, Fourth ed. Springﬁeld, IL: Charles C Thomas Publisher, 1983. 5. Leksell L. Cerebral radiosurgery I. Gamma thalamotomy in two cases of intractable pain. Acta Chir Scand 1968; 134:585–595.

6. Wu A, Maitz AH, Kalend AM, et al. Physics of gamma knife approach on convergent beams in stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 1990; 18:941–949. 7. Simonova G, Novotny J Jr, Vladyka V, et al. Fractionated stereotactic radiotherapy with the Leksell gamma knife: feasibility study. Radiother Oncol 1995; 37:108–116. 8. Poffenbarger B. Viability of an isocentric cobalt-60 teletherapy unit for stereotactic radiosurgery. MSc thesis, McGill University, Montreal, Canada, 1998. 9. Poffenbarger B. Viability of an isocentric cobalt-60 teletherapy unit for stereotactic radiosurgery. Med Phys 1998; 25:1935–1943. 10. Bellerive M, Kooy HM, Loefﬂer J. Linac radiosurgery at the Joint Center for Radiation Therapy. Med Dosim 1998; 23:187–199. 11. Betti OO, Derechinsky VE. Hyperselective encelphalic irradiation with linear accelerator. Acta Neurochir Suppl 1984; 33:385–390. 12. Colombo F, Benedetti A, Pozza F, et al. External stereotactic irradiation by linear accelerator. Neurosurgery 1985; 16:154–160. 13. Hamilton JA, Lulu BA, Fosmire H, et al. Preliminary clinical experience with linear accelerator based spinal stereotactic radiosurgery. Neurosurgery 1995; 36:311–318. 14. Lutz W, Winston KR, Maleki N. A system for stereotactic radiosurgery with a linear accelerator. Int J Radiat Oncol Biol Phys 1988; 14:373–381. 15. Larsson B. Dosimetry and radiobiology of protons as applied to cancer therapy and neurosurgery. In: Thomas RH, Perez-Mendez V, eds. Advances in Radiation Protection and Dosimetry in Medicine. New York: Plenum, 1980:367–394. 16. Frankel KA, Phillips MH. Charged particle method: protons and heavy charged particles. In: Phillips MH, ed. In Physical Aspects of Stereotactic Radiosurgery. New York: Plenum, 1993. 17. Kjellberg RN, Shintani NA, Frantz AG. Proton beams in acromegaly. N Engl J Med 1968; 278:689–695. 18. Larsson B, Leksell L, Rexed B, et al. The high energy proton beam as a neurosurgical tool. Nature 1958; 182:1222–1223. 19. Breuer H, Smit BJ. Proton Therapy and Radiosurgery. Berlin, Heidelberg, New York: Springer-Verlag, 2000. 20. Wieszczycka W, Scharf WH. Proton Radiotherapy Accelerators. Singapore: World Scientiﬁc, 2001. 21. Karzmark CJ, Nunan CS, Tanabe E. Medical Electron Accelerators. New York: McGraw-Hill, 1993. 22. Shiu AS, Kooy HM, Ewton JR, et al. Comparison of miniature multileaf collimation with circular collimation for stereotactic treatment. Int J Radiat Oncol Biol Phys 1997; 37:679–688. 23. Arcovito G, Piermattei A, D’Abramo G et al. Dose measurements and calculations of small radiation ﬁelds for 9-MV x rays. Med Phys 1985; 12:779–784. 24. Bova FJ. Radiation physics. Neurosurg Clin N Am 1990; 1:909– 931. 25. Houdek PV, VanBuren JM, Fayos JV. Dosimetry of small radiation ﬁelds for 10-MV x rays. Med Phys 1983; 10:333–336. 26. Rice RK, HansenJL, Svensson GH, et al. Measurements of dose distributions in small beams of 6 MV x-rays. Phys Med Biol 1987; 32:1087–1099. 27. Heydarian M, Hoban PW, Beddoe AH. A comparison of dosimetry techniques in stereotactic radiosurgery. Phys Med Biol 1996; 41:93–110. 28. Rustgi SN. Evaluation of dosimetric characteristics of a diamond detector for photon beam measurement. Med Phys 1995; 22:567–570. 29. Vatnitsky S, Jarvinen H. Application of natural diamond detector for the measurement of relative dose distribution in radiotherapy. Phys Med Biol 1993; 38:173–184. 30. Schell MC, Smith V, Larson DA, et al. Evaluation of radiosurgery techniques with cumulative dose volume histograms in linac-based stereotactic irradiation. Int J Radiat Oncol Biol Phys 1991; 20: 1325–1330.

4.

the physics of stereotactic radiosurgery

31. Serago CF, Houdek PV, Bauer-Kirpes B, et al. Stereotactic radiosurgery: dose volume analysis of linear accelerator techniques. Med Phys 1992; 19:181–185. 32. Friedman WA, Bova FJ. The University of Florida radiosurgery system. Surg Neurol 1989; 32:334–342. 33. Meeks SL, Bova FJ, Friedman WA, et al. Linac scalpel radiosurgery at the University of Florida. Med Dosim 1998; 23:177–185. 34. Webb S. Optimization by simulated annealing of three-dimensional conformal treatment planning for radiation ﬁelds deﬁned by a multileaf collimator. Phys Med Biol 1991; 36:1201–1226. 35. Adler JR, Cox RS. Preliminary clinical experience with the CyberKnife: image guided stereotactic radiosurgery. In: Kondziolka D, ed. Radiosurgery. Basel: Karger, 1995:317–326. 36. Maciunas RJ, Fitzpatrick M, Galloway RL, et al. Beyond stereotaxy: extreme levels of application accuracy are provided by implantable ﬁducial markers for interactive image guided neurosurgery. In: Maciunas RJ, ed. Interactive Image Guided Neurosurgery. Washington, DC: AANS, 1994:261–270. 37. Ma L, Kwok Y, Chin LS, et al. Comparative analyses of linac and Gamma Knife radiosurgery for trigeminal neuralgia treatments. Phys Med Biol 2005; 50:5217–5227. 38. Gerbi BJ, Higgins PD, Cho KH, et al. Linac-based stereotactic radiosurgery for treatment of trigeminal neuralgia. J Appl Clin Med Phys 2004; 5:80–90. 39. Verhey LJ, Smith V, Sergo CF. Comparison of radiosurgery treatment modalities based on physical dose distributions. Int J Radiat Oncol Biol Phys 1998; 40:495–505. 40. Plowman PN, Doughty D. Stereotactic radiosurgery, X: clinical isodosimetry of Gamma knife versus linear accelerator X-knife for pituitary and acoustic tumors. Clin Oncol 1999; 11:321–329. 41. Perks JR, St. George EJ, Hamri KE, et al. Stereotactic radiosurgery XVI: isodose comparison of photon stereotactic radiosurgery techniques (Gamma knife vs. micromultileaf collimator linear accelerator) for acoustic neuroma-and potential clinical importance. Int J Radiat Oncol Biol Phys 2003; 57:1450–1459. 42. Hartmann G, Lutz W, Arndt J, et al. Quality Assurance Program on Stereotactic Radiosurgery. Berlin: Springer-Verlag, 1995. 43. Schell M, Bova FJ, Larson DA, et al. Stereotactic radiosurgery. Report of Task Group 42. American Association of Physicists in Medicine (AAPM), Report No. 54. Medical Physics Publishing: Madison, WI, 1995. 44. Shaw E, Kline R, Gillin M, et al. Radiation therapy oncology group: radiosurgery quality assurance guidelines. Int J Radiat Oncol Biol Phys 1993; 27:1231–1239. 45. Fraass B, Doppke K, Hunt M, et al. American Association of Physicists in Medicine (AAPM): Radiation Therapy Committee Task Group 53: Quality assurance for clinical radiotherapy treatment planning. Med Phys 1998; 25:1773–1829. 46. Falco T, Lachaine M, Poffenbarger B, et al. Setup veriﬁcation in linac-based radiosurgery. Med Phys 1999; 26:1972–1978.

Glossary AAPM American Association of Physicists in Medicine accelerator a device that accelerates subatomic charged particles or nuclei to high energies alpha particle positively charged particle consisting of two protons and two neutrons Bq Becquerel, the Système International (SI) unit of radioactivity equal to one disintegration per second beta particle high-speed electron or positron binding energy net energy required to remove an atomic electron from its orbit

49

Bragg peak peak can be seen at the end of the depth dose curve of heavy charged particles such as a proton BTS beam transport system, a proton accelerator system that carries the protons exiting the accelerator to the patient site cavity a device through which power is coupled to the accelerating particles in a resonant mode Ci Curie, unit of radioactivity equal to 3.7 × 1010 disintegrations per second collimator a device capable of collimating radiation CT computed tomography cyclotron a particle accelerator in which charged particles are accelerated in a circular path by an alternating electric ﬁeld in a constant magnetic ﬁeld DD depth dose, relative dose level according to the depth of the material when the radiation beam is incident dipole a pair of electric charges or magnetic poles dose energy absorbed per unit mass by radiation DVH dose-volume histogram, accumulated volume according to dose for a volume of interest ﬁlter a device used to modify the intensity of radiation FWHM full width at half maximum gamma particle electromagnetic radiation particle (photon) emitted by radioactive decay Gy Gray, a unit of dose equivalent to joules per kilogram half-life length of time needed for a radioactive substance to lose half of its radioactivity from decay IMSR intensity-modulated stereotactic radiosurgery, a treatment technique in which radiation intensity is modulated within a ﬁeld to obtain an optimized dose distribution ionization chamber a chamber for determining the intensity of ionizing radiation by measuring ionization of the enclosed gas ionizing radiation electromagnetic or particulate radiation capable of producing ions in its passage through matter isocenter a point in the space that is the center of rotation of the gantry, collimator, and table of an isocentric medical linear accelerator isodose line a line made by connecting all points of the same dose kernel a function that describes how the dose is distributed within a medium keV kilo-electron-volts, a unit of energy of particles klystron a microwave ampliﬁer linear accelerator a particle accelerator in which the path of the particles is straight mA milliampere, a unit of electric current magnetron a diode-type electron tube that generates microwave pulses MeV million electron-volts, a unit of energy of particles mMLC micro-multileaf collimator, a special collimator consisting of many small collimator leaves that can be individually moved to conform to the shape of the target mR milliroentgen, a unit of exposure μSv microsievert, the SI unit for an equivalent radiation dose that accounts for biological effect MV million volts, a unit of beam quality of radiation therapy beams produced in linear accelerators OAR off-axis ratio, relative dose distribution following a line on a plane normal to the direction of the radiation beam, same as proﬁle

50

s. kim and j. palta

output magnitude of radiation produced PDD percent depth dose, relative dose level in percentile according to the depth of the material when the radiation beam is incident photon quantum of electromagnetic energy with no mass and charge considered to be both particle and wave positron positively charged particle of the same mass and magnitude of charge as an electron proﬁle relative dose distribution following a line on a plane normal to the direction of radiation beam, same as the offaxis ratio (OAR) proton stable and positively charged subatomic particle QA quality assurance radioactive decay spontaneous disintegration of a radionuclide followed by the emission of ionizing radiation in the form of alpha, beta, or gamma particles

radioactivity capability of radioactive decay radioisotope naturally or artiﬁcially produced radioactive isotope of an element range distance charged particles can travel resonance condition of being resonant in which an increase in the amplitude of a physical quantity can occur RF radiofrequency, a range of electromagnetic frequencies above sound and below visible light synchrotron a particle accelerator in which charged particles are accelerated around a ﬁxed circular path by an electric ﬁeld and held to the path by an increasing magnetic ﬁeld TERMA total energy released per unit mass, amount of energy released from radiation to the unit mass of the interaction material Z atomic number; that is, the number of protons in an element

5

Radiobiological Principles Underlying Stereotactic Radiation Therapy David J. Brenner

Introduction Since the Gamma Knife was ﬁrst conceived in 1968, primarily for arterial and functional lesions [1], single-fractioned stereotactic radiation therapy* has been increasingly used to treat a variety of cerebral lesions. By 1985, an alternative modality was available for stereotactic radiation therapy, using a linear accelerator (linac) and a stereotactic head frame [4, 5]. Recently, the CyberKnife, a frameless robotic system, has been developed for stereotactic radiation therapy [6], and intensity-modulated stereotactic radiation therapy [7, 8] is now entering clinical practice. In its early use, stereotactic radiation therapy was always applied in single fractions (i.e., radiosurgery), so its beneﬁts were entirely related to its ability to irradiate target volumes with excellent dose distributions. By about 1990, however, several groups [9–15] began to consider the potential biological advantages of fractionated stereotactic radiotherapy, stimulated also by the development of relocatable stereotactic head frames for linac-based treatments [11, 12, 16]. As we will discuss, new technology has made it increasingly practical to fractionate a stereotactic treatment, and the use of fractionated stereotactic radiotherapy has indeed increased steadily over the past 15 years, as illustrated by more than 700 publications documented in PubMed/Medline on this modality. We will review here the radiobiological principles underlying stereotactic radiation therapy and their applications to single or multifractioned radiotherapy of the three main types of lesions that might be treated with this modality: malignant tumors, benign tumors, and vascular disorders. In that little is known about the radiobiological rationale behind radiotherapy for functional disorders, this area will not be covered. A complementary review on the clinical aspects of fractionated stereo-

* Note the term stereotactic radiation therapy will be used here to apply both to single-fraction treatment (often called radiosurgery) and to multiple-fractioned stereotactic radiotherapy. There is still debate about the most appropriate terminology [2, 3].

tactic radiotherapy has been published by Tomé and colleagues [17].

The Three R’s of Radiobiology: Reoxygenation, Repair, and Repopulation All these three radiobiological phenomena relate ultimately to cell-killing processes, the assumption being that this is the primary (though not the only) mechanism by which all radiotherapy both produces tumor control and induces side effects. Of course, there are a variety of mechanisms that lead to cell killing, and a variety of relevant target cells, but underlying all radiotherapeutic response remains cell killing [18].

Reoxygenation The ﬁrst radiobiological principle of importance here is that malignant tumors, even those of limited size, almost always contain a proportion of hypoxic cells that, because of their deﬁciency in oxygen, are highly resistant to killing by X- or γrays [19]. Examples are shown in Figure 5-1, showing hypoxic regions in sections of small tumors derived from human glioma xenograph lines. Figure 5-2 shows a dose-response curve, derived from the classic studies of Powers and Tolmach [20], illustrating the fraction of cells surviving in very small tumors in a mouse irradiated in a single fraction in vivo, and subsequently assayed by transplantation to other animals. The cellular survival curve is characterized by two distinct components; the slopes of the two components differing by a factor of 2 to 3. Up to doses of several Gy, the response is dominated by the killing of aerobic cells, whereas, for higher doses, the killing of hypoxic cells dominates. It is apparent that irradiating partially hypoxic tumors with a single large dose of several tens of Gy is a futile exercise if the goal is sterilization, because the hypoxic cells will not be adequately depopulated with a dose of this size. Figure 5-3 gives rough estimates, based on in vitro data, of the single fraction dose to sterilize a 30-mm-diameter tumor, with and without a hypoxic component [21].

51

52

d.j. brenner 100 90 80 20% hypoxic

70

Dose (Gy)

60 50 40 30 Fully oxygenated

20 10

Tumors, however, exhibit a characteristic known as reoxygenation, whereby, between fractionated doses of X- or γ-rays, tumors tend to reestablish their original pattern and proportion of oxygenated and hypoxic cells [22] (Fig. 5-4). In a fractionated

head & neck

ovarian

renal

brain

prostate

breast

medullo

colo-rectal

NSCLC

melanoma

SCLC

0 sarcoma

FIGURE 5-1. Hypoxic regions are present even in small tumors. Imaged here is a section of a small tumor derived from a human glioma xenograft line, with the black areas indicating regions of hypoxia. (From Rijken PF, Peters JP, Van der Kogel AJ. Quantitative analysis of varying proﬁles of hypoxia in relation to functional vessels in different human glioma xenograft lines. Radiat Res 2002; 157:626–632. Used with permission.)

FIGURE 5-3. Rough estimate, based on in vitro data, of the singlefraction dose required to sterilize various 30-mm-diameter tumors, which are either fully oxygenated (lower) or contain 20% hypoxic cells (upper). (Adapted from Leith JT, Cook S, Chougule P, et al. Intrinsic and extrinsic characteristics of human tumors relevant to radiosurgery: comparative cellular radiosensitivity and hypoxic percentages. Acta Neurochir Suppl 1994; 62:18–27. With kind permission of Springer Science + Business Media.)

100 response dominated by aerated cells

10–1

Surviving Fraction

10–2

response dominated by 1–2% hypoxic cells

10–3 10–4 10–5 10–6 10–7 0

5

10 15 20 25 Dose (Gy) FIGURE 5-2. Fraction of surviving cells, after a single radiation dose, in small mouse lymphosarcomas irradiated in vivo. The shallow initial part of the curve is due to the presence of aerated cells, whereas the steeper slope at higher doses is caused by the presence of 1% to 2% of hypoxic cells. (Adapted from Hall EJ, Brenner DJ. The radiobiology of radiosurgery: rationale for different treatment regimes for AVMs and malignancies. Int J Radiat Oncol Biol Phys 1993; 25:381–385. Copyright 1993, with permission from Elsevier.)

regime, therefore, each X- or γ-ray dose predominately kills aerobic cells, and the interval between treatments allows hypoxic cells to reestablish their oxygenated state.

Repair The second radiobiological principle is repair. It has been well established for many decades that protracting or fractionating an acute exposure reduces the level of cell killing. An example is shown in Figure 5-5 [23]. If all tissues were equally affected by changes in protraction or fractionation, then there would be no radiotherapeutic signiﬁcance to fractionation beyond the need to increase the dose to compensate for the increased cellular repair. There is, however, a wealth of experimental evidence indicating that there is a difference in shape between the doseresponse relationship characteristic of early-responding tissues and tumors and late-responding tissues. The inference from experiments in animals, which is conﬁrmed in clinical practice, is that the dose-response relationship for late-responding tissues is more “curvy” than that for early-responding tissues, as shown in Figure 5-6. In mathematical terms, if the dose-survival relationship is expressed in terms of a linear-quadratic relation, S = exp(−αD − βD2),

5.

radiobiological principles underlying stereotactic radiation therapy

FIGURE 5-4. Schematic illustration of the process of reoxygenation: Almost all tumors initially contain a mixture of aerated and hypoxic cells. A single radiation dose will kill a larger fraction of aerated than hypoxic cells because they are more radiosensitive (see Fig. 5-2). Thus, after irradiation, a larger fraction of cells are hypoxic. But after a short time, the tumor tends to return to its original proportion of aerated and hypoxic cells, allowing many of the previously hypoxic (but now aerated) cells to be killed by a second, and subsequent, dose fraction. (Reprinted from Hall EJ, Brenner DJ. The radiobiology of radiosurgery: rationale for different treatment regimes for AVMs and malignancies. Int J Radiat Oncol Biol Phys 1993; 25:381–385. Copyright 1993, with permission from Elsevier.)

Fraction of Initial Colony-Forming Units (3.5 cells/col) Surviving

100 0.36 rad/min 10–1 0.86 rad/min

10–2

16 rad/min

10–3

30 rad/min

10–4 107 rad/min 10–5

500

1000 1500 2000 2500 Dose (rad) FIGURE 5-5. Illustrating that, as a given dose is protracted, either by lowering the dose rate (as here), or by increasing the number of fractions, the biological effect (in this case, cell killing of Chinese hamster cells) decreases, due to repair. Base on data from Bedford and Mitchell [23]. (Reprinted from Hall EJ, Brenner DJ. The dose-rate effect revisited: radiobiological considerations of importance in radiotherapy. Int J Radiat Oncol Biol Phys 1991; 21:1403–1414. Copyright 1991; with permission from Elsevier.)

53

FIGURE 5-6. The γ-ray dose-response curve for late-responding tissues (e.g., AVM obliteration, cerebral necrosis) is “curvy” (i.e., has a small α/β ratio); for early-responding end points such as tumor control, the dose-response curve is straighter (i.e., the α/β ratio is larger). Consequently, dose fractionation spares late-responding tissues more than early-responding tissues. (Reprinted from Hall EJ, Brenner DJ. The radiobiology of radiosurgery: rationale for different treatment regimes for AVMs and malignancies. Int J Radiat Oncol Biol Phys 1993; 25:381– 385. Copyright 1993, with permission from Elsevier.)

then the ratio α/β (the dose at which the cell killing by the linear and quadratic terms is equal) tends to be small for lateresponding tissues (less than ∼3 Gy) and larger (greater than ∼8 Gy) for early-responding tissues [18, 22, 24–27]. The practical consequence of the difference in shape between the dose-response curves for early and late responding tissues is a marked difference in the response to fractionation of these two types of tissues. Late-responding tissues are more sensitive to changes in fractionation than early-responding tissues. The overall outcome is illustrated in Figure 5-6. A fractionated regime spares late-responding normal tissues more than a single acute dose for a given level of tumor damage. It is pertinent to ask what exactly constitute “early-” and “late” responding tissues? Probably, early-responding tissues contain a large proportion of cycling cells, whereas lateresponding tissues contain large proportions of noncycling cells.

Accelerated Repopulation As a tumor shrinks during radiotherapy, the surviving clonogens are stimulated to proliferate at an accelerated rate [28, 29]. This accelerated repopulation typically does not commence for about 30 days after the beginning of treatment as illustrated in Figure 5-7 [30]; however, radiation treatments of duration longer than about 30 days need to increase the total dose to compensate for this clonogen proliferation. Note that it is unrealistic to categorize almost any tumor type as one that is not signiﬁcantly prone to repopulation—if research in predictive assays has taught us anything, it is that intertumor variation can be large; thus, in this context, some tumors will grow slowly, and some will grow very quickly.

54

d.j. brenner

remaining two R’s of radiotherapy (reoxygenation and repair) affect stereotactic radiotherapy for the three main target lesions: malignancies, arteriovenous malformations, and benign tumors.

Malignancies

FIGURE 5-7. Tumor-control probability for T3 laryngeal cancer, as a function of overall time and overall dose. The data points, at the top of the dashed vertical lines, are those reported by Slevin et al], and the surface is the result of a linear-quadratic (with repopulation) ﬁt] to the data, with repopulation effectively starting about 33 days after the beginning of the treatment. (From Brenner DJ. Accelerated repopulation during radiotherapy: quantitative evidence for delayed onset. Radiat Oncol Invest 1993; 1:167–172. Used with permission.)

How Do These Radiotherapeutic Principles Apply to Stereotactic Radiation Therapy? In general, these three phenomena control the dose delivery for cancer radiotherapy: 1. We must fractionate treatment to overcome hypoxia and for differential response with late effects. 2. We would like to prolong treatment to limit early sequelae. 3. We would like to shorten treatment to prevent accelerated repopulation. Prima facie, requirements (2) and (3) are mutually exclusive and so, in most radiotherapeutic situations, the overall treatment time represents a compromise between short treatment times to minimize tumor repopulation and long treatment times to prevent unacceptable early complications. In those protocols that have attempted to utilize shorter overall treatment times, it has been necessary to reduce the overall dose in order to avoid excessive early complications [31]. By contrast, stereotactic radiotherapy represents a situation where the contradictory aspects of overall treatment time can be resolved without the need for compromise. Because of the excellent dose distributions that can be obtained using stereotactic radiotherapy, the need for long overall treatment times to reduce early normal-tissue complications will rarely apply. Speciﬁcally, early skin or mucosal reactions, which are often a limiting factor in radiotherapy, will not be an issue for intracranial stereotactic radiotherapy. Thus, short overall treatment times become a practical possibility, with their attendant advantages in terms of limiting tumor repopulation. Inherently, then, stereotactic radiation therapy is advantageous in terms of repopulation. We discuss here how the

As we have discussed, it is very unlikely that a single-dose fraction could sterilize even a small tumor that contains hypoxic malignant cells. Thus, there is a clear rationale, based on the issue of hypoxia, for stating unequivocally that single-fraction radiotherapy (radiosurgery) is a suboptimal modality when the target is a malignant tumor [32]. In addition, because of the differential α/β ratio of the tumor and the surrounding late-responding normal tissue sequelae (such as delayed cerebral necrosis, or optic neuropathy), even in the rare situation when there is minimal hypoxia, single-fractioned treatment of a malignancy would be expected to give a suboptimal therapeutic ratio between tumor control and late complications. Thus, an optimal stereotactic radiotherapy protocol for a large intracranial malignancy would involve large numbers of fractions over a short time period—20 fractions in 2 weeks might be an appropriate target. For smaller tumors, where the relative dose to normal tissue surrounding the malignancy is smaller—though still, as always, dose limiting—a regime consisting of 5 or 10 fractions over 1 week would combine the clinical advantages of a very short overall time, the radiobiological advantages of reoxygenation and (partial) repair of lateresponding normal tissue damage, and the practical advantages of a small number of treatments. It has been argued that, even for malignancies, “when the treatment volume is small and contains little functioning brain tissue, the need for fractionation may not apply” [33]; however, the dose-limiting factor of any radiotherapeutic treatment of a malignancy must always be the normal-tissue tolerance, so fractionation should always allow a greater tumoricidal dose. Of course, one of the potential advantages of radiosurgery is that it may be able to overcome the radioresistance of tumors such as metastases from melanomas or renal cell sarcomas; however, by keeping the number of fractions low (perhaps to ﬁve or six), the radioresistance of such tumors will still be adequately addressed while maintaining the other biological advantages of fractionation that have been outlined here. The arguments presented here do not exclude the fact that good results may be obtained using radiosurgery for brain malignancies. Indeed, there are reports of good results with this technique. Rather, for any given situation, better results— in terms of therapeutic ratio between tumor control and complications—would always be expected by fractionating than can be obtained with a single fraction, an important consideration in the treatment of cerebral malignancies.

Arteriovenous Malformations In treating arteriovenous malformations (AVMs), the goal is to produce progressive luminal closure through an inﬂammatory reaction in the vessel walls of the malformation, by irradiating the constituent epithelial cells [34]. This is a classic “late”

5.

radiobiological principles underlying stereotactic radiation therapy

response, occurring weeks to months after the radiation treatment and thus, prima facie, one would expect a relevant α/β ratio similar to that for other late-responding tissue. The issue here is that the dose-limiting side effect of the treatment, generally delayed cerebral radionecrosis, is, of course, also a late response [35–39]. Thus, in the treatment of AVMs, the tissue that it is desired to damage and the surrounding normal tissue that it is desired to spare are both of the same radiobiological type (i.e., they are both late-responding tissues). Consequently, one would expect that nothing is to be gained by a fractionated course relative to a single dose. In other words, a change in fractionation pattern will not preferentially produce more damage in the AVM than in the surrounding normal tissues. The ratio of damage between the AVM and the surrounding normal tissue would be the same whether the dose is delivered in a single or in multiple exposures. Given that the issue of hypoxic cells is not relevant here, this is a strong rationale for radiosurgery (i.e., a single, highly localized radiation treatment). In fact, there have been few estimates of α/β ratios reported in the literature corresponding with the end point of AVM obliteration, as few dose-response data have been established. An early estimate [32] of 0.6 Gy (range, 0.2 to 5 Gy) was reported based on limited dose-response results for complete AVM obliteration after stereotactic irradiation with helium ions. Using the more recent dose-response data from Touboul et al. [40] and Flickinger [41], we estimated values of 5.2 Gy from long-term obliteration data and 2.4 Gy (range, 0.9 to 8 Gy) from short-term obliteration data. A recent analysis by Kocher et al. [42] of most available data resulted in an overall best estimate of 3.5 Gy, and an estimate of 4.6 to 6.4 Gy for small AVMs (Fig. 5-8). All these values are consistent with those of a classic late response. In summary, in the treatment of AVMs, there is probably nothing to be gained by a fractionated course relative to a single

FIGURE 5-8. Linear quadratic ﬁt to combined long-term AVM obliteration data from Paris [39] and Pittsburgh [40]. The estimated α/β ratio from this analysis is 5.2 Gy, consistent with a recent estimate of 3.5 Gy by Kocher et al. [41], based on most currently available data, which is in turn consistent with a typical α/β value for a late-responding tissue.

55

dose. Thus, the optimal treatment for AVMs is expected to be single-fractionated stereotactic radiotherapy or radiosurgery; that is, a single, highly localized radiation treatment at, of course, the appropriate dose.

Benign Tumors Less is known about the sensitivity to fractionation of benign tumors. It is less likely that hypoxia will be a signiﬁcant role here; one estimate of the α/β ratio for benign parasellar meningiomas is 3.3 Gy [43], similar to that for late-responding normal tissues. For benign tumors as with AVMs, the most likely scenario is that both the target and the surrounding normal tissue respond in a similar way to fractionation, like that of a lateresponding normal tissue, suggesting that there is little to be gained from fractionation. For example, if the dose-limiting late sequelae were radiation-induced optic neuropathy, which is caused by damage to the arteries supplying the optic nerve, this almost certainly responds to changes in fractionation like a classic late-responding tissue [44, 45]. In summary, in the treatment of benign tumors, there is probably nothing to be gained by a fractionated course relative to a single dose; however we certainly do not have adequate data for a variety of benign tumors to consider this conclusion deﬁnitive.

How Many Fractions Should Be Used in Stereotactic Radiation Therapy of Cerebral Malignancies? As we have discussed, any multifraction scheme (with appropriately chosen doses) would be expected [32] to be superior to a single fraction for treating malignancies. Even apart from the issue of sparing late-responding normal tissue, if a tumor contains hypoxic cells, it would be most unlikely that it could be sterilized by a single dose of radiation. Most tumors, even small ones, probably contain hypoxic cells [22], so why take that risk? Given, then, that one should fractionate, how many fractions is appropriate? Five? Ten? Thirty? Consider the use of 30 fractions: when critical structures such as the optic nerve are likely to receive large doses, large fraction numbers such as 30 are indicated. On the other hand, most institutions treat solely with single fractions, and the use of, say, 10 fractions is certainly preferable to that. When critical structures are not considered to be at risk, however, then accelerated fractionation may become the treatment of choice [46] to avoid accelerated tumor repopulation, and as small a fraction number as ﬁve is probably appropriate to overcome the problem of hypoxic cells. Given that some tumors will show rapid repopulation, when critical structures are not a problem, why take the risk of a prolonged treatment? On a pragmatic note, whereas a few institutions are in a position to use, say, 30 fractions, most of the rapidly increasing number of centers using a Gamma Knife or a linac simply are not. On the other hand, smaller numbers of fractions could realistically become a generally acceptable option, with attendant beneﬁts compared with a single fraction.

56

d.j. brenner

Is Fractionated Stereotactic Radiotherapy Practical? The very earliest fractionated stereotactic radiotherapy treatments were accomplished using an invasive stereotactic head frame that was kept on for the entire course of treatment [5, 9, 15]. This approach, however, was difﬁcult for the patient and led to signiﬁcant infection issues at the pin sites. The ﬁrst major advance, in the early 1990s, was the development of noninvasive relocatable head frames [11, 12, 16], and versions of the Gill-Thomas-Cosman relocatable frame [16], which uses a bite block system, have been widely used [47–51]. Other systems are based on other anatomic features such as the ear canals/nasal bridge (the Laitinen stereoadapter [12, 52–54]) or a facial mask [55, 56]. Another recent system involves the use of one-time invasive placement of screws into the skull [57]. By contrast, the Gamma Knife is rather less suited to fractionated treatment. An invasive method based on inserting four screws into the skull has been described and tested by Walton et al. [58], but no subsequent clinical results have been reported. More recently, frameless, noninvasive optical systems have been introduced for use with linac systems, which show great potential [59–61]. Localization is accomplished through detection of four passive markers attached to a custom biteplate linked to the maxillary dentition of the patient. Translations and rotations of the patient’s isocenter are tracked in real time using a charge-coupled device (CCD)-based optical system detecting infrared light that is reﬂected off the four passive markers. A logical extension of frameless, noninvasive optical systems is the CyberKnife system [6]. The CyberKnife, developed at Stanford University, mounts a lightweight 6-MV linear accelerator on a computer-controlled robotic arm with 6 degrees of freedom, which irradiates the target while being guided in real time by computer image tracking technology in a frameless environment. Tracking is achieved using two ceiling-mounted diagnostic X-ray cameras with corresponding orthogonal and ﬂoor-mounted amorphous silicon detectors for real-time digital imaging. With such a system, a fractionation exposure presents no more technical difﬁculties than a single fraction, and the use of the CyberKnife for fractionated treatments has indeed been recently reported [62–66]. A ﬁnal approach to stereotactic radiation therapy may be the one with the greatest potential: to extend the current technology of intensity-modulated radiation therapy to apply to stereotactic radiotherapy. Intensity-modulated radiation therapy refers to linear accelerator–based radiation therapy in which individual small beams (typically 10 × 10 mm) within the larger ﬁeld are modulated with a multileaf collimator to produce highly conformal dose distributions. By modulating the intensity of large numbers of small beamlets within a larger open ﬁeld, an intensity variation is created that will generate an “optimized” dose distribution. To adapt this widely used technique to stereotactic radiotherapy, beamlet widths and multileaf collimator leaf widths needed to be reduced from ∼10 mm to ∼3 mm [67, 68], and this technology is now commercially available, leading to the use of both single-fraction [7, 68, 69] and multiple-fraction [70–73] intensity-modulated stereotactic radiation therapy.

How to Calculate the Appropriate Multifractioned Dose That Is Biologically Equivalent to a Given Single-Fractioned Dose The standard technique for making such isoeffect calculations is the linear-quadratic (L-Q) model, which was spelled out in detail more than 50 years ago by Lea and Catcheside [74, 75], based on a mechanistic analysis of chromosomal aberration induction in Tradescantia spores. The application of this formalism to radiation therapy has been reviewed extensively [18, 25, 76, 77]. Central to the approach is that radiotherapeutic response is primarily related to cell survival (or perhaps survival of groups of cells). This concept is itself not necessarily true, but there is now a wealth of evidence that cell killing is the dominant determinant of radiotherapeutic response, both for early- and lateresponding end points [18]. In the most basic linear-quadratic approach, cellular survival S at a dose D is written as S(D) = exp(−αD − βD²).

(5-1)

The mechanistic interpretation of Eq. (5-1) is that cell killing results from the interaction of two elementary damaged species (probably DNA double-strand breaks) to produce a species (perhaps a dicentric chromosomal aberration) that may cause lethality. The two terms in Eq. (5-1) indicate that the two elementary damaged species may be produced by the passage of the same track of radiation (linear term in dose) or by two different tracks (quadratic term). Clearly, if some time elapses between the passage of the ﬁrst and second tracks, there exists the possibility of the ﬁrst damaged site being repaired before interacting with the second. This repair will result in the a reduction of the second, quadratic term in Eq. (5-1) (but not the ﬁrst), by a factor denoted G by Lea and Catcheside [75]: S = exp(−αD − GβD²),

(5-2)

where, for acute exposures, G → 1, and for very long exposures, G → 0. In this context, “acute” and “long” are deﬁned relative to the half time for repair of sublethal damage (T1/2). In general, the G factor in Eq. (5-2) will depend on the details of the temporal distribution of the dose, as well as on T1/2. As discussed before, for many simple cases, G can be calculated analytically. For example, for n short, equal fractions, where the separation between fractions is much longer than T1/2, then G ≈ 1/n: S = exp(−αD − βD²/n),

(5-3)

Formulae for G for many other standard radiotherapeutic regimens have also been derived [75, 76], as has a general formalism for any possible regime [78]. It is important to note the mechanistic basis of Eq. (5-2) so that it is not simply an equation that happens to ﬁt cellular survival curves. It has been suggested, for example, that the L-Q model can be considered simply as the ﬁrst two terms (i.e., dose and dose squared) of a power-series expansion [79]. If L-Q were just another empirical model, there would be no good reason for considering the linear dose term to be independent of protraction/fractionation, and the quadratic term in dose to be fractionation dependent. This distinction between the linear

radiobiological principles underlying stereotactic radiation therapy

and the quadratic terms is at the heart of the L-Q model and its application. Based on Eq. (5-2), we can equate schemes; that is, produce a regimen with the same, say, tumor response, as a “tried and tested” regimen. Thus, assuming tumor repopulation (see above) is negligible (which will be true for a 5- or 10-fraction treatment), to match a new fractionation scheme (labeled 2, with n2 fractions) to a given (“old”) fractionation scheme (labeled 1, with n1 fractions), we can calculate the dose (D2) in scheme 2 such that αD2 + βD22/n2 = αD1 + βD12/n1.

57

60 multi-fractioned dose (Gy)

5.

50 40 30 20 10 0

And if the “tried and tested” regime is a single fraction (i.e., n1 = 1) then, dividing by β, we have (α/β)D2 + D22/n2 = (α/β)D1 + D12, which is a simple quadratic equation, easily solved to yield D2, assuming a given value of α/β. Some equivalences are shown in Table 5-1 and Figure 5-9 [13], and similar results have recently been published by Liu et al. [80]. It is important to be cautious in using isoeffect calculations to extrapolate from single-fraction radiosurgical protocols of 15 to 20 Gy to more radiobiologically appropriate multifractioned regimens. Indeed, the results of all isoeffect calculations in radiotherapy should not be accepted uncritically. There is, however, good evidence that the results are reasonable. Figure 5-10, for example, shows some isoeffect results from Van der Kogel [81] for late-responding damage to the rat spinal cord and from Douglas and Fowler [82] for acute damage in mouse skin. The form of the plots is such that, if the L-Q formalism applies, the data would fall on a straight line [82]. Although there are more sophisticated methods available for assessing agreement with the L-Q model [83], given the inherent uncer-

9

8

7 6 5 #o 4 f fra 3 ctio ns

2 1

25 21 23 19 ) 17 Gy 15 se ( 13 n do o i 11 t ac le fr sing

FIGURE 5-9. Gamma-ray doses for multifractioned stereotactic radiotherapy, which are calculated to be equivalent, in terms of tumor control, to single-fractioned radiosurgical doses. See also Table 5-1. (Reprinted from Hall EJ, Brenner DJ. The radiobiology of radiosurgery: rationale for different treatment regimes for AVMs and malignancies. Int J Radiat Oncol Biol Phys 1993; 25:381–385. Copyright 1993, with permission from Elsevier.)

tainties in the data, it is clear that all these data, including those for single fractions of ∼20 Gy, are consistent with the L-Q model. Although it is important to be appropriately critical of the L-Q model and its application to radiotherapy, it is equally important to recognize that it is the best model we have. It is a

TABLE 5-1. Total gamma-ray doses for multifractioned stereotactic radiotherapy, which are calculated [13] to be equivalent, in terms of tumor control, to single-fractioned radiosurgical doses (see also Fig. 5-9). Single-fraction dose (Gy)

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Fx = 2 (Gy)

Fx = 3 (Gy)

Fx = 5 (Gy)

Fx = 9 (Gy)

12.7 14.0 15.4 16.8 18.1 19.5 20.9 22.3 23.6 25.0 26.4 27.8 29.2 30.6 32.0 33.4

14.3 15.9 17.6 19.2 20.8 22.5 24.1 25.8 27.4 29.1 30.8 32.5 34.1 35.8 37.5 39.2

16.5 18.4 20.4 22.4 24.4 26.5 28.5 30.6 32.6 34.7 36.8 38.9 41.0 43.2 45.3 47.4

18.9 21.3 23.7 26.2 28.7 31.2 33.8 36.4 39.0 41.6 44.3 47.0 49.7 52.4 55.1 57.8

FIGURE 5-10. Isoeffect data for late response from three (䊐, 䊊, 䉭) different regions of the rat spinal cord [80], and for acute skin reactions (䉬) in mice [81]. All the points at ≥20 Gy/fraction correspond with single acute exposures. The data are plotted in a form such that, if they follow a linear-quadratic relationship, the points would fall on a straight line. The highlighted data points, which are for single-fractioned exposures of 20 to 25 Gy, are clearly consistent with the linear-quadratic formalism. (Reprinted from Hall EJ, Brenner DJ. In response to Dr. Marks. Int J Radiat Oncol Biol Phys 1995; 32:275–276. Copyright 1995, with permission from Elsevier.)

58

d.j. brenner

mechanistically based model of cell killing [84], and there is a wealth of evidence [18, 22] that cell killing dominates all radiotherapeutic response, including, for example, late vascular damage. Of course, the simplest form of the L-Q model is not necessarily the most appropriate to apply. When repopulation is important, the L-Q formalism can be appropriately modiﬁed [29, 85]. If redistribution or reoxygenation is important, the L-Q formalism can again be appropriately modiﬁed [86]. Similarly, if there is evidence that the L-Q model is underpredicting survival at high doses (though from results such as in Fig. 5-10, it appears that this is not the case at doses ≤20 Gy), appropriate saturation-related modiﬁcations to the L-Q formalism have been described [87].

Conclusion Malignant Tumors • Radiobiological and clinical arguments, based on the need to overcome hypoxia and the potential for a differential response between the tumor and the surrounding normal tissue, strongly suggest that fractionation will result in improved outcome. • For most malignancies, even when good dose distributions can be obtained, single-fraction radiation therapy has not been in general use for more than 70 years. • For most sites, only a few fractions (5 to 10) would be optimal for treating malignant tumors. • Modern technology has made fractionated stereotactic radiotherapy increasingly practical.

Arteriovenous Malformations • The arguments in favor of fractionation probably do not hold, and single-fraction treatment is generally optimal. • There is no hypoxic cell issue to overcome, and both the AVM and the surrounding normal tissue probably respond the same way to changes in fractionation.

Benign Tumors • There is some evidence that benign tumors respond to fractionation like a late-responding normal tissue, in which case fractionation would not improve outcome, unless hypoxic cells were an issue. • Overall, the radiobiological evidence is probably insufﬁcient to make recommendations on appropriate fractionation schemes for benign tumors.

References 1. Leksell L. Cerebral radiosurgery. I. Gammathalanotomy in two cases of intractable pain. Acta Chir Scand 1968; 134:585–595. 2. Lunsford LD, Flickinger JC, Larson D. Regarding: Rosenthal DI, Glatstein E. “We’ve Got a Treatment, but What’s the Disease?” Oncologist 1996; 1. Oncologist 1997; 2:59–61. 3. Adler JR Jr, Colombo F, Heilbrun MB, et al. Toward an expanded view of radiosurgery. Neurosurgery 2004; 55:1374–1376.

4. Colombo F, Benedetti A, Pozza F, et al. External stereotactic irradiation by linear accelerator. Neurosurgery 1985; 16:154–160. 5. Houdek PV, Fayos JV, Van Buren JM, et al. Stereotaxic radiotherapy technique for small intracranial lesions. Med Phys 1985; 12:469–472. 6. Adler JR Jr, Chang SD, Murphy MJ, et al. The CyberKnife: a frameless robotic system for radiosurgery. Stereotact Funct Neurosurg 1997; 69:124–128. 7. Benedict SH, Cardinale RM, Wu Q, et al. Intensity-modulated stereotactic radiosurgery using dynamic micro-multileaf collimation. Int J Radiat Oncol Biol Phys 2001; 50:751–758. 8. Leavitt DD, Watson G, Tobler M, et al. Intensity-modulated radiosurgery/radiotherapy using a micromultileaf collimator. Med Dosim 2001; 26:143–150. 9. Pozza F, Colombo F, Chierego G, et al. Low-grade astrocytomas: treatment with unconventionally fractionated external beam stereotactic radiation therapy. Radiology 1989; 171:565–569. 10. Heifetz MD, Whiting J, Bernstein H, et al. Stereotactic radiosurgery for fractionated radiation: a proposal applicable to linear accelerator and proton beam programs. Stereotact Funct Neurosurg 1989; 53:167–177. 11. Hariz MI, Henriksson R, Lofroth PO, et al. A non-invasive method for fractionated stereotactic irradiation of brain tumors with linear accelerator. Radiother Oncol 1990; 17:57–72. 12. Delannes M, Daly NJ, Bonnet J, et al. Fractionated radiotherapy of small inoperable lesions of the brain using a non-invasive stereotactic frame. Int J Radiat Oncol Biol Phys 1991; 21:749– 755. 13. Brenner DJ, Martel MK, Hall EJ. Fractionated regimens for stereotactic radiotherapy of recurrent tumors in the brain. Int J Radiat Oncol Biol Phys 1991; 21:819–824. 14. Souhami L, Olivier A, Podgorsak EB, et al. Fractionated stereotactic radiation therapy for intracranial tumors. Cancer 1991; 68: 2101–2108. 15. Schwade JG, Houdek PV, Landy HJ, et al. Small-ﬁeld stereotactic external-beam radiation therapy of intracranial lesions: fractionated treatment with a ﬁxed-halo immobilization device. Radiology 1990; 176:563–565. 16. Gill SS, Thomas DG, Warrington AP, et al. Relocatable frame for stereotactic external beam radiotherapy. Int J Radiat Oncol Biol Phys 1991; 20:599–603. 17. Tomé WA, Mehta MP, Meeks SL, et al. Fractionated stereotactic radiotherapy: a short review. Technol Cancer Res Treat 2002; 1: 153–172. 18. Thames HD, Hendry JH. Fractionation in Radiotherapy. London: Taylor & Francis, 1987. 19. Overgaard J, Horsman MR. Modiﬁcation of hypoxia-induced radioresistance in tumors by the use of oxygen and sensitizers. Semin Radiat Oncol 1996; 6:10–21. 20. Powers WE, Tolmach LJ. A multicomponent x-ray survival curve for mouse lymphosarcoma cells irradiated in vivo. Nature 1963; 197:710–711. 21. Leith JT, Cook S, Chougule P, et al. Intrinsic and extrinsic characteristics of human tumors relevant to radiosurgery: comparative cellular radiosensitivity and hypoxic percentages. Acta Neurochir Suppl 1994; 62:18–27. 22. Hall EJ. Radiobiology for the Radiologist. Philadelphia: Lippincott Williams & Wilkins, 2000. 23. Bedford JS, Mitchell JB. Dose-rate effects in synchronous mammalian cells in culture. Radiat Res 1973; 54:316–327. 24. Brenner DJ, Hall EJ. Conditions for the equivalence of continuous to pulsed low dose rate brachytherapy. Int J Radiat Oncol Biol Phys 1991; 20:181–190. 25. Fowler JF. The linear-quadratic formula and progress in fractionated radiotherapy. Br J Radiol 1989; 62:679–694.

5.

radiobiological principles underlying stereotactic radiation therapy

26. Thames HD, Bentzen SM, Turesson I, et al. Fractionation parameters for human tissues and tumors. Int J Radiat Biol 1989; 56: 701–710. 27. Withers HR, Thames HD Jr, Flow BL, et al. The relationship of acute to late skin injury in 2 and 5 fraction/week gamma-ray therapy. Int J Radiat Oncol Biol Phys 1978; 4:595–601. 28. Withers HR, Taylor JM, Maciejewski B. The hazard of accelerated tumor clonogen repopulation during radiotherapy. Acta Oncol 1988; 27:131–146. 29. Brenner DJ. Accelerated repopulation during radiotherapy— evidence for delayed onset. Radiat Oncol Invest 1993; 1:167– 172. 30. Slevin NJ, Hendry JH, Roberts SA, et al. The effect of increasing the treatment time beyond three weeks on the control of T2 and T3 laryngeal cancer using radiotherapy. Radiother Oncol 1992; 24:215–220. 31. Saunders MI, Dische S, Grosch EJ, et al. Experience with CHART. Int J Radiat Oncol Biol Phys 1991; 21:871–878. 32. Hall EJ, Brenner DJ. The radiobiology of radiosurgery: rationale for different treatment regimes for AVMs and malignancies. Int J Radiat Oncol Biol Phys 1993; 25:381–385. 33. Loefﬂer JS, Kooy HM, Wen PY, et al. The treatment of recurrent brain metastases with stereotactic radiosurgery. J Clin Oncol 1990; 8:576–582. 34. Ogilvy CS. Radiation therapy for arteriovenous malformations: a review. Neurosurgery 1990; 26:725–735. 35. Kjellberg RN, Hanamura T, Davis KR, et al. Bragg-peak protonbeam therapy for arteriovenous malformations of the brain. N Engl J Med 1983; 309:269–274. 36. Marks LB, Spencer DP. The inﬂuence of volume on the tolerance of the brain to radiosurgery. J Neurosurg 1991; 75:177–180. 37. Nedzi LA, Kooy H, Alexander III E, et al. Variables associated with the development of complications from radiosurgery of intracranial tumors. Int J Radiat Oncol Biol Phys 1991; 21:591– 599. 38. Sheline GE, Wara WM, Smith V. Therapeutic irradiation and brain injury. Int J Radiat Oncol Biol Phys 1980; 6:1215–1228. 39. Wowra B, Schmitt HP, Sturm V. Incidence of late radiation necrosis with transient mass effect after interstitial low dose rate radiotherapy for cerebral gliomas. Acta Neurochir (Wien) 1989; 99:104–108. 40. Touboul E, Al Halabi A, Buffat L, et al. Single-fraction stereotactic radiotherapy: a dose-response analysis of arteriovenous malformation obliteration. Int J Radiat Oncol Biol Phys 1998; 41:855–861. 41. Flickinger JC, Kondziolka D, Maitz AH, et al. An analysis of the dose-response for arteriovenous malformation radiosurgery and other factors affecting obliteration. Radiother Oncol 2002; 63:347– 354. 42. Kocher M, Wilms M, Makoski HB, et al. Alpha/beta ratio for arteriovenous malformations estimated from obliteration rates after fractionated and single-dose irradiation. Radiother Oncol 2004; 71:109–114. 43. Shrieve DC, Hazard L, Boucher L, et al. Dose fractionation in stereotactic radiotherapy for parasellar meningiomas: radiobiological considerations of efﬁcacy and optic nerve tolerance. J Neurosurg 2004; 101(Suppl 3):390–395. 44. Goldsmith BJ, Rosenthal SA, Wara WM, et al. Optic neuropathy after irradiation of meningioma. Radiology 1992; 185:71–76. 45. Bhandare N, Monroe AT, Morris CG, et al. Does altered fractionation inﬂuence the risk of radiation-induced optic neuropathy? Int J Radiat Oncol Biol Phys 2005; 62:1070–1077. 46. Brenner DJ, Hall EJ. Stereotactic radiotherapy of intracranial tumors—an ideal candidate for accelerated treatment. Int J Radiat Oncol Biol Phys 1994; 28:1039–1041; discussion 1047.

59

47. Kooy HM, Dunbar SF, Tarbell NJ, et al. Adaptation and veriﬁcation of the relocatable Gill-Thomas-Cosman frame in stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 1994; 30:685–691. 48. Kitchen ND, Thomas DG. Minimally invasive stereotaxy: clinical use of the Gill-Thomas-Cosman (GTC) repeat stereotactic localiser. Minim Invasive Neurosurg 1994; 37:61–63. 49. Fairclough-Tompa L, Larsen T, Jaywant SM. Immobilization in stereotactic radiotherapy: the head and neck localizer frame. Med Dosim 2001; 26:267–273. 50. Burton KE, Thomas SJ, Whitney D, et al. Accuracy of a relocatable stereotactic radiotherapy head frame evaluated by use of a depth helmet. Clin Oncol (R Coll Radiol) 2002; 14:31–39. 51. Kumar S, Burke K, Nalder C, et al. Treatment accuracy of fractionated stereotactic radiotherapy. Radiother Oncol 2005; 74:53–59. 52. Miranpuri AS, Tome WA, Paliwal BR, et al. Assessment of patient-independent intrinsic error for a noninvasive frame for fractionated stereotactic radiotherapy. Int J Cancer 2001; 96:320–325. 53. Kalapurakal JA, Ilahi Z, Kepka AG, et al. Repositioning accuracy with the Laitinen frame for fractionated stereotactic radiation therapy in adult and pediatric brain tumors: preliminary report. Radiology 2001; 218:157–161. 54. Ashamalla H, Addeo D, Ikoro NC, et al. Commissioning and clinical results utilizing the Gildenberg-Laitinen adapter device for X-ray in fractionated stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 2003; 56:592–598. 55. Alheit H, Dornfeld S, Dawel M, et al. Patient position reproducibility in fractionated stereotactically guided conformal radiotherapy using the BrainLab mask system. Strahlenther Onkol 2001; 177:264–268. 56. Lopatta E, Liesenfeld SM, Bank P, et al. Improved patient repositioning accuracy by integrating an additional jaw ﬁxation into a high precision face mask system in stereotactic radiotherapy of the head. Strahlenther Onkol 2003; 179:571–575. 57. Salter BJ, Fuss F, Vollmer DG, et al. The TALON removable head frame system for stereotactic radiosurgery/radiotherapy: measurement of the repositioning accuracy. Int J Radiat Oncol Biol Phys 2001; 51:555–562. 58. Walton L, Hampshire A, Roper A, et al. Development of a relocatable frame technique for gamma knife radiosurgery. Technical note. J Neurosurg 2000; 93(Suppl 3):198–202. 59. Kai J, Shiomi H, Sasama T, et al. Optical high-precision threedimensional position measurement system suitable for head motion tracking in frameless stereotactic radiosurgery. Comput Aided Surg 1998; 3:257–263. 60. Kamath R, Ryken TC, Meeks SL, et al. Initial clinical experience with frameless radiosurgery for patients with intracranial metastases. Int J Radiat Oncol Biol Phys 2005; 61:1467–1472. 61. Meeks SL, Bova FJ, Wagner TH, et al. Image localization for frameless stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 2000; 46:1291–1299. 62. Mehta VK, Lee QT, Chang SD, et al. Image guided stereotactic radiosurgery for lesions in proximity to the anterior visual pathways: a preliminary report. Technol Cancer Res Treat 2002; 1:173–180. 63. Ishihara H, Saito K, Nishizaki T, et al. CyberKnife radiosurgery for vestibular schwannoma. Minim Invasive Neurosurg 2004; 47: 290–293. 64. Pham CJ, Chang SD, Gibbs IC, et al. Preliminary visual ﬁeld preservation after staged CyberKnife radiosurgery for perioptic lesions. Neurosurgery 2004; 54:799–810; discussion 810–792. 65. Kajiwara K, Saito K, Yoshikawa K, et al. Image-guided stereotactic radiosurgery with the CyberKnife for pituitary adenomas. Minim Invasive Neurosurg 2005; 48:91–96.

60

d.j. brenner

66. Chang SD, Gibbs IC, Sakamoto GT, et al. Staged stereotactic irradiation for acoustic neuroma. Neurosurgery 2005; 56:1254– 1261. 67. Xia P, Geis P, Xing L, et al. Physical characteristics of a miniature multileaf collimator. Med Phys 1999; 26:65–70. 68. Cosgrove VP, Jahn U, Pfaender M, et al. Commissioning of a micro multi-leaf collimator and planning system for stereotactic radiosurgery. Radiother Oncol 1999; 50:325–336. 69. Georg D, Dieckmann K, Bogner J, et al. Impact of a micromultileaf collimator on stereotactic radiotherapy of uveal melanoma. Int J Radiat Oncol Biol Phys 2003; 55:881–891. 70. Baumert BG, Norton IA, Davis JB. Intensity-modulated stereotactic radiotherapy vs. stereotactic conformal radiotherapy for the treatment of meningioma located predominantly in the skull base. Int J Radiat Oncol Biol Phys 2003; 57:580–592. 71. Tobler M, Leavitt DD, Watson G. Optimization of the primary collimator settings for fractionated IMRT stereotactic radiotherapy. Med Dosim 2004; 29:72–79. 72. Jin JY, Yin FF, Ryu S, et al. Dosimetric study using different leaf-width MLCs for treatment planning of dynamic conformal arcs and intensity-modulated radiosurgery. Med Phys 2005; 32:405– 411. 73. Fuss M, Salter BJ, Sadeghi A, et al. Fractionated stereotactic intensity-modulated radiotherapy (FS-IMRT) for small acoustic neuromas. Med Dosim 2002; 27:147–154. 74. Lea DE. Actions of Radiation on Living Cells. Cambridge: Cambridge University Press, 1946. 75. Lea DE, Catcheside DG. The mechanism of the induction by radiation of chromosome aberrations in tradescantia. J Genet 1942; 44:216–245. 76. Brenner DJ, Herbert DE. The use of the linear-quadratic model in clinical radiation oncology can be defended on the basis of empirical evidence and theoretical argument. Med Phys 1997; 24:1245–1248.

77. Dale RG. The application of the linear-quadratic dose-effect equation to fractionated and protracted radiotherapy. Br J Radiol 1985; 58:515–528. 78. Brenner BJ, Huang Y, Hall EJ. Fractionated high dose-rate versus low dose-rate regimens for intracavitary brachytherapy of the cervix: equivalent regimens for combined brachytherapy and external irradiation. Int J Radiat Oncol Biol Phys 1991; 21:1415–1423. 79. Yaes RJ, Patel P, Maruyama Y. On using the linear-quadratic model in daily clinical practice. Int J Radiat Oncol Biol Phys 1991; 20:1353–1362. 80. Liu L, Bassano DA, Prasad SC, et al. The linear-quadratic model and fractionated stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 2003; 57:827–832. 81. van der Kogel AJ. Chronic effects of neutrons and charged particles on spinal cord, lung, and rectum. Radiat Res Suppl 1985; 8: S208–216. 82. Douglas BG, Fowler JF. The effect of multiple small doses of x rays on skin reactions in the mouse and a basic interpretation. Radiat Res 1976; 66:401–426. 83. Tucker SL. Tests for the ﬁt of the linear-quadratic model to radiation isoeffect data. Int J Radiat Oncol Biol Phys 1984; 10:1933– 1939. 84. Sachs RK, Hahnfeld P, Brenner DJ. The link between low-LET dose-response relations and the underlying kinetics of damage production/repair/misrepair. Int J Radiat Biol 1997; 72:351–374. 85. Travis EL, Tucker SL. Isoeffect models and fractionated radiation therapy. Int J Radiat Oncol Biol Phys 1987; 13:283–287. 86. Brenner DJ, Hlatky LR, Hahnfeldt PJ, et al. A convenient extension of the linear-quadratic model to include redistribution and reoxygenation. Int J Radiat Oncol Biol Phys 1995; 32:379–390. 87. Zaider M, Rossi HH. Saturation effects for sparsely ionizing particles. New York: Columbia University, Radiological Research Laboratory Annual Report (COO-4733-3:126-134), 1980.

6

Experimental Radiosurgery Models Ajay Niranjan and Douglas Kondziolka

Introduction The ﬁeld of stereotactic radiosurgery represents one of the fundamental shifts in neurologic surgery over the past two decades. Compared with conventional invasive surgery techniques, radiosurgery is minimally invasive and relies on biological response of tissues in order to eradicate or inactivate them. Radiosurgery is conceptually different from fractionated radiation therapy. The efﬁcacy of large-ﬁeld fractionated radiotherapy to treat brain tumors is dependent on biological differences between normal and tumor cells. Fractionated radiotherapy exploits these differences to limit the risk of normal tissue injury in patients with malignant brain tumors, thus it can increase the therapeutic ratio, which is equivalent to the rate of tumor control divided by the rate of complications. Radiosurgery, in contrast with conventional radiotherapy, uses a single high dose of radiation. Normal tissue effects are limited by the highly focused nature of the radiosurgical beams. In addition, unlike radiotherapy, radiosurgery manages smallvolume targets using much higher doses. Finally, whereas fractionated radiotherapy is generally most effective in killing rapidly dividing cells, radiosurgery induces biological responses irrespective of the mitotic activity, oxygenation, and inherent radiosensitivity of target cells. Considering the unique biological response of tissues to radiosurgery, it is important to study the biological effects of radiosurgery in both normal and pathologic nervous system tissues in animal models. Information gained from radiosurgical research studies would be useful in devising strategies to avoid, prevent, or ameliorate damage to normal tissue without compromising treatment efﬁcacy. As radiosurgery evolves from a treatment speciﬁcally for brain tumors into a widely available treatment modality for a variety of intracranial lesions, understanding of biological responses using animal investigations becomes crucial.

proton beam from a 230-cm synchrocyclotron) [1]. The early histologic results (3rd to 8th day) showed complete transection of the rabbit spinal cord with 40,000 rads (400 Gy) using 1.5-mm beam diameter and with 200 Gy using 10-mm beam diameter. These investigators also used a goat brain model to document sharply deﬁned lesions in deep parts of brain 4 to 7 weeks after 200 Gy of stereotactic multiple-port proton beam radiation. Rexed et al. performed proton beam radiosurgery on rabbit brain to study the long-term effects (2 to 56 weeks) of irradiation [2]. Using a 1.5-mm collimator, 20,000 rads (200 Gy) was delivered to the anterior part of rabbit brain. Serial histology up to 3 months showed a well-demarcated lesion in the path of the beam. After 3 months, a lesion broader than the beam size was noted. Leksell et al. investigated the features of a radiolesion in the depth of brain produced by cross-ﬁred irradiation with a narrow beam of high-energy [3]. Their results showed that with 200 Gy, well-circumscribed intracerebral lesions of appropriate size and shape could be created. Andersson et al. performed protons radiosurgery on goat brain to study the late histologic effects of the cross-ﬁred beams [4]. These investigators did not detect untoward changes in or around the lesion (e.g., elements resembling neoplasm, hemorrhage, or telangiectasis) 1.5 to 4 years after 200 Gy radiosurgery. Nilsson et al. irradiated (100 to 300 Gy) the basilar artery of cats by stereotactic technique using a 179-source cobalt-60 prototype gamma unit [5]. Histology demonstrated vascular lesions such as vacuolization, degeneration, and desquamation of the endothelium and hyalinization and necrosis of the muscular coat. The reparatory reactions were relatively sparse, and thrombosis was completely absent. These investigations demonstrated that radiosurgery could potentially be used to create sharply deﬁned lesions in deep parts of the brain (Table 6-1).

Experimental Models for Investigations into Effects of Normal Brain Radiosurgery Experimental Models to Investigate Radiosurgery as a Neurosurgical Tool In the initial studies involving focused radiation as a neurosurgical tool, rabbit and goat central nervous system (CNS) models were used. Larsson et al. used a rabbit spinal cord model to investigate the effect of the high-energy proton beam (185-MeV

The effect of fractionated radiation therapy on the nervous system depends on both the dose delivered and the time elapsed [6–8]. Although all brain regions are affected, the radiation response tends to be most severe in white-matter regions [9–11]. A dose-related variable latency period after irradiation can last from months to years [6–8]. A single radiation dose of 20 Gy to

61

62

a. niranjan and d. kondziolka

TABLE 6-1. Radiosurgery as a neurosurgical tool. First author and year

Maximum dose (Gy)

Region(s) irradiated

Irradiation technique

Animal model

Results

Larsson, 1958 [1]

20 krad

Internal capsule

Proton beam

Rabbit and goat

Rexed, 1960 [2]

20 krad

Forebrain

Proton beam

Rabbit

Leksell, 1960 [3]

20–38 krad

Internal capsule

Proton beam

Goat

Anderson, 1970 [4]

15–20 krad

Basal ganglia, internal capsule, optic chiasm, optic tract

Proton beam

Goat

Stereotactic proton beam radiation can produce restricted brain and spinal cord lesions in 4–6 weeks High-energy proton irradiation can cause a welldeﬁned lesion in 2–56 weeks Well-circumscribed radiolesions of appropriate size and shape can be produced with a suitable dose (20 krad) No untoward late effects (neoplasia, telangiectasis, hemorrhage) were noted 1.5 to 4 years after proton beam radiosurgery

rat brain creates lesions that are primarily conﬁned to the vasculature at latencies of more than 12 months [12, 13]. Radiation-induced vascular changes in the CNS include perivascular ﬁbrosis and ﬁbrinoid necrosis of vessel walls, hyaline degeneration, edema, telangiectasia, thrombosis, and hemorrhage [14]. At higher doses of around 25 Gy, white-matter lesions predominate in irradiated rat brain at latencies of less than 12 months. Radiation-induced lesions in the white matter can range from demyelination to malacia [14]. Radiosurgery models used to study the biologic response of radiosurgery on the CNS are listed in Table 6-1. In initial studies, Lunsford et al. [15] and Kondziolka et al. [16] studied the radiobiological effects of stereotactic radiosurgery using a baboon model. These investigators delivered a central dose of 150 Gy using an 8-mm collimator to the caudate, thalamus, or pons regions using the Gamma Knife. No changes were noted by computed tomography (CT) or by T1-weighted, T2-weighted, and gadolinium-enhanced magnetic resonance imaging (MRI) at 4 weeks after irradiation. A circumscribed, contrast-enhanced lesion was seen by 6 to 8 weeks, which was characterized by demyelination, microvascular damage, hemorrhage, and astrocytosis. The edema was ﬁrst evident at 8 weeks. Frank necrosis appeared in the irradiated region by 24 weeks. Kondziolka et al. [17] used a rat brain model to study the histologic changes in rat brain 90 days after radiosurgery (ﬁxed latency, variable dose). The frontal lobe of rats was irradiated with maximal doses of 30 to 200 Gy using a 4-mm collimator. Detectable histologic alterations were noted with doses of more than 70 Gy. Necrosis was seen only in tissues irradiated with more than 100 Gy. Blatt et al. [18] evaluated serial tissue changes after 125 Gy linear accelerator (linac) radiosurgery of internal capsule of cats (variable latency, ﬁxed dose). MRI and histopathologic evaluations were performed serially for 1 year starting at 3.5 weeks after irradiation. Tissue necrosis was evident in the cat brain by 3.5 weeks and was accompanied by vascular proliferation and edema. The lesions showed increased vascularity and microglial inﬁltration that resolved by 12 to 29 weeks. In a study evaluating the effects of radiation dose and time after treatment on the radiosensitivity of brain, Kamiryo et al. irradiated rat brain with maximum doses of 50, 75, or 120 Gy and

analyzed for histologic changes and blood-brain barrier integrity up to 12 months (variable latency, variable dose) [19], Whereas higher doses (120 Gy) induced alterations in astrocytic morphology by just 3 days after treatment, such changes were not observed until 3 months with lower doses (50 Gy). Bloodbrain barrier breakdown as assessed by Evans blue leakage was evident within 3 weeks of 120-Gy irradiation but was not seen across 12 months after 50 Gy. These ﬁndings indicate that the latent period between irradiation and detection of pathologic alterations is dependent on both the dose and the biological end point used. Such ﬁndings are consistent with the results of studies using a more conventional radiation source (60Co) to irradiate the rat spinal cord. In this model of radiation-induced CNS injury, latency to paralysis after irradiation of an 8- or 16mm segment of cervical spinal cord decreased as dose increased [20]. Also, the ED50 for paralysis after 4 mm of spinal cord irradiation was 51 Gy, whereas the ED50 for vascular damage was only 25.6 Gy. The impact of dose and biological end points on latency was also reported by Karger et al. [21], who evaluated the rat brain using T1- and T2-weighted, gadoliniumenhanced MRI at 15, 17, or 20 months after treatment with 26 to 50 Gy of linear accelerator–based radiosurgery. A 3-mm collimator was employed to deliver the radiosurgical dose using a convergent arc technique and resulted in an 80% isodose distribution of 4.7 mm in diameter. No radiation-induced affect was noted on MRI at any time point for doses less than 30 Gy. After 40-Gy radiosurgery, the latency of detectable MRI changes was approximately 19 to 20 weeks, whereas the latency after 50 Gy was 15 to 16 weeks. In addition, T1-weighted changes in the MRI signal had a shorter latency than T2-weighted changes. Considering that the changes in T1-weighted images are due to leakage of gadolinium-DTPA across the blood-brain barrier, the results of this study point to the likely role of vascular damage in radiation-induced injury. The importance of the vasculature in radiation-induced brain injury is well recognized; a prevalent hypothesis regarding the pathogenesis after conventional radiotherapy is that damage to capillary endothelium and/or supporting cells ultimately interrupts blood ﬂow resulting in secondary ischemic necrosis. In a report focusing on vascular changes after a maximum dose of 75 Gy delivered

6.

63

experimental radiosurgery models

TABLE 6-2. Dose and latency effects of radiosurgery on central nervous system. First author and year

Maximum dose (Gy)

Region(s) irradiated

Irradiation technique

Animal model

Results

Lunsford, 1990 [15]

150

Caudate, thalamus, pons R. frontal lobe

Gamma Knife, 8-mm collimator Gamma Knife, 4-mm collimator

Baboon

Internal capsule

Linac, 10-mm collimator

Cat

75

Parietal cortex

Gamma Knife, 4-mm collimator

Rat

26–50

Parietal cortex

Linac, 3-mm collimator

Rat

MRI and histology showed lesion 45–60 days posttreatment Histopathology at 90 days showed tissue changes at doses lower than 60 Gy, necrosis at doses of 100 Gy MRI and serial histopathology showed mass effect and neurologic deﬁcits at 3.5–4.5 weeks, some necrosis 12–29 weeks, and late resorption of necrosis Electron microscopy at 3.5 months showed decreased vascularity and increased capillary diameter in irradiated regions; basement membrane changes precede vascular damage MRI showed contrast enhancement at 15 weeks after 50-Gy and 19 weeks after 40Gy radiosurgery

Kondziolka, 1992 [17]

Blatt, 1994 [18]

Kamiryo, 2001 [22]

Karger, 2002 [21]

30–200

149

to the rat brain using a Gamma Knife, it was noted that vascular changes, speciﬁcally alterations in the basement membrane, preceded changes in necrosis [22]. This ﬁnding suggests that vascular damage is also an important component in biological response after radiosurgery. Although radiosurgery generally involves the use of higher single doses and smaller treatment volumes than conventional irradiation, the histologic effects of these two methodologies appear similar. The biggest differences are that the latency period after radiosurgery is shorter, and the major histologic ﬁnding is vascular damage (Table 6-2).

Experimental Models Exploring Strategies to Enhance Tumoricidal Effect of Radiosurgery Although benign tumor radiosurgery is associated with high tumor control rates, malignant glial tumors often recur. Additional strategies to improve cell kill of malignant brain tumors and to protect normal surrounding brain tissue are needed. A few strategies for radioprotection of normal tissue and radiosensitization of tumor tissue have already been explored.

Rat

breaks and initiate lipid peroxidation of vascular membranes, ultimately leading to membrane lysis and cell death. As a lipid antioxidant and free radical scavenger, 21-AS inhibit oxygen radical–initiated peroxidation of vascular membrane. 21Aminosteroids also block the release of free arachidonic acid from cell membrane, thereby inhibiting activation of the proinﬂammatory cyclooxygenase pathway. These properties of 21AS are thought to protect cerebral vessels from injury and prevent cerebral edema. The effects of the 21-AS compound U-74389G on radiation injury have been evaluated in both rat and cat models. Bernstein et al. [26] reported that U-74389G reduced brachytherapy-induced brain injury in the rat. Buatti et al. [27] found that this same agent also protected the cat brain from injury due to radiosurgery and was signiﬁcantly more effective than corticosteroids. In our own studies, 15 mg/kg but not 5 mg/kg U-74389G was effective at reducing brain injury in the rat when administered 1 hour before radiosurgery. U74389G ameliorated vasculopathy and regional edema and delayed the onset of necrosis while gliosis was unaffected [28]. Preliminary results suggest that this agent may act through reduction of the cytokines induced by brain irradiation. An alternative strategy for radiation protection seeks to repair radiation-induced brain damage.

Radiation Protection and Repair The initial strategies included use of cerebral protective agents while delivering a high dose to tumor cells. Oldﬁeld et al. [23] noted protection from radiation-induced brain injury using pentobarbital. The 21-aminosteroids (21-AS) have been evaluated as potential radioprotective agents. The 21-AS, commonly known as lazaroids, have been advocated as cerebral protective agents in patients with head trauma or subarachnoid hemorrhage [24]. The 21-AS act as antioxidants [25], and much of the damage from radiation is due to the production of oxygen free radicals, which can induce DNA modiﬁcations and strand

Radiation Potentiation We studied the synergistic effect of tumor necrosis factor-α (TNF-α) on enhancing the tumor response to radiosurgery. TNF-α can act as a tumoricidal agent with direct cytotoxicity mediated through binding to its cognate cell-surface receptors and a variety of activities triggering a multifaceted immune attack on tumors [29–34]. In addition, locally produced TNF-α has been reported to enhance the sensitivity of tumors to radiation in nude mice [29]. We employed a replication-defective herpes simplex virus (HSV) as a vector to deliver thymidine

64

a. niranjan and d. kondziolka

TABLE 6-3. Effects of radiosurgery on malignant brain tumors. First author and year

Tumor margin dose (Gy)

Tumor model

Animal model

Irradiation technique

Experimental treatment

Results

Kondziolka, 1992 [38]

15–50

C6 glioma

Rat

Gamma Knife, 4-mm collimator

Radiosurgery

Kondziolka, 1999 [39]

35 Gy

C6 glioma

Rat

Gamma Knife, 4-mm collimator

Radiosurgery + 21-aminosteroid

Niranjan, 2000 [35]

15

U 87 MG

Nude mouse

Gamma Knife, 4-mm collimator

Radiosurgery + HSV TK-GCV+TNF

Nakahara, 2001 [40]

16

MADB 106 cells

Rat

Gamma Knife, 4-mm collimator

Radiosurgery + cytokine transduced tumor cell vaccine

Niranjan, 2003 [41]

15

9L gliosarcoma

Rat

Gamma Knife, 4-mm collimator

Radiosurgery + HSV TK-GCV+TNF+ Connexin

Treated animals survived 39 days (control 29 days). Treated tumors had hypocellular appearance with cellular edema. 21-Aminosteroid exhibited a radioprotectant effect on normal brain tissue, but did not protect the tumor The combination treatment enhanced median survival (75 days) with 89% animal surviving The combination treatment signiﬁcantly prolonged animal survival and protected animals from a subsequent challenge by parental tumor cells placed in the CNS The combination of radiosurgery and multigene therapy enhanced median animal survival (150 days) with 75% of animals surviving

kinase (TK) and/or TNF-α genes to U-87 MG tumors in nude mice. Radiosurgery was performed 48 hours after gene transfer using 15 Gy to the tumor margin (21.4 Gy to the center). Daily ganciclovir (GCV) therapy was started after gene transfer and continued for 10 days. The combination of radiosurgery with TNF-α or with HSV-TK-GCV (suicide gene therapy) and TNFα signiﬁcantly improved median survival of animals [35]. In additional experiments, the connexin-43 gene was added to enhance the formation of gap junctions between tumor cells, which should facilitate the intercellular dissemination of TKactivated GCV from virus-infected cells to noninfected surrounding cells. This creates a bystander effect that can improve tumor cell killing [36]. Addition of connexin-43 gene to this paradigm (TK-GCV + TNF-α + radiosurgery) further improved survival (90% survival in tumor-bearing mice). We also studied this strategy in a 9L rat glioma model and found that addition of radiosurgery to suicide gene therapy (SGT) signiﬁcantly improved animal survival compared with SGT alone. The combination of HSV-based SGT (TK-GCV), TNF-α gene transfer, and radiosurgery was more effective than SGT or radiosurgery alone. The combination of SGT with radiosurgery was also more effective than SGT or radiosurgery alone. Although the exact mechanism of this effect is unclear and remains the subject of future investigations, these experiments indicate that gene therapy could be an effective strategy for enhancing the radiobiological impact of radiosurgery. In other studies, tumor sensitization to radiation was apparently mediated by extracellular TNF-α promoting the destruction of tumor vessels, whereas HSV-vector–mediated TNF-α–enhanced killing of malignant glioma cell cultures is presumably a consequence of an intracellular TNF-α activity (Table 6-3) [34, 37].

Experimental Models for Functional Brain Radiosurgery Radiosurgery is rapidly expanding beyond its use as a treatment of brain tumors and arteriovenous malformations (AVMs). It has been found effective for other neurologic disorders, such as epilepsy, movement disorders, and trigeminal neuralgia. The promise of “functional” radiosurgery has led to a need to investigate its efﬁcacy, limitations, and potential drawbacks.

Hippocampal Radiosurgery for Epilepsy The potential efﬁcacy of radiosurgery for the treatment of epilepsy has been evaluated using rat models. Kainic acid reproducibly induces epilepsy in rats when injected into the hippocampus. Mori et al. [42] treated kainic acid–induced epilepsy in rats with doses of 20 to 100 Gy radiosurgery using Gamma Knife. The efﬁcacy of the treatment on epilepsy was evaluated by direct observation and scalp EEG for 42 days. Even 20 Gy signiﬁcantly reduced the number of seizures, and the efﬁcacy improved with increasing dose. Only doses >60 Gy induced histologic changes. Maesawa et al. [43] treated epileptic rats with a single dose of 30 or 60 Gy. Both doses signiﬁcantly reduced EEG-deﬁned seizures. The latency to this effect was less after the higher dose (5 to 9 weeks for 60 Gy vs. 7 to 9 Gy for 30 Gy). Whereas kainic acid injection alone reduced performance of rats on the water maze task, the performance of rats that were treated by radiosurgery after kainic acid administration was not different from controls. Liscak et al. [44] evaluated the effects of radiosurgery on normal hippocampus in an effort to identify potential normal

6.

65

experimental radiosurgery models

tissue complications and determine dose limits for hippocampal radiosurgery. This study employed four separate 4-mm isocenters to irradiate the entire hippocampus with 25 to 100 Gy. Doses <50 Gy did not cause any perceptible changes based on histology, MRI, and Morris water maze testing. In contrast, the performance on the Morris water maze was signiﬁcantly worse for animals treated with >50 Gy. These investigations support the concept that radiosurgery may be an effective method for treating epilepsy, but they also suggest that doses to the hippocampus should be limited to reduce potential effects on learning and memory.

Thalamic Radiosurgery for Movement Disorders The effect of radiosurgery on potential targets for the treatment of movement disorders has been evaluated. De Salles and colleagues [45] used a linear accelerator and 3-mm collimator to deliver a maximal dose of 150 Gy to the subthalamic nucleus of one vervet monkey and to the substantia nigra of another. Follow-up MRI detected a 3-mm lesion that did not increase in size throughout the course of the study. Kondziolka et al. [46] examined the effects of thalamic radiosurgery in a baboon model and reported that a dose of 100 Gy (central dose using 4-mm collimator) was sufﬁcient to induce contrast enhance-

ment of magnetic resonance images and coagulative necrosis as evaluated by histology.

Trigeminal Nerve Radiosurgery for Trigeminal Neuralgia Radiosurgery has signiﬁcant potential as an effective, noninvasive method for treatment of trigeminal neuralgia, and the effect of Gamma Knife irradiation on the trigeminal nerve has been evaluated in the baboon [47]. We irradiated the normal proximal trigeminal nerve with 80 or 100 Gy using a 4-mm collimator. A 4-mm region of contrast enhancement was visible by MRI at 6 months after treatment. Both large and small ﬁbers were affected with axonal degeneration occurring after 80 Gy and necrosis after 100 Gy. Neither dose was effective at selectively damaging ﬁbers responsible for transmission of pain while maintaining those responsible for other sensations, which would be optimal for effective treatment of trigeminal neuralgia. Nevertheless, this study does demonstrate that it is possible to noninvasively and precisely affect speciﬁc nerves using the Gamma Knife. Whether other dose regimens might cause selective damage to pain ﬁbers will require further investigation (Table 6-4).

TABLE 6-4. Central nervous system response to functional radiosurgery. First author and year

Ishikawa, 1999 [48]

Maximum dose (Gy)

200

Region(s) irradiated

Irradiation technique

Animal model

Results

Medial temporal lobe

Gamma Knife, 4-mm collimator

Rat

Gamma Knife, 4-mm collimator Gamma Knife, 4-mm collimator

Rat

Gamma Knife, 4-mm collimator Gamma Knife, 4-mm collimator

Baboon

Linac, 3-mm collimator

Monkey

Baboon

Caudate-putamen complex

Gamma Knife, 4-mm collimator Gamma Knife, 4-mm collimator Gamma Knife, 4-mm collimator

Hippocampus

Proton beam

Rat

Medial hypothalamus

Gamma Knife, 4-mm collimator

Rat

Sequential MRI and histopathology showed consistent necrosis at 2 weeks after 200-Gy radiosurgery Seizure frequency decreased after ≥20-Gy radiosurgery Seizure frequency decreased after 30–60 Gy radiosurgery. Shorter latency after higher dose. Learning and memory unaffected MRI, histology at 6 months showed axonal degeneration at all doses Subnecrotic (20–40 Gy) radiosurgery substantially reduces seizure frequency and duration MRI and histology showed that necrotic lesion remained at <3-mm size MRI, histology showed necrosis at 6 months More than 50 altered memory performance 6-OHDA–induced hemiparkinsonian behavior was signiﬁcantly reduced. Necrotic lesions were surrounded by regions that were highly positive for GDNF Doses 90 CGE or higher resulted in adverse behavioral effects and necrosis in 3 months 30 or 60 CGE radiosurgery led to marked increase in HSP-72 staining but no necrosis Signiﬁcant and sustained reductions in weight set-point after a latency of 7 weeks was noted

Mori, 2000 [42]

20–100

Hippocampus

Maesawa, 2000 [43]

30–60

Hippocampus

Kondziolka, 2000 [47] Chen, 2001 [49]

80–100

Trigeminal nerve

20–40

Hippocampus

De Salles, 2001 [45]

150

Kondziolka, 2002 [46] Liscak, 2002 [44]

100

Zerris, 2002 [50]

140

Brisman, 2003 [51]

Vincent, 2005 [52]

25–150

5–130 CGE

40 Gy

Subthalamic nucleus, substantia nigra Thalamus Hippocampus

Rat

Rat

Rat Rat

66

a. niranjan and d. kondziolka

Hypothalamic Radiosurgery for Obesity Vincent et al. studied the effect of subnecrotic hypothalamic radiosurgery on body weight set-point [52]. These investigators performed hypothalamic radiosurgery on genetically obese Zucker rats using a total dose of 40 Gy delivered to two nearby targets in the medial hypothalamus. These investigators noted signiﬁcant and sustained reductions in weight set-point for animals that received radiosurgery compared with sham-treated animals after a latency of 7 weeks. No gross behavioral abnormalities were noted. Histopathologic analysis showed no abnormalities except a small area of necrosis in one animal. These investigations indicate the potential role of a single dose of irradiation to the hypothalamus in producing sustained reduction in the weight set-point. At the University of Pittsburgh, the authors are currently investigating the effects of hypothalamic radiosurgery on obesity using a primate model.

Future Models of Experimental Radiosurgery Rapid developments in neuroimaging, stereotactic techniques, and robotic technology in the past decade have contributed to improved results and wider applications of radiosurgery. The role of radiosurgery has expanded well beyond its initial application for functional neurosurgery, pain management, AVMs, and selected skull base tumors. The clinical spectrum now includes a wide variety of rare skull base neoplasms, serves as the primary treatment of metastatic brain cancer, and provides adjuvant management of malignant primary brain tumors. Although radiosurgery provides survival beneﬁts in diffuse malignant brain tumors, cure is still not possible. Microscopic tumor inﬁltration into surrounding normal tissue is the main cause of recurrence. Additional strategies are needed to speciﬁcally target tumor cells. In the future, gene transfer to sensitize malignant tumor cells to radiosurgery may provide enhanced tumor cell kill while radioprotective agents will prevent damage to surrounding normal tissue. Although the nature of brain injury after radiosurgery appears similar to that seen after conventional radiation treatments, there remain a number of questions concerning the effects and the pathogenesis of such effects after both forms of radiation. At present, the cellular target that is primarily responsible for radiation-induced breakdown of normal tissue is unclear. The white matter and the cerebral vasculature appear to be particularly susceptible to radiation, which suggests that oligodendrocytes and endothelial cells may be critical targets of radiation. Recent studies have also implicated a potential role for neural progenitors in radiation-induced brain injury [53, 54]. The role of neural and endothelial precursors in repairing radiation-induced brain damage is under evaluation. Neural stem cells can be isolated from normal adult mammalian brain and can be induced to differentiate into neurons or glia. In the future, if implanted neural stem cells could prevent or repair radiation-induced damage to normal brain, then the tumors could be targeted with higher radiosurgery doses. These higher doses may prove lethal to tumor cells while stem cells will prevent damage to surrounding normal tissue. While radiosurgery usage continues to expand as we sort out the roles of precision radiation, we must strive to understand the mechanism of biological response of CNS tissues to

radiation as well as the potential of long-term adverse effects including the risk of delayed oncogenesis. Radiosurgery can affect the cerebral microenvironment. The role of radiosurgery in altering the local immune response by activating microglia and stimulating cytokines needs to be studied in order to develop strategies to treat brain tumors. Further research to answer these questions is needed to maximize the effectiveness of radiosurgery on target regions and to minimize injury to other areas.

References 1. Larsson B, Leksell L, Rexed B, et al. The high-energy proton beam as neurosurgical tool. Nature 1958; 182(6):1222–1223. 2. Rexed B, Mair W, Sourander P, et al. Effect of high-energy protons on the rabbit. Acta Radiol Oncol Radiat Phys Biol 1960; 53:289–299. 3. Leksell L, Larsson B, Anderson B, et al. Lesions in the depth of the brain produced by a beam of high-energy protons. Acta Radiol Ther Phys Biol 1960; 53:251–264. 4. Andersson B, Larsson B, Leksell L, et al. Histopathology of late local radiolesions in the goat brain. Acta Radiol Ther Phys Biol 1970; 9(5):385–394. 5. Nilsson A, Wennerstrand J, Leksell D, et al. Stereotactic gamma irradiation of basilar artery in cat. Preliminary experiences. Acta Radiol Oncol Radiat Phys Biol 1978; 17(2):150–160. 6. Calvo W, Hopewell JW, Reinhold HS, et al. Time- and doserelated changes in the white matter of the rat brain after single doses of X rays. Br J Radiol 1988; 61:1043–1052. 7. Fike J, Gobbel G. Central nervous system radiation injury in large animal models. In: Gutin P, Leibel S, Sheline G, eds. Radiation Injury to the Nervous System. New York: Raven Press; 1991:113– 135. 8. van der Kogel A. Central nervous sytem radiation injury in small animal models. In: Gutin P, Leibel S, Sheline G, eds. Radiation Injury to the Nervous System. New York: Raven Press; 1991:113– 135. 9. Calvo W, Hopewell J, Reinhold H, et al. Dose-dependent and time-dependent changes in the choroid plexus of the irradiated rat brain. Br J Radiol 1987; 60:1109–1117. 10. Hopewell J, Calvo W, Campling D, et al. The role of the vasculature in normal tissue responses. In: Fielden E, Fowler J, Hendry J, Scott D, eds. Radiation Research, Proceedings of the 8th International Congress of Radiation Research. London: Taylor & Francis; 1987:789–794. 11. Zeman W, Samorajski T. Effects of irradiation on the nervous system. In: Berdjis CC, ed. Pathology of Irradiation. Baltimore: Williams & Wilkins, 1971:213–277. 12. Hopewell JW, Wright EA. The nature of latent cerebral irradiation damage and it modiﬁcation by hypertension. Br J Radiol 1970; 43:161–167. 13. Reinhold H, Hopewell J. Late changes in the architecture of blood vessels of the rat brain after irradiation. Br J Radiol 1980; 53: 693–696. 14. Schultheiss T, Stephens L. Permanent radiation myelopathy. Br J Radiol 1992; 65:737–753. 15. Lunsford LD, Altschuler EM, Flickinger JC, et al. In vivo biological effects of stereotactic radiosurgery: a primate model. Neurosurgery 1990; 27(3):373–382. 16. Kondziolka D, Lunsford LD, Altschuler EM, et al. Biological effects of stereotactic radiosurgery in the normal primate brainstem. In: Lunsford LD, ed. Stereotactic Radiosurgery Update. New York: Elsevier; 1992:291–294. 17. Kondziolka D, Lunsford LD, Claassen D, et al. Radiobiology of radiosurgery: Part I. The normal rat brain model. Neurosurgery 1992; 31(2):271–279.

6.

experimental radiosurgery models

18. Blatt DR, Friedman WA, Bova FJ, et al. Temporal characteristics of radiosurgical lesions in an animal model. J Neurosurg 1994; 80(6):1046–1055. 19. Kamiryo T, Kassell NF, Thai QA, et al. Histological changes in the normal rat brain after gamma irradiation. Acta Neurochir (Wien) 1996; 138(4):451–459. 20. Hopewell JW, Wright EA. The effects of dose and ﬁeld size on late radiation damage to the rat spinal cord. Int J Radiat Biol Relat Stud Phys Chem Med 1975; 28(4):325–333. 21. Karger CP, Munter MW, Heiland S, et al. Dose-response curves and tolerance doses for late functional changes in the normal rat brain after stereotactic radiosurgery evaluated by magnetic resonance imaging: inﬂuence of end points and follow-up time. Radiat Res 2002; 157(6):617–625. 22. Kamiryo T, Lopes MB, Kassell NF, et al. Radiosurgery-induced microvascular alterations precede necrosis of the brain neuropil. Neurosurgery 2001; 49(2):409–414; discussion 14–15. 23. Oldﬁeld EH, Friedman R, Kinsella T, et al. Reduction in radiation-induced brain injury by use of pentobarbital or lidocaine protection. J Neurosurg 1990; 72(5):737–744. 24. Smith SL, Scherch HM, Hall ED. Protective effects of tirilazad mesylate and metabolite U-89678 against blood-brain barrier damage after subarachnoid hemorrhage and lipid peroxidative neuronal injury. J Neurosurg 1996; 84(2):229–233. 25. Braughler JM. Lipid peroxidation-induced inhibition of gammaaminobutyric acid uptake in rat brain synaptosomes: protection by glucocorticoids. J Neurochem 1985; 44(4):1282–1288. 26. Bernstein M, Ginsberg H, Glen J. Protection of iodine-125 brachytherapy brain injury in the rat with the 21-aminosteroid U-74389F. Neurosurgery 1992; 31(5):923–927; discussion 7–8. 27. Buatti JM, Friedman WA, Theele DP, et al. The lazaroid U74389G protects normal brain from stereotactic radiosurgery-induced radiation injury. Int J Radiat Oncol Biol Phys 1996; 34(3):591– 597. 28. Kondziolka D, Somaza S, Martinez AJ, et al. Radioprotective effects of the 21-aminosteroid U-74389G for stereotactic radiosurgery. Neurosurgery 1997; 41(1):203–208. 29. Staba MJ, Mauceri HJ, Kufe DW, et al. Adenoviral TNFalpha gene therapy and radiation damage tumor vasculature in a human malignant glioma xenograft. Gene Therapy 1998; 5(3):293– 300. 30. Cao G, Kuriyama S, Du P, et al. Complete regression of established murine hepatocellular carcinoma by in vivo tumor necrosis factor alpha gene transfer.[comment]. Gastroenterology 1997; 112(2):501–510. 31. Han SK, Brody SL, Crystal RG. Suppression of in vivo tumorigenicity of human lung cancer cells by retrovirus-mediated transfer of the human tumor necrosis factor-alpha cDNA. Am J Respir Cell Mol Biol 1994; 11(3):270–278. 32. Ostensen ME, Thiele DL, Lipsky PE. Enhancement of human natural killer cell function by the combined effects of tumor necrosis factor alpha or interleukin-1 and interferon-alpha or interleukin-2. J Biol Response Modiﬁers 1989; 8(1):53–61. 33. Owen-Schaub LB, Gutterman JU, Grimm EA. Synergy of tumor necrosis factor and interleukin 2 in the activation of human cytotoxic lymphocytes: effect of tumor necrosis factor alpha and interleukin 2 in the generation of human lymphokine-activated killer cell cytotoxicity. Cancer Res 1988; 48(4):788–792. 34. Gridley DS, Archambeau JO, Andres MA, et al. Tumor necrosis factor-alpha enhances antitumor effects of radiation against glioma xenografts. Oncol Res 1997; 9(5):217–227. 35. Niranjan A, Moriuchi S, Lunsford LD, et al. Effective treatment of experimental glioblastoma by HSV vector-mediated TNF alpha and HSV-tk gene transfer in combination with radiosurgery and ganciclovir administration. Mol Ther 2000; 2(2):114– 120.

67

36. Marconi P, Tamura M, Moriuchi S, et al. Connexin 43-enhanced suicide gene therapy using herpesviral vectors. Mol Ther 2000; 1(1):71–81. 37. Moriuchi S, Oligino T, Krisky D, et al. Enhanced tumor cell killing in the presence of ganciclovir by herpes simplex virus type 1 vectordirected coexpression of human tumor necrosis factor-alpha and herpes simplex virus thymidine kinase. Cancer Res 1998; 58(24): 5731–5737. 38. Kondziolka D, Lunsford LD, Claassen D, et al. Radiobiology of radiosurgery: Part II. The rat C6 glioma model. Neurosurgery 1992; 31(2):280–287; discussion 7–8. 39. Kondziolka D, Mori Y, Martinez AJ, et al. Beneﬁcial effects of the radioprotectant 21-aminosteroid U-74389G in a radiosurgery rat malignant glioma model. Int J Radiat Oncol Bio Phys 1999; 44(1):179–184. 40. Nakahara N, Okada H, Witham TF, et al. Combination of stereotactic radiosurgery and cytokine gene-transduced tumor cell vaccination: a new strategy against metastatic brain tumors. J Neurosurg 2001; 95(6):984–989. 41. Niranjan A, Wolfe D, Tamura M, et al. Treatment of rat gliosarcoma brain tumors by HSV-based multigene therapy combined with radiosurgery. Mol Ther 2003; 8(8):530–542. 42. Mori Y, Kondziolka D, Balzer J, et al. Effects of stereotactic radiosurgery on an animal model of hippocampal epilepsy. Neurosurgery 2000; 46(1):157–165; discussion 65–68. 43. Maesawa S, Kondziolka D, Dixon CE, et al. Subnecrotic stereotactic radiosurgery controlling epilepsy produced by kainic acid injection in rats. J Neurosurg 2000; 93(6):1033–1040. 44. Liscak R, Vladyka V, Novotny J Jr, et al. Leksell gamma knife lesioning of the rat hippocampus: the relationship between radiation dose and functional and structural damage. J Neurosurg 2002; 97(5 Suppl):666–673. 45. De Salles AA, Melega WP, Lacan G, et al. Radiosurgery performed with the aid of a 3-mm collimator in the subthalamic nucleus and substantia nigra of the vervet monkey. J Neurosurg 2001; 95(6):990–997. 46. Kondziolka D, Conce M, Niranjan A, et al. Histology of the 100 Gy thalomotomy in the baboon. Radiosurgery 2002; 4(4):279–284. 47. Kondziolka D, Lacomis D, Niranjan A, et al. Histological effects of trigeminal nerve radiosurgery in a primate model: implications for trigeminal neuralgia radiosurgery. Neurosurgery 2000; 46(4):971–976; discussion 6–7. 48. Ishikawa S, Otsuki T, Kaneki M, et al. Dose-related effects of single focal irradiation in the medial temporal lobe structures in rats—magnetic resonance imaging and histological study. Neurol Med Chir (Tokyo) 1999; 39(1):1–7. 49. Chen ZF, Kamiryo T, Henson SL, et al. Anticonvulsant effects of gamma surgery in a model of chronic spontaneous limbic epilepsy in rats. J Neurosurg 2001; 94(2):270–280. 50. Zerris VA, Zheng Z, Noren G, et al. Radiation and regeneration: behavioral improvement and GDNF expression after Gamma Knife radiosurgery in the 6-OHDA rodent model of hemi-parkinsonism. Acta Neurochir Suppl 2002; 84:99–105. 51. Brisman JL, Cole AJ, Cosgrove GR, et al. Radiosurgery of the rat hippocampus: magnetic resonance imaging, neurophysiological, histological, and behavioral studies. Neurosurgery 2003; 53(4):951– 961; discussion 61–62. 52. Vincent DA, Alden TD, Kamiryo T, et al. The baromodulatory effect of gamma knife irradiation of the hypothalamus in the obese Zucker rat. Stereotact Funct Neurosurg 2005; 83(1):6–11. 53. Tada E, Yang C, Gobbel GT, et al. Long-term impairment of subependymal repopulation following damage by ionizing irradiation. Exp Neurol 1999; 160(1):66–77. 54. Tada E, Parent JM, Lowenstein DH, et al. X-irradiation causes a prolonged reduction in cell proliferation in the dentate gyrus of adult rats. Neuroscience 2000; 99(1):33–41.

7

Treatment Planning for Stereotactic Radiosurgery David M. Shepard, Cedric Yu, Martin Murphy, Marc R. Bussière, and Frank J. Bova

Introduction This chapter provides an introduction to treatment-planning procedures for stereotactic radiosurgery (SRS). The chapter begins with a brief history of SRS planning. Next, the basic steps followed in the development of an SRS treatment plan are described including imaging, contouring, selection of plan parameters, and evaluating treatment plan quality. The remainder of the chapter focuses on treatment planning for speciﬁc SRS delivery techniques including the Gamma Knife, linear accelerator–based SRS, CyberKnife, and proton radiosurgery.

Background on SRS Planning During the past three decades, radiosurgery has been transformed from a curiosity available at a single institution to a routinely used therapy available at medical facilities throughout the world. During this increase in utilization, radiosurgical planning and delivery techniques have undergone signiﬁcant technical improvements including (1) availability of true threedimensional image data, (2) improved plan delivery techniques, and (3) increased sophistication of planning algorithms. The most critical improvements have come about through the availability of true three-dimensional image data provided primarily through computed tomography (CT) and magnetic resonance imaging (MRI) scanning techniques. Today’s practicing radiosurgeon has the advantage of viewing target and nontarget tissues with increased contrast resolution and with dramatically improved spatial resolution. Clinicians have also beneﬁted from the development of tools for registering or fusing single and multimodality data sets. The second area that has undergone signiﬁcantly improvement is that of plan delivery. For several years, the only radiosurgical device available was the Gamma Knife. The Gamma Knife has now been joined by several designs that use linear accelerators as the radiation source. Whereas the initial linear accelerator (linac)-based systems were less accurate than the Gamma Knife, devices were developed in the late 1980s that provided signiﬁcant improvement in the delivery accuracy of linac-based radiosurgical units. Recent years have seen additional advancements in basic linac technology

resulting in highly accurate base linear accelerators. These devices provide efﬁcient radiosurgical delivery and have made it possible to extend the beneﬁts of radiosurgery to targets outside the cranium. The third area of technological advancement in radiosurgery has been the increased sophistication of the planning algorithms that has been made possible by the tremendous increase in computer processing power over the past three decades. The ability to process images, register images, and develop plans has gone from a process that required many hours to one that can be accomplished in a matter of seconds. This has allowed radiosurgical teams to not only develop and evaluate their ﬁrst best guess at a plan but also to iterate through many different plans in order to arrive at a more optimal plan. Although these planning and delivery tools help provide improved treatment plans, the basics principles set forth by Lars Leksell in the 1950s remain the foundation for radiosurgery [1–4]. Leksell’s basic idea was to place the target tissues in the center of a large number of beams of radiation. The arrangement of beams was designed so that the beams only intersected over the target tissues and they quickly diverged from the target in all directions. The majority of clinical data published on radiosurgery is based on a planning technique known as sphere packing. This technique was initially developed by Leksell and constitutes the basis of the Gamma Knife planning process. It has also been used in conjunction with many of the linac-based delivery systems. In this technique, a set of beams are aimed at a point in space, know as the isocenter. The beams are selected so that they approach and leave the isocenter through unique paths, providing both geometric concentration and a high-dose gradient. Much of the accuracy discussed in the radiosurgery literature regards the measure of delivery systems’ ability to precisely target radiation beams at this “point” in space. The resultant dose distribution is approximately spherical and therefore provides a reasonable plan for spherical target volumes. For more complex target shapes, a technique known as sphere packing is employed. In this technique, the planner places the initial dose sphere inside the target volume. The initial dose sphere is typically selected to be the largest sphere that the system can produce and that can ﬁt “inside” the target volume. The planner then continues to pack the target volume with spheres of equal

69

70

d.m. shepard et al.

from basic radiosurgical principles and begins to depend more upon radiologic principles. While this may be appropriate in certain clinical situations, a cautious approach is warranted.

Basic Steps in Developing an SRS Treatment Plan The basic steps followed in radiosurgery treatment planning are illustrated in Figure 7-1.

Stereotactic Localization Because SRS involves the delivery of a very high dose of radiation in either a single delivery or a small number of fractions, it is essential to achieve a high degree of dose conformity and accurate patient positioning. This critical dosimetric requirement is accomplished by stereotactically focusing many convergent radiation beams on the target. By cross-ﬁring a large number of beams, a high dose is delivered in the region of intersection with a rapid fall-off in dose outside of the target. One approach to accurate stereotactic localization is achieved with ﬁxation frames such as the frame shown in Figure 7-2. On the day of the procedure, this ﬁxation frame is ﬁxed to the patient’s skull under local anesthetic. The frame serves two important purposes. First, it provides a rigid ﬁxation system that ensures that the patient cannot move during the delivery. Second, it provides a frame of reference whereby the tumor location can be determined relative to the frame and relative to the delivery unit. FIGURE 7-1. The basic steps in SRS treatment planning.

or smaller diameter until adequate target coverage is achieved. An advantage of a linear accelerator–based system over the Gamma Knife is the ability to produce larger spheres and therefore pack a speciﬁc volume with fewer total spheres or isocenters. These spheres of dose are created using beam sets that deliver radiation from a large number of beam angles. The Gamma Knife makes use of 201 individual beams whereas multiple arcs are used in the linear accelerator systems. It has been shown that in order to produce the concentration and gradient, which is responsible for a vast majority of all published radiosurgical clinical outcomes, at least 15 beams must be spread over approximately 2π space [5]. In recent years, SRS delivery techniques have expanded beyond the Gamma Knife and linac-based delivery using circular collimators to include the CyberKnife, tomotherapy, and the adaptation of linacs with micro-multileaf collimators. Although each of these systems offers the radiosurgeon special features, they are all bound by the same principle that a large number of beam paths are required to concentrate the dose in the target and create steep dose gradients. It is important to note that as new SRS delivery techniques emerge, clinicians must examine the impact on dose conformity compared with historical SRS delivery techniques using the Gamma Knife or linac-based delivery with circular collimators. A loss in dose conformity may force the user to adopt a treatment schedule that deviates

FIGURE 7-2. Gamma Knife coordinate frame. (Image Courtesy of Elekta, Inc.)

7.

treatment planning for stereotactic radiosurgery

71

FIGURE 7-3. (a) The ﬁducial indicator box for CT imaging. (b) The patient is imaged with a ﬁducial box attached to the frame. (Image Courtesy of Elekta, Inc.)

One approach for facilitating the registration of the patient images is to attach a ﬁducial box (Fig. 7-3) to the stereotactic head frame before patient imaging. The box includes tubes ﬁlled with a high-contrast solution that provides easy to delineate ﬁducial marks on all of the patient’s images (Fig. 7-4). After the imaging is complete, the patient images are loaded into the planning system. The images are registered with respect to the ﬁducial marks. The markings on the ﬁducial box provide a coordinate system that makes it possible to precisely determine the position of the treatment volume with respect to the stereotactic head frame. In addition to the ﬁxation frames described above, frameless SRS techniques have been developed that maintain a highly precise delivery [6–12]. For example, researchers at the

FIGURE 7-5. Repeat ﬁxations system that allows the biteplate to be reinserted a number of times and judges the reproducibility against a ﬁxed head band. (Image courtesy of Varian Medical Systems. Copyright 2006, Varian Medical Systems. All rights reserved.)

Patient’s Positional Indicator (left/anterior in this axial image) Fiducials

Third Plate

Image

University of Florida have developed and commercialized a biteplate system (Fig. 7-5). Frameless SRS is typically the preferred option in fractionated radiosurgery where the invasive nature of ﬁxation frames makes them a less-viable option. Fractionation radiosurgery, known as stereotactic radiotherapy (SRT), is commonly delivered to patients where the likelihood of late toxicity makes it inadvisable to treat a large target volume in a single high-dose fraction [7, 12–14].

Patient Imaging

Fiducials Right Left Fiducial Fiducial Markers Markers FIGURE 7-4. The ﬁducial marks appear as white dots on this MRI image of a Gamma Knife patient. (Image Courtesy of Elekta, Inc.)

The three primary imaging techniques for SRS are magnetic resonance imaging (MRI), computed tomography (CT), and planar angiography. MRI is an excellent technique for imaging soft tissue contrast. MRI, however, does not provide attenuation coefﬁcients and must be fused with CT images if one wishes to make appropriate heterogeneities in the dose calculation. It should be noted that in most MRI units, the spatial uniformity of the images degrade as the radius of the images increase. Because most ﬁducial systems place the ﬁducial markers at the outermost extent of the MRI image volume, the accuracy obtained in mapping the imaged voxels to stereotactic space can be compromised. This is one of several reasons

72

d.m. shepard et al.

FIGURE 7-6. One approach to image registration.

some radiosurgical practitioners have opted for image fusion technologies as opposed to direct stereotactic MRI scanning (Fig. 7-6). CT imaging provides inferior soft tissue contrast compared with MRI scanning; however, the spatial resolution of CT is extremely reliable. CT also provides attenuation coefﬁcients that make it possible to make accurate heterogeneity corrections in the dose calculation. Planar angiography is an imaging option that can be used for imaging patients with arteriovenous malformations (AVMs). Although plane ﬁlm angiography was the gold standard for AVM nidus identiﬁcation, cut ﬁlms have almost been fully replaced by electronically derived images. For the past few decades, the primary mode of electronic image acquisition has been the image intensiﬁer. Although this device is know for its excellent image contrast, it is equally known for its poor spatial accuracy. Most image intensiﬁers have multiple levels of image distortion. More recently, solid-state image acquisition systems have been developed. These systems have all but eliminated concerns regarding spatially nonuniform images. An additional problem with angiography is that although plane ﬁlm angiography can provide images in multiple planes, the basic image set that the clinician has to work with is twodimensional. This two-dimensional image data set can lead to inaccuracies in the target deﬁnition. It is for this reason that most targeting for vascular radiosurgery targets is obtained either through CT angiography, MRI angiography, or both.

Registering the Images The general planning procedure starts with the acquisition of three-dimensional image information and the import of patient images into the treatment planning system via a computer network or data storage media. On the planning computer, these images are typically displayed in two-dimensional slices along the axial, sagittal, or coronal planes [15]. A screen capture from the GammaPlan system shown in Figure 7-4 illustrates the process of aligning the ﬁducial marks in the SRS planning system. The planning system must be able to accurately identify the ﬁducial marks and have internal mechanisms for validating their consistency. Once the ﬁducial markers are identiﬁed, the patient anatomy shown on the images is placed in the coordinate system deﬁned by the stereotactic localization device. Some systems use ﬁducial markers that fully deﬁne the stereotactic reference system whereas others require speciﬁc information from the scanner. For example, a fully deﬁned system

does not require information on the axial slice position. This is because it can derive this information from the actual scan image. Consequently, this approach eliminates the need for QA measurements to guarantee accurate scanner information such as linearity of table movement or degree of gantry tilt. Frameless systems also need to be able to map the threedimensional data set into a rigid and deﬁnable coordinate system. To achieve this, stereotactic systems have been designed that incorporate surrogate ﬁducial markers into the CT or MRI scans. These markers can also be referenced at the time of therapy through mechanical, electrical, or optical means. A critical feature of frameless systems is the ability to demonstrate that the reference can in fact be reliably and precisely ﬁxed to the patient. Many systems offer testing only at the time of design and development. These tests are usually at the hands of the developer, an expert with extensive experience. It is critical for a repeat ﬁxation system to also provide a methodology of testing its accuracy for each patient. Figure 7-5 shows the RadioCameras Treatment Guidance System (Varian, Palo Alto, CA). This system provides patient-speciﬁc statistics on the device’s ability to be precisely reapplied. Two basic approaches are used in the registration of images. The ﬁrst is to use direct stereotactic MRI and CT scanning. In this procedure, ﬁducial markers are incorporated into both image data sets and each data set is independently mapped into stereotactic space. In the second approach, stereotactic CT scanning is performed using a ﬁducial system when acquiring the images. Nonstereotactic MRI scans are also obtained (see Fig. 7-6). The MRI scans are then mapped into stereotactic space by registration of the MRI data set to the CT data set. In most cases, rigid registration techniques are used. Typically, the clinician visually inspects the results of the registration to verify the quality of the ﬁt. An advantage of this second approach is that it allows the MRI scan to be obtained prior to the placement of any stereotactic ring. Consequently, the clinician can exam the actual scan that will be used for scanning prior to deciding that radiosurgery is the appropriate mode for treatment.

Contouring Structures After the images have been imported into the planning system and the ﬁducial marks have been registered, the target to be treated is then identiﬁed. To aid target delineation, fusion of images from different imaging modalities is possible with many SRS planning systems. Whereas some planning systems may not require the users to outline the target and critical structures, contouring is required if plan evaluation tools such dose-volume histograms (DVHs) are to be used. Contouring is also required by systems that employ inverse planning. The patient’s external contour can either be manually contoured or identiﬁed automatically by the planning system based on grayscale information from the images. For Gamma Knife planning, the skin contour or scalp of the patient’s head is measured with a skull-scaling device (Fig. 7-7). The location of the surface is important for computing the penetration depth for each beam during the dose calculation. The skull-scaling device used for Gamma Knife planning makes it possible to know the penetration depth for each photon beam even if the entire skull has not been imaged.

7.

treatment planning for stereotactic radiosurgery

73

80% and 40% isodose surfaces. Therefore, prescribing to the 80% isodose line minimizes the integral dose to all normal tissues (the tissues outside the target volume). Compared with the Gamma Knife, single-isocenter plans are used more frequently for radiosurgery treatments on linear accelerators. Gamma Knife units are limited to four focusing helmets (shot sizes), which is signiﬁcantly fewer than the typical number of cones that are available on a linear accelerator–based system. Additionally, multileaf collimator–based deliveries generally use only one isocenter. For multiple-isocenter plans, this optimal surface shifts to the 70% isodose surface and at times can be extended to isodose surfaces as low as the 50% dose surface, as is often the case in Gamma Knife plans.

Designing the Treatment Plan

FIGURE 7-7. Skull scaling device used with the Gamma Knife to provide mapping of the skull. (Image Courtesy of Elekta, Inc.)

Deﬁning the Prescription The dosimetric characteristics of the treatment plan vary from one delivery approach to another. With the Gamma Knife and linac-based SRS, the use of multiple isocenters means that multiple spherical isodose distributions have to be packed into the target volume. Overlaps of such high-dose spheres are inevitable. As a result, the dose is highly nonuniform in the target. The isodose surface that encloses the target (which is often taken as the prescription dose) is typically 50% of the maximum dose in the target. For linac-based SRS, the target dose is highly uniform when a single isocenter is used. The isodose line that encompasses the target can be greater than 80% of the maximum dose in the target. In SRS, the prescription dose is normally set to the dose level that conforms to the target, or the minimum target dose. Compared with fractionated radiation treatments, radiosurgery treatment plans have signiﬁcantly less normal tissue included in the prescription isodose volume. For this reason, the typical restrictions on dose uniformity (a foundation of fractionated radiation therapy) do not apply in radiosurgery cases. The radiosurgeon is more concerned with the dose to normal tissue outside the target volume. Consequently, the goal in treatment planning is to achieve maximum normal tissue sparing rather than maximum target dose uniformity. The radiosurgeon can minimize the dose to normal tissue by designing a plan where the isodose that just covers the target surface is along the steepest portion of the dose gradient. It can be shown that for single-isocenter plans, this places the optimal dose prescription at the 80% isodose surface [16]. This is because the dose gradient falls off most quickly between the

For optimal planning, the neurosurgeon, radiation oncologist, and physicist or dosimetrist should perform the treatment planning as a team and bring all aspects of expertise to bear on the problem. After the images have been imported and registered and contouring of structures has been completed, one can proceed with the task of formulating a treatment plan that meets the dosimetric requirements speciﬁed by the physician. The planning steps that are followed differ from one delivery technique to the next. For the Gamma Knife and for linacbased SRS with circular collimators, the task of planning is to ﬁnd a set of isocenter locations, the size of the collimators to use for each location, and the weights of the isocenters. Multiple isocenters are commonly used in a Gamma Knife treatment plan due to the limited number of collimator sizes to choose from and the relative efﬁciency with which each isocenter can be delivered [17–24]. For linac-based SRS with circular collimators, the planner strives to minimize the number of isocenters in a plan to compensate for the relative inefﬁciency of the beam delivery [25, 26]. Linac systems, however, have the advantage of a much wider range of collimators usually extending from 5 mm to 50 mm in diameter. For linac-based SRS with micro-multileaf collimators, only one isocenter is used [25, 27–37], and the planning methods are similar to those used for external beam radiation therapy. The selection of beam directions and ﬁeld shapes is aided by the use of tools such as beam’s-eye-view (BEV) visualization and digitally reconstructed radiographs (DRRs).

Dose Calculation A variety of dose calculation techniques have been employed for SRS treatment planning [38–48]. Simple empirical dose calculation methods were employed with many of the earlier SRS planning systems [17]. For Gamma Knife radiosurgery, a standard set of beam data is included with the system, and the user veriﬁes the dose proﬁles of individual beams for each of the focusing helmets. The use of a standard data set is possible because all Gamma Knife systems use the same design with only slight variations in the single-beam dosimetry. The dose distribution in the patient is calculated by adding the dose distributions for all 201 beams [20]. For linac-based SRS with circular collimators, the tissue-maximum ratios (TMRs) and dose proﬁles can be measured for each collimator size as a function of depth. For a given arc, the mean TMR for all beam directions

74

d.m. shepard et al.

is calculated and the total dose contributed by the arc can be calculated. Dose distributions can be computed by approximating an arc as a series of ﬁxed beams and summing the dose distributions from all of the beams. For intracranial SRS, heterogeneity corrections are generally not applied in the dose calculation. Corrections to account for variations in the attenuating properties of tissue, bone, and air are not needed because of the simple geometry that is employed [44, 49]. The magnitude of the resulting error can be estimated by assuming that the average beam passes through 5 mm of skull with an average density of 1.2 g/cm3. Assuming an attenuation of 4% per centimeter for a 6-MV beam, the maximum error in absolute TMR that would result from ignoring heterogeneity corrections is less than 1%. However, when SRS is extended to extracranial applications, traditional empirical dose calculation methods are no longer adequate for accurate dose calculations due to the presence of bone and air. Therefore, many treatment-planning systems designed for planning extracranial SRS employ three-dimensional pencil beam dose calculation methods.

Forward Versus Inverse Planning Forward planning is the most common planning approach for SRS. With forward planning, the treatment plan is developed through an iterative trial-and-error approach. From one iteration to the next, the planner attempts to determine a set of parameters (such as beam angle, beam weights, etc.) that give an acceptable plan. The iterative process is usually stopped when the planner is no longer able to make noticeable improvements in the plan quality. With inverse treatment planning, the user begins by outlining the target and any sensitive structures. A series of treatment goals are then deﬁned, and an optimization algorithm determines the plan parameters that provide a plan that best satisﬁes these goals. The quality of the plan is scored using an objective function and constraints. An objective function reduces an entire treatment plan into a single numerical value. The job of the optimizer is to either minimize or maximize the value of the objective function. Typically, an optimization will also include

one or more constraints. A constraint is a condition that must be satisﬁed in order for a solution to be considered feasible. The most basic constraint in any radiotherapy optimization is that the beam weights must be nonnegative. Compared with forward planning, inverse planning provides two potential advantages. First, the time required for planning can be reduced because much of the trial and error is removed. Second, inverse planning should lead to improved plan quality due to the ability of the optimizer to consider thousands of plan conﬁgurations in selecting the optimal plan. Unfortunately, the quality of inverse planning tools for SRS varies signiﬁcantly from one delivery technique to the next and one planning system to the next. Users of some systems will ﬁnd that an experienced planner using forward planning techniques produces the highest quality plans.

Evaluation of Plan Quality The most common tool for evaluating plan quality is the visualization of isodose curves superimposed on the patient’s anatomic images. Figure 7-8a shows a typical isodose plot for a Gamma Knife patient. In this case, the isodose curves are plotted as a percentage of the maximum target dose. Isodose curves can also be plotted as a percentage of the prescribed dose or as absolute dose lines. When evaluating plan quality, it is common practice to visualize isodose curves in the axial, sagittal, and coronal planes. Some systems can also display three-dimensional dose clouds that make it possible to quickly assess the quality of the dose coverage. DVHs also serve as an important tool for analyzing plan quality. A DVH plot from a Gamma Knife patient is shown in Figure 7-8b. For each structure, the DVH plots the fraction of the volume covered by each dose level. Both the dose and the volume can be expressed as either absolute or relative values. DVHs are particularly useful because they reduce a three-dimensional treatment plan into an easy-to-read twodimensional plot. The comparison of multiple plans is also simpliﬁed by overlying DVHs on the same plot. The disadvantage of DVHs is the lack spatial information. A DVH will indicate the presence of hot or cold spots, but it does not specify where

FIGURE 7-8. (a) A Gamma Knife isodose plot with the target outlined in red, the 50% isodose line in yellow, and the 30% isodose line in green. (b) Dose-volume histogram for target in this case.

7.

treatment planning for stereotactic radiosurgery

in the structure of interest this underdosage or underdosage is located. Consequently, slice-by-slice evaluation of a plan is critical in making a ﬁnal decision regarding a treatment plan. Additional parameters have been deﬁned to score both dose conformity and target coverage for SRS treatment plans [50–58]. The most commonly used parameter for scoring dose conformity is the conformity index deﬁned in the radiosurgery quality assurance guidelines of the Radiation Therapy Oncology Group (RTOG) [52]. The RTOG deﬁned the conformity index as the volume of the prescription isodose surface divided by the target volume. For a perfectly conformal plan where the prescription isodose line exactly matches the target volume, the conformity index would equal one. A case is considered to be per protocol if this ratio falls between 1.0 and 2.0. A shortcoming of this index is that it does not consider the degree of overlap between the prescription isodose curve and the target. A plan with a complete geometric miss of the target could still give a perfect conformity index. As an alternative, Lomax and Scheib have suggested a conformity index deﬁned as “the ratio of the volume within the target irradiated to at least the prescription isodose over the total volume enclosed by the prescription isodose” [55]. Consequently, this value ranges from 0 (no conformity) to 1.0 (for perfect conformation, where the prescription isodose is identical to the target volume). As a planning goal, Lomax and Scheib suggest a conformity index of 0.6 or higher. The RTOG has also deﬁned a homogeneity index that is equal to the maximum dose in the treatment volume divided by the prescription dose. A case is considered per protocol if this ratio is less than or equal to 2.0. In terms of target coverage, the RTOG considers a case to be per protocol if the isodose line equal to 90% of the prescribed dose completely encompasses the target. Lomax and Scheib suggest an alternative volumetric deﬁnition of coverage where target coverage is deﬁned as the percentage of the target volume covered by the prescription [55].

75

FIGURE 7-9. A patient positioned for treatment on a Gamma Knife. (Image Courtesy of Elekta, Inc.)

shaping of the dose distribution. The beneﬁts of plugging are illustrated in Figure 7-12. Figure 7-12a shows the isodose curve for a single 4-mm shot for a trigeminal neuralgia patient. In Figure 7-12b, a plugging pattern has been used to lower the dose to the brain stem. The plugging pattern is shown in Figure 7-12c. After the imaging is complete, the patient images are loaded into the Gamma Knife’s treatment planning system (GammaPlan). The images are registered with respect to the ﬁducial marks, and the treatment volume is outlined by a physician.

Gamma Knife Gamma Knife Unit The ﬁrst Gamma Knife was built in 1967 under the direction of Lars Leksell of the Karolinska Institute in Stockholm, Sweden. Currently, there are more than 200 units worldwide with more than 35,000 patients treated annually [59]. Inside of the shielded treatment unit of the Gamma Knife (Fig. 7-9), the beams from 201 radioactive sources are focused so that they intersect at a single location (Fig. 7-10). The result is an elliptical region of high dose with a rapid fall-off in dose outside of the boundaries of the ellipse. Each exposure to an elliptical region of high dose is referred to as a “shot” of radiation. For each exposure, the focusing helmet dictates the size of the high-dose region. A focusing helmet incorporates a separate collimator for each of the 201 cobalt-60 sources (Fig. 7-11a). The four focusing helmets provided with the Gamma Knife can be used to produce a shot of radiation that is 4 mm, 8 mm, 14 mm, or 18 mm in diameter (Fig. 7-11b). Within each helmet, individual collimators may be plugged to provide further

FIGURE 7-10. The 201 collimators focus the beam to a single intersection point. (Image Courtesy of Elekta, Inc.)

76

d.m. shepard et al.

are adjusted to place the center of the planned shot of radiation at the focal point of the 201 beams of radiation. After this, the personnel leave the room, the exposure time is programmed, the door of the shielding unit is opened, and the couch is advanced into the high-radiation ﬁeld. Next, the treatment couch is docked, and the shot of radiation is delivered. After the exposure is complete, the couch retracts and the door closes. If more than 1 shot of radiation is to be delivered, the process is repeated with the appropriate focusing helmet and x, y, and z positioning. The model C Gamma Knife provides an automated positioning system in which the shot positions and exposure times are directly transferred to a record-and-verify system, and the machine sets the coordinates in an automated fashion.

Treatment Planning

FIGURE 7-11. The four focusing helmets dictate the size of each shot of radiation. (Image Courtesy of Elekta, Inc.)

At the time of treatment, the patient lies on the couch of the treatment unit, and the appropriate focusing helmet is afﬁxed to the table. The patient’s stereotactic head frame is then attached to the focusing helmet, and the x, y, and z coordinates

For some cases, the treatment planning process is relatively straightforward. This is particularly true for small, spherical targets. For example, Figure 7-13 shows a case where the treatment volume is relatively spherical and approximately 6 mm in diameter. An 8-mm shot of radiation was placed so that it covers the entire tumor volume. The treatment planning process becomes much more complex when the tumor volume is large or irregularly shaped. These cases typically require several shots of radiation. Through an iterative process, the planner must determine the number of shots of radiation that are required along with the size, the location, and the weight that should be assigned to each. In Figure 7-14, a simple two-dimensional bean-shaped target is used to illustrate the general planning procedure [22]. Each frame illustrates the dosimetric impact of an added shot of radiation. In this case, 5 shots of radiation provide a conformal treatment plan. Note that three different shot sizes were used in creating this plan. Figure 7-15 shows a screen capture from the GammaPlan system. During the process of treatment planning, the plan quality is evaluated through the use of isodose plots and DVHs. In the GammaPlan system, users typically normalize to the maximum dose and seek to cover the target with the 50% isodose line. An imaging study with SRS ﬁducials is required for Gamma Knife planning. For cases where the proximity of a sensitive structure leads to concern over ability to produce an acceptable plan, a “mock” treatment planning session can be helpful. To achieve this, an imaging study is performed with the patient’s head positioned within the ﬁducial box without a frame.

Viewed from behind left

right

back FIGURE 7-12. The isodose curves for a trigeminal neuralgia patient (a) before and (b) after plugging. (c) The plugging pattern: 35 of the 201 sources have been blocked (blocked collimators are shown in black).

plan that is produced can depend upon both the experience and the patience of the user. Because of these factors, researchers have sought to develop an automated process for creating Gamma Knife treatment plans. A number of researchers have investigated techniques for automating the Gamma Knife treatment planning process [18, 19, 21–24, 60–64]. One approach incorporates the assumption that each shot of radiation can be modeled as a rigid sphere. The problem is then reduced to one of geometric coverage, and a ball-packing approach can be used to determine the shot locations and sizes [22, 60, 63, 64]. Researchers have also examined the optimization of plug patterns for Gamma Knife treatment plans [23]. By selectively plugging a subset of the 201 beams, one can provide further sparing of adjacent sensitive structures. In addition, Elekta has provided an automated planning solution as an add-on to the GammaPlan system, called the GammaPlan Wizard. In the research of Shepard et al. [22, 62], the dose distribution is modeled and a formal constrained optimization is used to determine the treatment plan. With this technique, the shot sizes, locations, and weights are optimized simultaneously in less than 10 minutes. The optimization does not require the user to provide initial shot locations, and the optimization model can include dose constraints applied to both the target and the sensitive structures. The treatment plan optimization is based on the use of migrating shot locations and a nonlinear programming approach. The clinical signiﬁcance of this automated system was assessed by comparing an optimized plan with a manual plan for 10 consecutive patients treated at our Gamma Knife facility. Each treatment plan was analyzed using DVHs in conjunction with the conformity index, the minimum target dose, and the integral normal tissue dose. The results are summarized in Table 7-1. The results demonstrate that the quality of treatment plan produced by the inverse planning tool consistently matches

FIGURE 7-13. A single-shot treatment plan for a Gamma Knife treatment. The target is outlined in blue, the 50% isodose line in yellow, and the 30% isodose line in green.

Cushioning around the patient’s head is used to maintain stable positioning. This imaging study is then used to analyze the quality plan that can be achieved.

Inverse Planning For some patients, the treatment planning process can become tedious and time consuming, and the quality of the treatment

100

0.9

80

0.8

0.8

70

0.7

0.7

50 40

2.5 centimeters

5

0.6 0.5

2.5

0.4

0.6 0.5

2.5

0.4

30

0.3

0.3

20

0.2

0.2

10

0.1

0.1

0

0

2.5 centimeters

a 5

1

5

0.9

60 2.5

1

5

90

centimeters

centimeters

5

77

treatment planning for stereotactic radiosurgery

centimeters

7.

5

2.5 centimeters

b

0

c

5

1.2

5

5 1.2

1.2

1.0

1.0

2.5

0.6

2.5 centimeters

0.8 2.5

0.6

centimeters

0.8

centimeters

centimeters

1.0

0.8 2.5 0.6

0.4

0.4

0.4

0.2

0.2

0.2

0

0 2.5 centimeters

5

2.5 centimeters

FIGURE 7-14. (a) A bean-shaped target. (b–f) The process of adding shots to create a treatment plan is illustrated.

5

0

78

d.m. shepard et al.

FIGURE 7-15. Screen capture from the GammaPlan system.

TABLE 7-1. Comparison of manual plans with those created using an inverse planning tool.

Patient number

1 2 3 4 5 6 7 8 9 10 Average

Number of shots

Conformity index

Volume of 30%

Percent coverage 90%

Minimum target dose

Target volume (cm3)

Manual

Opt

Manual

Opt

Manual

Opt

Manual

Opt

Manual

Opt

12.1 3.4 3.9 11.4 7.6 6.7 9.0 37.1 7.2 3.8 10.2

6 4 4 6 8 6 5 12 6 9 6.6

6 4 4 6 8 6 5 12 6 9 6.6

1.67 1.35 2.12 1.19 1.76 1.62 1.64 1.15 1.51 1.16 1.52

1.61 1.38 1.64 1.34 1.39 1.60 1.56 1.25 1.29 1.53 1.46

44.4 12.4 21.6 34.7 33.6 28.0 35.4 105.7 30.2 12.0 35.8

41.9 11.8 16.1 35.8 24.2 27.4 31.9 102.8 25.3 13.7 33.1

100 100 99 100 100 100 100 97 100 99 99.5

100 97 100 100 100 100 100 99 100 100 99.6

48% 46% 40% 45% 49% 52% 47% 39% 49% 43% 45.8

47% 44% 48% 49% 46% 52% 48% 43% 48% 48% 47.3

7.

treatment planning for stereotactic radiosurgery

79

or beats the corresponding plan developed by an experienced physician.

Linac-Based SRS with Circular Cones Initial developments in the use of linac-based SRS techniques were centered on the use of multiple converging arc treatments delivered with circular collimators. For each treatment plan, the user must select the number of arcs, the angular arrangement of each arc, the weight assigned to each arc, and the number of isocenters. When one isocenter is used, the high isodose levels at the isocenter are nearly spherical. A variety of arc arrangements have been reported [26]. Compared with the Gamma Knife, a greater number collimator sizes is available with circular collimators. By adjusting the weightings of the arcs or through an asymmetric arc arrangement, ellipsoidal instead of spherical isodose distributions can be created. For tumors that are small and convex in shape, it is often possible to treat with only one isocenter. When one isocenter is used, the dose uniformity in the target is high and the dose can be prescribed to a high isodose level. When the target volume is large or it deviates from a sphere or an ellipsoid, multiple isocenters are typically required to achieve adequate target dose conformity. The spherical high-dose regions must overlap in order to avoid leaving a cold spot in the target. Consequently, one must accept a decreased target dose uniformity compared with plans using a single isocenter. To date, no negative clinical consequences have been reported as a result of the lack of dose uniformity seen in Gamma Knife and linac-based circular collimator treatments. When multiple isocenters are used, the treatment planning problem is similar to the sphere-packing problem of the Gamma Knife [60], where spherical dose distributions are packed together to achieve a composite dose distribution conforming to the target volume. Because the setup and delivery time per isocenter is typically longer for linac-based radiosurgery compared with Gamma Knife–based radiosurgery, it is generally desirable to deliver a limited number of isocenters; however, the greater selection of collimator sizes and the availability of larger collimator sizes compared with the Gamma Knife reduces the need for the use of a large number of isocenters. Overall, the Gamma Knife centers and linac-based SRS centers have reported similar cures rates and complications levels.

Linac-Based SRS with Micro-Multileaf Collimators A multileaf collimator (MLC) is a ﬁeld-shaping device that uses movable leaves made out of a highly attenuating material such as tungsten in order to generate arbitrary ﬁeld shapes (see Fig. 7-16). MLCs used for routine external beam delivery typically have leaf widths that project to 1 cm at isocenter. These MLCs lack the geometric precision for shaping small irregularly shaped ﬁelds such as those in commonly encountered in SRS. With micro-multileaf collimators (mMLCs), each leaf projects to a width of between 2 and 5 mm at isocenter. mMLCs are suitable for SRS applications, because they are capable of

FIGURE 7-16. A photo of Varian’s Millennium mMLC. (Image courtesy of Varian Medical Systems. Copyright 2006, Varian Medical Systems. All rights reserved.)

shaping small, irregular ﬁelds with acceptable geometric error. Typical mMLCs have between 20 and 80 leaves, arranged in pairs. The maximum ﬁeld size of mMLCs generally varies from 8 to 20 cm, much greater than those available with traditional circular collimators. As a result, extracranial SRS and stereotactic body radiation therapy (SBRT) can be delivered. In addition, linear accelerator vendors now offer MLCs that incorporate leaves of varying widths. For example, Varian’s Millennium MLC has 120 leaves (60 leaves in each leaf bank). Over the central 20 cm of the ﬁeld, the leaves project to 0.5 cm at isocenter while at the edges of the ﬁeld the leaves project to 1 cm at isocenter. This type of MLC can be used for either traditional fractionated radiotherapy or SRS. Due to their ease of use and wide range of possible applications, the advent of mMLCs has led to a decline in popularity of SRS delivery using circular collimators. Compared with Gamma Knife and linac-based SRS with circular collimators, the use of a mMLC and a single isocenter can lead to plans that are less conformal with a less steep dose gradient at the target edge. The decreased dose gradient is in part the result of the use of larger beams and fewer unique beam paths. As mentioned previously, a routine Gamma Knife plan will involve as many as 201 beams, and a typical plan for a linac-based cone-beam treatment would include ﬁve arcs each covering approximately 100° of gantry rotation. By contrast, a typical micro-multileaf plan may have as few as six beams. Target size is a second issue that affects one’s ability to create a conformal treatment plan using a mMLC. For small targets, such as in the treatment of trigeminal neuralgia, high precision in target localization and positioning is required. mMLCs may not be suitable for target sizes signiﬁcantly less than 1 cm due to the undulating ﬁeld edges caused by the ﬁnite leaf width. For larger targets, mMLC can provide efﬁcient beam

80

d.m. shepard et al.

delivery and dose uniformity due to the use of a single isocenter [50]. However, the dose to surrounding structures increases as the size of the target increases [50]. As a result, the dose to normal structures may exceed the acceptable limits if one attempts to treat a large tumor with a single fraction. Generally, multileaf collimators (including mMLCs) can be used for three delivery approaches: (1) ﬁxed ﬁeld delivery; (2) conformal arc delivery where the leaves of the MLC are adjusted continuously to match the BEV projection of a target volume; and (3) intensity-modulated delivery.

Fixed Fields With mMLC, three-dimensional (3D) conformal therapy treatment planning techniques can be applied to SRS planning. If there are a sufﬁcient number of mMLC-shaped beams, the dose distribution quality can rival that obtained with multiple arcs using circular collimators and multiple isocenters. This is especially true when the targets are larger and nonspherical in shape. Because only one isocenter is needed, the dose uniformity created with mMLCs is generally better than that achieved with the use of circular collimators and multiple isocenters. With the exception of the need for stereotactic reconstruction from stereotactic frames and markers, the planning method for SRS essentially matches that for 3D conformal therapy. Each non-coplanar ﬁeld is shaped based on the BEV of the target with additional margins to account for the width of the beam penumbra, which is wider than the penumbra from circular collimators and increases with leaf width. The use of ﬁxed ﬁelds delivered with a mMLC is best suited for convex target shapes. With ﬁxed ﬁelds, there is only a limited ability to spare normal tissues in the center of a concave volume.

Dynamic Conformal Arcs Many of the commercially available mMLCs are capable of dynamic beam delivery where the ﬁeld shape changes continuously to conform to the BEV of the target during arced beam delivery. Dynamic conformal arc delivery combines the dosimetric advantages of arcs (reduced hot-streaks of dose through the patient) with the dose conformity that is possible with mMLC beam shaping. Additionally, this approach only requires a single isocenter. With the use of three or four non-coplanar arcs, a uniform high-dose volume can be created that conforms to the target. Treatment planning for dynamic conformal arcs requires the delineation of the target and critical structures. The planner must determine the number of arcs, their length and their arrangements. For planning and delivery control, each arc is approximated as a series of ﬁxed ﬁelds. The shapes are typically set based on the BEV of the target and critical structures. For example, one can set each ﬁeld shape contained within an arc to match the BEV of the target plus a 3-mm margin. Additionally, one may choose to use the BEV of a critical structure to design ﬁeld shapes that block that structure throughout the arc path. If the gantry of the accelerator rotates with a constant speed and maintains a constant dose rate during rotation, one must assign the same number of monitor units to each beam angle within a given arc.

The dose calculation for conformal arcs is more complicated than that for conventional arcs due to fact that the ﬁeld shapes changes while the beam is on. The planning system must calculate the dose contributions from a large number of irregularly shaped ﬁelds. In determining the angular spacing of the ﬁxed ﬁelds, one must balance the need for accuracy with the desire for a reasonably quick dose calculation time. When an arc is approximated with ﬁelds spaced more than 5° apart from one another, the lack of sufﬁcient ﬁeld overlap results in undulating features in the lower isodose lines away from the focal region. These features are not reﬂected in the actual delivery. Finer spacing of the beams will reduce such artifacts in the displayed isodose lines. In some treatment planning system, a long arc can also be broken into multiple sub-arcs. The weights of these sub-arcs can be optimized based on the user-deﬁned dose-volume constraints [66]. During delivery, each of these sub-arcs is treated as a separate beam. Generally, this technique works best on convex targets and less well on targets with concave surfaces.

Intensity-Modulated Fields With intensity-modulated radiation therapy (IMRT), a modulated intensity pattern is delivered from each beam direction. Consequently, radiation can be delivered to the target through preferred locations within each beam. From each beam direction, the dose delivered to the target is nonuniform. However, all of the beams in combination produce a highly conformal dose distribution. For both external beam radiation therapy and SRS, IMRT improves dose conformity compared with the use of conventional unmodulated beams [67]. It should be noted that for small intracranial targets, IMRT delivered with a mMLC may be unable to achieve the dose conformality obtained using sphere packing due to the fact the width of the leaves of the mMLC are of the order the size of the targets to be treated. The IMRT planning problem is modeled by subdividing each beam into a grid of beamlets. The weight, or intensity, of each of the beamlets is then determined. Because of the complexity of determining the appropriate beamlet weights, inverse (automated) planning techniques are employed. IMRT planning for SRS and IMRT planning for external beam radiation therapy share the same general procedures. The user deﬁnes a series of treatment goals, and an optimization algorithm determines the plan parameters that lead to a plan that best satisﬁes those goals.

CyberKnife The CyberKnife (Fig. 7-17) is a fully integrated radiosurgery system that uses dual kilovoltage imaging devices to locate the treatment site and direct the external treatment beam to it [68]. The treatment beam is provided by a linac mounted to a robotic arm that is capable of maneuvering and pointing the linac with nearly complete freedom within the treatment workspace (with the exception of the angles blocked by the integrated imaging system). During treatment, the imaging system repeatedly acquires and analyzes targeting radiographs, supplying updated target coordinates automatically through a control loop to the robotic arm. This enables the system to maintain alignment

7.

treatment planning for stereotactic radiosurgery

FIGURE 7-17. A CyberKnife.

of the beam with the target even while the patient moves [11, 69]. The maneuverability of the beam, unconstrained by an isocenter, combined with the capacity to adjust alignment during treatment, endow the CyberKnife with unique dose delivery capabilities. These capabilities enable (1) the delivery of highly conformal dose distributions to irregular target volumes, (2) the delivery of fractionated stereotactic radiotherapy treatments, and (3) the treatment of extracranial sites that are not amenable to localization and/or ﬁxation using conventional stereotactic frames. They are exploited in a treatment planning system that has been designed speciﬁcally for the CyberKnife.

Beam Characteristics The treatment beam is provided by a 6-MV X-band linear accelerator that does not employ a ﬂattening ﬁlter. The beam is collimated to a circular cross-section by one of a set of 12 interchangeable collimators. At a source-to-surface distance (SSD) of 80 cm, these collimators provide a beam diameter that ranges from 5 mm to 60 mm. The SSD of the beam can be varied from 60 cm to 100 cm. A continuous range of beam diameters can be achieved by varying the collimator size and SSD.

Treatment Sites, Planning Scenarios, and Imaging Requirements All treatment planning for the CyberKnife is based on a CT study used alone or in conjunction with supplemental diagnostic imaging such as MRI. As with most radiosurgical and stereo-

81

tactic radiotherapy systems, the CT study is utilized for both target delineation and dose calculation; however, with the CyberKnife, the CT also plays a critical role in the imageguidance aspect of the delivery process. If MRI images are to be used to enhance anatomic delineation, then the MRI study must be fused to the CT study prior to beginning the planning process. The integrated image-guidance system allows the CyberKnife to target any treatment site that can be located via radiographic landmarks. Targeting is achieved by registering the landmarks in the kilovoltagae (kV) alignment images with their counterparts in digitally reconstructed radiographs (DRRs) derived from the treatment planning CT study. There are two basic strategies for localization. Sites that maintain a rigid relationship with bony landmarks that are easy to image (such as intracranial lesions) can be located by measuring the position of the skeletal features in the treatment room. Sites in soft tissue or along the spine are located with the assistance of artiﬁcial ﬁducial markers implanted near the lesion in a manner similar to those being employed by other extracranial systems. This targeting capability enables the user of the CyberKnife to plan treatments for central nervous system (CNS) lesions in the cranium and anywhere along the spine, as well as soft-tissue sites in the thorax, abdomen, and pelvis. Although the general treatment planning procedure is not site-speciﬁc, each site introduces some distinctive elements to the planning process. The important role that the planning CT study plays in target localization and beam alignment during treatment places additional demands on its spatial resolution. To achieve optimal targeting accuracy [70], it is recommended that the CT slice thickness not exceed 1.25 mm [71]. Although this slice thickness is commonly used for intracranial lesions, it is a higher spatial resolution than is typically used for routine diagnostic studies of extracranial sites. Central nervous system lesions, including tumors and AVMs, are the most commonly treated sites. As examples of CNS applications, we take note of two specialized treatment planning problems: trigeminal neuralgia and spinal lesions. The trigeminal nerve is visualized for planning using CT cisternography [72]. The patient is scanned over the full head in the Trendelenberg position. Typically, 64 cGy is prescribed to the 80% isodose line, encompassing approximately 8 mm of nerve, and is delivered in one fraction. Beam alignment and tracking during treatment is based on the position of the cranium visualized in the treatment room radiographs. Treatments of lesions along the spine are delivered in 1 to 5 fractions. Targeting is based on ﬁducials implanted into the spine near the treatment site. The planning CT study (Fig. 7-18) is set up to visualize the region in and around the lesion, plus the targeting ﬁducials, with the patient lying supine in an alpha cradle. The typical treatment dose is 1200 to 2000 cGy prescribed to the 80% isodose line. Dose to the cord is limited to 800 cGy. Experience from numerous treatments has shown that patients resting supine in an alpha cradle move less than 3 mm over the course of a 30-minute treatment fraction [9]. The image-guidance system detects and corrects for intrafraction movement. In addition to CNS sites, the CyberKnife has been used to treat pancreatic, lung, prostate and other soft tissue tumors. These represent innovations in the application of radiosurgery

82

d.m. shepard et al.

FIGURE 7-18. Optimized plan for a spinal lesion.

and thus have not yet evolved standard protocols for treatment planning.

The Planning Process The CyberKnife treatment planning process combines a variety of user-managed and automatic planning operations to arrive at satisfactory deliverable plans. The planning system uses a CT study as the primary planning resource. The dose calculation process makes a coarse accommodation of anatomic inhomogeneities by distinguishing air, tissue, and bone based on the Hounsﬁeld number. The robotic manipulator has a great deal of ﬂexibility in positioning and aiming the treatment beam. This makes it possible to deliver both isocentric and non-isocentric dose distributions. To utilize the full beneﬁt of this ﬂexibility and at the same time make planning calculations tractable, the planning system works with a discrete set of linac positions (called nodes) and a discrete set of pointing directions at each node analogous to a multitude of isocenter-gantry-table positions in routine radiosurgery planning. During the delivery process, the robotic manipulator moves the linac from one node to another. At each node, the robot stops, aims the linac in the selected directions, and delivers dose increments in a “stop and shoot” sequence. The beam is not on while the robot is in motion.

A typical treatment plan has at its disposal a set of up to 110 nodes distributed approximately uniformly over about onehalf of a sphere centered on the treatment site. For each plan, these nodes are selected from a total of more than 300 possible linac positions. For non-isocentric delivery, the planning system deﬁnes up to 12 discrete pointing directions (vectors) at each node. These pointing vectors are aligned to designated points within the treatment volume to produce a set of overlapping beams designed to optimally cover the target volume while avoiding critical structures. Figure 7-19 illustrates this approach schematically. The combination of nodes and pointing vectors provides altogether a set of 1320 “beams” from which to construct a plan. The plan is developed by selecting beams (i.e., nodes and pointing vectors) from among the 1320 available and assigning them “weights” corresponding with the amount of dose each beam is to deliver. The planning process consists of a sequence of planning choices by the clinician combined with a set of constraints on the dose actually delivered by the plan. The ﬁrst step is to identify the treatment site. Because each anatomic location is optimally treated by a particular conﬁguration of nodes, the planning system has predetermined groups of nodes that are considered the best choices for each treatment site. These groups of nodes are called “paths.” There are 80 to 110 nodes in each path, chosen from among 300+ possible linac positions. When the

FIGURE 7-19. (a, b) Illustration of isocentric versus non-isocentric delivery.

treatment planning for stereotactic radiosurgery

clinician designates the treatment site, the planning system selects the appropriate path. In the second step, the clinician chooses to construct either a single-isocenter, a multiple-isocenter, or a non-isocentric plan. Although the isocentric option is reasonable for compact semispherical lesions, the non-isocentric technique is most commonly employed due to its greater ﬂexibility. The third step is to choose either forward or inverse planning. Non-isocentric plans are difﬁcult to design using forward planning and so inverse planning is typically used. It will be assumed from this point on that the clinician is interested in inverse planning. One or more collimators can be employed in a plan. Typically, for isocentric plans, the collimator diameter is matched to the diameter of the lesion. In non-isocentric delivery, the clinician has more freedom in selecting the collimator diameter(s). Typically, the collimator diameter is chosen to be approximately 60% of the lesion’s smallest projected cross-section. This makes it possible to obtain acceptable dose conformality, dose homogeneity, and dose fall-off at the boundary of the target without unduly lengthening the treatment time. As mentioned above, the beam diameter can be ﬁne-tuned by adjusting the SSD away from its nominal distance of 80 cm. Inverse planning works within a set of constraints to ﬁnd the best deliverable plan. The clinician speciﬁes a minimum target dose and a maximum dose to each critical structure that can potentially be transited by a beam. Additionally, the planner can create artiﬁcial dose constraint structures to further inﬂuence the plan. These structures can be used to attenuate or block beams so as to shield critical structures. They can also be used to modulate the dose reaching the lesion. The effective use of artiﬁcial constraints is a valuable acquired skill. In the contouring process, the clinician outlines the target volume, all relevant critical structures, and any artiﬁcial structures needed to help manage the dose distribution. Part of this step involves identifying up to 12 locations within the target lesion that serve as end points for the pointing vectors at each node. After the contouring is complete, the planning system has up to 1320 beams to choose from when optimizing the plan. The inverse calculation then selects from among the available beams to get optimal target coverage and critical structure avoidance. The optimizer then assigns beam weights that meet the dose constraints. If constraints applied in an inverse planning formulation are too strict, the system may be unable to ﬁnd a plan that simultaneously satisﬁes all of the constraints. Experienced CyberKnife planners have found it most effective to begin by deﬁning minimum target doses combined with fairly generous maximum doses to the critical structures. If the system can ﬁnd a solution for this ﬁrst approximation, the clinician then reduces the maximum doses to the critical structures until either the necessary constraints are met or a solution is no longer possible. If the obtainable solutions cannot quite meet the original critical structure constraints, then the clinician must decide if the best obtainable solution is good enough. After an optimized plan has been calculated, the clinician has an opportunity to ﬁne-tune it in a pseudo-forward planning step by selectively turning individual beams on and off. The ﬁnal decision made in the planning process is whether the original fractionation scheme should be modiﬁed in light of

83

the achievable dose limits to the target volume and critical structures. This is an important consideration for spine treatments because there is incomplete knowledge of the spinal cord’s tolerance of single-fraction doses.

Proton SRS There are currently 23 clinical proton radiotherapy facilities in the world (9 proposed over the next 4 years) [73]. The upfront costs of around $100 million as well as the operational costs are the primary reasons there are so few proton facilities. Of those centers offering proton therapy, fewer than half have an active radiosurgery program. By comparison, there are approximately 200 Gamma Knife centers worldwide [59] and a much larger number of linac-based SRS programs. The major advantage of high-energy protons and other heavy charged particles is the characteristic distribution of dose with depth (Fig. 7-20). As the beam passes through tissue, the dose initially remains approximately constant. Near the end of the range, however, the dose quickly increases to its peak value then rapidly falls off to near zero dose. The region of high dose at the end of the particle range is called the Bragg peak [74]. By adjusting the incident energy of the beam, the position of the Bragg peak can be controlled to match the depth of the lesions being treated. This is, however, a case of too much of a good thing. The Bragg peak is usually much narrower than the span of the average target. This in turn requires that the peak be “spread out” or modulated. The effect of this modulation is an increase in the surface dose of the entrance beam (Fig. 7-21). Proton delivery systems are not compact like conventional linacs or Gamma Knife units. Until recently, proton facilities that were originally designed for nuclear research used retroﬁtted equipment to bring the radiation to the patients. This meant using ﬁxed beamlines and innovative methods to deliver ﬁelds from various directions. Modern proton facilities designed and built for radiation therapy have incorporated gantry systems making their use similar to linacs [75, 76] (Fig. 7-22).

Dose Modeling Doses from proton beams are clinically described in cobalt Gray equivalent (CGE). The CGE represents the equivalent Photon & Proton Depth Doses 190 MeV Photons

100

170 MeV Photons 150 MeV Photons

Relative Dose (%)

7.

130 MeV Photons

75

6 MeV Photons

50 25 0 0

5

10 15 20 25 Depth in Water (cm) FIGURE 7-20. Dose distributions for 6-MV photons and protons of various energies.

84

d.m. shepard et al.

FIGURE 7-21. (a) SOBP dose distribution designed with a 16-cm range to the distal 90% and 3-cm modulation. The SOPB consists of ﬁve pristine peaks with respective ranges of 16.0, 15.4, 14.8, 14.2, and 13.6 cm, with relative weights of 0.488, 0.168, 0.137, 0.098, and 0.109. The proximal tail of the individual pristine peaks is omitted on all but two curves to improve the visual clarity of the graph. (b) Pristine Bragg peaks with the same energy but varying ﬁeld diameters from 1.0 to 4.8 cm. The smaller ﬁelds lack lateral dose equilibrium because more protons diverge from the central axis than converge to the central axis. Thus, for the same delivered monitor units, fewer protons reach the isocenter with small ﬁelds compared with large ﬁelds. The ﬁeld size effect is also energy dependent.

100

Dose

80

60

Relative Dose (%)

90

40 45 Ø 10 – 48 mm 20

20

5 16 Depth in Water (cm) 13 FIGURE 7-22. (a) The STAR (Stereotactic Alignment Radiosurgery) isocentric patient positioning device designed and built by Product Genesis Inc. and used by the Massachusetts General Hospital (Boston) group in conjunction with horizontal ﬁxed proton beamlines. The patient can be rotated enabling any portal combination aimed from the

top cranial hemisphere. (b) A 110-ton isocentric gantry and 6-axis robotic patient positioner designed and built as a collaborative effort between General Atomic Inc. and Ion Beam Application Inc. and used for proton SRT and SRS at the Northeast Proton Therapy Center, Massachusetts General Hospital, Boston.

7.

treatment planning for stereotactic radiosurgery

biological dose to achieve the same cell-kill as 60Co gamma rays. The ratio of the physical to biological dose is called the radiobiological effectiveness (RBE). Clinical applications of protons assume an RBE of 1.1 relative to 60Co [77, 78]. This means that 10% less physical dose is required when delivering protons beams compared with X-rays. The Bragg peak measured for a single energy beam is referred to as a “pristine” Bragg peak. Figure 7-20 shows pristine Bragg peaks for beams of various energies measured in a water phantom. The width of a pristine Bragg peak as measured from the proximal to distal 90% dose varies with energy. The width is narrower for shallow ﬁelds and broader for deeper ﬁelds, however, a typical width is around 0.5 cm. In order to treat larger lesions, multiple pristine Bragg peaks are combined to create a spread out Bragg peak (SOBP) (see Fig. 7-21a). The SOPB 90–90% width (also known as the modulation width) can be customized to the thickness of any lesion by varying the depth and weight of each pristine peak. It is important to note that the relative weights and depth pullbacks of the individual pristine peaks needed to generate a ﬂat SOBP are dependent not only on the 90–90% width but also on their entire depth dose proﬁle. The pristine peak shape is also affected by the ﬁeld diameter (see Fig. 7-21b). Therefore, for a speciﬁc beamline, both the energy and the ﬁeld diameter must be modeled in the treatment planning algorithm to properly generate the desired SOBP. A SOPB can be generated by delivering the dose from each pristine Bragg peak separately or by using beam spreading devices such as such as spinning absorber wheels or ridge ﬁlters [79]. The lateral dose penumbra of proton beams depend both on the inherent beam source geometry as well as individual treatment ﬁeld settings. Beamlines designed to treat large ﬁelds generally use a double scattering system, which can result in a 5-cm FWHM source size. When this source size is combined with a source-to-axis-distance (SAD) of 220 cm that is typical for a proton gantry system, the result is an 80–20% isocenter plane penumbra of 2.8 mm for a range of 8 cm, and 6.3 mm for a range of 16 cm. In a small-ﬁeld single-scattering system with an SAD of 450 cm, the source size is around 1 cm FWHM, and the penumbra is reduced to 2.3 mm for a range of 8 cm, and 5.3 mm for a range of 16 cm. Treatment planning systems with integrated proton beam algorithms, such as CMS Focus (CMS, Inc.), must be able to model both the lateral and depth dose proﬁles based on as many clinical variables as possible. This is achieved by using pencil beam algorithms that model multiple Coulomb scattering within beam-modifying devices as well as the patient [80, 81].

Preplanning Consideration Compared with linac and Gamma Knife delivery, protons result in a lower integral dose to the patient due to the sharp dose fall-off beyond the Bragg peak. Larger lesions beneﬁt the most from the reduction of dose to normal brain from protons compared with other modalities. Protons also make it possible to deliver a uniform target dose even with large lesions. This may be considered an advantage when treating AVM, where the target and normal brain are intertwined; however, additional clinical results are needed to demonstrate this conclusively.

85

The immobilization hardware and alignment techniques used for linac SRS can be applied to proton radiosurgery particularly when a gantry-based beamline is used. In ﬁxed beamlines, the need to rotate the patient can lead to torque on the patient’s head that can result in slight shifts of the head relative to an external reference frame. In such cases, it would be unwise to rely on conventional stereotactic external reference frames. The insertion of three 1/16-inch-diameter, 316LVM-grade stainless steel spheres in the outer table of the patient’s skull used in conjunction with a diagnostic-quality X-ray treatment alignment system provides a reliable reference coordinate system [82]. A beneﬁt of using an internal marker system is the ﬂexibility to image, plan, and treat on different days. This is an important consideration for proton facilities that have signiﬁcant overhead due to the need to fabricate custom beam shaping devices, which require pretreatment quality assurance/quality control and possibly dosimetric veriﬁcation. Patients also beneﬁt from not having to wear a stereotactic frame while waiting for their treatments as is necessary with same-day CT-plan-treat schedules. Using internal alignment markers does not necessarily preclude the use of stereotactic bony ﬁxation for CT imaging and/or the treatment. It does, however, provide the ﬂexibility to use stable noninvasive immobilization devices as assessed on an individual patient basis. Treatment planning for proton therapy requires CT imaging because the dose algorithms convert the pixel densities to proton stopping power. The stopping powers are used to calculate the penetration of the protons inside the patient. When required, secondary imaging studies such as positron emission tomography, MRI, and angiography can be fused or merged with the primary planning CT as is done with conventional treatments [83, 84]. Because of the proton penetration conversion from CT densities and the sharp dose fall-off beyond the Bragg peak, special precautions are necessary to ensure that the planning CT’s densities are not artiﬁcially altered. This means that if a CT requires contrast for target and normal structure delineation, it may be necessary to ﬁrst obtain a noncontrast CT to be used for treatment planning. A secondary scan performed with contrast is used for structure delineation and fused to the noncontrast planning CT. This is especially important when treating vascular lesions, where using the contrast CT scan to calculate the proton stopping powers would overestimate the required proton range and modulation and result in an increased treatment volume.

Clinical Treatment Planning Proton arc therapy is not feasible nor is it necessary to generate conformal plans. Three to ﬁve static ports are generally sufﬁcient to obtain satisfactory target dose conformity. Using BEV projections, the user determines the best combination of ports to optimize the dose conformality. When practical, the use of orthogonal beams minimizes overlap regions between ﬁelds. Brass apertures are custom milled to the shape of the lesion’s BEV projection (Fig. 7-23). The expansion of the aperture openings accounts for the penumbra of the speciﬁc portal. Blocking of speciﬁc critical structures is made easy using the BEV approach. Custom Lucite range compensators are fabricated to enable dose shaping to conform to the distal edge of the target.

86

d.m. shepard et al.

FIGURE 7-23. A custom brass aperture and Lucite compensator used to ensure a conformal proton beam.

Although the beam directions are determined on a case-bycase basis, an effort is made to use standard beam arrangements. Pituitary adenomas are located in and around sella and have similar dose constraints. A standard approach at the Massachusetts General Hospital (MGH) for pituitary lesions has been to use four ﬁelds (RL, LL, ASO, PSO) (Fig. 7-24). The lateral ﬁelds avoid the optic structures and brain stem while passing through the temporal lobes; the PSO avoids the optic structures and temporal lobes while passing through the brain stem; and the reduced-weight ASO avoids the temporal lobes and brain stem while passing through the optic structures. Another standardized approach is used to treat acoustic neuromas using three to four ﬁelds (L/RPO, L/RAO, L/RSO, L/RPSO). This ﬁeld combination limits the dose to the brain stem, cerebellum, and temporal lobe while avoiding bone heterogeneities as much as possible. In other cases such as AVMs, the variability in location, shape, and size make standardization difﬁcult (Fig. 7-25). In order to maximize dose uniformity across targets, it has been the MGH practice, for proton SRS, to avoid truncating a target so as to treat it using abutting or patch ﬁelds [85]. These techniques are regularly used when treating with fractionated proton SRT to improve dose conformality of very irregular targets; however, they create small, local high-dose points within the treatment volume.

FIGURE 7-24. A standard four-ﬁeld pituitary beam combination used for PSRS. The ﬁrst image shows a 3D view of the ports and the structures of interest. Typical doses are 18 to 20 CGE at 90% to the target with constraints of 8 and 12 CGE to the surface of the optic structures and brain stem, respectively.

7.

treatment planning for stereotactic radiosurgery

87

FIGURE 7-25. Six-ﬁeld PSRS radiosurgery isodose plan for a 9.6-cm3 right thalamic AVM in a 17-year-old patient. The treatment was delivered in two sessions, separated by 2 weeks, each delivering 8 CGE at

90% for a total of 16 CGE at 90%. Three ﬁelds (LL, RPO, ASO) were delivered at the ﬁrst session and three different ﬁelds (RL, LAO, LSO) were delivered at the second session.

Patient demographics are dependent on both the practice referral patterns as well as the facility’s ability to offer alternative treatments. Proton facilities that do not have an alternative modality option are likely to have a signiﬁcant percentage (~25%) of patients with metastatic disease compared with very few if an alternative such as linac SRS is available. At the MGH, 47% of PSRS patients have AVMs, 19% pituitary or cavernous sinus lesions, 13% acoustic neuromas, 11% menin-

giomas, 4% extracranial, and 6% have miscellaneous other disease.

Other Considerations When one is considering the overall advantage of protons for SRS, one should compare the dose gradient relative to standard Gamma Knife or linac-based systems. As with any

88

d.m. shepard et al.

such treatment device, there exist unique planning situations at which it will excel. When one considers, however, that the Gamma Knife and the linac-based system routinely produce plans that expose as little as 5% to 10% of normal nontarget tissues in the prescription isodose shell while providing dose gradients that drop the dose from the prescription dose to half of the prescription dose in 2.5 to 4 mm, the margin for improvement is relatively small. The image-guidance techniques available with linac-based systems such as cone-beam CT could also impact the degree of the advantage seen with proton therapy for SRS and SBRT.

Conclusion Technological advances in recent years have created a new array of options for SRS. The dosimetric advantage of SRS has also been gradually extended to extracranial sites. The introduction of micro-multileaf collimators, IMRT, and inverse planning into SRS gives us more control in shaping the highdose volume to conform to the target. Online or real-time image guidance also provides a new alternative to the use of traditional stereotactic frames. With these advances, fractionated treatments can now be reliably administered. Although the dosimetric goals and basic principles of SRS planning have not changed, the means of achieving the dosimetric goals has shifted more toward automation and computer optimization.

References 1. Ganz JC. Gamma Knife Surgery, 2nd ed. New York: SpringerVerlag, 1997:205. 2. Leksell L. Stereotactic radiosurgery. J Neurol Neurosurg Psychiatry 1983; 46(9):797–803. 3. Leksell L. Cerebral radiosurgery. I. Gammathalanotomy in two cases of intractable pain. Acta Chir Scand 1968; 134(8):585–595. 4. Leksell L. The stereotaxic method and radiosurgery of the brain. Acta Chir Scand 1951; 102(4):316–319. 5. Wagner TH. Optimal delivery techniques for intracranial stereotactic radiosurgery using circular and multileaf collimators. In: Nuclear and Radiological Engineering. Gainesville, FL: University of Florida, 2000. 6. Takeuchi H, Yoshida M, Kubota T, et al. Frameless stereotactic radiosurgery with mobile CT, mask immobilization and micromultileaf collimators. Minim Invasive Neurosurg 2003; 46(2): 82–85. 7. Ryken TC, Meeks SL, Pennington EC, et al. Initial clinical experience with frameless stereotactic radiosurgery: analysis of accuracy and feasibility. Int J Radiat Oncol Biol Phys 2001; 51(4):1152– 1158. 8. Murphy MJ, Cox RS. The accuracy of dose localization for an image-guided frameless radiosurgery system. Med Phys 1996; 23(12):2043–2049. 9. Murphy MJ, Chang SD, Gibbs IC, et al. Patterns of patient movement during frameless image-guided radiosurgery. Int J Radiat Oncol Biol Phys 2003; 55(5):1400–1408. 10. Kamath R, Ryken TC, Meeks SL, et al. Initial clinical experience with frameless radiosurgery for patients with intracranial metastases. Int J Radiat Oncol Biol Phys 2005; 61(5):1467–1472. 11. Gerszten PC, Ozhasoglu C, Burton SA, et al. Evaluation of CyberKnife frameless real-time image-guided stereotactic radiosurgery for spinal lesions. Stereotact Funct Neurosurg 2003; 81(1–4):84–89.

12. Bova FJ, Buatti JM, Friedman WA, et al. The University of Florida frameless high-precision stereotactic radiotherapy system. Int J Radiat Oncol Biol Phys 1997; 38(4):875–882. 13. Meeks SL, Bova FJ, Wagner TH, et al. Image localization for frameless stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 2000; 46(5):1291–1299. 14. Buatti JM, Bova FJ, Friedman, et al. Preliminary experience with frameless stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 1998; 42(3):591–599. 15. Gehring MA, Mackie TR, Kubsad SS, et al. A three-dimensional volume visualization package applied to stereotactic radiosurgery treatment planning. Int J Radiat Oncol Biol Phys 1991; 21(2): 491–500. 16. Friedman WA, Bova FJ. The University of Florida radiosurgery system. Surg Neurol 1989; 32(5):334–342. 17. Coy SR, Houdek PV. Radiosurgery treatment planning. Semin Radiat Oncol 1995; 5(3):213–219. 18. Ferris MC, Lim J, Shepard DM. Radiosurgery treatment planning via nonlinear programming. Ann Operations Res 2003; 119:247– 260. 19. Ferris MC, Shepard DM. Optimization of Gamma Knife radiosurgery. In: Du Pardolas DZ, Wang J, eds. Discrete Mathematical Problems with Medical Applications. Providence, RI: AMS, 2000. 20. Flickinger JC, Lunsford LD, Wu A, et al. Treatment planning for gamma knife radiosurgery with multiple isocenters. Int J Radiat Oncol Biol Phys 1990; 18(6):1495–1501. 21. Luo L, Shu H, Yu W, et al. Optimizing computerized treatment planning for the Gamma Knife by source culling. Int J Radiat Oncol Biol Phys 1999; 45(5):1339–1346. 22. Shepard DM, Ferris MC, Ove R, Ma L. Inverse treatment planning for Gamma Knife radiosurgery. Med Phys 2000; 27(12): 2748–2756. 23. Zhang P, Wu J, Dean D, et al. Plug pattern optimization for gamma knife radiosurgery treatment planning. Int J Radiat Oncol Biol Phys 2003; 55(2):420–427. 24. Zhang P, Dean D, Metzger A, Sibata C. Optimization of Gamma knife treatment planning via guided evolutionary simulated annealing. Med Phys 2001; 28(8):1746–1752. 25. Kubo HD, Pappas CT, Wilder RB. A comparison of arc-based and static mini-multileaf collimator-based radiosurgery treatment plans. Radiother Oncol 1997; 45(1):89–93. 26. Podgorsak EB, Pike GB, Olivier A, et al. Radiosurgery with high energy photon beams: a comparison among techniques. Int J Radiat Oncol Biol Phys 1989; 16(3):857–865. 27. Alheit H, Dornfeld S, Dawel M, et al. Patient position reproducibility in fractionated stereotactically guided conformal radiotherapy using the BrainLab mask system. Strahlenther Onkol 2001; 177(5):264–268. 28. Bourland JD, McCollough KP. Static ﬁeld conformal stereotactic radiosurgery: physical techniques. Int J Radiat Oncol Biol Phys 1994; 28(2):471–479. 29. Cardinale RM, Benedict SH, Wu Q, et al. A comparison of three stereotactic radiotherapy techniques; ARCS vs. noncoplanar ﬁxed ﬁelds vs. intensity modulation. Int J Radiat Oncol Biol Phys 1998; 42(2):431–436. 30. Grebe G, Pfaender M, Roll M, et al. Dynamic arc radiosurgery and radiotherapy: commissioning and veriﬁcation of dose distributions. Int J Radiat Oncol Biol Phys 2001; 49(5):1451–1460. 31. Hamilton RJ, Kuchnir FT, Sweeney SJ, et al. Comparison of static conformal ﬁeld with multiple noncoplanar arc techniques for stereotactic radiosurgery or stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 1995; 33(5):1221–1228. 32. Laing RW, Bentley RE, Nahum AE, et al. Stereotactic radiotherapy of irregular targets: a comparison between static conformal beams and non-coplanar arcs. Radiother Oncol 1993; 28(3):241– 246.

7.

treatment planning for stereotactic radiosurgery

33. Shiu AS, Kooy HM, Ewton JR, et al. Comparison of miniature multileaf collimation (MMLC) with circular collimation for stereotactic treatment. Int J Radiat Oncol Biol Phys 1997; 37(3): 679–688. 34. Soanes T, Hampshire A, Vaughan C, et al. The commissioning and quality assurance of the Automatic Positioning System on the Leksell gamma knife. J Neurosurg 2002; 97(5 Suppl):574– 578. 35. Tsai JS, Engler MJ, Ling MN, et al. A non-invasive immobilization system and related quality assurance for dynamic intensity modulated radiation therapy of intracranial and head and neck disease. Int J Radiat Oncol Biol Phys 1999; 43(2):455–467. 36. Tome WA, Meeks SL, Buatti JM, et al. A high-precision system for conformal intracranial radiotherapy. Int J Radiat Oncol Biol Phys 2000; 47(4):1137–1143. 37. Urie MM, Lo YC, Litofsky S, FitzGerald TJ. Miniature multileaf collimator as an alternative to traditional circular collimators for stereotactic radiosurgery and stereotactic radiotherapy. Stereotact Funct Neurosurg 2001; 76(1):47–62. 38. Haedinger U, Krieger T, Flentje M, Wulf J. Inﬂuence of calculation model on dose distribution in stereotactic radiotherapy for pulmonary targets. Int J Radiat Oncol Biol Phys 2005; 61(1):239– 249. 39. Chaves A, Lopes MC, Alves CC, et al. A Monte Carlo multiple source model applied to radiosurgery narrow photon beams. Med Phys 2004; 31(8):2192–2204. 40. Pike B, Peters TM, Podgorsak E, et al. Stereotactic external beam calculations for radiosurgical treatment of brain lesions. Appl Neurophysiol 1987; 50(1–6):269–273. 41. Pike B, Podgorsak EB, Peters TM, Pla C. Dose distributions in dynamic stereotactic radiosurgery. Med Phys 1987; 14(5):780– 789. 42. Wu A, Lindner G, Maitz AH, et al. Physics of gamma knife approach on convergent beams in stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 1990; 18(4):941–949. 43. Wu X, Ting JY, Markoe AM, et al. Stereotactic dose computation and plan optimization using the convolution theorem. I. Dose computation. Stereotact Funct Neurosurg 1996; 66(Suppl 1):302– 308. 44. Solberg TD, Holly FE, De Salles AA, et al. Implications of tissue heterogeneity for radiosurgery in head and neck tumors. Int J Radiat Oncol Biol Phys 1995; 32(1):235–239. 45. Kubsad SS, Mackie TR, Gehring MA, et al. Monte Carlo and convolution dosimetry for stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 1990; 19(4):1027–1035. 46. Dong, L, Shiu A, Tung S, Hogstrom K. A pencil-beam photon dose algorithm for stereotactic radiosurgery using a miniature multileaf collimator. Med Phys 1998; 25(6):841–850. 47. Bardash M, Amols HI, Kohn S, et al. Rapid dose calculations for stereotactic radiosurgery. Med Phys 1992; 19(4):965–970. 48. Ayyangar KM, Jiang SB. Do we need Monte Carlo treatment planning for linac based radiosurgery? A case study. Med Dosim 1998; 23(3):161–168. 49. Verellen D, Linthout N, Bel A, et al. Assessment of the uncertainties in dose delivery of a commercial system for linac-based stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 1999; 44(2):421–433. 50. Verhey LJ, Smith V, Serago CF. Comparison of radiosurgery treatment modalities based on physical dose distributions. Int J Radiat Oncol Biol Phys 1998; 40(2):497–505. 51. Smith V, Verhey L, Serago CF. Comparison of radiosurgery treatment modalities based on complication and control probabilities. Int J Radiat Oncol Biol Phys 1998; 40(2):507–513. 52. Shaw E, Kline R, Gillin M, et al. Radiation Therapy Oncology Group: radiosurgery quality assurance guidelines. Int J Radiat Oncol Biol Phys 1993; 27(5):1231–1239.

89

53. Paddick I. A simple scoring ratio to index the conformity of radiosurgical treatment plans. Technical note. J Neurosurg 2000; 93(Suppl 3):219–222. 54. Nedzi LA, Kooy HM, Alexander E 3rd, et al. Dynamic ﬁeld shaping for stereotactic radiosurgery: a modeling study. Int J Radiat Oncol Biol Phys 1993; 25(5):859–869. 55. Lomax NJ, Scheib SG. Quantifying the degree of conformity in radiosurgery treatment planning. Int J Radiat Oncol Biol Phys 2003; 55(5):1409–1419. 56. Leung LH, Chua DT, Wu PM. A new tool for dose conformity evaluation of radiosurgery treatment plans. Int J Radiat Oncol Biol Phys 1999; 45(1):233–241. 57. Knoos T, Kristensen I, Nilsson P. Volumetric and dosimetric evaluation of radiation treatment plans: radiation conformity index. Int J Radiat Oncol Biol Phys 1998; 42(5):1169–1176. 58. Borden JA, Mahajan A, Tsai JS. A quality factor to compare the dosimetry of gamma knife radiosurgery and intensity-modulated radiation therapy quantitatively as a function of target volume and shape. Technical note. J Neurosurg 2000; 93(Suppl 3):228– 232. 59. Elekta AB. 60. Cho PS, Kuterdem HG, Marks RJ 2nd. A spherical dose model for radiosurgery plan optimization. Phys Med Biol 1998; 43(10):3145– 3148. 61. Ferris MC, Lim J, Shepard DM. An optimization approach for radiosurgery treatment planning. SIAM J Optimiz 2003; 13(3):921– 937. 62. Shepard DM, Chin LS, DiBiase SJ, et al. Clinical implementation of an automated planning system for gamma knife radiosurgery. Int J Radiat Oncol Biol Phys 2003; 56(5):1488–1494. 63. Wu QJ, Bourland JD. Three-dimensional skeletonization for computer-assisted treatment planning in radiosurgery. Comput Med Imaging Graph 2000; 24(4):243–251. 64. Wu QJ, Bourland JD. Morphology-guided radiosurgery treatment planning and optimization for multiple isocenters. Med Phys 1999; 26(10):2151–2160. 65. Lutz W, Winston KR, Maleki N. A system for stereotactic radiosurgery with a linear accelerator. Int J Radiat Oncol Biol Phys 1988; 14(2):373–381. 66. Ma L, Yu CX, Earl M, et al. Optimized intensity-modulated arc therapy for prostate cancer treatment. Int J Cancer 2001; 96(6): 379–384. 67. Kramer BA, Wazer DE, Engler MJ, et al. Dosimetric comparison of stereotactic radiosurgery to intensity modulated radiotherapy. Radiat Oncol Invest 1998; 6(1):18–25. 68. Adler JR Jr, Murphy MJ, Chang SD, Hancock SL. Image-guided robotic radiosurgery. Neurosurgery 1999; 44(6):1299–1306; discussion 1306–1307. 69. Chang SD, Murphy M, Geis DP, et al. Clinical experience with image-guided robotic radiosurgery (the CyberKnife) in the treatment of brain and spinal cord tumors. Neurol Med Chir (Tokyo) 1998; 38(11):780–783. 70. Chang SD, Main W, Martin DP, et al. An analysis of the accuracy of the CyberKnife: a robotic frameless stereotactic radiosurgical system. Neurosurgery 2003; 52(1):140–146; discussion 146–147. 71. Heilbrun MP. CyberKnife Radiosurgery: A Practical Guide. The CyberKnife Society, 2003. 72. Mawad ME, Silver AJ, Hilal SK, Ganti SR. Computed tomography of the brain stem with intrathecal metrizamide. Part I: the normal brain stem. AJR Am J Roentgenol 1983; 140(3):553– 563. 73. Sisterson J. Sponsored by the Particle Therapy Co-Operative Group (PTCOG). Particles Newsletter 2005; 35. 74. Kahn FM. The Physics of Radiation Therapy, 3rd ed. Philadelphia: Lippincott Williams & Wilkins, 2003:560.

90

d.m. shepard et al.

75. Slater J, Miller D, Archambeau J. Development of a hospital based proton therapy center. Int J Radiat Oncol Biol Phys 1988; 14(4):761–775. 76. Flanz J, Bailey J, Bradley S, et al. Recent performance of the NPTC equipment compared with the clinical speciﬁcations. In: Proceedings of 15th Conference on the Application of Accelerators in Research and Industry. AIP Press, 1999. 77. Raju MR. Proton radiobiology, radiosurgery and radiotherapy. Int J Radiat Biol 1995; 67(3):237–259. 78. Paganetti H, Niemierko A, Ancukiewicz M, et al. Relative biological effectiveness (RBE) values for proton beam therapy. Int J Radiat Oncol Biol Phys 2002; 53(2):407–421. 79. Chu W, Ludewigt B, Renner T. Instrumentation for treatment of cancer using proton and light-ion beams. Rev Sci Instrum 1993; 64(8):2055–2122. 80. Russell KR, Isacsson U, Saxner M, et al. Implementation of pencil kernel and depth penetration algorithms for treatment planning of proton beams. Phys Med Biol 2000; 45(1):9–27. 81. Hong L, Goitein M, Bucciolini M, et al. A pencil beam algorithm for proton dose calculations. Phys Med Biol 1996; 41(8):1305– 1330. 82. Gall KP, Verhey LJ, Wagner M. Computer-assisted positioning of radiotherapy patients using implanted radiopaque ﬁducials. Med Phys 1993; 20(4):1153–1159. 83. Wells WM 3rd, Viola H, Atsumi S, et al. Multi-modal volume registration by maximization of mutual information. Med Image Anal 1996; 1(1):35–51. 84. van Herk M, Kooy HM. Automatic three-dimensional correlation of CT-CT, CT-MRI, and CT-SPECT using chamfer matching. Med Phys 1994; 21(7):1163–1178. 85. Bussiere MR, Adams JA. Treatment planning for conformal proton radiation therapy. Technol Cancer Res Treat 2003; 2(5):389–399.

Glossary conformity index deﬁned by the RTOG as the volume of the prescription isodose divided by the target volume constraint a condition that must be satisﬁed during inverse treatment planning for a plan to be considered feasible; the most basic constraint is that all beam weights must be nonnegative dose-volume histogram (DVH) a plot of the volume versus dose used to analyze the dose distribution on a structure by structure basis forward treatment planning an iterative approach to planning where the user manually changes each of the plan parameters until an acceptable plan is obtained homogeneity index deﬁned by the RTOG as the maximum dose in the treatment volume divided by the prescription dose inverse treatment planning an automated approach to planning where the user deﬁnes the treatment goals and an optimization algorithm is run that determines the parameters that best meet the goals micro-multileaf collimator (mMLC) a device attached to or incorporated into the head of a linear accelerator used to deﬁne ﬁeld shapes; compared with a conventional multileaf collimator (MLC), the leaves of a mMLC are less wide and project to 5 mm or less in width at the isocenter objective function a scoring function that reduces the entire treatment plan into a single numerical value that is to be either minimized or maximized

8

Designing, Building and Installing a Stereotactic Radiosurgery Unit Lijun Ma and Martin Murphy

Design Principles The ﬁrst stereotactic radiosurgery (SRS) unit was designed by Swedish neurosurgeon Dr. Lars Leksell in the 1950s [1]. The term stereotactic literally means “spatially ﬁxed.” In general, a SRS procedure involves delivering a single fraction of high-dose radiation, usually with the guidance of a rigid ﬁxation device (i.e., a stereotactic frame). The purpose of the frame is to map out the coordinate system of the target for accurate reference of the radiation beams [2–6]. Common types of radiation used for SRS are high-energy gamma rays (e.g., 60Co), high-energy x-rays, and charged particles such as protons. Building a photon-based (x-rays and gamma rays) or an ion-based (protons and heavy charged particles) SRS unit is different in principle. The goal of the photon-based units is to converge a large number of beams to the isocenter in order to produce rapid dose fall-off outside the focal area. In contrast, the ion-based units use a limited number of beams in order to spread the Bragg peak across the target area. The rapid dose fall-off is largely created via the distal fall-off of the Bragg peak [7–12]. Currently, the photon-based units employ radioactive sources (60Co) and electron linear accelerators. The ion-based units employ high-energy cyclotrons or synchrocyclotrons. Due to the high cost of these heavy-particle accelerators, the use of the ion-based SRS units are limited. In this chapter, we discuss the photon-based SRS units only. The major photon-based SRS units include Gamma Knife, conventional linear accelerator (linac)-based units, CyberKnife, and integrated x-ray systems such as Tomotherapy and Varian Trilogy units. Four design principles are used in building a SRS unit: 1. Fix the source of the radiation and ﬁx the patient position when beam is on. 2. Move the source of the radiation and ﬁx the patient position when beam is on. 3. Move the source of the radiation and also move the patient when beam is on. 4. Fix the source of the radiation and move the patient when beam is on. The ﬁrst principle is used in the Gamma Knife and early proton and heavy-ion systems; the second principle is used in

most linac-based SRS systems including the CyberKnife; the third principle is used in Tomotherapy and dynamic arc linacbased systems; and the fourth principle is uncommon but has been reported in the linac-based rotating-chair SRS system [13].

Major Components and Functions Gamma Knife Gamma Knife is a 60Co unit speciﬁcally designed for intracranial SRS treatments 14–17]. It is solely manufactured by the Elekta company (Stockholm, Sweden). The unit contains 201 60 Co sources ﬁxed on a hemispherical surface to deliver gamma rays of 1.25 MeV. The center of the hemisphere is the isocenter where all the beams from the sources intersect. The initial activity of a single source is about 3.0 Ci. Therefore, the total activity from all the sources exceeds 6000 Ci, which generates a starting dose rate of approximately 300 cGy/min at the isocenter. The source-to-isocenter distance is 40.1 cm for the Gamma Knife unit. This distance is signiﬁcantly shorter than the 80- to 100-cm source-to-axis distance (SAD) of the linear accelerators. Short source-to-focal distance is one of the distinct design characteristics of the Gamma Knife unit. This not only reduces the total source activity for high-dose delivery but also facilitates high mechanical precision near the isocenter. For example, if a small milling error occurs in the primary collimator, the error would be projected signiﬁcantly smaller at a short sourceto-isocenter distance than at a large source-to-isocenter distance. High source activity demands high-density materials for proper shielding of the Gamma Knife unit. With the shielding and the collimator assembly, a Gamma Knife unit weighs about 20 tons. Heavy and robust structure, ﬁxed source conﬁguration, and short source-to-focal distance are three major factors contributing to the high mechanical accuracy of the Gamma Knife unit. Historically, there are several models of Gamma Knife units (model U, model B, and model C). In the United States, the ﬁrst Gamma Knife (model U) was installed at the

91

92

l. ma and m. murphy

FIGURE 8-1. The initial model U Gamma Knife installed in the United States.

University of Pittsburgh in 1987. An illustration of the model U unit is given in Figure 8-1. Following U.S. FDA approval of later models (models B and C), the original model U is being gradually replaced. Major differences between the original and the new models include the couch driving mechanism (hydraulic vs. electrical), patient positioning assembly (manual vs. automatic), collimator design, and source patterns. Despite these differences, the structural layout of the Gamma Knife SRS remains identical for all of the models. Figure 8-2 illustrates the major components of the model C Gamma Knife unit. The key components of a Gamma Knife unit include a shielded source head housing 201 60Co sources, a movable couch, four helmets (18 mm, 14 mm, 8 mm, and 4 mm in beam aperture), removable plug collimators (201 for each helmet), console control units, patient positioning devices, audio/video monitoring devices, and treatment planning systems. The patient positioning device of the Gamma Knife consists of a Leksell frame (pinned to the patient skull) and its attachments to the helmets. The frame is locked onto the helmet through a pair of bars (i.e., trunnions) or the automatic positioning system (APS). The details of the APS device attached to a helmet are illustrated in Figure 8-3. The APS device uses

FIGURE 8-3. Detailed illustration of the patient APS with the mounted helmet: a, Leksell frame; b, shielding door to the source; c, APS device (right side); d, helmet (4-mm shown). (Courtesy of Elekta AB, Stockholm, Sweden.)

a micro step motor to position the Leksell frame to precise localization coordinates. The positional accuracy of the APS system is speciﬁed to be less than 0.2 mm. Figure 8-4 shows four plug collimators on each helmet of the Gamma Knife unit (i.e., 18 mm, 14 mm, 8 mm, and 4 mm). The size of these plug collimators approximately equals the full width at half maximum (FWHM) of the beam proﬁles measured along the major axes at the isocenter. Precise alignment of the plug collimator directly affects the precision of the beam. Manufacturing >800 individual plug collimators comprised a major part of the effort in building the Gamma Knife unit. The treatment planning system is a separate component of the Gamma Knife unit. The approved treatment parameters from the treatment planning system are uploaded directly onto the console control computer via a serial cable. The salient features of the system as well as the three-dimensional dose calculation algorithms are discussed in a separate chapter.

Linear Accelerator–Based SRS Units Historically, the linac-based SRS units were built based on the existing linear accelerators using an add-on collimation system and couch stabilization devices [18–29]. The majority of the

FIGURE 8-2. Illustration of the key components of the model C Gamma Knife unit: a, source head unit; b, helmet changer; c, position control pendant; d, docking veriﬁcation pendant; e, helmet, f, patient automatic positioning system (APS) device. (Courtesy of Elekta AB, Stockholm, Sweden.)

FIGURE 8-4. Plug collimators of the Gamma Knife unit. From left to right: 4-mm, 8-mm, 14-mm, and 18-mm collimators.

8.

designing, building and installing a srs unit

93

FIGURE 8-5. A set of tertiary cones for traditional linac-based SRS deliveries. From left to right, the cone diameters are 2 cm, 3 cm, 4 cm, and 5 cm.

linac SRS units still use a set of tertiary cones or add-on micromultileaf collimators to deliver non-coplanar converging arc beams at the isocenter. The arrangement of these non-coplanar arc beams is to spread the beam to form a spherical isodose distribution similar to that of a single-shot delivery of Gamma Knife SRS. Figure 8-5 shows a set of tertiary collimator cones for the linac-based SRS units. The comparison of a 2-cm cone with an 18-mm Gamma Knife plug collimator is shown in Figure 8-6. As illustrated in Figures 8-5 and 8-6, the linac-based SRS cones are signiﬁcantly larger in size than the Gamma Knife collimator for similar beam aperture sizes. The large size of the linac cones is intended for extended source-to-diaphragm distance in order to sharpen the beam penumbra at the isocenter. Unlike Gamma Knife units and early linac systems where the frame isocenter is mechanically ﬁxed at the center of the beam collimators, most modern linac-based SRS units rely on wall-mount lasers and a special localizer box attached to the stereotactic frame for aligning the target coordinates. Therefore, the accuracy of the laser system is critical to the overall treatment accuracy of the linac-based SRS units. As a general rule, a pretreatment isocenter alignment check is carried out for each patient treatment.

FIGURE 8-6. The 18-mm plug collimator of the Gamma Knife unit compared with the 2-cm cone collimator for a linac-based SRS unit.

Figure 8-7 shows the isocenter alignment devices for the linac-based unit and the Gamma Knife unit. In the Gamma Knife alignment tool (Fig. 8-7b), a pin pricks the ﬁlm at the isocenter set by the mechanical coordinates. The ﬁlm is then exposed and the pin-pricked mark is then compared with the center of the beam proﬁle to determine mechanical and radiologic isocenter alignment [4, 21, 26, 30, 31]. For linac-based units, several methods have been reported. The most common method is the Winston-Lutz test method [24, 29]. To carry out Winston-Lutz test, a metallic spherical ball ﬁxed at the tip of a wire is typically used (Fig. 8-7a). The ball is ﬁrst aligned with the isocenter using wall-mount lasers. A series of port ﬁlms are then exposed downstream for a “dry-run” of actual beam deliveries. Typically, the ﬁlm is placed on a special holder attached to the gantry port extended underneath the ball thus allowing its shallow projection onto the ﬁlm. If the linac is equipped with a high-resolution electronic portal imager (EPID) [32], the image can then be taken with the device in

FIGURE 8-7. Isocenter test tools for (a) linac-based SRS units and (b) for the Gamma Knife unit.

94

l. ma and m. murphy

lieu of the ﬁlm. Once exposed, the projected ball center is compared with the ﬁeld edge to detect any deviations from the center of the ﬁeld. The Winston-Lutz test is especially useful for linac-based SRS because it checks the isocenter alignment and also checks potential collisions among the gantry, couch, and tertiary collimator setups during the patient deliveries. Once the alignment between the laser and mechanical isocenter is veriﬁed, the target coordinates for the linac-based SRS treatments are set via a laser-based localizer. A picture of the localizer is given in Figure 8-8. To facilitate alignment of the isocenter, the localizer can be adjusted with 6 degrees of freedom (x, y, z translation and pitch, yaw, and roll) to correct for non-ﬂat couch top and small rotational errors. The coordinates on the localizer are commonly set in positive numbers in order to avoid misreading of positive or negative signs in setting up the coordinates. Common techniques for linac-based SRS use either noncoplanar converging arc beams or dynamic rotation beams. In dynamic rotation delivery, the gantry rotates while the couch rotates (∼150o) at the same time. The advantage of the dynamic arc-beam delivery is that only a single treatment setup is required and no parallel-opposed beam occurs during the entire delivery [26, 27]. With the advent of precise and high-output linear accelerators, the ﬁxed-cone collimator is gradually being replaced by micro-multileaf collimators (mMLCs). In practice, these highresolution multileaf collimators (MLCs) shape the beam aperture conformally in the beam’s-eye view while the gantry rotates for the linac-based SRS delivery. The use of the shaped beams rather than the ﬁxed cones represents a signiﬁcant paradigm shift in the linac-based SRS deliveries as the dose distributions from superposing arcs are no longer elliptical in shape analogous to the Gamma Knife shot delivery. Many investigators classiﬁed the MLC-based SRS as the shaped-beam SRS delivery to distinguish it from the traditional ﬁxed-cone linac-based SRS deliveries [23, 25, 28, 33–35].

Mini-MLC

FIGURE 8-9. A dedicated linear accelerator unit with built-in multileaf collimator for shaped-beam SRS delivery.

For most large tumor treatments, the shaped-beam SRS is capable of delivering the treatments with single isocenter rather than packing multiple spherical shot-like dose distributions inside the target volume. This improves the delivery efﬁciency in terms of total monitor units (MU) as well as the total setup time. With optimized beam weights, the shaped-ﬁeld SRS is expected to deliver more uniform and more conformal dose distribution for large and irregular targets in comparison with the conventional ﬁxed-ﬁeld approach [28]. A picture of a dedicated shaped-beam SRS unit is shown in Figure 8-9. In addition to beam-shaping capability using a built-in MLC or mMLC, a dedicated SRS unit also incorporates the capability of performing online imaging guidance of the delivery. The goal is to register and verify the target locations and to carry out real-time adjustment of the treatment setup. The mechanical accuracy of the unit has been reported to be of the order 0.5 mm [36]. There have been preliminary reports of the use of such dedicated SRS units for functional SRS deliveries such as trigeminal neuralgia treatments [35].

CyberKnife

FIGURE 8-8. A laser-based stereotactic localizer for linac-based SRS treatments.

The CyberKnife [37] is a type of linac-based stereotactic radiosurgery system; however, it differs in two fundamental ways from most conventional linear accelerator systems. First, it does not rely on a stereotactic frame for target localization. Instead, the system employs an integrated X-ray image-guidance system to locate and monitor radiographic landmarks such as the cranium that indicate the target position. Second, the linac is supported and maneuvered by a robotic arm instead of a rotating gantry. This allows the treatment beam to be positioned and aimed with 6 degrees of freedom, which in turn enables the system to adapt to arbitrary patient positions. With this capability, it is not necessary to immobilize the patient at a predeﬁned treatment isocenter. By combining image guidance with complete beam maneuverability, the CyberKnife was the ﬁrst radio-

8.

designing, building and installing a srs unit

surgical system to achieve stereotactic targeting precision without using externally attached devices for either targeting or immobilization. Figure 8-10 shows a typical CyberKnife system. The robotic arm supports an X-band 6-MV linear accelerator. The treatment beam is shaped by cylindrical collimators similar to those used in conventional linac SRS systems. Two orthogonally aligned diagnostic X-ray sources and ﬂat-panel imaging detectors are positioned on either side of the patient. Prior to treatment, the patient undergoes a computed tomography (CT) exam for treatment planning purposes. This CT study also serves as the reference image for the X-ray guidance system. At the time of treatment, the patient lies on the couch in a position that closely but not exactly duplicates his position in the treatment planning CT study. The imaging system acquires a pair of X-ray images of the patient’s anatomy that are compared with a corresponding pair of digitally reconstructed radiographs (DRRs) that have been calculated from the planning CT study. The comparison process establishes the difference between the patient’s pose (position and orientation) in the CT study and his pose at the moment of treatment by matching either bony anatomy or internal radiopaque ﬁducials in the images. Then, rather than use this information to correct the patient position so as to reproduce the planning pose, the position coordinates are automatically sent to the robotic arm, which adjusts the alignment of the treatment beam instead. Thus, the planned beam positions are adapted to the patient position, rather than vice versa. This strategy avoids the need to put the patient at a precisely deﬁned treatment position. Furthermore, it allows the system to adapt to patient movement by continuing to take positioning images throughout treatment. This eliminates the need to rigidly immobilize the patient. The CyberKnife’s frameless alignment capability makes the concept of a ﬁxed mechanical isocenter irrelevant. This carries over to the planning of dose distributions. The treatment beams can be arranged in complex non-isocentric patterns of overlapping rays rather than in the spherically symmetric pattern required by isocentric delivery systems with ﬁxed collimators. This improves the uniformity and conformality of the dose for irregular target volumes.

FIGURE 8-10. The CyberKnife system: 1, diagnostic X-ray sources; 2, 6-MV X-band linac; 3, ﬂat-panel X-ray imaging detectors.

95

Treatment delivery proceeds in a stop, align, and shoot sequence. The total dose for the treatment plan is divided among 80 to 110 linac positions distributed spherically around the patient to obtain the maximum geometric distribution of beams. The robotic arm moves the linac to a designated position (called a node), where it stops while X-ray images are acquired. After automatically processing the images to determine the target position, the linac beam alignment is adjusted and the beam is turned on to deliver a dose increment. The system then advances to the next node and repeats the process. The entire sequence of steps at each node (positioning, image acquisition and analysis, beam realignment, and dose delivery) typically takes 10 to 20 seconds, resulting in a total fraction duration of 20 to 30 minutes. The design, installation, and commissioning of a CyberKnife facility differ in several ways from conventional linear accelerator facilities. For example, the robot can point the beam anywhere in the treatment room. This must be accommodated in the shielding design. Unlike the vault for a gantry-mounted linac, which can concentrate shielding in the plane of gantry rotation, a CyberKnife vault requires a more uniform conﬁguration of shielding. At the same time, though, the beam dwell time in any particular direction is typically less than for a gantry system delivering a plan with the same number of monitor units. This can reduce the required thickness of shielding. Because the robot can move freely in three dimensions, the positions of all physical objects within its workspace (e.g., couch, cameras, etc.) must be programmed into the robot’s workspace model to avoid collisions. If any new object is moved into the workspace without including it in the workspace model, or if an existing object is moved away from its prescribed position, there is risk of a collision. The complete CyberKnife system (robot, linac, imaging system, and couch) is installed as an integrated unit. Because the imaging system is solely responsible for determining beam alignment with the patient, its position calibration with respect to the linac delivery system is critically important. The coalignment of the imaging and robot coordinate frames proceeds in three steps. First, the robot is calibrated to identify a point in its own coordinate frame that is near the center of the imaging coordinate frame. When this step is done, the robot can align the linac beam to within 0.7 mm (RMS variation) of a ﬁxed point from anywhere in its workspace [38]. The second calibration step measures the intrinsic and extrinsic camera models describing the magniﬁcation, distortion, image plane alignment, and other characteristics of the imaging system. This measurement is made with a phantom consisting of a planar grid of radiopaque balls placed in the ﬁeld of view of the two imaging systems. The resulting camera model is used to create the DRRs used for image-guided alignment. Using a well-calibrated camera model, the DRRs can reproduce actual X-ray images with 0.1-mm precision. The last calibration step establishes the position of the imaging coordinate origin within the coordinate frame of the robotic arm. The imaging origin is referred to as a virtual isocenter to distinguish it from the ﬁxed mechanical isocenter of the Gamma Knife or a gantry-based linac SRS system. The virtual isocenter position is ﬁrst estimated using laser and image alignment tools. Then it is measured precisely using a combination dosimetric/imaging phantom. This phantom consists of an

96

l. ma and m. murphy

isocenter within the robot workspace. The mean offset in each direction is used to reﬁne the position of the isocenter in the robot coordinate frame. The random scatter around the mean is then identiﬁed as the residual overall dose alignment error of the system. Through successive reﬁnements in the calibration and analysis process, this residual error has been reduced from 1.7 mm RMS variation [38] to 1.1 mm [39]. Once the system alignment is calibrated, any residual offset (as well as any later perturbation) of the imaging and robot frames appears as a systematic dose offset detectable via the end-to-end phantom calibration process just described. Therefore, it is important to continue these measurements on a regular quality assurance schedule after installation, to detect irregularities in system alignment, and to further reduce any residual systematic offset.

Integrated Units FIGURE 8-11. An imaging/dosimetry phantom used to calibrate and verify the placement of doses within the robotic delivery system frame of reference.

anthropomorphic skull with a cube of radiochromic ﬁlm layers ﬁxed inside (Fig. 8-11). The ﬁlm cube records a threedimensional image of a dose distribution relative to internal imaging landmarks. The dosimetric phantom is scanned in a CT exam and then planned to receive a spherical dose distribution. The phantom is placed on the treatment couch near the center of the imaging system, where it is located via the image-guidance system before receiving the planned dose exposure. Then the phantom is disassembled to measure the center of the actual delivered dose distribution for comparison with the planned position. The deviation is recorded as an offset in (x, y, z). This procedure exactly emulates a treatment from end-to-end and therefore detects all the imprecision in the planning, alignment, and delivery processes. The end-to-end calibration test is repeated a number of times, resulting in a scattered distribution of dose offsets, usually with a nonzero mean. The offset of the mean from zero corresponds with a systematic error in locating the virtual imaging

Several integrated units have recently emerged as SRS delivery units. The purpose of the integrated units is to deliver SRS as well as modern treatments such as adaptive imaging-guided radiation therapy [40, 41]. Two distinct units are the Tomotherapy unit (Tomotherapy, Madison, WI) and the Varian Trilogy Unit (Varian Oncology, Palo Alto, CA). The Tomotherapy unit employs a compact 6-MeV S-band linac mounted on a rotating ring gantry (Fig. 8-12). There are 64 individual binary (either open or shut completely) collimators that shape a fan-beam X-ray across the axial plane. During the delivery, the patient is transported through the bore as the beam rotates around to form a spiral trajectory around the patient. The unit is equipped with a large array of xenon detectors and a kilovoltage (kV) X-ray tube that are capable of performing online CT of the patient. The Trilogy unit is built upon the high-end performance of the Varian Clinac series. In addition to conventional linac ﬂattening ﬁlters, the unit has a separate ﬁlter assembly for delivering enhanced dose rates (>1000 MU/min) to accommodate SRS treatments. The unit is also equipped with either tertiary cones or a built-in multileaf collimator. In addition, the system has a side-mounted kV X-ray tube and a ﬂat-panel imager to provide cone-beam CT imaging capabilities (Fig. 8-12b).

Slip Ring Gantry

Gantry

kV Source Flat panel detector Couch CT Couch

FIGURE 8-12. Two integrated SRS units with online CT capabilities: (a) the Tomotherapy unit; (b) the Trilogy unit. (Part A courtesy of Tomotherapy. Part B courtesy of Varian.)

8.

97

designing, building and installing a srs unit

TABLE 8-1. Estimated overall uncertainties in intracranial SRS deliveries. Source of errors

Imaging studies (resolution and distortions) Mechanical plus setup errors Tissue/target motion Treatment planning Total

Estimated error (mm)

1.0–3.0 0.3–2.0 0.5–1.0 0.5 1.3–3.8

Physically, both the Tomotherapy and the Trilogy systems are suitable for SRS treatments. The Tomotherapy unit uses a fan beam to irradiate the target in the axial plane via slice-byslice fashion; while the Trilogy unit makes use of a cone-beam to irradiate the target from coplanar or nonplanar directions. The mechanical accuracy of these units is estimated to be of the order 0.5 to 1.0 mm. More experiences with these new units for SRS treatments are expected in the future.

Accuracy The treatment accuracy is the distinguishing characteristic of all SRS units. The mechanical accuracy is reported to be less than 0.3 mm for Gamma Knife, 0.5 mm for dedicated SRS units, 0.5 to 0.7 mm for CyberKnife and Tomotherapy units, and 1.0 mm for standard linac-based SRS units. To evaluate the overall physical accuracy of the delivery, other sources of errors also need to be considered. These include (1) uncertainties in the diagnostic imaging studies, (2) dose calculation uncertainties, and (3) target motion and setup errors. In general, the imaging resolutions of CT and magnetic resonance (MR) studies and the spatial distortions in the imaging acquisition process are the common source of uncertainties for all SRS units. This is largely caused by the irregularities in the gradient ﬁelds and the susceptibility from the stereotactic frame in the MR studies for SRS treatments. Table 8-1 summarizes the estimated error levels for the SRS deliveries. The total error shown in Table 8-1 was calculated via quadrature sum of the errors from the individual sources. Overall, the imaging studies and setup variations are the major sources of uncertainties for most SRS deliveries.

Measurements have been carried out to document such dose for both Gamma Knife and linac-based SRS units in the treatment of intracranial lesions [4, 42]. As percentage of the target dose, the Gamma Knife and linac-based units produced ∼0.4% to blood forming organs, ∼0.5% to thyroid, ∼0.04% to gonads, and ∼0.05% to breast or thorax region. Because dmax for the 60Co unit is 0.5 cm and for 4 to 6 MV is in the range of 1.0 to 1.5 cm, the lens dose is expected to be slightly higher for the 60Co units. The lens dose for both units is estimated to ∼ 2.5% of the target dose. Such a percentage could be important for treatments involving large dose deliveries such as trigeminal neuralgia where 70 to 90 Gy may be used. For room shielding purpose, the bunker design depends on numerous factors that include the machine load, how frequently a wall is irradiated, the distance between the wall and the isocenter, and so forth. To ameliorate the requirements for primary irradiations, it is often advantageous to construct a bunker in the basement and direct the primary beam toward the underground earth. Figure 8-13 shows a sample ﬂoor plan hosting a SRS unit. As an illustration, we here give an example calculation for the shielding design of the ﬂoor plan in Figure 8-13. The equation for shielding design calculations is as follows: R = WUT × (d0/d)2 × 10(−t/TVL) × S, where R is the exposure per week, W is the work load (i.e., maximum dose per week), U is the use factor (i.e., fraction of the time that the beam is directed or scattered toward the barrier; U = 1 if scattered radiation is considered), T is the occupancy factor (i.e., fraction of the time a person is present at the point), d0 is the reference dose calibration distance (e.g., d0 = 100 cm for the linac-based SRS unit), t is the thickness of the barrier materials, TVL is the tenth-value layer of the barrier material, and S is the scatter factor (e.g., S = 1 if primary is used, S = 0.001 if scatter radiation is considered). If we need to calculate the required wall thickness near the console operator at the point A, the calculation parameters are determined as follows: W = maximum 5 patients per day × 5 days per week × 5000 cGy per treatment = 125,000 cGy/week; TVL = 13.5 inches for the concrete of the wall; d0 = 1 m and d = 4 m; U = 1; T = 1, and S = 1 for the case; and the exposure limit is 0.002 rad/week. We have

C storage

B

Computer Room

Treatment Room

a

A

CONSOLE

One of the major tasks in building and installing a SRS unit is to design a bunker room and to enforce radiation protection rules. There are two aspects of source shielding and radiation protection for the SRS delivery: (1) reduce extrafocal radiation that contributes to the dose deposition outside of the target volume and (2) reduce radiation exposure to the operators and the general public outside of the treatment room. The radiation to be shielded is generated either from the primary beam directly hitting outside of the target or from the scattered radiation and the leakage radiation from the head of the SRS unit. When considering extrafocal radiation internal to the patient, the concern is the secondary malignancy via carcinogenesis effects, particularly in the treatment of young patients.

SRS Unit D

EARTH BELOW GRADE

Shielding Design

Alley E

FIGURE 8-13. A sample ﬂoor plan for installing a dedicated SRS unit. The designated points (A, B, C, D, E) are selected calculation points for barrier transmissions.

98

l. ma and m. murphy

R = 125,000 × 1 × 1 × (1/4)2 × 10(−t/13.5) × 1 ≤ 0.002. Solving for t of the above equation, we have t ≥ 89 inches. This means that the minimum concrete thickness at the point A should be about 7.5 feet in order to meet the exposure limit of 2 mR/week. It is clear that increasing the source-to-barrier distance or adding high-density shielding materials such as lead in the wall (reducing TVL value) can signiﬁcantly shorten the wall thickness therefore increase the room size. In case of need, the workload can be restricted to maintain the exposure level to a satisfactory level.

Radiation Safety Program Before building or installing a SRS unit, an institutional radiation safety committee should be formed to implement a radiation safety program. The goal of the program is to maintain the radiation exposures to all employees and individual members of the public to be “As Low as Reasonably Achievable” (ALARA). This is commonly referred to as the ALARA principle. To implement the program, a general rule is to monitor the radiation level that is signiﬁcantly below the regulated legal dose limits (e.g., a factor of 10 lower) In general, ﬁlm badges are used to monitor the radiation exposure to all SRS operators and visitors. A radiation safety ofﬁcer (RSO) should be designated for the SRS treatments. The key responsibilities of RSOs are to oversee the radiation exposure reports (e.g., monthly) and to provide regular (e.g., annual) refresher radiation safety training to the authorized users. The RSO needs to be responsive to emergency situations and serve as the contact person between the institution and the federal and the state agencies. For some states in the United States, the RSO is also required to be a qualiﬁed operator of the SRS unit. In case of pregnancy of an authorized user, the RSO needs to determine whether it is appropriate for her to continue to operate the unit. In general, a declared worker may work as long as the total exposure to the embryo or fetus is maintained within the ALARA dose limits.

FIGURE 8-14. Example of calibration phantoms for SRS units: (a) for linac-based SRS units; (b) for Gamma Knife. These phantoms are also used for regular quality assurance measurements or serve as the scat-

Installation and Acceptance Installation and acceptance of a SRS unit requires extensive tests of the functionalities of the unit. Common to all SRS units, three categories of tests should be performed: (1) dose rate calibration and radiation survey; (2) radiation and mechanical isocenter alignment; and (3) imaging acquisition and treatment planning process. For example, the absolute dose rate for a newly installed Gamma Knife should exceed 300 cGy/min with the 18-mm helmet. This is important because this dose rate not only affects the delivery efﬁciency but it also ensures adequate source life and extends the use time before the next reloading is needed. The dose calibration geometry should use tissue-equivalent phantoms of standard shapes as illustrated in Figure 8-14. Another important test is the radiation and the mechanical isocenter alignment. This is performed using the special mechanical alignment tool for the Gamma Knife unit, the Winston-Lutz test method for the linac-based units, and the dosimetry phantom for the CyberKnife, as described above. In general, the smallest ﬁeld size should be measured to ensure maximum accuracy. Example measurement results for a Gamma Knife unit and a linac-based SRS unit are given in Figures 8-15 and 8-16. The most important test in installing a SRS unit is to ensure that the delivered dose proﬁles matched the prescribed ones. As a general practice, the beam proﬁles along all major axes of a single-isocenter delivery should compare with the calculated proﬁles. The agreement should satisfy minimum manufacturer speciﬁcations such as 2% at the central axis and 2 mm in the dose gradient region. For accurate delivery using small collimators (e.g., 4 to 5 mm in diameter), the requirements can be more stringent such as 1% or 1 mm in the dose proﬁles. Such dose proﬁles can be measured using radiochromic ﬁlms with the insert pieces ﬁt into the standard calibration phantoms as illustrated in Figure 8-14. The exposed radiochromic ﬁlms need to be scanned using laser densitometers with resolution of 100 μm or better. An example result for the measured dose proﬁle is given in Figure 8-17.

tering medium for radiation survey tasks. (Part A courtesy of Lucy phantom.)

8.

99

designing, building and installing a srs unit

FIGURE 8-15. Results of radiation and mechanical alignment tests for the new Gamma Knife unit (model C): (a) 4-mm shot; (b) 8-mm shot. The agreement between the pin-pricked center and ﬁeld edge outline was found within 0.25 mm.

FIGURE 8-16. Results of Winston-Lutz test for isocenter alignment checks for a linac-based SRS unit: (a) acquired using electronic portal imager (EPID) with MLC-shaped square ﬁeld; (b) acquired using radiographic ﬁlm for a circular cone. Both ﬁeld shapes can be used for the test.

110 Measurements Reference

RELATIVE DOSE (%)

100 90 80 70 60 50 40 30 20 10

FIGURE 8-17. Measured dose proﬁles compared with the calculated proﬁle in accepting a GKSRS unit. The agreement was found to be within 1% and 1-m requirements for the 4-mm helmet shown here.

0 85

90

95

100 Z-AXIS

105

110

115

100

l. ma and m. murphy TABLE 8-2. Servicing period for major components of the SRS units. Recommended check and servicing frequency

Major components:

Couch motion mechanism (all units) APS (GK) Microswitches (GK) Source activity (GK) Stereotactic frames and ﬁducial boxes (all units) Alarm and interlocks (all) Audio/video system (all) Laser systems (LB, TU, CK) Gantry rotation (LB, TU) Localizing camera systems (LB, CK) Robotic arms (CK)

Annually Annually Monthly Every 7–10 years Weekly Daily Annually Daily Annually Weekly Quarterly

GK, Gamma Knife; LB, Trilogy/linac-based units; CK, CyberKnife; TU, Tomotherapy unit.

FIGURE 8-18. The grid phantom for detecting spatial distortion of the SRS imaging studies.

Besides these major tests, additional acceptance tests may be carried out to satisfy federal and state regulations and institutional protocols for installing a SRS unit. Most of these tests are also part of a quality assurance program in operating the SRS unit. Detailed tasks are recommended in the American Association of Physicists in Medicine (AAPM) task group TG42 report [4]. Example additional tests include timer or monitor chamber linearity tests, radiation survey, wipe tests (for Gamma Knife units), emergency beam-off switch, door interlocks, patient audio-video monitoring and communication circuits, and so forth. One important but often neglected test in accepting a SRS unit is to check the quality and the accuracy of imaging modalities for the treatment planning. To carry out such a task, a MR or CT compatible phantom with precalibrated grid pattern should be scanned (preferably with the ﬁducial localizer box) and imported into the treatment planning system. The procedure aims to validate the image resolution and detect any spatial distortion for the imaging studies. One standard test phantom is shown in Figure 8-18. In general, it is recommended that additional contrast anthropomorphic head phantoms should be scanned for all imaging modalities including CT, MR, angiogram, and so forth, with the acceptance of a SRS unit.

ment tools such as laser systems or patient positioning devices. Because the skull surface measurement can be visually checked against the patient imaging studies, the chance of operation error and malfunction is small. If an error occurs, it can be reversed and corrected without signiﬁcant risk of injuring the patient. However, malfunctions in the laser system and patient positioning device will directly affect the treatment locations and may produce unrecoverable errors. As general practice, regular testing and servicing of high-risk components of SRS units is required (Table 8-2). Most service and maintenance jobs listed above are performed in the treatment room except the source loading and reloading task of the Gamma Knife, because that unit requires manipulation of >1200 Ci of 60Co sources. The radiation exposure is generally higher than the design limit of a treatment room. Typically, a special hot-cell bunker is constructed for carrying out such tasks. The hot-cell should be large enough to accommodate the source head and the source loading unit. An illustration of the source loading unit with an opened unit head is shown in Figure 8-19. In particular, the door to the hot cell should be shielded GK Source Ball

Loading Unit

Manipulators

Service and Maintenance Proper functioning of a SRS unit is only as good as its weakest link. To maintain high-performance quality, routine service and maintenance is important. Before building and installing a SRS unit, it is recommended that a comprehensive probable risk analysis (PRA) be carried out. The goal of PRA is to assign priorities or possible risk factors to individual components, steps, or sequences of SRS unit operations. This aims to eliminate potential causes leading to unrecoverable or disastrous events. For example, a relatively low priority can be assigned to functionality of the skull surface measurement device while a high priority can be given to the functionality of isocenter align-

FIGURE 8-19. Source loading unit with opened unit head for handling high-activity 60Co sources. (Courtesy of Elekta AB, Stockholm, Sweden.)

8.

designing, building and installing a srs unit

TABLE 8-3. Minimum requirements of a qualiﬁed user for performing SRS treatments. Member of the team

Qualiﬁcations

Radiation oncologist

Certiﬁcation (e.g., American Board of Radiology or equivalent) SRS training during or after residency Certiﬁcation (e.g., American Board of Neurological Surgery) SRS training during or after residency Certiﬁcation (e.g., American Board of Radiology or equivalent) SRS physics training and unit operations and quality assurance State licensing and certiﬁcation (e.g., American Registry of Radiological Technologists, ARRT) Unit operation training

Neurosurgeon

Qualiﬁed physicist

Radiation therapist

but allow operation of the robotic manipulators by an operator from the outside. Lead bricks are typically used for this purpose.

Personnel Training and Qualiﬁcations Operation of the SRS units requires qualiﬁed users. The qualiﬁcations and the training of the personnel should follow the practice guideline from the American College of Radiology (ACR). The minimum requirements for the key members of the team are summarized in Table 8-3. All team members should receive basic operation instructions and radiation safety training. The team may also include nursing staff, neuroradiologists, and anesthesiologists in case a young child needs to be treated.

101

with the Winston-Lutz tests, and so forth. The routine QA for SRS deliveries generally include unit functionality, safety interlocks, radiation monitor, and imaging modality functionality and accuracy. In addition to general procedures, qualiﬁed operators must carry out unit-speciﬁc QA tests. For example, Gamma Knife uses radioactive 60Co sources. Therefore, Nuclear Regulatory Commission (NRC) and state regulations mandate speciﬁc tests for Gamma Knife units. One of these tests often includes annual wipe testing of all the helmets to ensure no source leakage occurs during routine operations. A QMP is also required for most SRS deliveries. Major components of such a program include documentation of patient identiﬁcation by at least two methods (name, photo, etc.), as well as records showing that the treatments followed written directives and were completed without deviations. In case of any deviations, the reporting level for misadministration and variance is very strict for SRS. The misadministration of a single-fraction SRS delivery is normally deﬁned as the incidence when the treatment dose differs from the prescribed dose by 10% or more. Because of such stringent criteria, small deviations from the standard protocol are likely to result in reportable misadministrations. Therefore, most SRS programs require qualiﬁed physicists to supervise all QA tasks for the procedure. In-treatment checks such as validation that the target coordinates match the treatment plan must be carried out. Redundant QA checks by other team members are recommended. All qualiﬁed SRS users should receive regular refreshing training. The training should cover radiation safety instructions and reporting requirements for radiation workers as well as QA chart reviews. One important part of training is to refresh emergency procedures such as when to and how to act in case of major equipment failures or medical emergencies.

Cost and Budget Quality Control, Quality Assurance, and Quality Management Quality control is the process to validate whether a SRS unit complies with the design requirements. Once a SRS unit has been built and installed, a series of tests needs to be performed to ensure initial quality of the unit. Quality control processes govern all the manufacturing tests involved in testing the quality of the unit. Because SRS units are medical devices, FDA regulates the manufacturing quality control process. For the unit to be used in human treatments, manufacturers of the SRS units need to submit proper documentation showing good manufacturing practice (GMP) in designing and producing the device. Gamma Knife and linac-based SRS units involve moving components such as gantry, couch, and patient positioning devices. Strict quality assurance (QA) programs should be enforced. The QA program can be separated into routine QA and patient-speciﬁc QA procedures. The routine QA procedures include daily, monthly, and semi-annual/annual QA procedures, user refreshing training, and emergency procedures. Patient-speciﬁc QA procedures include quality management program (QMP) checks, pretreatment checks such as dry-runs

Budgeting SRS units is a complex issue. On appearance, purchasing add-on options to retroﬁt a conventional linear accelerator is appealing because the accelerator was already in use; however, it is a misconception that any conventional linear accelerator can be retroﬁtted for SRS delivery. Along with the fact that isocenter precision is a factor of 2 or 3 higher for the SRS units, the SRS deliveries demand high dose rate and stable output of a large number of monitor units. Wear and tear on the major linear accelerator components such as the klystron, magnetron, and wave guide can be high for such treatments. For gantry-based SRS delivery, the speed of the gantry should be over 5 MU/degree and preferably 10 MU/degree or higher to allow efﬁcient treatment deliveries. Extra cost considerations also include pretreatment QA effort, machine scheduling, patient ﬂow logistics and room design, and so forth. Today, many centers opt to purchase a dedicated SRS unit or integrated SRS unit in order to maximize SRS treatment capabilities. A list of estimated startup costs for various SRS units is summarized in Table 8-4. In general, the acquisition and the ownership cost are higher for the dedicated SRS units than for standard linear accelerators. Aside from operation logistics, patient care and patient

102

l. ma and m. murphy

TABLE 8-4. Total estimated acquisition cost of different SRS units. Units

Estimated cost (US$)

Gamma Knife

$3 million to $4 million (includes source units, planning software, bunker, service contract and training); additional $500,000 to $1.0 million for source reloading with 1–2 weeks down time $2 million to $4 million (includes dedicated accelerator, treatment planning system, bunker, add-on imaging components, service contract and training) $3 million (includes accelerator units, online-CT, treatment planning system, service and training) $3 million (includes X-band accelerator units, bunker, imaging devices, service contract and training)

Dedicated linac-based unit (e.g., Novalis unit)

Integrated unit (e.g. Tomotherapy or Trilogy units) CyberKnife

quality of life should be the driving factor for purchasing an SRS unit. It is evident that the patients are more likely to pursue an SRS option that yields the best possible results while providing the most agreeable treatment experiences.

References 1. Leksell L. The stereotaxic method and radiosurgery of the brain, Acta Chir Scand 1951; 102 (4):316–9. 2. Andrews DW, Scott CB, Sperduto PW, et al. Whole brain radiation therapy with or without stereotactic radiosurgery boost for patients with one to three brain metastases: phase III results of the RTOG 9508 randomised trial. Lancet 2004; 363 (9422): 1665–1672. 3. Corn BW, Curran WJ Jr, Shrieve DC, et al. Stereotactic radiosurgery and radiotherapy: new developments and new directions. Semin Oncol 1997; 24 (6):707–714. 4. Schell MC, Bova FJ, Larson DA, et al. Stereotactic Radiosurgery, Report of the American Association of Physicists in Medicine Task Group No. 42. College Park: American Institute of Physics, 1995. 5. Phillips MH, Stelzer KJ, Grifﬁn TW, et al. Stereotactic radiosurgery: a review and comparison of methods. J Clin Oncol 1994; 12(5):1085–1099. 6. Loefﬂer JS, Shrieve DC, Wen PY, et al. Radiosurgery for intracranial malignancies. Semin Radiat Oncol 1995; 5(3):225–234. 7. Harsh GR, Thornton AF, Chapman PH, et al. Proton beam stereotactic radiosurgery of vestibular schwannomas. Int J Radiat Oncol Biol Phys 2002; 54(1):35–44. 8. Larsson B, Leksell L, Rexed B, et al. The high-energy proton beam as a neurosurgical tool. Nature 1958; 182(4644):1222–1223. 9. Larsson B, Sarby B. Equipment for radiation surgery using narrow 185 MeV proton beams. Dosimetry and design. Acta Oncol 1987; 26(2):143–158. 10. Lawrence JH, Tobias CA, Linfoot JA, et al. Heavy particles and the Bragg peak in therapy. Ann Intern Med 1965; 62:400–407. 11. Levy RP, Fabrikant JI, Frankel KA, et al. Heavy-charged-particle radiosurgery of the pituitary gland: clinical results of 840 patients. Stereotact Funct Neurosurg 1991; 57(1–2):22–35. 12. Weber DC, Chan AW, Bussiere MR, et al. Proton beam radiosurgery for vestibular schwannoma: tumor control and cranial nerve toxicity. Neurosurgery 2003; 53(3):577–586; discussion 586– 588.

13. McGinley PH, Butker EK, Crocker IR, et al. A patient rotator for stereotactic radiosurgery. Phys Med Biol 1990; 35(5):649–657. 14. Leksell DG. Stereotactic radiosurgery. Present status and future trends. Neurol Res 1987; 9(2):60–68. 15. Lindquist C. Gamma Knife radiosurgery. Semin Radiat Oncol 1995; 5:197–202. 16. Maitz AH, Wu A, Lunsford LD, et al. Quality assurance for gamma knife stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 1995; 32(5):1465–1471. 17. Wu A, Lindner G, Maitz H, et al. Physics of gamma knife approach on convergent beams in stereotactic radiosurgery. Int J Radiati Oncol Biol Phys 1990; 18:941–949. 18. Bourland JD, McCollough KP. Static ﬁeld conformal stereotactic radiosurgery: physical techniques. Int J Radiat Oncol Biol Phys 1994; 28(2):471–479. 19. Bova FJ, Friedman WA, Mendenhall WM. Stereotactic radiosurgery. Med Prog Technol 1992; 18(4):239–251. 20. Colombo F, Benedetti A, Pozza F, et al. Stereotactic radiosurgery utilizing a linear accelerator. Appl Neurophysiol 1985; 48(1–6):133– 145. 21. Falco T, Lachaine M, Poffenbarger B, et al. Setup veriﬁcation in linac-based radiosurgery. Med Phys 1999; 26(9):1972–1978. 22. Friedman WA, Bova FJ, Spiegelmann R. Linear accelerator radiosurgery at the University of Florida. Neurosurg Clin N Am 1992; 3(1):141–166. 23. Leavitt DD, Watson G, Tobler M, et al. Intensity-modulated radiosurgery/radiotherapy using a micromultileaf collimator. Med Dosim 2001; 26(2):143–150. 24. Lutz W, Winston KR, Maleki N. A system for stereotactic radiosurgery with a linear accelerator. Int J Radiat Oncol Biol Phys 1988; 14(2):373–381. 25. Nedzi LA, Kooy HM, Alexander E 3rd, et al. Dynamic ﬁeld shaping for stereotactic radiosurgery: a modeling study. Int J Radiat Oncol Biol Phys 1993; 25(5):859–869. 26. Podgorsak EB, Olivier A, Pla M, J. Hazel, et al. Physical aspects of dynamic stereotactic radiosurgery. Appl Neurophysiol 1987; 50(1–6):263–268. 27. Podgorsak EB, Olivier A, Pla M, et al. Dynamic stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 1988; 14(1):115–126. 28. Solberg TD, Boedeker KL, Fogg R, et al. Dynamic arc radiosurgery ﬁeld shaping: a comparison with static ﬁeld conformal and noncoplanar circular arcs. Int J Radiat Oncol Biol Phys 2001; 49(5):1481–1491. 29. Winston KR, Lutz W. Linear accelerator as a neurosurgical tool for stereotactic radiosurgery. Neurosurgery 1988; 22(3):454–464. 30. Colombo F, Francescon P, Cora S, et al. A simple method to verify in vivo the accuracy of target coordinates in linear accelerator radiosurgery. Int J Radiat Oncol Biol Phys 1998; 41(4):951–954. 31. Gibbs FA Jr, Buechler D, Leavitt DD, et al. Measurement of mechanical accuracy of isocenter in conventional linearaccelerator-based radiosurgery. Int J Radiat Oncol Biol Phys 1993; 25(1):117–122. 32. Boyer AL, Antonuk L, Fenster A, et al. A review of electronic portal imaging devices (EPIDs). Med Phys 1992; 19(1):1–16. 33. Leavitt DD, Gibbs FA Jr, Heilbrun MP. Dynamic ﬁeld shaping to optimize stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 1991; 21(5):1247–1255. 34. Pedroso AG, De Salles AA, Tajik K. Novalis shaped beam radiosurgery of arteriovenous malformations. J Neurosurg 2004; 101(Suppl 3):425–434. 35. Smith ZA, De Salles AA, Frighetto L. Dedicated linear accelerator radiosurgery for the treatment of trigeminal neuralgia. J Neurosurg 2003; 99(3):511–516. 36. Rahimian J, Chen JC, Rao AA. Geometrical accuracy of the Novalis stereotactic radiosurgery system for trigeminal neuralgia. J Neurosurg 2004; 101(Suppl 3):351–355.

8.

designing, building and installing a srs unit

37. Adler JR, Murphy MJ, Chang SD, et al. Image-guided robotic radiosurgery. Neurosurgery 1999; 44:299–306. 38. Murphy MJ, Cox RS. Dose localization accuracy for an imageguided frameless radiosurgery system. Med Phys 1996; 23(12):2043– 2049. 39. Chang SD, Main W, Martin DP, et al. An analysis of the accuracy of the CyberKnife: a robotic frameless stereotactic radiosurgery system. Neurosurgery 2003; 52:140–147.

103

40. Mackie TR, Balog J, Ruchala K, et al. Tomotherapy. Semin Radiat Oncol 1999; 9(1):108–117. 41. Mackie TR, Holmes T, Swerdloff S, et al. Tomotherapy: a new concept for the delivery of dynamic conformal radiotherapy. Med Phys 1993; 20(6):1709–1719. 42. Berk HW, Larner JM, Spaulding C, et al. Extracranial absorbed doses with Gamma Knife radiosurgery. Stereotact Funct Neurosurg 1993; 61(Suppl 1):164–172.

PA R T III

Stereotactic Radiosurgery Techniques

105

9

Gamma Knife Radiosurgery Ajay Niranjan, Sait Sirin, John C. Flickinger, Ann Maitz, Douglas Kondziolka, and L. Dade Lunsford

Historical Review Professor Lars Leksell ﬁrst coupled an orthovoltage X-ray tube with his ﬁrst-generation guiding device to focus radiation on the Gasserian ganglion to treat facial pain. He subsequently investigated cross-ﬁred protons as well as X-rays from an early-generation linear accelerator (linac) for radiosurgery. In the 1960s, he became dissatisﬁed with the cumbersome nature of crossﬁred proton beams and the poor reliability and wobble of thenexisting linear accelerators. Leksell and Larsson ﬁnally selected cobalt-60 as the ideal photon radiation source and developed the Gamma Knife [1, 2]. They placed 179 60Co sources in a hemispherical array so that all gamma rays (radiation from the decay of 60Co) focused on a single point thereby creating cumulative radiation isocenters of variable volume depending on the beam diameter. The ﬁrst Gamma Knife created a discoidshaped lesion suitable for neurosurgical treatment of movement disorders and intractable pain management. Clinical work with the Gamma Knife began in 1967 at the manufacturing site, the Motala AB workshop near Linköping, Sweden. The ﬁrst patient had a craniopharyngioma. The patient’s head was immobilized using a plaster-molded headpiece. In 1975, a series of surgical pioneers at the Karolinska Hospital, Stockholm, began to use a reengineered Gamma Knife (spheroidal lesion) for the treatment of intracranial tumors and vascular malformations. Units 3 and 4 were placed in Buenos Aires and Shefﬁeld England in the early 1980s. Lunsford et al. introduced the ﬁrst clinical 201-source Gamma Knife unit to North America (the ﬁfth gamma unit worldwide). Lunsford ﬁrst performed Gamma Knife radiosurgery in August 1987 at University of Pittsburgh Medical Center. In the United States, based on the available published literature, arteriovenous malformations (AVMs) and skull base tumors that failed other treatments were considered the initial indications for radiosurgery. A cautious approach was adopted while waiting for increased scientiﬁc documentation. The encouraging results of radiosurgery for benign tumors and vascular malformations led to an exponential rise of radiosurgery cases and sales of radiosurgical units (Tables 9-1, 9-2, and 9-3). In recent years, metastatic brain tumors have become the most common indica-

tion for radiosurgery. Brain metastases now comprise 30% to 50% of radiosurgery cases at busy centers.

The Evolution of Gamma Knife: Models A, B, and C The Gamma Knife has evolved steadily since 1967. Three commercially produced models are now used worldwide. In the ﬁrst models (model U or A), 201 cobalt sources were arranged in a hemispherical array. These units present challenging 60Co loading and reloading issues. To eliminate this problem, the unit was redesigned so that sources were arranged in a circular (O-ring) conﬁguration (models B, C, and 4-C) (Fig. 9-1). Gamma Knife radiosurgery usually involves multiple isocenters of different beam diameters to achieve a treatment plan that conforms to the irregular three-dimensional volumes of most lesions. The total number of isocenters may vary depending upon the size, shape, and location of the target. Each isocenter has a set of three x, y, z stereotactic coordinates corresponding with its location in three-dimensional space as deﬁned using a rigidly ﬁxed skull stereotactic frame. In terms of actual dose delivery, this means several changes in the patient’s head position within the helmet. In 1999, the model C Gamma Knife was introduced. The ﬁrst model C in the United States was installed at the University of Pittsburgh Medical Center in March 2000. This technology combines advances in dose planning with robotic engineering and uses a submillimeter accuracy automatic positioning system (APS). This technology obviates the need to manually adjust each set of coordinates in a multiple-isocenter plan. The robotic positioning system (Fig. 9-2) moves the patient’s head to the target coordinates deﬁned in the treatment plan. The robot eliminates the time spent removing the patient from the helmet, setting the new coordinates for each isocenter, and repositioning the patient in the helmet. This has signiﬁcantly reduced the total time spent to complete the treatment. Because the treatment time is shortened, a precise three-dimensional (3D) plan can be generated using multiple smaller beams achieving volumetric conformality. Such an

107

108

a. niranjan et al.

TABLE 9-1. Numbers of active Gamma Knife units worldwide. Continent

Country

Asia

China Japan India Korea Philippines Singapore Taiwan Thailand Canada Mexico United States Austria Belgium Croatia Czech Republic Egypt France Germany Greece Iran Italy Jordan The Netherlands Norway Romania Spain Sweden Switzerland Turkey United Kingdom Argentina Brazil

North America

Europe and Middle East

Latin America

Total units

Active units

17 47 3 7 1 1 6 1 3 2 90 2 1 1 1 1 2 5 1 1 4 1 1 1 1 1 2 1 2 3 1 1

TABLE 9-2. Brain disorders treated worldwide using Gamma Knife radiosurgery by December 2004. Brain disorder

Indications

Vascular disorders

AVM

Benign tumors

Malignant tumors

Functional targets

Ocular disorders

212

approach results in a steeper dose fall-off extending beyond the target (higher selectivity). The other features of the model C unit include an integral helmet changer, dedicated helmet installation trolleys, and color-coded collimators. In 2005, the fourth-generation Leksell Gamma Knife, model 4-C, was introduced. The ﬁrst unit was installed at the University of

Number of patients treated

39,847

Aneurysm Cavernous malformation Other vascular Vestibular schwannoma Meningioma Pituitary adenoma Pineal region tumor Craniopharyngioma Hemangioblastoma Trigeminal schwannoma Chordoma Other benign tumors Glial tumors (grade I–II) Glial tumors (grade III–IV) Metastatic tumor Chondrosarcoma Nasopharyngeal carcinoma Other malignant tumors Trigeminal neuralgia Parkinson disease Pain Epilepsy Obsessive compulsive disorder Other functional targets Uveal melanoma Glaucoma Other ocular disorders

177 437 3,328 28,306 36,602 24,604 2,619 2,748 1,296 1,781 1,336 4,408 505 20,614 100,098 273 1,087 4,070 17,799 1,208 491 1,879 117 646 1,062 158 33

Total indications

297,529

Pittsburgh in January 2005. The model 4-C is equipped with enhancements designed to improve workﬂow, increase accuracy, and provide integrated imaging capabilities. The integrated imaging, powered by Leksell GammaPlan, offers the ability to fuse images from multiple sources. These images can also be exported to a CD-ROM, so the referring physician can

TABLE 9-3. Peer-reviewed publications on the outcome of Gamma Knife radiosurgery. Disease category

Diagnosis

Vascular malformation

Arteriovenous malformation Cavernous malformation Acoustic neuroma Meningioma Pituitary adenoma Metastases Glial tumors Craniopharyngioma Non–acoustic schwannoma Glomus tumor Pineal tumor Hemangioma Hemangioblastoma Trigeminal neuralgia Movement disorders Epilepsy Obsessive compulsive disorder

Benign tumors

Malignant tumors Other tumors

Functional disorders

Number of Gamma Knife publications

Years of experience

Multi-institutional trials

Randomized controlled trial

85 14 124 60 49 130 46 19 22 13 7 11 13 75 37 48 5

1989–2005 1995–2005 1969–2005 1991–2005 1993–2005 1990–2005 1992–2005 1994–2004 1993–2004 1997–2005 1990–2005 1999–2005 1996–2004 1991–2005 1991–2005 1991–2005 1991–2005

1 0 0 1 0 1 0 0 0 0 0 0 0 2 0 4 0

1 0 0 0 0 3 1 0 0 0 0 0 0 1 0 0 0

9.

109

gamma knife radiosurgery

Intergrated Shielding Plastic Automatic Beam LCD screen doors cover Positioning SystemTM channel

Cobalt-60 Shielding sources

Helmet in treatment position

Protection panels

FIGURE 9-1. Schematic diagram of model 4-C Gamma Knife unit. (Courtesy of Elekta AB, Stockholm, Sweden.)

receive pre- or postoperative images for reference and followup. The planning information can be viewed on both sides of the treatment couch. The helmet changer and robotic APS are faster and reduce total treatment time.

The Radiosurgery Procedure Following are the basic steps of Gamma Knife radiosurgery: 1. Daily quality assurance of the radiosurgery system. 2. Application of the stereotactic guiding device to the patient’s head. 3. Stereotactic brain imaging using magnetic resonance imaging (MRI), computed tomography (CT), and/or an angiogram. 4. Quality assurance of images. 5. Determination of target volume(s). 6. Conformal radiosurgery dose planning by the radiosurgery team. 7. Stereotactic delivery of radiation to the target volume by positioning the patient’s head inside a collimator helmet (Gamma Knife), or on treatment couch (linac-based systems). 8. Removal of stereotactic guiding device.

Daily Quality Assurance of the Radiosurgery System FIGURE 9-2. Automated positioning system (APS). APS is a robotic device that positions the patient’s head at planned target coordinates.

Gamma Knife quality assurance testing is performed by an authorized medical physicist every morning. The medical

110

a. niranjan et al.

physicist ensures that the system tests required by U.S. Nuclear Regulatory Commission (NRC) regulations are performed and functioning properly. These tests include the permanently mounted radiation monitor and its remote indicator, handheld radiation monitor, patient viewing and communication systems, door interlock, timer termination of exposure, emergency stops, beam status indicators, availability of the release rods for the emergency removal of a patient, test run of the automatic patient positioning unit, microswitch test, and function of the helmet hoist. Apart from daily quality assurance tests, monthly and annual quality assurance is also performed in addition to preventative maintenance of the Gamma Knife unit.

Application of the Stereotactic Guiding Device For Gamma Knife radiosurgery, appropriate stereotactic frame placement is the initial critical aspects of the procedure. Prior to frame placement, the radiosurgery team should review the preoperative images and discuss optimal frame placement strategy. The preoperative images should be kept in plain view while applying the head frame. An effort should be made to keep the lesion as close to the center of the frame as possible. The possibility of collision by the frame base ring, the posts/pins assembly, or the patient’s head with the collimator helmet during treatment should also be considered prior to the frame application. Steps to avoid the possible collision should be taken during frame placement. The authors use the following strategies for optimal frame placement. To target lower lesions (skull base tumors, acoustic tumors, cerebellar lesions), the frame is positioned lower by placing the ear bars in the top holes of the earpieces on the Leksell G frame. For higher lesions, (sagittal sinus meningioma, metastases high in the frontal or parietal lobe), the frame is positioned higher by placing the ear bars in the bottom holes of the earpieces. For anterior targets, (anterior frontal lobe tumors, cavernous sinus tumors, sellar lesion and lesion anterior to sella, and anterior temporal lesions), the frame is shifted forward by placing earpieces posteriorly on the base ring of the frame. The posterior edge of the earpiece is kept at 90 to 75 mm on the ydimension of the head frame (instead of 95 to 100 mm) depending upon the shift that is needed to bring the lesion closest to the center of the frame. For anterior target, short posterior posts are preferred to avoid collision of the posterior post/pin assembly with the collimator helmet. To target posterior lesions (occipital lobe tumors, transverse sinus tumors, cerebellar lesions), the frame is shifted backward by positioning the earpieces forward. The posterior edge of the earpiece is kept at 110 to 125 mm instead of 95 to 100 mm. The anterior posts are positioned as low as possible on the supraorbital region to avoid collision of the frontal post/pin assembly with the collimator helmet. For radiosurgery planning, a higher gamma angle (120° to 140°) is used if a potential collision is detected at the default angle of 90°. To reach lateral targets (lateral metastases, convexity tumors, far lateral tumors), the frame is shifted laterally (right or left) toward the lesion. While shifting the frame laterally, it is important to make sure that there is enough space on the contralateral side to allow positioning of the ﬁducial box on the base ring of the frame. The MRI or CT ﬁducial should be tried on the frame prior to sending the patient to the MRI unit. If the ﬁducial box does not ﬁt on to the frame due to excessive shifting of the frame, the frame will have to be repositioned.

Stereotactic Brain Imaging Techniques Aside from the frame application, the next most important aspect of radiosurgery is accurate imaging of the target. MRI is the preferred imaging modality. CT is used when MRI is not possible. Angiograms are used in conjunction with MRI for AVM radiosurgery.

Stereotactic MRI The highlights of stereotactic imaging include optimal contrast between normal and abnormal tissues in addition to high spatial resolution, short scan time, and thin slices so that accurate target localization can be achieved. The use of MRI in stereotactic planning has enhanced accurate targeting of lesions that are usually not adequately deﬁned by any other imaging modality. Some physicians prefer the fusion of magnetic resonance and CT images for stereotactic guidance, as they believe that in certain type of scanners, distortion may affect the accuracy of target localization in MRI. For the initial 2 years, the authors used both MRI and CT for stereotactic planning. Signiﬁcant target coordinate differences were not observed using the Leksell stereotactic system. Since 1993, MRI has been used for stereotactic radiosurgery planning in almost all eligible cases using a 1.5-tesla unit. In addition, arteriovenous malformations are imaged also by biplane angiography. At our institution, high-resolution, gadolinium-enhanced 3D localizer (T2* images) image sequence is used ﬁrst to localize the tumor in axial, sagittal, and coronal images. This sequence (3-mm-thick slices 2-mm apart) only takes 45 seconds for 11 axial, 11 sagittal, and 11 coronal slices. Using the axial images, the ﬁducials can be measured and compared with the opposite side to exclude the possibility of magnetic resonance (MR) artifacts and to conﬁrm that there is no angulation or head tilt. Alternatively, T1-weighted sagittal scout images (3mm-thick slices with 1 mm) using spin echo pulse sequence can be obtained for lesion localization. The average time for this sequence is approximately 1.5 minutes. For stereotactic imaging of most lesions, a 3D-volume acquisition using fast spoiledGRASS (gradient recalled acquisition in steady state) sequence at 512 × 256 matrix and 2 NEX (number of excitations) covering the entire lesion and surrounding critical structures is preferred. To deﬁne the radiosurgery target, this volume is displayed as 1- or 1.5-mm-thick axial slices. The ﬁeld of view (FOV) is kept at 25 cm × 25 cm in order to visualize all ﬁducials. The approximate imaging time for this sequence is 6 to 8 minutes. We generally prefer 3D spoiled-GRASS sequence for most lesions. Additional sequences are performed when more information is needed. Pituitary lesions are particularly difﬁcult to image especially if there has been prior surgery. A half dose of paramagnetic contrast is usually given to image pituitary adenomas. For residual pituitary tumors, after trans-sphenoid resection, a fat-suppression SPGR sequence is recommended in order to differentiate tumor from the fat packed in the resection cavity. For cavernous malformations, an additional variable echo multi planar (VEMP) imaging is obtained to deﬁne the hemosiderin rim. For thalamotomy planning, an additional fast inversion recovery sequence is performed to differentiate basal ganglia from white-matter tracts. Brain metastases patients receive a double dose of contrast agent, and the entire brain is imaged by 2 mm slices to identify all of the lesions. Before

9.

gamma knife radiosurgery

removing the patients from the MR scanner, the images must be checked for accuracy.

Stereotactic CT Imaging When using CT imaging instead of MRI, it is advisable to use short posterior posts to avoid artifacts from the posts and pins. Care should also be taken in deciding the optimal place for the pins because they cause artifacts on CT. An effort should be made to keep the lesion away from the pin artifacts. With modern CT scanners, 1- or 2-mm-thick slices (depending upon the size of the lesion) without any gap can be obtained in 4 to 5 minutes. Before removing the patients from the CT table, accuracy checks are performed to make sure that images would be accepted by the planning system and the lesions have coordinates that are achievable in the Gamma Knife unit.

Stereotactic Angiography Angiography is the gold standard for AVM radiosurgery planning. It should be used in conjunction with MR or CT imaging to provide the third imaging dimension. The technique of angiography differs slightly from the conventional digital angiography as the stereotactic angiographic images are used not only for AVM nidus deﬁnition but also to guide radiation to the target. The orthogonal images (instead of oblique or rotated) are preferred but are not necessary. For AVM nidi that are not properly visualized in orthogonal planes, a rotation of up to 10° in two dimensions or aspects can be used without compromising the accuracy of radiation delivery. Before removing the angiography catheter, the images should be reviewed to make sure that all the ﬁducials are seen on the images. Digital subtraction techniques, despite a potential radial distortion error, have proved satisfactory.

Quality Assurance of Images Regular quality control checks of the MR unit are performed in order to maintain accuracy of images. With a properly shimmed magnet, regular servicing, and strict quality assurance on the unit as well as on the images, MRI provides high-resolution images with accurate target localization. A special frame holder is used in order to avoid head movement during MRI. Accuracy of images is checked for each image sequence by comparing the known frame measurement with image measurements in addition to the distance from the posterior ﬁducials to the middle ﬁducials. The images are exported from Imaging Suite via the Ethernet. Images are deﬁned using Leksell GammaPlan software (LGP) after the images are transferred to the LGP computer. The measurements are again checked and compared with the known frame measurement and also the distance to the middle ﬁducial in order to check for any distortion during image transfer.

111

and outlining by surgeon or oncologist becomes an important step. The surgeon’s input is required to deﬁne radiosurgery targets for patients with AVMs, tumors, and functional neurosurgery as used by some centers. Deﬁning the target volume also helps in calculating conformality index.

Techniques of Conformal Dose Planning In the process of treatment planning, several strategies can be used. The model C allows treatment using robotic automatic patient positioning system (APS mode), manual positioning (trunnion mode), or mixed treatment (some isocenters in APS mode and some in trunnion mode). A different approach would be used if only a trunnion treatment plan was possible versus an APS treatment plan. Universally, in LGP one can start planning from the middle of the target and then move to the top and bottom of the target. Another approach is to start at the bottom or top and build from the starting point. Some surgeons prefer to outline the target volume before planning the treatment volume. Beginners can also use the inverse dose-planning algorithm (Wizard) to create a plan and then optimize it manually. When planning a treatment using trunnions only, one might tend to use larger collimators (especially for larger lesions) to reduce the time and maximize coverage of the target. For example, for a medium-size acoustic tumor, in trunnion mode, one might use a few 14-mm collimators for the majority of the tumor and a few 4-mm collimators for the intracanalicular portion of the tumor. In APS mode, however, one would most likely use multiple 8-mm isocenters for the majority of the tumor and 4-mm isocenters for the intracanalicular portion because the total time spent would be less. There would be no need to go into the treatment room to set each isocenter. As long as the isocenters are in close proximity to one another, the software would automatically put them into the same treatment run and the patient would move from one set of coordinates to the next until all isocenters of one collimator size were treated. The conformal dose planning is enhanced by the use of multiple small collimators. There are other tricks to use in planning; for example, using a steep (125°) gamma angle for posterior lesions (cerebellar or occipital) to avoid frame collisions. Another technique available for single isocenters lesions is to match the gamma angle to the angle of the target. In APS mode, during the set-up phase of planning, the idea is to try to group as many isocenters in the same run as possible, even if it means changing all of the isocenters to high docking or a different gamma angle than the default of 90°. If a different gamma angle is used, the plan must be rechecked for accuracy and adjusted if necessary. In the current version of LGP, multiple targets (multiple tumors) can be treated using different isodose prescription and different central doses with the use of multiple matrices.

Determination of Target Volume(s)

Techniques of Stereotactic Radiation Delivery Using Gamma Knife

Target determination is an important step in order to make a conformal plan. Target volume can be outlined using the LGP software (manual or semiautomatic mode). Experienced surgeons can create conformal dose plans without outlining the target; however, for new centers especially where physicists assume the initial responsibility for planning, target deﬁnition

The model C Gamma Knife allows radiation delivery using trunnion mode (manual patient positioning) or APS mode (robotic positioning) or a combination of the two (mixed treatment). In trunnion treatment, the x, y, and z of each isocenter are set manually and double-checked to avoid errors. The same coordinates and the time obtained from LGP are entered into

112

a. niranjan et al.

the control console, and radiation is administered. The APS plan is transferred directly from planning computer to control computer. The operator selects the run (a combination of isocenters of same beam diameter) that matches the collimator helmet on the Gamma Knife unit. The APS is moved to the dock position, and the patient’s head frame is ﬁxed into the APS. The accuracy of the docking position is checked. The system prompts the user to perform clearance checks ﬁrst for all those planned isocenters in which the pins, posts, frame, or patient’s head would be less than 12 mm away from the inner surface of the collimator helmet (even though they may not match with the collimator size that is being used for ﬁrst run). The clearance check is performed by moving the patient to those positions under APS manual control and by checking the risk of collision with the collimator helmet. After the clearance check, the system prompts the surgeon to carry out position checks. In the position checks, all the isocenters using the same helmet are checked, one by one, by moving the patient’s head to these positions using APS manual control. The team moves out after the position checks are completed, and the radiosurgical dose is administered. The APS moves the patient to all the planned positions, one by one, until the isocenters using that size of collimator helmet are completed. The team monitors the patient and the coordinates of different isocenters on the control computer. If other runs using a different gamma angle but using the same helmet are planned, then the patient is taken out, the next run is selected, APS is moved to the dock position, and the patient’s head is again ﬁxed in the APS using the planned angle (72°, 90°, 110°, or 125°). Position checks are performed, and procedure commences is begun. Similarly, if additional runs using different helmets are planned, the helmet is changed, the patient’s head is positioned in the APS, and the position checks for all the isocenters for that helmet and gamma angle are performed.

Long-Term Outcome After Gamma Knife Radiosurgery Radiosurgery for Brain Vascular Malformations AVM Radiosurgery Untreated, patients with AVMs have a 2% to 4% annual risk of hemorrhage and a signiﬁcant lifetime risk of death if hemorrhage occurs. Approximately 40% to 50% of patients with untreated AVMs will experience some physical deterioration in their working capacity during a 20- to 40-year period after presentation. The goals of radiosurgery are to achieve complete AVM obliteration, to improve symptoms, and to preserve existing neurologic function. Obliteration is a process resulting from endothelial proliferation within the AVM blood vessel walls, supplemented by myoﬁbroblast proliferation. This leads to contraction and eventual obliteration of the AVM blood vessel lumens. The process is cumulative, with earliest obliterations noted within 2 to 3 months, 50% of the effect often seen within 1 year, 80% within 2 years, and 90% within 3 years. Radiosurgery is an effective primary management strategy for patients with small to moderate size (less than 10 cm3)

AVMs (Fig. 9-3). Current AVM radiosurgical studies report obliteration rates of 50% to 95% after radiosurgery (Table 9-4) [3–12]. Dose-volume guidelines for AVM management have been published [5]. AVM outcomes are best for those patients with small-volume AVMs located in noncritical locations. Children may respond faster than adults in terms of the obliteration rate. Within 3 years, radiosurgery offers the potential advantage of complete AVM obliteration in 80% to 95% of small AVMs. Small size and noncritical location predict good outcomes with either AVM resection or radiosurgery. Pollock et al. reported clinical and angiographic variables that affect the results of AVM radiosurgery [13]. When 220 patients were subjected to a multivariate analysis with patient outcomes as the dependent variable, four factors were associated with successful AVM radiosurgery: smaller AVM volume, fewer draining veins, younger patient age, and hemispheric AVM location. Preradiosurgical embolization was a negative predictor of successful AVM radiosurgery. These investigators concluded that AVM obliteration without new neurologic deﬁcits can be achieved in at least 80% of patients with small-volume, hemispheric AVMs after single-session AVM radiosurgery [13]. Multimodality strategies employing combinations of embolization, microsurgery, and radiosurgery may allow greater numbers of patients with AVMs to be successfully treated. Multimodality management is especially needed for large AVMs. Embolization is used as an adjunct to radiosurgery at some centers. The purpose of embolizing large AVMs prior to radiosurgery is to permanently decrease the volume of the AVM. With a smaller AVM, a higher and more effective radiation dose can be administered. We have not found embolization overly effective for radiosurgery. There are several reasons for this. Embolization can only be an effective adjunct to radiosurgery if it results in permanent reduction of a deﬁnitive nidus volume. Reduction in ﬂow or obliteration of a small part within the overall AVM does not help reduce the subsequent radiosurgery volume. Partial embolization is often helpful prior to microsurgery and facilitates bloodless surgery. For large AVMs (more than 15 cm3) where an effective radiosurgery dose cannot be given for fear of causing radiation-related complications, we recommend prospectively staged (volumetric) radiosurgery instead of adjuvant embolization [14]. In staged radiosurgery, a component of AVM nidus is treated using an effective radiosurgery dose, which cannot be prescribed to a larger volume. The remaining volume undergoes second-stage radiosurgery at 3 to 6 months. Staged radiosurgery may allow for the elimination of larger AVMs over a period of 4 to 5 years. In general, most reports indicate that patients remain at risk for hemorrhage during the latency interval until AVM obliteration is complete [15, 16]; however, the long-term results of radiosurgery (5- to 14-year results after Gamma Knife radiosurgery) suggest that the majority of AVM patients (73%) have reduced risk of future hemorrhage after 2 years. If at the end of 3 years residual AVM is identiﬁed by imaging, repeat radiosurgery may be considered (as may other management strategies designed to complete obliteration of the AVM) [17].

Cavernous Malformation Radiosurgery Cavernous malformation radiosurgery has provided a therapeutic option for patients with symptomatic, hemorrhagic

9.

113

gamma knife radiosurgery

FIGURE 9-3. (a) Posterior-anterior (top left) and lateral (top right) carotid angiograms showing large left parietal AVM at the time of Gamma Knife radiosurgery. Axial MR poster (bottom) showing the GammaPlan with 50% isodose line covering AVM nidus. (b) Axial contrast-enhanced T1-weighted MR image (top left) and T2-weighted

image (top right) of the same patient 3 years after radiosurgery showing some radiation-related changes and absence of ﬂow voids signal suggesting nidus obliteration. Posterior-anterior (bottom left) and lateral (bottom right) carotid angiograms showing complete nidus obliteration after 3 years.

malformations in high-risk brain locations not amenable to microsurgery [18, 19]. Radiosurgery is performed for patients with symptomatic, imaging-conﬁrmed hemorrhages for which resection is believed to be associated with high risk. Hasegawa et al. studied the long-term hemorrhage rate in 82 patients after cavernous malformation radiosurgery [19]. Most patients had multiple hemorrhages from brain stem or diencephalic cavernous malformations. During an average observation period of 4.33 years (for a total of 354 patient-years) before treatment, 202 hemorrhages were noted, for an annual hemorrhage rate of 33.9%, excluding the ﬁrst hemorrhage. After radiosurgery, 19 hemorrhages were identiﬁed during an average of 5 years (for

a total of 401 patient-years). The annual hemorrhage rate was 12.3% per year for the ﬁrst 2 years after radiosurgery, followed by 0.76% per year from years 2 to 12. Eleven patients had new neurologic symptoms without hemorrhage after radiosurgery (13.4%). A signiﬁcant decrease in the symptomatic hemorrhage rate after stereotactic radiosurgery of cerebral cavernous malformations indicates radiosurgery is an effective management strategy for patients with hemorrhagic malformations in high-risk brain locations. When patients serve as their own control, the pre-radiosurgery hemorrhage rates fall dramatically 2 to 5 years after radiosurgery.

TABLE 9-4. Results of Gamma Knife radiosurgery for intracranial AVMs. First author

Year

No. of patients

Volume (cm3)

Margin dose (Gy)

Follow-up (months)

Obliteration rate (%)

Morbidity (%)

Re-bleed rate (%)

Chang [3] Regis [9]

2000 2001

254 45

12.1 0.55

16.2 23

24.7 18

78.2 (4 year) 82

5 4.4

Kurita [6] Coffey [4] Pollock [8] Steiner [10] Flickinger [5] Maruyama [7] Shin [12]

2000 1995 2003 1992 2002 2004

30 121 144 247 351 50

1.35 — 5.7 — 5.7 1.5

18.4 — 18 — 20 20

52.2 — 86 — 51 72

52.2 (3 year) 72 64 81 75 66 (6 year)

6.9 2.5 5 3.2 — 16

17.2 4 8 1.9/year — 1.7/year

— — 4 — — 0

2004

400

1.9

20

65

72 (3 year)

6.9

1.9/year

—

6.8 4.4

Cyst formation (%)

0.4 —

114

a. niranjan et al.

TABLE 9-5. Results of Gamma Knife radiosurgery for acoustic tumors.

First author

Year

No. of patients

Median volume (cm3)

Margin dose (Gy)

Follow-up (months)

Tumor control (%)

Flickinger [29] Delbrouck [30] Unger [21] Lunsford [27] Kondziolka [28] Noren [23] Niranjan [24]

2004 2003 2002 2005 1998 1998 1999

313 95 100 829 162 254 29

1.1 — 3.4 2.5 — — 0.4

13 — 13 13 16 13.6 14

24 12 76 120 60–120 33

98.6 — 96 97 98 93 100

Nakamura [25] Prasad [22] van Eck [20] Muacevic [26]

2000 2000 2005 2004

78 153 78 219

— — 2.28 —

— — 13 —

— — 22 72

81 92 97.5 97

Radiosurgery for Brain Tumors Vestibular Schwannoma Radiosurgery The management options for vestibular schwannomas include observation, microsurgery, and radiosurgery. Vestibular schwannoma radiosurgery using the Gamma Knife unit was ﬁrst performed by Leksell in 1969. Acoustic tumor radiosurgery has evolved steadily in the past decade [20–30]. Advanced dose planning software, MR-guided dose planning, and dose optimization reﬂect the evolution of this technology. The goal of radiosurgery is tumor growth control with preservation of cochlear, facial, and trigeminal nerve functions. The results after radiosurgery of acoustic tumors have established it as the effective minimally invasive alternative to microsurgery. In a review of his 17 years’ experience with radiosurgery in more than 1000 acoustic neuromas patients, Lunsford et al. reported that more than 97% of patients had tumor growth control [27]. Patients with acoustic tumors are evaluated with high-resolution MRI scan and undergo clinical evaluation as well as audiologic tests that include pure tone average (PTA) and speech discrimination score (SDS). Hearing is graded using the Gardner-Robertson’s modiﬁcation of the Silverstein and Norell classiﬁcation [31] and facial nerve function is assessed according to the House-Brackmann grading system [32]. “Serviceable” hearing (class I and II) is deﬁned as a PTA or speech reception threshold (SRT) lower than 50 dB and a speech discrimination score better than 50%. After radiosurgery, all patients are followed up with serial gadolinium-enhanced MRI scans, which are requested at 6-month intervals for 2 years. If there is no appreciable change in tumor size, subsequent MRIs are requested at 2-year intervals. All patients who have some preserved hearing are advised to obtain audiologic evaluation (PTA and SDS) near the time of their MRI follow-up. TUMOR GROWTH CONTROL Recent reports suggest a tumor control rate of 93% to 100% after radiosurgery (Table 9-5). Kondziolka et al. studied 5- to 10-year outcomes in 162 acoustic tumor patients who had radiosurgery at the University of Pittsburgh [28]. In this study, a long-term 98% tumor control rate was reported (Fig. 9-4).

Cranial nerve preservation (%) VIII

78.6 67 55 50–77 47 60 73 (<14 Gy, 100) — 65 83.40 90

VII

100 — 94.00 99 79 86 100 — — 98.70 99.50

V

95 — 95 97 73 92 100 — — 97.40 95

Sixty-two percent of tumors became smaller, 33% remained unchanged, and 6% became slightly larger. Some tumors initially enlarged 1 to 2 mm during the ﬁrst 6 to 12 months after radiosurgery as they lost their central contrast enhancement. Such tumors generally regressed in volume compared with their size before radiosurgery. Only 2% of patients required tumor resection after radiosurgery. Niranjan et al. analyzed outcome of intracanalicular tumor radiosurgery performed at the University of Pittsburgh. All patients (100%) had imaging documented tumor growth control [24]. Flickinger et al. performed an outcome analysis of acoustic neuroma patients treated between August 1992 and August 1997 at the University of Pittsburgh [29]. The actuarial 5-year clinical tumor control rate (no requirement for surgical intervention) was 99.4 + 0.6% [29, 33]. The long-term (10- to 15-year) outcome of benign tumor radiosurgery has been evaluated. In a study that included 157 patients with acoustic tumors, the median follow-up for the patients still living at the time of the study (n = 136) was 10.2 years. Serial imaging studies after radiosurgery (n = 157) showed a decrease in tumor size in 114 (73%) patients, no change in 40 (25.5%) patients, and an increase in 3 (1.9%) patients who had later resection [34]. No patient developed a radiation-associated malignant or benign tumor (deﬁned as a histologically

FIGURE 9-4. (Left) Axial contrast-enhanced T1-weighted MR image showing a right-sided acoustic tumor. Gamma Knife radiosurgery was performed using 12.5 Gy to 50% isodose line. (Right) Two-year followup axial contrast-enhanced T1-weighted MR image showing tumor shrinkage.

9.

gamma knife radiosurgery

conﬁrmed and distinct neoplasm arising in the initial radiation ﬁeld after at least 2 years had passed). HEARING PRESERVATION Pre-radiosurgery hearing now can be preserved in 60% to 70% of patients. In a long-term (5- to 10-year follow-up) study conducted at the University of Pittsburgh, 51% of the patients had no change in hearing ability [28, 33]. All patients (100%) who were treated with a margin dose of 14 Gy or less maintained serviceable level of hearing after intracanalicular tumor radiosurgery [24]. Among patients treated after 1992, the 5-year actuarial rates of hearing level preservation and speech preservation were 75.2% and 89.2%, respectively, for patients (n = 89) treated with 13-Gy tumor margin dose. The 5-year actuarial rates of hearing level preservation and speech preservation were 68.8% and 86.3.2%, respectively, for patients (n = 103) treated with >14 Gy as tumor margin dose [29]. Unlike microsurgery, immediate hearing loss is uncommon after radiosurgery. If hearing impairment is noted, it is gradual over 6 to 24 months. Early hearing loss after radiosurgery (within 3 months) is rare and may result from neural edema, inﬂammation, or demyelination. The exact mechanism of delayed hearing loss after radiosurgery is still unclear. Perhaps gradual obliteration of microvessels, inﬂammatory response, or even direct radiation axonal or cochlear injury is implicated. The effect of radiation on normal microvessels supplying the cochlear nerve or cochlea itself is not known; however, with doses as low as 12 to 13 Gy (which are sufﬁcient to halt the tumor growth), vascular obliteration of normal vessels seems less likely. This dose probably does not adversely affect the vessels as well as the axons. Although with current imaging techniques the cochlear nerve cannot be well visualized, effort should be made to achieve high conformality at the anterior and inferior margins of the tumor. Conformal dose planning using 4-mm collimator for the intracanalicular tumor may prevent further injury to the cochlear nerve. FACIAL NERVE AND TRIGEMINAL NERVE PRESERVATION Facial and trigeminal nerve function can now be preserved in the majority of patients (>99%). In the early experience at the University of Pittsburgh, normal facial function was preserved in 79% of patients after 5 years and normal trigeminal nerve function was preserved in 73% [28]. These facial and trigeminal nerve preservation rates reﬂected the higher tumor margin dose of 18 to 20 Gy used during the CT-based planning era before 1991. In a recent study using MR-based dose planning, 13-Gy tumor margin dose was associated with no risk of new facial weakness and 3.1% risk of facial numbness (5-year actuarial rates). A margin dose of >14 Gy was associated with 2.5% risk of new facial weakness and 3.9% risk of facial numbness (5-year actuarial rates) [29]. None of the patients who had radiosurgery for intracanalicular tumor developed new facial or trigeminal neuropathy. NEUROFIBROMATOSIS TYPE 2 Patients with acoustic neuroma associated with neuroﬁbromatosis type 2 (NF 2) represent a special challenge because of the risk of complete deafness. Unlike the solitary sporadic tumors that tend to displace the cochlear nerve, tumors associated with NF 2 tend to form nodular clusters that engulf or even inﬁltrate

115

the cochlear nerve. Complete resection may not be always possible. Radiosurgery has been performed for patients with NF 2. Subach et al. studied 40 patients (with 45 tumors) who were treated with radiosurgery for NF 2 [35]. Serviceable hearing was preserved in 6 of 14 (43%) patients, and this rate improved to 67% after modiﬁcations made in technique in 1992. The overall tumor control rate was 98% [36]. Only one patient showed imaging-documented growth. Normal facial nerve function and trigeminal nerve function was preserved in 81% and 94% of patients, respectively. It now appears that preservation of serviceable hearing in patients with NF 2 is an attainable goal with modern radiosurgery technique.

Meningioma Radiosurgery The optimal treatment for meningioma is complete resection of tumor with its dural base; however, when meningiomas are attached to skull base cranial nerves or vascular structures, complete resection may not be possible. Multimodality management should then be considered. Recurrence rates are higher for meningiomas in critical locations where only subtotal resections are possible due to limited access and involvement of the critical structures. Radiosurgery offers an attractive option for patients with residual or recurrent meningioma as well as for patients in whom complete resection of tumor is considered attainable but only with unacceptable morbidity. Table 9-6 shows recent results of meningioma radiosurgery from various institutions [37–58]. Tumor control rates ranged from 98% (at 2 years) to 75% (at 8 years). Excellent clinical outcomes after skull base meningioma radiosurgery have been reported (Fig. 9-5). Meningiomas attached to major venous sinuses can be successfully treated by radiosurgery. Tumor regression may occur slowly over several years after radiosurgery. Radiosurgery provides long-term tumor control associated with high rate of neurologic preservation and patient satisfaction. Surgical excision is the preferred ﬁrst-line approach for convexity, anterior fossa, or lateral sphenoid ridge meningiomas, which can be easily approached. For meningiomas at all other intracranial locations, radiosurgery can be offered as the ﬁrst management approach unless the tumor needs debulking because of mass effect. Larger tumors involving critical locations such as optic chiasm may require combined approaches. Malignant meningiomas especially require multimodality management that includes resection, radiosurgery, and radiation therapy.

Pituitary Adenoma Radiosurgery Multimodality management is needed for patients with pituitary tumors. The primary aim of treatment for clinically nonfunctioning pituitary macroadenomas is tumor removal and preservation of visual function. Transsphenoidal surgery is the preferred approach for managing most pituitary adenomas. Radiosurgery is often indicated as an adjuvant management after partial resection or later recurrence of nonfunctioning pituitary adenomas; however, radiosurgery can be performed as the primary management of nonfunctioning adenomas in carefully selected patients such as those who have major surgical risks or for patients who decline microsurgery. Cavernous sinus invasion can occur de novo in patients with large pituitary macroadenomas but is more commonly seen as a residual tumor

116

a. niranjan et al.

TABLE 9-6. Results of Gamma Knife radiosurgery for meningiomas. No. of patients

Author

Year

Liscak [50]

1999

53

Aichholzer [37] Roche [58] Muthukumar [52] Huffmann [43] Kim [44] Feigl [41] Kondziolka [48] Pan [55] DiBiase [39] Flickinger [42] Iwai [44]

2000 2003 1998 2005 2005 2005 1999 1998 2004 2003 2003

46 32 41 15 23 127 99 80 121 219 42

Chang [38] Nicolato [53] Eustacchio [40] Lee [49]

2003 2002 2002 2002

187 156 121 159

Kobayashi [46] Morita [51]

2001 1999

87 88

Tumor location

Cavernous sinus Skull base Petroclival Tentorial Atypical Superﬁcial Mixed Mixed Mixed Mixed Mixed Cavernous sinus Mixed Mixed Skull base Cavernous sinus Mixed Skull base

Mean follow-up (months)

19 48 56 36 35 32 29.3 42 21 54 29 49.4 37.3 48.9 60–118 35 24.2 35

after attempted microsurgical resection. The cranial nerve complication risks and cerebrovascular risks of cavernous sinus microsurgery warrant consideration of radiosurgery. In many cases, the cavernous sinus mass can be treated while selectively sparing not only the optic apparatus but also the pituitary stalk and residual pituitary gland within the sella. Tumor growth control rates of 90% to 100% have now been conﬁrmed by multiple centers after pituitary adenoma radiosurgery [59]. The antiproliferative effect of radiosurgery has been reported in nearly all patients who underwent Gamma Knife radiosurgery. Relatively few patients (who usually had received lower margin doses) eventually required additional treatment. For secretory adenomas, medical management is extremely useful as either ﬁrst-line therapy or as an adjunct in a combined multimodality approach to overall patient management. Tumor

FIGURE 9-5. (Left) Axial contrast-enhanced T1-weighted MR image showing a skull base meningioma at the time of radiosurgery. (Right) Three-year follow-up axial contrast-enhanced T1-weighted MR image showing complete tumor disappearance.

Temporary

Persistent

100.0

3.8

0.0

15.9 13 15.3 16 16 13.8 16 12–20 14 14 11

96.0 100.0 97.5 86.6 95.6 96.4 95.0 91.0 91.7 93.2 90.5

2.0 6.2 — 6.6 43.0 2.5 14 5.0 — — 4.7

9.0 6.2 2.5 0.0 0.0 1.2 5 2.5 8.3 5.3 0.0

10.1 8.3 6.8 6.5

15.1 14.6 13 13

97.1 96.0 99.2 93.1

10.7 4.0 3.3 1.8

0.0 1.0 1.7 5.0

Diameter 25.8 10

14.5 16

93.1 95.0

10.3 2.2

3.4 12.5

Mean volume (cm )

7.8 Diameter 23.5 mm — Diameter 20 mm — 4.7 5.9 4.7 — 4.5 5 14.7

Mean margin dose (Gy)

Complications (%)

Tumor control (%)

3

12

resection is the preferred management strategy when medical management fails to normalize pituitary function. Radiosurgery is often indicated as an adjunct to control residual or recurrent secretory adenoma. The initial ﬁrst stage extracavernous microsurgery is often optimal in order to reduce the subsequent tumor volume, create space between the tumor and the optic apparatus, and thus allow safe delivery of the highest dose of radiosurgery possible. The goal of radiosurgery for functional adenomas is pituitary hormone normalization. Biochemical remission for growth hormone (GH)-secreting adenomas is deﬁned as GH level suppressed to below 1 μg/L on oral glucose tolerance test (OGTT) and normal age-related serum insulin-like growth factor-1 (IGF-1) levels. OGTT remains the gold standard for deﬁning a cure of acromegaly. The IGF-1, however, is far more practical. Decrease of random GH to less than 2.5 μg/L is achieved more frequently than the normalization of IGF-1 but it is necessary to obtain the fulﬁllment of both criteria. Hormonal normalization after radiosurgery was achieved in 23% to 82% of cases in the published series [60–64]. The suppression of hormonal hyperactivity is more effective when higher doses of radiation are used. In a study at the University of Pittsburgh, 38% of recurrent tumor patients were cured (GH ≤1 μg/L), and overall, 66% had growth hormone levels ≤5 μg/L 3 to 5 years after radiosurgery [65]. The impact of radiosurgery has a latency of about 20 to 28 months [66, 67]. During this interval, hormone-suppressive medications may be beneﬁcial. Because hormone-suppressive medication during radiosurgery may act as a radioprotective agent, this medication should be discontinued at least 6 to 8 weeks prior to the radiosurgery and may be resumed after a week. Patients with Cushing disease (adrenocorticotropic hormone [ACTH]-secreting adenomas) respond to radiosurgery, but

9.

gamma knife radiosurgery

more than one procedure may be needed. Often, tumor cannot be well deﬁned during the initial imaging. In addition, there is a latency of about 14 to 18 months for maximal therapeutic response [66, 67]. In various published series, 38% to 83% hormone normalization after radiosurgery has been observed [68–71]. Most prolactinomas can be controlled successfully by dopaminergic suppressive therapy. Surgery is indicated for cases of intolerance to medical treatment, in cases where women desire to have children, or when patients are dopamine agonist resistant (5% to 10% of patients). Some patients prefer microsurgery or radiosurgery to the need for highly expensive lifelong medical treatment. In published studies of patients treated with radiosurgery, 25% to 29% showed normalization [60, 61]. The possible radioprotective effect of dopaminergic drugs should be taken into account. Because patients treated on dopamine agonist during radiosurgery had lower remission rate, it is therefore recommended that prolactinoma radiosurgery be performed during a period of drug withdrawal. New pituitary hormone deﬁciency has been reported in 0 to 30% of patients after radiosurgery for functional pituitary adenomas [60, 61]. The most important factor inﬂuencing hypopituitarism after radiosurgery seems to be the mean dose to the hypophysis (pituitary stalk). Vladyka et al. observed some worsening of gonadotropic, corticotropic, or thyrotropic functions 12 to 87 months after radiosurgery, usually within 4 to 5 years after radiosurgery [72]. Deterioration in pituitary functions was observed when pituitary stalk received higher doses (>15 Gy). The risk for hypopituitarism after stereotactic radiosurgery thus becomes a primary function of the anatomy of the tumor and the dose prescribed. For recurrent tumors primarily where the pituitary stalk (and even at times the residual pituitary gland) is separate from the tumor, is easily visualized, and can be excluded from higher dose, the risk of hypopituitarism is extremely small. For adenomas that cannot be visually separated from the normal gland, particularly if they extend upward to involve or compress the pituitary stalk, the risk is predominately related to the dose necessary to effectively achieve all outcome goals for the functional status of the tumor (higher for secretory than nonsecretory adenomas). Gamma Knife radiosurgery is superior to radiation therapy because there is a faster response and fewer adverse radiation effects. Response to radiosurgery is best with ACTH-producing tumors, followed by GH-producing tumors, prolactinomas having the poorest response usually because they have failed prior medical management due to their invasive nature. Hypopituitarism can be expected to occur in up to 30% within 4 to 5 years but can be avoided by minimizing radiation to pituitary stalk and hypothalamus. Somatostatin analogues and dopamine agonists may have a radioprotective effect [60, 73]. Although the radioprotective effect of these drugs was not conﬁrmed in subsequent studies [62–64], it is advisable to stop these drugs prior to radiosurgery. Short-acting form of somatostatin analogues can be given until 2 weeks prior to GK. Long-acting somatostatin analogues should be discontinued 4 months prior to GK. Dopamine agonists should be discontinued 2 months prior to radiosurgery. After radiosurgery, once hormone levels are normal on medical therapy, somatostatin analogues should be stopped for 4 months each year to assess for biochemical cure. Similarly, dopamine agonists should be stopped for 2

117

months. A panel of tests to detect hypopituitarism should be done at 6-month intervals for the ﬁrst 5 years and then yearly.

Radiosurgery of Glial Tumors MALIGNANT GLIOMAS Malignant gliomas continue to represent one of the most serious challenges in neurosurgery. Radiation therapy has become the mainstay of the treatment. The observation that local control and median survival can be improved through the dose escalation is the basis for the application of radiosurgery to malignant gliomas. Radiosurgery is used for boost irradiation of patients with malignant glial tumors in addition to conventional widemargin fractionated radiotherapy. It has been used mainly for patients with tumors >3.5 cm in diameter as part of a multimodality approach to malignant gliomas. Early radiosurgery reports widely varied in the outcomes for malignant gliomas with a median survival for GBM patients ranging from 9.5 months to 26 months. These variations could result from patient selection biases and other prognostic factors [74]. We performed a retrospective study to evaluate the result of radiosurgery on 64 GBM and 43 anaplastic astrocytoma patients. The median survival for the GBM patients was 16 months after radiosurgery and 26 months after diagnosis. A 2-year survival rate was 51%. For patients with anaplastic astrocytomas, median survival after radiosurgery was 21 months and after diagnosis was 32 months. A 2-year survival rate after diagnosis was 67%. Other centers have recently reported survival rates that seem signiﬁcantly improved compared with 9-month median survival and 10% 2-year actuarial rate reported for standard therapy. Nwokedi et al. compared survival between 33 patients treated with external beam radiation therapy (EBRT) alone (group 1) and 31 patients managed with EBRT plus a Gamma Knife radiosurgery (GK-SRS) boost (group 2) [75]. GK-SRS was administered to most patients within 6 weeks of the completion of EBRT. The median EBRT dose was 59.7 Gy (range, 28 to 70.2 Gy), and the median GK-SRS dose to the prescription volume was 17.1 Gy (range, 10 to 28 Gy). Both groups were comparable in age, Karnofsky performance status, extent of resection, and tumor volume. The median survival was signiﬁcantly better in patients treated with EBRT plus GK-SRS (13 months in EBRT alone vs. 25 months in EBRT plus GK-SRS). Age, Karnofsky performance status, and the addition of GKSRS were all found to be signiﬁcant predictors of overall survival. No acute grade 3 or grade 4 toxicity was encountered. There is a signiﬁcant survival advantage using radiosurgery boost in patients with malignant glioma, especially if appropriately used with surgery and other adjuvant therapies; however, a carefully designed prospective randomized trial is needed to reliably establish survival beneﬁt from radiosurgical boost for malignant gliomas. LOWER-GRADE GLIOMAS Low-grade gliomas have been treated with radiosurgery. Simonova et al. treated 68 patients with low-grade gliomas using Gamma Knife surgery [76]. The median patient age was 17 years and median target volume was 4200 mm3. The median marginal prescription dose was 25 Gy. Ninety-ﬁve percent of patients were treated in ﬁve daily stages. These authors reported 83% rate of partial or complete tumor regression with a median

118

a. niranjan et al.

time to response of 18 months. In this series, the progressionfree survival was 92% at 3 years and 88% at 5 years. Moderate acute or late toxicity was noted in 5% of patients. Kida et al. treated 51 patients harboring low-grade gliomas (12 grade I astrocytomas, 39 grade II astrocytomas) using Gamma Knife [77]. The mean margin dose was 12.5 Gy for grade I and 15.7 Gy for grade II tumors. In the mean follow-up of 27.6 months, grade I astrocytomas had a response rate of 50% and a control rate of 91.7%. Grade II astrocytomas had a 46.2% response rate and an 87.2% control rate. Despite the favorable histologic characteristics and prognosis of pilocytic astrocytomas, some patients may not be cured after microsurgery because of an adverse location, recurrence, or tumor progression. Radiosurgery is an effective alternative therapeutic approach for these patients. Hadjipanayis et al. reported outcome in 38 patients harboring unresectable pilocytic astrocytomas who were treated with radiosurgery [78]. The median radiosurgical dose to the tumor margin was 15 Gy (range, 9.6 to 22.5 Gy). After radiosurgery, serial imaging demonstrated complete tumor resolution in 10 patients, reduced tumor volume in 8, stable tumor volume in 7, and delayed tumor progression in 12 patients. Three patients died of local tumor progression. Stereotactic radiosurgery is a valuable adjunctive strategy in the management of recurrent or unresectable pilocytic astrocytomas especially small-volume, sharp-bordered tumors. Radiosurgery can play an important role in the treatment of low-grade astrocytomas, and complete cure of these tumors has been achieved in at least some of the cases [78–80]. Acute complications after radiosurgery are unusual and limited to exacerbation of existing symptoms. The most frequently seen delayed complication of radiosurgical boost is tumor swelling radiation reaction in the tumor or surrounding brain swelling. Symptoms are usually controllable by steroid therapy. The reported incidence of radiation necrosis ranges from 2% to 22%. Reoperation rates ranging from 21% to 33% have been reported after radiosurgery. Neither radiation necrosis nor reoperation is associated with diminished length of survival.

Brain Metastases Radiosurgery The best initial management for brain metastases patients remains to be deﬁned. Current options include fractionated radiation therapy alone, surgery alone, radiosurgery alone, surgery plus radiation therapy, or radiosurgery plus radiation therapy. There are several features that make brain metastases the most common indication for radiosurgery. Most brain metastases are roughly spherical and therefore can be easily targeted by radiosurgery. Brain metastases are compact targets. Although peritumoral microscopic spread is likely, conformal radiosurgery provides additional therapeutic beneﬁt because of the fall-off zone of radiation outside the imaging-deﬁned margin. Advances in neuroimaging have led to early diagnosis of metastases while these are still small and without signiﬁcant mass effect and symptoms. Whereas single brain metastasis without mass effect is the ideal indication of radiosurgery, multiple metastases are treated when the total target volume allows for safe and effective dose delivery. Radiosurgery is not recommended for patients with large metastatic tumors causing signiﬁcant mass effect. Such patients should undergo surgical excision.

FIGURE 9-6. (Left) Axial contrast-enhanced T1-weighted MR image showing left frontal brain metastases with signiﬁcant surrounding edema. (Right) Six-month follow-up axial contrast-enhanced T1weighted MR image showing signiﬁcant tumor shrinkage and no edema in surrounding brain parenchyma.

A large number of scientiﬁc publications deﬁne the effectiveness of radiosurgery for brain metastases (Fig. 9-6). Table 9-7 lists several large representative series of patients [81–92]. These reports include patients with various primary histologies. The local tumor control ranges from 25% to 97%, and median survival ranges from 6 to 27 months. The Gamma Knife Users Group studied the outcomes of radiosurgery in 116 patients with solitary brain metastases. Radiosurgery was part of the initial management of 71 patients, and 45 patients had recurrent tumors after prior whole-brain fractionated radiation therapy. In this study, actuarial local control rate of 67% at 2 years was reported. Shiau et al. recently reported their radiosurgery experience in 219 brain metastases in 100 patients [93]. The actuarial tumor control, deﬁned as freedom from progression, was 82% and 77% at 6 and 12 months, respectively. These data substantially validate the clinical observation of improved local control after radiosurgery (Fig. 9-7). Although local tumor control rates have improved, mortality is usually related to the uncontrolled primary tumor or metastatic spread to other organs. In general, survival and morbidity results of radiosurgery are superior to those reported for surgical resection followed by whole-brain radiation therapy. The results show that radiosurgery is associated with high local control and low morbidity in comparison with surgical resection. When interpreting radiosurgery results, one should also take into account the facts that patients in surgical series are selected for their suitable locations and good general condition, whereas no such selection is performed for radiosurgery. On the contrary, those who are not suitable candidates for surgery either due to the eloquent brain locations or poor medical condition are included in the radiosurgery series. Rutigliano et al. performed a cost-beneﬁt comparison of Gamma Knife radiosurgery and surgical resection for solitary brain metastases and concluded that radiosurgery had a lower uncomplicated procedure cost, a lower average complication per procedure, was more effective, and had a better incremental cost effectiveness per life-year [93].

Radiosurgery for Pineal Gland Tumors Management of pineal region tumors remains a signiﬁcant challenge because of the anatomic complexity of the area and the

9.

119

gamma knife radiosurgery

TABLE 9-7. Results of Gamma Knife radiosurgery for metastatic brain tumors. Author

Year

No. of patients

Serizawa [90] Lippitz [86]

2005 2004

521 215

Nam [87] Pan [88] Gerosa [82] Gerosa [83] Petrovich [89] Sheehan [91]

2005 2005 2002 2005 2002 2002

130 191 804 504 458 273

Amendola [81]

2000

68

Hasegawa [84] Hoffman [85]

2003 2001

121 113

Simonova [92]

2000

237

Median survival (months)

9 Multiple 7.8 Single 13.7 9 14 13.5 14.5 9 7 7.8 8 12

6–12

Systemic control

Margin dose (Gy)

TPFSR 1-year

70% lung —

13% —

Median 20 Median 22

95.70% 93.9%

23% 20%

73% 74% — — 25% 78%

45% lung All lung 88% lung All NSLC 50% melanoma All NSLC

62% — 29% 31% 43% —

Mean 17.9 Mean 18 Mean 20.6 Mean 21.4 Median 18 Median 16

63.90% 91% 94% 95% 87% 86%

65% 46% 55% 60% 58%

53%

All breast

97%

15–24

0% 17%

35% lung All lung

36% 61%

Mean 18.5 Median 18

0%

43% lung

100%

84% local control 79% 81% GK alone 86% with XRT 95.3%

WBRT

Primary cancer

No No

Median 21.5

Single tumor

>50% 22% 80% 45%

100%

WBRT, whole-brain radiation therapy; TPFSR 1-year, one-year tumor progression free survival; NSLC, non–small cell lung cancer.

presence of critical brain and vascular structures. Microsurgical techniques are often successful in obtaining a tissue diagnosis; however, the likelihood of curative resection remains low. There are only few published reports on radiosurgery for pineal tumors [95, 96]. At the authors’ institution, 14 patients with parenchymal pineal tumors were treated between 1989 and 1997. Local tumor control was achieved in 13 patients while one died of tumor progressions despite chemotherapy and craniospinal irradiation prior to radiosurgery. Neuroimaging followup showed complete disappearance of tumor in 3 patients, decrease in tumor size in 7, no change in tumor size in 3, and tumor growth in 1 patient.

Radiosurgery for Skull Base Tumors Radiosurgery is a primary and adjuvant management for tumors of skull base [97–103] (Table 9-8). From September 1987 through December 2004, 238 miscellaneous skull base tumors were treated with Gamma Knife radiosurgery at the University

FIGURE 9-7. (Left) Axial contrast-enhanced T1-weighted MR image of an 80-year-old man showing a large hemangioblastoma after attempted tumor resection. Gamma Knife radiosurgery was performed using 15 Gy prescribed to the tumor margin (tumor volume, 16.6 cm3). (Right) Follow-up axial MR image shows signiﬁcant regression of the tumor 4 years after Gamma Knife radiosurgery.

of Pittsburgh Medical Center. These tumors and their subsequent management are described below in more detail.

Non–Acoustic Schwannomas Thirty-ﬁve patients underwent radiosurgery for trigeminal nerve sheath tumors deﬁned by clinical examination, high-resolution intraoperative imaging, and in selected cases prior surgery. Our results of trigeminal schwannoma have been recently published [99]. The records of 23 patients were reviewed with a median follow-up of 40 months. Twenty of 23 (91%) patients had tumor growth control, with regression noted in 15 and no further tumor growth in 5. Patients who had subsequent tumor enlargement underwent a second radiosurgical procedure. Twelve of 23 (52%) trigeminal nerve sheath tumor patients reported systemic improvement. Nine (39%) patients had no change in their symptoms. Only two patients noted new neurologic complaints such as facial weakness (one patient) and worsening of the pre-radiosurgical facial numbness (one additional patient). Of interest, trigeminal nerve sheath tumors have a much higher likelihood of developing transient but occasionally impressive short-term swelling of the tumor in the ﬁrst year after radiosurgery. This is quite distinct from those patients who have undergone acoustic tumor radiosurgery. In the majority of trigeminal neuroma patients, transient swelling is followed by delayed shrinkage, often of profound degree. Therefore, it is critical that patients and referring doctors do not despair during this transient tumor enlargement phase identiﬁed by imaging and sometimes associated with temporary concomitant neurologic symptoms. Most such symptoms will resolve as the tumors regress during the next 3 to 6 months. Radiosurgery using the Gamma Knife proved to be an effective management strategy for those patients who had undergone both primary as well as adjuvant (post-microsurgery) radiosurgery [100]. Three patients underwent Gamma Knife radiosurgery for facial schwannomas, all identiﬁed at the time of prior

120

a. niranjan et al.

TABLE 9-8. Results of Gamma Knife radiosurgery for skull base tumors. First author

Year

Diagnosis

Saringer [102] Zhang [103] Nettel [99] Pollock [101] Pan [100] Muthukumar [98] Miller [97]

2001 2002 2004 2002 2005 1998 1997

Glomus tumor Jugular foramen schwannomas Trigeminal schwannomas Non–vestibular schwannomas Trigeminal schwannomas Chordomas, chondromas Carcinomas, sarcomas

No. of patients

Mean follow-up (months)

Tumor control (%)

Complications (%)

13 27 23 23 46 15 32

60 38.7 40 43 68 40 27

100 92.5 91 96 93 73 91

0 0 8 17 8 0 3

microsurgery and associated with recurrence or subtotal prior resection. Tumors of the ninth and tenth cranial nerves pose special challenges. Twenty-six patients with jugular bulb schwannomas underwent radiosurgery between August 1987 and September 2004. Most such patients present with symptoms related to imbalance, incoordination, dysphasia, or hearing loss. A total of 12 patients had previously undergone gross total resection with tumor recurrence, and 4 patients had undergone prior partial resection. Results to date show a high likelihood of long-term tumor growth control for such tumors. In an earlier report including 17 patients, we reported a tumor control rate of 94% (8 decreased and 8 were stable in size) after jugular foramen schwannoma radiosurgery [104]. Zhang et al. reported 96% (26/27) tumor growth control with a follow-up period of 38.7 months [103]. In the series of non–vestibular schwannomas, Pollock et al. reported 96% (22/23) tumor growth control after Gamma Knife radiosurgery [101].

Craniopharyngioma Multimodality therapy is often necessary for craniopharyngioma patients because of the development of refractory cystic components of their tumors. Radiosurgery is usually part of a multimodality management when prior therapies have failed [105, 106]. Forty-three patients have undergone Gamma Knife radiosurgery as part of a primary or adjuvant management strategy for craniopharyngioma. Long-term follow-up in our patient series was available in 29 patients. The median tumor volume was 0.4 (range, 0.12 to 6.36) cm3. One to nine isocenters of different beam diameters were used. The median dose to the tumor margin was 12.5 Gy (range, 9 to 20 Gy), and the maximum dose was 25 Gy (range, 21.8 to 40 Gy). The dose to the optic apparatus was limited to less than 8 Gy. Clinical and imaging follow-up data were obtained at a median of 24 months (range, 13 to 150) from radiosurgery. Overall, 14 of 29 tumors regressed or vanished, and 10 remained stable after radiosurgery. Further tumor growth was noted in ﬁve patients, three of who underwent surgical resection and one who had repeat radiosurgery. Two additional patients needed management for cyst enlargement. One patient with prior visual defect had further vision deterioration 9 months after radiosurgery. No patient developed new-onset diabetes insipidus. We found that stereotactic radiosurgery was a reasonable option in selected patients with small recurrent or residual craniopharyngiomas. Adverse radiation risks related to adjacent cranial nerve structures or the development of new extraocular movement deﬁcits are rare, providing that the optic nerve and

tract dose is kept lower than 8 Gy or less in a single procedure. In general, we prefer the use of multimodality management including microsurgery, radiosurgery, and intracavitary radiation rather than stereotactic or fractionated radiation therapy. The goal has been to maintain endocrinologic function whenever possible, reduce the risks of visual dysfunction, and subsequently control tumor growth. There are other reports that have similar results in the management of craniopharyngiomas using Gamma Knife [107–109].

Glomus Tumors Radiosurgery using the Gamma Knife has been performed in 16 patients in a 17-year interval. This sparse number of patients (of 7200 who had Gamma Knife radiosurgery) is accounted for by the tendency of such tumors to extend well below the skull base. When surgical resection is not feasible, we consider staged radiosurgery technologies such as linac-based radiosurgery for the extracranial component and the Gamma Knife for the intracranial portion. Some patients also have undergone elective embolization for shrinkage of their tumor or subtotal microsurgical resection. Only one patient in our series had a glomus tympanicum tumor. Gamma Knife radiosurgery appears to have a long-term, high tumor control rate of glomus tumors, paralleling the beneﬁt provided by fractionated radiation therapy. However, the Gamma Knife provides a superior biological effective tumor dose, with better dose sparing of the adjacent brain stem and cranial nerve structures. Pollock et al. in a series including 42 patients reported 98% tumor control after glomus jugulare radiosurgery at a mean follow-up of 44 months [110]. Neurologic improvement or stability was observed in the majority of patients in published series. Centers using linac-based radiosurgery continue to support radiosurgery as an effective and safe method of treatment for glomus jugulare tumors that results in low rates of morbidity.

Hemangiomas Radiosurgery for hemangioma was performed in seven patients. Hemangiomas of neurosurgical interest are histologically benign vascular epithelial cell origin tumors that most often occur in the orbit or cavernous sinus or both. These patients tend to present with ocular symptoms or signs such as orbital pain, ophthalmoplegia, proptosis, or impaired visual acuity. They can, in fact, be diagnosed by their characteristic imaging appearance by MRI. Because they may hemorrhage dramatically at the time of attempted removal, it is prudent for

9.

gamma knife radiosurgery

121

surgeons considering biopsy or resection of such tumors to get the appropriate imaging in advance. Asymptomatic lesions do not require intervention but are often approached surgically in pursuit of a diagnosis. Symptomatic lesions require treatment. Options include en bloc resection, embolization, or radiation. Radiosurgery is a better option. In our relatively limited experience, some patients have had incomplete resection because of excessive blood loss, and one patient had undergone unsuccessful embolization. We recently reported the outcomes of four patients treated with radiosurgery with tumor doses ranging from 14 to 19 Gy at the margin [111]. All patients had symptomatic improvement, and all had shown a dramatic reduction in the overall volume of the tumor. One patient had persistent diplopia. In our early experience, stereotactic radiosurgery proved to be a very effective management strategy, which avoided potentially serious complications associated with skull base microsurgery or embolization. The other reports including 3 to 5 cases in each also achieved reduction in tumor volume after radiosurgery [112–114].

Radiosurgery appears to be a safe and effective management for small-volume tumors, but over the course of many years, especially from years 5 to 10 after initial surgery and radiosurgery, recurrence rates continue to increase [98, 118]. In such cases, repeat radiosurgery or perhaps fractionated radiation therapy and repeat radiosurgery should be considered. Most such patients require one or more microsurgical approaches for tumor cytoreduction. More recently, we have embarked on the usage of endoscopic transsphenoidal resection followed by radiosurgery. In our 1998 report, with an average of 4 years of experience, 15 patients were evaluated. In 13 cases, it was used as an adjunctive treatment and in two patients as an alternative to microsurgical resection. Eight patients had clinical improvement, three remained stable, and four died. Two of the four patients who died had tumor progression outside of the radiosurgical volumes, but two patients died of unrelated disorders. Tumor reduction was noted in 5 of 11 patients. Five patients had deﬁned additional growth and underwent repeat resection [98].

Hemangioblastoma

Invasive Skull Base Cancers

Thirty-six patients with intracranial hemangioblastomas, usually in conjunction with the syndrome of von Hippel–Lindau disease (VHL), have been treated by radiosurgery at our center. Early experience from several centers indicated that radiosurgery could lead to tumor control or regression [115, 116] (Fig. 9-7). For the most part, we have treated tumors with documented tumor growth, which are usually solid, and almost exclusively located in the posterior fossa, cerebellum, and brain stem. Such tumors are generally treated when they have shown evidence of objective growth and neurologic symptoms develop. Prophylactic radiosurgery for hemangioblastomas in the case of VHL is not performed unless tumor growth or new symptoms are documented. Multifocality is often a characteristic of the 20% of hemangioblastomas that are associated with VHL. Radiosurgery is a potential therapeutic option for these patients where resection of multiple tumors might be precluded because of brain location. For those patients with cystic hemangioblastomas, we have less optimism related to the overall role of radiosurgery at least as a single option. Cyst-associated tumors with nonenhancing cyst cavities were controlled by including only the enhancing nodule in the target volume; however, surgical removal of a large cystic component of a tumor producing mass effect symptoms is usually appropriate followed by radiosurgery for any residual solid component. In selected cases, stereotactic aspiration of the cyst followed by subsequent radiosurgery is feasible. Repeat radiosurgery may be required over many years when other tumors show additional growth [117].

After combined otolaryngological and neurosurgical procedures, we have used adjuvant radiosurgery for invasive skull base cancers (28 patients over the past 17 years). Fourteen patients had adenocarcinomas, 13 squamous cell carcinomas, and 1 patient had a metastatic neuroendocrine tumor. In such cases, radiosurgery has been used as an adjuvant or in combination with external beam fractionated radiation therapy. Many reports have documented the role of radiosurgery as salvage procedure for malignant tumors involving the skull base [119–121].

Chordoma and Chondrosarcoma During our 17-year experience, 26 patients with chordoma and 17 patients with chondrosarcomas have undergone management with radiosurgery. We continue to regard these tumors as difﬁcult tumors to manage. Almost invariably, they require multimodality management over the course of many years. These invasive tumors provide a management challenge because of their critical location and their tendency to aggressively recur locally despite multimodality treatment. Radiosurgery has been used both as a primary and adjuvant management strategy.

Radiosurgery for Functional Brain Disorders Trigeminal Neuralgia Radiosurgery Our current experience included 513 patients, managed since 1992. There were 305 (60%) women and 208 men. The mean age was 68 years (range, 16 to 92 years), and the mean duration of symptoms was 8 years. Our last detailed review studied 220 consecutive radiosurgery procedures for typical trigeminal neuralgia, all performed between 1992 and 1998 [122]. All 220 patients had trigeminal neuralgia that was idiopathic, longstanding, and refractory to medication therapy. Most of the patients had a long history of medical treatment with the median symptom duration of 96 months (range, 3 to 564 months). Pain was predominately distributed in the V2 and V3 distributions of the trigeminal nerve (29.5%), followed by V2 alone (22.3%) and V3 alone (13.2%). Prior surgery was performed in 135 (61.4%) patients, including microvascular decompression, glycerol rhizotomy, radiofrequency rhizotomy, balloon microcompression, peripheral neurectomy, or ethanol injections. Thus, the majority of patients represented both medical and surgical failures. In the remaining 85 (38.6%) patients, radiosurgery was the ﬁrst surgical procedure performed. The median central dose at trigeminal nerve was 80 Gy (Fig. 9-8). The pain relief after radiosurgery was graded into four categories: excellent, good, fair, and poor. Complete pain relief without the use of any

122

a. niranjan et al.

FIGURE 9-8. (Left) Axial contrast-enhanced T1-weighted MR image showing trigeminal neuralgia radiosurgery dose plan. (Right) Axial contrast-enhanced T1-weighted MR image of the same patient 6 months later shows contrast enhancement at the site of radiosurgery.

medication was deﬁned as an excellent outcome. We recommended all patients with complete pain relief to taper off their medications, and some patients were in the process of tapering at the time of evaluation (or refused to taper off because of the fear of a recurrence). Those patients with complete pain relief but who were still using some medication were considered as good outcomes. Patients with partial pain relief (more than 50% pain relief) were considered to have a fair outcome [11]. No pain relief or less than 50% pain relief were considered as poor. Placement within a category was decided by the patient rather than by the physician. Criteria for improvement included a reduction in both the frequency and severity of pain attacks. Of the 220 patients, 47 (25.1%) required further additional surgical procedures because of poor pain control. These patients were considered as treatment failures (poor outcome), and the results after the additional procedure were excluded from this analysis. Most of the patients responded to radiosurgery within 6 months of the procedure (median, 2 months). The ﬁrst evaluation was performed for all patients within 6 months after radiosurgery. At the initial follow-up assessment, excellent results were obtained in 105 (47.7%) patients, and excellent plus good results were found in 139 (63.2%) patients. More than 50% pain relief (excellent, good, or fair) was noted in 181 (82.3%) patients. At the last follow-up evaluation, 88 (40%) patients had excellent outcomes, 121 (55.9%) patients had excellent plus good outcomes, and 152 (69.1%) patients were fair or better. Thirty (13.6%) patients had recurrence of pain after the initial achievement of pain relief (25 patients after complete relief, 5 patients after more than 50% relief) between 2 and 58 months after radiosurgery. Recurrences occurred at a mean of 15.4 months from irradiation. The median time to achieving more than 50% pain relief (excellent, good, or fair) was 2 months (2.0 ± 0.05), and median time to achieving complete pain relief (good or excellent) was also 2 months (2.0 ± 5.1). At 6 months after treatment, 81.4 ± 2.6% patients had achieved more than 50% pain relief, and by 12 months, 85.6 ± 2.47% (actuarial statistics). Complete pain relief (good or excellent) was achieved in 64.9 ± 3.2% of the patients at 6 months, 70.3 ± 3.16% by 1 year, and in 75.4 ± 3.49% of patients by 33 months.

Prior authors, including our group, noted a latency interval to pain relief of approximately 1 to 2 months; however, approximately 15% of patients had no improvement in their pain even after 12 months. The duration of pain relief after initial response in all patients was also analyzed. Patients who never responded to radiosurgery were recorded as having a relief duration of zero months. More than 50% pain relief (excellent, good, or fair) was achieved and maintained in 76% of patients at 1 year, 71% of patients at 2 years, 67% of patients at 3 years, 65% of patients at 3.5 years, and 56% of patients at 5 years. Complete pain relief (excellent or good) was achieved and maintained in 63.6 ± 3.3% of patients at 1 year, 59.2 ± 3.5% of patients at 2 years, and 56.6 ± 3.8% of patients at 3 years. A history of no prior surgery was the only factor signiﬁcantly associated (p = 0.01) with achieving and maintaining complete pain relief. No patient sustained an early complication after any radiosurgery procedure. Seventeen patients (7.7%) developed increased facial paresthesia and/or facial numbness that lasted more than 6 months. Others have noted a dry eye, without signiﬁcant facial numbness. The median time to developing paresthesia was 8 months (range, 1 to 19 months). After 19 months, no patient developed any new sensory symptoms. No patient developed a mastication deﬁcit after radiosurgery or noted problems in facial motor function. One patient (0.4%) developed deafferentation pain after radiosurgery. The low incidence of complications is the greatest advantage of stereotactic radiosurgery compared with all other surgical options. In this study, less than 10% of patients developed increased facial paresthesia and/or facial sensory loss. The majority of our patients described their numbness or paresthesia as minor and not bothersome. Radiosurgery can be repeated if pain returns after initial relief. We advocate repeat radiosurgery only if complete pain relief had been achieved with subsequent recurrence [123]. We advocate a maximum dose of 50 to 60 Gy at a second procedure, and usually target a volume anterior to the prior target. Doing so has led to a pain response similar to that after primary radiosurgery in properly selected patients.

Movement Disorder Radiosurgery There is a small subset of movement disorder patients who have conditions that may make them unacceptable candidates for invasive stereotactic neurosurgical intervention. Such conditions are chronic use of anticoagulants and severe cardiac or respiratory disease. In addition, very elderly and noncompliant patients are usually considered poor surgical candidates. Finally, some patients voluntarily choose a less-invasive alternative to open stereotactic technique. Stereotactic radiosurgery is an option for this subset of patients with movement disorders. Gamma Knife is the preferred radiosurgical tool for treatment of movement disorders.

Radiosurgical Targets for Functional Disorders VIM nucleus of the thalamus is targeted for tremor patients. In our series, we determined the VIM coordinates based on the position of the nucleus relative to the AC-PC line and the anatomic information gathered from very-high-resolution MRI.

9.

gamma knife radiosurgery

Current planning software incorporates computerized atlases, which can be registered with the images and can be projected on MR images. The x, y, and z coordinates can be determined for the target using either SurgiPlan (Elekta Corp, Georgia), GammaPlan (Elekta Corp) or Multiview (Elekta Corp). Duma et al. reported results of radiosurgical pallidotomy for Parkinson disease. The target localization of globus pallidus interna (GPi) was determined by coordinates based on anatomic information gathered from very-high-resolution MRI and subjective surgeon correlation with the Schaltenbrand atlas. The 50% isodose line of a single or double isocenter 4-mm collimator plan was placed at the center of GPi. Keep et al. performed radiosurgical subthalamotomy using a primary target of 13 mm lateral to, 2 mm posterior to the midpoint of, and 5 mm inferior to the AC-PC line. This was optimized using atlas reference and experience. We are hesitant to perform radiosurgical pallidotomy because of the nearby optic tract. Friehs et al. reported targeting the center of the heads of the caudate nuclei bilaterally to treat the bradykinesia and rigidity of parkinsonism, and Pan et al. targeted the anterior portion of the VL nucleus for dystonia.

Gamma Knife Thalamotomy We have previously reported our experience with the treatment of essential tremor (ET) and Parkinson disease (PD) tremor using Gamma Knife radiosurgery. Niranjan et al. evaluated 11 patients managed with Gamma Knife thalamotomy for essential and multiple sclerosis (MS)–related tremor [124]. All patients noted improvement in action tremor. Six of eight ET patients had complete tremor arrest, and the violent action tremor in all three patients with MS was improved. One patient developed transient arm weakness. Duma et al. treated 42 patients with tremor from PD or ET with VIM thalamotomy using Gamma Knife. Median time of onset of improvement was 2 months (range, 1 week to 8 months) [125]. No change in tremor occurred in four Gamma Knife thalamotomies (8.6%), “mild” improvement was seen in 4 (8.6%), “good” improvement was seen in 13 (28%), and “excellent” improvement in 13 (28%). In 12 thalamotomies (26%), the tremor was eliminated completely. The high-dose (160 Gy mean maximum dose) thalamotomy lesion was more effective at reducing tremor than the low dose (120 Gy mean maximum dose). One patient, after bilateral treatment, suffered a mild acute dysarthria 1 week after GK thalamotomy. Ohye et al. reported 36 Gamma Knife thalamotomies in 31 patients. Maximum dose was 150 Gy in the ﬁrst 6 cases, which was subsequently reduced to 130 Gy [126, 127]. In two patients undergoing repeat procedures, the dose was decreased to 120 Gy. In all cases except one, a single 4-mm isocenter was used. In their 15 cases with more than 2 years follow-up, a clinically good result was seen in 87%, with no noticeable side effects. In a more recent report, these authors have compared the results of 51 patients who had thalamotomy after reloading of Gamma Knife with that of previous patients. The authors conﬁrmed two different patterns of post-radiosurgical lesions on follow-up imaging. One was a round punchedout lesion with enhancing borders with good symmetry, 7 to 8 mm in diameter to the enhancing edges. The second type of lesion seen extended to surrounding areas including the capsule with “rail-like” high signal along the border of the thalamus and

123

GP. Young et al. in a large series of patients reviewed their use of GK thalamotomy for the treatment of tremor [128]. Their series included 102 patients with parkinsonian tremor, 52 patients with essential tremor, and 4 patients with tremor of other etiology. The single 4-mm collimator was used with doses varying from 110 to 160 Gy. At a median follow-up of 52.5 months (range, 11 to 93 months), 76% were tremor free, and 12% were “nearly free of tremor.” Thus, there was failure in 12%. In 52 patients with disabling ET, median follow-up was 26 months. At 1 year, 92% were completely or nearly tremor free; after 4 years of follow-up this percentage decreased to 88%. One patient experienced a transient complication of contralateral balance disturbance; one patient had mild contralateral paresthesia in the face and upper extremity without detectable sensory deﬁcit and no impairment of function. A third patient had a mild weakness and dysphasia. All complications were believed to be due to lesions that became larger than expected. The overall complication rate was 1.3%.

Gamma Knife Pallidotomy Duma et al. performed Gamma Knife pallidotomy on 18 patients with medically recalcitrant and disabling symptoms of PD. Fifteen patients were treated using a single 4-mm collimator with a median central dose of 160 Gy (range, 90 to 165 Gy) [125]. Three patients were treated using a combination of two 4-mm shots with a dose of 160 Gy. Only 6 patients (33%) showed improvement in rigidity and dyskinesia. Three patients (17%) were unchanged, and nine patients (50%) worsened. Of the six patients with improvement, two exhibited visual ﬁeld deﬁcits. Overall, four (22%) patients had a visual ﬁeld deﬁcit, three patients had speech and/or swallowing difﬁculties, three had worsening of their gait, and one patient had numbness in the contralateral hemibody. Nine patients (50%) had one or more complications related to treatment. Okun et al. reported similar complications of GK pallidotomy in a report describing eight patients seen in an 8-month period referred for complications of GK radiosurgery [129]. Complications included hemiplegia, homonymous ﬁeld cut, weakness, dysarthria, hypophonia, aphasia, hemihypesthesia, and pseudobulbar laughter. Friedman et al. had similar experience [130]. They described their results in four patients using Gamma Knife pallidotomy in advanced disease. No patient improved in a signiﬁcant manner within the follow-up interval of 18 months. One patient experienced an improvement in his dyskinesia, but also became transiently psychotic and demented. The other three patients suffered no adverse effects.

Other Functional Targets Friehs et al. reported the efﬁcacy of GKRS caudatotomy for the treatment of the bradykinesia and rigidity of parkinsonism [131]. One month after treatment, 6 of 10 patients showed clear beneﬁt from bilateral 4-mm head of caudate lesions without any treatment-related complications. Keep et al. reported radiosurgical subthalamotomy using the GK in a single case report [132]. The 73-year-old patient received 120 Gy central dose using the 4-mm collimator helmet. At 2 weeks, she was able to reduce her Sinemet dose. At 5 weeks, she had no tremor, rigidity, or dyskinesia and walked easily with improved balance while using only a one-point cane for support. At 3

124

a. niranjan et al.

months, she had partial return of increased motor tone and cogwheel rigidity. At 1 year, after medication adjustments, she was able to move with ease and had no tremor. Imaging at 42 months showed a well-demarcated signal focus corresponding with the subthalamic nucleus. Pan et al. reported two patients who underwent radiosurgery for torsion spasm to evaluate the efﬁcacy of Gamma Knife radiosurgery as an alternative treatment. The target was located at the anterior portion of the ventrolateral nucleus. The maximum doses were 150 Gy and 145 Gy, respectively. Double isocenters with a 4-mm collimator were used. Follow-up lasted for 18 months and 8 months, respectively. Both patients had excellent clinical improvement 2 to 3 months after Gamma Knife radiosurgery, respectively. The authors concluded that Gamma Knife radiosurgery might be a safe and efﬁcient treatment for torsion spasm. Gamma Knife radiosurgical thalamotomy is a safe and effective alternative to invasive radiofrequency or DBS. This is not the case with radiosurgical pallidotomy. The paucity of radiosurgical pallidotomy reports in the literature reﬂects a lack of enthusiasm in the procedure. Subtle differences in lesion targeting have the potential to affect outcome. Without physiologic feedback, differentiation of internal and external globus pallidus is impossible during gamma pallidotomy. The lack of clinical improvement may therefore have been attributable to inaccurate physiologic lesioning within the GP without physiologic monitoring. The high complication rate of 50% in pallidotomy series is likely due to the variability and unpredictability of the lesion size when the globus pallidus serves as the target. This unpredictability and variability is not seen in the VIM thalamotomy series. It seems that there is a differential sensitivity to radiation between these two locations. Historically, the pallidum has exhibited a “supersensitivity” to hypoxia, and this may be the reason for higher complication rate. The pallidum is known to contain high levels of iron, which typically rises with age. It has been hypothesized that the presence of iron within this structure may catalyze free-radical reactions causing toxicity to the aging brain.

Conclusion In the past 15 years, we have witnessed dramatic improvements in the stereotactic radiosurgery technologies. Gamma Knife radiosurgery now offers better image-handling features including image fusion; faster, more compact platforms that make the calculations almost real-time; automated patient positioning reducing the potential for human error; and the inverse treatment planning. In the future, more accurate imaging techniques and improved software to handle those images as well as advanced inverse planning software will provide better treatment resulting in better patient outcomes.

References 1. Lunsford LD. Lars Leksell. Notes at the side of a raconteur. Stereotact Funct Neurosurg 1996; 67(3–4):153–168. 2. Ammar A. Lars Leksell’s vision—radiosurgery. Acta Neurochir Suppl 1994; 62:1–4. 3. Chang JH, Chang JW, Park YG, Chung SS. Factors related to complete occlusion of arteriovenous malformations after gamma knife radiosurgery. J Neurosurg 2000; 93(Suppl 3):96–101.

4. Coffey RJ, Nichols DA, Shaw EG. Stereotactic radiosurgical treatment of cerebral arteriovenous malformations. Gamma Unit Radiosurgery Study Group. Mayo Clin Proc 1995; 70(3):214–222. 5. Flickinger JC, Kondziolka D, Maitz AH, Lunsford LD. An analysis of the dose-response for arteriovenous malformation radiosurgery and other factors affecting obliteration. Radiother Oncol 2002; 63(3):347–354. 6. Kurita H, Kawamoto S, Sasaki T, et al. Results of radiosurgery for brain stem arteriovenous malformations. J Neurol Neurosurg Psychiatry 2000; 68(5):563–570. 7. Maruyama K, Kondziolka D, Niranjan A, Flickinger JC, Lunsford LD. Stereotactic radiosurgery for brainstem arteriovenous malformations: factors affecting outcome. J Neurosurg 2004; 100(3):407– 413. 8. Pollock BE, Gorman DA, Coffey RJ. Patient outcomes after arteriovenous malformation radiosurgical management: results based on a 5- to 14-year follow-up study. Neurosurgery 2003; 52(6):1291–1296. 9. Regis J, Massager N, Levrier O, et al. (Gamma-knife radiosurgery for brainstem arteriovenous malformations. Preliminary results]. Neurochirurgie 2001; 47(2–3 Pt 2):291–297. 10. Steiner L, Lindquist C, Adler JR, Torner JC, Alves W, Steiner M. Clinical outcome of radiosurgery for cerebral arteriovenous malformations. J Neurosurg 1992; 77(1):1–8. 11. Szeifert GT, Major O, Kemeny AA. Ultrastructural changes in arteriovenous malformations after gamma knife surgery: an electron microscopic study. J Neurosurg 2005; 102(Suppl):289–292. 12. Shin M, Maruyama K, Kurita H, et al. Analysis of nidus obliteration rates after gamma knife surgery for arteriovenous malformations based on long-term follow-up data: the University of Tokyo experience. J Neurosurg 2004; 101(1):18–24. 13. Pollock BE, Flickinger JC, Lunsford LD, Maitz A, Kondziolka D. Factors associated with successful arteriovenous malformation radiosurgery. Neurosurgery 1998; 42(6):1239–1244. 14. Pollock BE, Kline RW, Stafford SL, Foote RL, Schomberg PJ. The rationale and technique of staged-volume arteriovenous malformation radiosurgery. Int J Radiat Oncol Biol Phys 2000; 48(3):817–824. 15. Pollock BE, Flickinger JC, Lunsford LD, Bissonette DJ, Kondziolka D. Hemorrhage risk after stereotactic radiosurgery of cerebral arteriovenous malformations. Neurosurgery 1996; 38(4): 652–659. 16. Izawa M, Hayashi M, Chernov M, et al. Long-term complications after gamma knife surgery for arteriovenous malformations. J Neurosurg 2005; 102(Suppl):34–37. 17. Maesawa S, Flickinger JC, Kondziolka D, Lunsford LD. Repeated radiosurgery for incompletely obliterated arteriovenous malformations. J Neurosurg 2000; 92(6):961–970. 18. Liscak R, Vladyka V, Simonova G, Vymazal J, Novotny J Jr. Gamma knife surgery of brain cavernous hemangiomas. J Neurosurg 2005; 102(Suppl):207–213. 19. Hasegawa T, McInerney J, Kondziolka D, Lee JY, Flickinger JC, Lunsford LD. Long-term results after stereotactic radiosurgery for patients with cavernous malformations. Neurosurgery 2002; 50(6):1190–1197. 20. van Eck AT, Horstmann GA. Increased preservation of functional hearing after gamma knife surgery for vestibular schwannoma. J Neurosurg 2005; 102(Suppl):204–206. 21. Unger F, Walch C, Schrottner O, Eustacchio S, Sutter B, Pendl G. Cranial nerve preservation after radiosurgery of vestibular schwannomas. Acta Neurochir Suppl 2002; 84:77–83. 22. Prasad D, Steiner M, Steiner L. Gamma surgery for vestibular schwannoma.(see comment]. J Neurosurg 2000; 92(5):745–759. 23. Noren G. Long-term complications following gamma knife radiosurgery of vestibular schwannomas. Stereotact Funct Neurosurg 1998; 70(Suppl 1):65–73.

9.

gamma knife radiosurgery

24. Niranjan A, Lunsford LD, Flickinger JC, Maitz A, Kondziolka D. Dose reduction improves hearing preservation rates after intracanalicular acoustic tumor radiosurgery. Neurosurgery 1999; 45(4):753–762. 25. Nakamura H, Jokura H, Takahashi K, Boku N, Akabane A, Yoshimoto T. Serial follow-up MR imaging after gamma knife radiosurgery for vestibular schwannoma. Ajnr: Am J Neuroradiol 2000; 21(8):1540–1546. 26. Muacevic A, Jess-Hempen A, Tonn JC, Wowra B. Results of outpatient gamma knife radiosurgery for primary therapy of acoustic neuromas. Acta Neurochir Suppl 2004; 91:75–78. 27. Lunsford LD, Niranjan A, Flickinger JC, Maitz A, Kondziolka D. Radiosurgery of vestibular schwannomas: summary of experience in 829 cases. J Neurosurg 2005; 102(Suppl):195–199. 28. Kondziolka D, Lunsford LD, McLaughlin MR, Flickinger JC. Long-term outcomes after radiosurgery for acoustic neuromas.[see comment]. N Engl J Med 1998; 339(20):1426–1433. 29. Flickinger JC, Kondziolka D, Niranjan A, Maitz A, Voynov G, Lunsford LD. Acoustic neuroma radiosurgery with marginal tumor doses of 12 to 13 Gy. Int J Radiat Oncol Biol Phys 2004; 60(1):225–230. 30. Delbrouck C, Hassid S, Massager N, et al. Preservation of hearing in vestibular schwannomas treated by radiosurgery using Leksell Gamma Knife: preliminary report of a prospective Belgian clinical study. Acta Otorhinolaryngol Belg 2003; 57(3):197–204. 31. Gardner G, Robertson JH. Hearing preservation in unilateral acoustic neuroma surgery. Ann Otol Rhinol Laryngol 1988; 97(1):55–66. 32. House JW, Brackmann DE. Facial nerve grading system. Otolaryngol Head Neck Surg 1985; 93(2):146–147. 33. Flickinger JC, Kondziolka D, Niranjan A, Lunsford LD. Results of acoustic neuroma radiosurgery: an analysis of 5 years’ experience using current methods.[see comment]. J Neurosurg 2001; 94(1):1–6. 34. Kondziolka D, Nathoo N, Flickinger JC, Niranjan A, Maitz AH, Lunsford LD. Long-term results after radiosurgery for benign intracranial tumors. Neurosurgery 2003; 53(4):815–821. 35. Subach BR, Kondziolka D, Lunsford LD, Bissonette DJ, Flickinger JC, Maitz AH. Stereotactic radiosurgery in the management of acoustic neuromas associated with neuroﬁbromatosis Type 2. J Neurosurg 1999; 90(5):815–822. 36. Subach BR, Kondziolka D, Lunsford LD, Bissonette DJ, Flickinger JC, Maitz AH. Stereotactic radiosurgery in the management of acoustic neuromas associated with neuroﬁbromatosis Type 2.[see comment]. J Neurosurg 1999; 90(5):815–822. 37. Aichholzer M, Bertalanffy A, Dietrich W, et al. Gamma knife radiosurgery of skull base meningiomas. Acta Neurochir (Wien) 2000; 142(6):647–652. 38. Chang JH, Chang JW, Choi JY, Park YG, Chung SS. Complications after gamma knife radiosurgery for benign meningiomas. J Neurol Neurosurg Psychiatry 2003; 74(2):226–230. 39. DiBiase SJ, Kwok Y, Yovino S, et al. Factors predicting local tumor control after gamma knife stereotactic radiosurgery for benign intracranial meningiomas. Int J Radiat Oncol Biol Phys 2004; 60(5):1515–1519. 40. Eustacchio S, Trummer M, Fuchs I, Schrottner O, Sutter B, Pendl G. Preservation of cranial nerve function following Gamma Knife radiosurgery for benign skull base meningiomas: experience in 121 patients with follow-up of 5 to 9.8 years. Acta Neurochir Suppl 2002; 84:71–76. 41. Feigl GC, Bundschuh O, Gharabaghi A, Samii M, Horstmann GA. Volume reduction in meningiomas after gamma knife surgery. J Neurosurg 2005; 102(Suppl):189–194. 42. Flickinger JC, Kondziolka D, Maitz AH, Lunsford LD. Gamma knife radiosurgery of imaging-diagnosed intracranial meningioma. Int J Radiat Oncol Biol Phys 2003; 56(3):801–806.

125

43. Huffmann BC, Reinacher PC, Gilsbach JM. Gamma knife surgery for atypical meningiomas. J Neurosurg 2005; 102(Suppl):283–286. 44. Iwai Y, Yamanaka K, Ishiguro T. Gamma knife radiosurgery for the treatment of cavernous sinus meningiomas. Neurosurgery 2003; 52(3):517–524. 45. Kim DG, Kim Ch H, Chung HT, et al. Gamma knife surgery of superﬁcially located meningioma. J Neurosurg 2005; 102(Suppl): 255–258. 46. Kobayashi T, Kida Y, Mori Y. Long-term results of stereotactic gamma radiosurgery of meningiomas. Surg Neurol 2001; 55(6): 325–331. 47. Kondziolka D, Flickinger JC, Perez B. Judicious resection and/or radiosurgery for parasagittal meningiomas: outcomes from a multicenter review. Gamma Knife Meningioma Study Group. Neurosurgery 1998; 43(3):405–413. 48. Kondziolka D, Levy EI, Niranjan A, Flickinger JC, Lunsford LD. Long-term outcomes after meningioma radiosurgery: physician and patient perspectives. J Neurosurg 1999; 91(1):44–50. 49. Lee JY, Niranjan A, McInerney J, Kondziolka D, Flickinger JC, Lunsford LD. Stereotactic radiosurgery providing long-term tumor control of cavernous sinus meningiomas. J Neurosurg 2002; 97(1):65–72. 50. Liscak R, Simonova G, Vymazal J, Janouskova L, Vladyka V. Gamma knife radiosurgery of meningiomas in the cavernous sinus region. Acta Neurochir (Wien) 1999; 141(5):473–480. 51. Morita A, Coffey RJ, Foote RL, Schiff D, Gorman D. Risk of injury to cranial nerves after gamma knife radiosurgery for skull base meningiomas: experience in 88 patients. J Neurosurg 1999; 90(1):42–49. 52. Muthukumar N, Kondziolka D, Lunsford LD, Flickinger JC. Stereotactic radiosurgery for tentorial meningiomas. Acta Neurochir (Wien) 1998; 140(4):315–320; discussion 20–21. 53. Nicolato A, Foroni R, Alessandrini F, Maluta S, Bricolo A, Gerosa M. The role of Gamma Knife radiosurgery in the management of cavernous sinus meningiomas. Int J Radiat Oncol Biol Phys 2002; 53(4):992–1000. 54. Nicolato A, Foroni R, Alessandrini F, Bricolo A, Gerosa M. Radiosurgical treatment of cavernous sinus meningiomas: experience with 122 treated patients. Neurosurgery 2002; 51(5):1153– 1159. 55. Pan DH, Guo WY, Chang YC, et al. The effectiveness and factors related to treatment results of gamma knife radiosurgery for meningiomas. Stereotact Funct Neurosurg 1998; 70(Suppl 1):19–32. 56. Pendl G, Schrottner O, Eustacchio S, Feichtinger K, Ganz J. Stereotactic radiosurgery of skull base meningiomas. Minim Invasive Neurosurg 1997; 40(3):87–90. 57. Pendl G, Eustacchio S, Unger F. Radiosurgery as alternative treatment for skull base meningiomas. J Clin Neurosci 2001; 8(Suppl 1):12–14. 58. Roche PH, Pellet W, Fuentes S, Thomassin JM, Regis J. Gamma knife radiosurgical management of petroclival meningiomas results and indications. Acta Neurochir (Wien) 2003; 145(10):883– 888. 59. Sheehan JP, Kondziolka D, Flickinger J, Lunsford LD. Radiosurgery for nonfunctioning pituitary adenoma. Neurosurg Focus 2003; 14(5):1–9. 60. Pollock BE, Nippoldt TB, Stafford SL, Foote RL, Abboud CF. Results of stereotactic radiosurgery in patients with hormoneproducing pituitary adenomas: factors associated with endocrine normalization. J Neurosurg 2002; 97(3):525–530. 61. Landolt AM, Lomax N. Gamma knife radiosurgery for prolactinomas. J Neurosurg 2000; 93(Suppl 3):14–18. 62. Attanasio R, Epaminonda P, Motti E, et al. Gamma-knife radiosurgery in acromegaly: a 4-year follow-up study. J Clin Endocrinol Metab 2003; 88(7):3105–3112.

126

a. niranjan et al.

63. Castinetti F, Taieb D, Kuhn JM, et al. Outcome of gamma knife radiosurgery in 82 patients with acromegaly: correlation with initial hypersecretion. J Clin Endocrinol Metab 2005; 90(8):4483– 4488. 64. Gutt B, Wowra B, Alexandrov R, et al. Gamma-knife surgery is effective in normalising plasma insulin-like growth factor I in patients with acromegaly. Exp Clin Endocrinol Diabetes 2005; 113(4):219–224. 65. Niranjan A, Szeifert G, Kondziolka D, Flickinger JC, Maitz A, Lunsford LD, eds. Gamma Knife Radiosurgery for Growth Hormone-Secreting Pituitary Adenoma. Basal: Karger, 2002. 66. Jane JA Jr, Vance ML, Woodburn CJ, Laws ER Jr. Stereotactic radiosurgery for Hypersecreting pituitary tumors: part of multimodality approach. Neurosurgery Focus 2003; 14(14):1–5. 67. Laws ER Jr, Vance ML. Radiosurgery for pituitary tumors and craniopharyngiomas. Neurosurg Clin N Am 1999; 10(2):327–336. 68. Rahn T, Thoren M, Hall K, Backlund EO. Stereotactic radiosurgery in Cushing’s syndrome: acute radiation effects. Surg Neurol 1980; 14(2):85–92. 69. Sheehan JM, Vance ML, Sheehan JP, Ellegala DB, Laws ER Jr. Radiosurgery for Cushing’s disease after failed transsphenoidal surgery. J Neurosurg 2000; 93(5):738–742. 70. Kobayashi T, Kida Y, Mori Y. Gamma knife radiosurgery in the treatment of Cushing disease: long-term results. J Neurosurg 2002; 97(5 Suppl):422–428. 71. Pollock BE, Young WF Jr. Stereotactic radiosurgery for patients with ACTH-producing pituitary adenomas after prior adrenalectomy. Int J Radiat Oncol Biol Phys 2002; 54(3):839–841. 72. Vladyka V, Liscak R, Novotny J Jr, Marek J, Jezkova J. Radiation tolerance of functioning pituitary tissue in gamma knife surgery for pituitary adenomas. Neurosurgery 2003; 52(2):309–316. 73. Landolt AM, Haller D, Lomax N, et al. Octreotide may act as a radioprotective agent in acromegaly. J Clin Endocrinol Metab 2000; 85(3):1287–1289. 74. Larson DA, Gutin PH, McDermott M, et al. Gamma knife for glioma: selection factors and survival. Int J Radiat Oncol Biol Phys 1996; 36(5):1045–1053. 75. Nwokedi EC, DiBiase SJ, Jabbour S, Herman J, Amin P, Chin LS. Gamma knife stereotactic radiosurgery for patients with glioblastoma multiforme. Neurosurgery 2002; 50(1):41–46. 76. Simonova G, Novotny J Jr, Liscak R. Low-grade gliomas treated by fractionated gamma knife surgery. J Neurosurg 2005; 102(Suppl):19–24. 77. Kida Y, Kobayashi T, Mori Y. Gamma knife radiosurgery for low-grade astrocytomas: results of long-term follow up. J Neurosurg 2000; 93(Suppl 3):42–46. 78. Hadjipanayis CG, Kondziolka D, Gardner P, et al. Stereotactic radiosurgery for pilocytic astrocytomas when multimodal therapy is necessary. J Neurosurg 2002; 97(1):56–64. 79. Hadjipanayis CG, Kondziolka D, Flickinger JC, Lunsford LD. The role of stereotactic radiosurgery for low-grade astrocytomas. Neurosurg Focus 2003; 14(5):1–15. 80. Hadjipanayis CG, Niranjan A, Tyler-Kabara E, Kondziolka D, Flickinger JC, Lunsford LD. Stereotactic radiosurgery for wellcircumscribed ﬁbrillary grade II astrocytomas: an initial experience. Stereotact Funct Neurosurg 2002; 79(1):13–24. 81. Amendola BE, Wolf AL, Coy SR, Amendola M, Bloch L. Gamma knife radiosurgery in the treatment of patients with single and multiple brain metastases from carcinoma of the breast. Cancer J 2000; 6(2):88–92. 82. Gerosa M, Nicolato A, Foroni R, et al. Gamma knife radiosurgery for brain metastases: a primary therapeutic option. J Neurosurg 2002; 97(5 Suppl):515–524. 83. Gerosa M, Nicolato A, Foroni R, Tomazzoli L, Bricolo A. Analysis of long-term outcomes and prognostic factors in patients with

84.

85.

86.

87.

88.

89.

90.

91.

92.

93.

94.

95.

96. 97.

98.

99.

100.

101.

102.

non-small cell lung cancer brain metastases treated by gamma knife radiosurgery. J Neurosurg 2005; 102(Suppl):75–80. Hasegawa T, Kondziolka D, Flickinger JC, Germanwala A, Lunsford LD. Brain metastases treated with radiosurgery alone: an alternative to whole brain radiotherapy? Neurosurgery 2003; 52(6):1318–1326. Hoffman R, Sneed PK, McDermott MW, et al. Radiosurgery for brain metastases from primary lung carcinoma. Cancer J 2001; 7(2):121–131. Lippitz BE, Kraepelien T, Hautanen K, et al. Gamma knife radiosurgery for patients with multiple cerebral metastases. Acta Neurochir Suppl 2004; 91:79–87. Nam TK, Lee JI, Jung YJ, et al. Gamma knife surgery for brain metastases in patients harboring four or more lesions: survival and prognostic factors. J Neurosurg 2005; 102(Suppl): 147–150. Pan HC, Sheehan J, Stroila M, Steiner M, Steiner L. Gamma knife surgery for brain metastases from lung cancer. J Neurosurg 2005; 102(Suppl):128–133. Petrovich Z, Yu C, Giannotta SL, O’Day S, Apuzzo ML. Survival and pattern of failure in brain metastasis treated with stereotactic gamma knife radiosurgery. J Neurosurg 2002; 97(5 Suppl):499– 506. Serizawa T, Saeki N, Higuchi Y, et al. Gamma knife surgery for brain metastases: indications for and limitations of a local treatment protocol. Acta Neurochir (Wien) 2005; 147(7):721–726. Sheehan JP, Sun MH, Kondziolka D, Flickinger J, Lunsford LD. Radiosurgery for non-small cell lung carcinoma metastatic to the brain: long-term outcomes and prognostic factors inﬂuencing patient survival time and local tumor control. J Neurosurg 2002; 97(6):1276–1281. Simonova G, Liscak R, Novotny J Jr, Novotny J. Solitary brain metastases treated with the Leksell gamma knife: prognostic factors for patients. Radiother Oncol 2000; 57(2):207–213. Shiau CY, Sneed PK, Shu HK, et al. Radiosurgery for brain metastases: relationship of dose and pattern of enhancement to local control. Int J Radiat Oncol Biol Phys 1997; 37(2):375–383. Rutigliano MJ, Lunsford LD, Kondziolka D, Strauss MJ, Khanna V, Green M. The cost effectiveness of stereotactic radiosurgery versus surgical resection in the treatment of solitary metastatic brain tumors. Neurosurgery 1995; 37(3):445–453; discussion 53– 55. Reyns N, Hayashi M, Chinot O, et al. The role of Gamma Knife radiosurgery in the treatment of pineal parenchymal tumours. Acta Neurochir (Wien) 2006; 148(1):5–11. Kobayashi T, Kida Y, Mori Y. Stereotactic gamma radiosurgery for pineal and related tumors. J Neurooncol 2001; 54(3):301–309. Miller RC, Foote RL, Coffey RJ, et al. The role of stereotactic radiosurgery in the treatment of malignant skull base tumors. Int J Radiat Oncol Biol Phys 1997; 39(5):977–981. Muthukumar N, Kondziolka D, Lunsford LD, Flickinger JC. Stereotactic radiosurgery for chordoma and chondrosarcoma: further experiences. Int J Radiat Oncol Biol Phys 1998; 41(2):387–392. Nettel B, Niranjan A, Martin JJ, et al. Gamma knife radiosurgery for trigeminal schwannomas. Surg Neurol 2004; 62(5):435–444; discussion 44–46. Pan L, Wang EM, Zhang N, et al. Long-term results of Leksell gamma knife surgery for trigeminal schwannomas. J Neurosurg 2005; 102(Suppl):220–224. Pollock BE, Foote RL, Stafford SL. Stereotactic radiosurgery: the preferred management for patients with nonvestibular schwannomas? Int J Radiat Oncol Biol Phys 2002; 52(4):1002–1007. Saringer W, Khayal H, Ertl A, Schoeggl A, Kitz K. Efﬁciency of gamma knife radiosurgery in the treatment of glomus jugulare tumors. Minim Invasive Neurosurg 2001; 44(3):141–146.

9.

gamma knife radiosurgery

103. Zhang N, Pan L, Dai JZ, Wang BJ, Wang EM, Cai PW. Gamma knife radiosurgery for jugular foramen schwannomas. J Neurosurg 2002; 97(5 Suppl):456–458. 104. Muthukumar N, Kondziolka D, Lunsford LD, Flickinger JC. Stereotactic radiosurgery for jugular foramen schwannomas. Surg Neurol 1999; 52(2):172–179. 105. Chiou SM, Lunsford LD, Niranjan A, Kondziolka D, Flickinger JC. Stereotactic radiosurgery of residual or recurrent craniopharyngioma, after surgery, with or without radiation therapy. NeuroOncology 2001; 3(3):159–166. 106. Hasegawa T, Kondziolka D, Hadjipanayis CG, Lunsford LD. Management of cystic craniopharyngiomas with phosphorus-32 intracavitary irradiation. Neurosurgery 2004; 54(4):813–820. 107. Chung WY, Pan DH, Shiau CY, Guo WY, Wang LW. Gamma knife radiosurgery for craniopharyngiomas. J Neurosurg 2000; 93(Suppl 3):47–56. 108. Amendola BE, Wolf A, Coy SR, Amendola MA. Role of radiosurgery in craniopharyngiomas: a preliminary report. Med Pediatr Oncol 2003; 41(2):123–127. 109. Ulfarsson E, Lindquist C, Roberts M, et al. Gamma knife radiosurgery for craniopharyngiomas: long-term results in the ﬁrst Swedish patients. J Neurosurg 2002; 97(5 Suppl):613–622. 110. Pollock BE. Stereotactic radiosurgery in patients with glomus jugulare tumors. Neurosurg Focus 2004; 17(1–3):E10. 111. Thompson TP, Lunsford LD, Flickinger JC. Radiosurgery for hemangiomas of the cavernous sinus and orbit: technical case report. Neurosurgery 2000; 47(3):778–783. 112. Peker S, Kilic T, Sengoz M, Pamir MN. Radiosurgical treatment of cavernous sinus cavernous haemangiomas. Acta Neurochir 2004; 146(4):337–341. 113. Nakamura N, Shin M, Tago M, et al. Gamma knife radiosurgery for cavernous hemangiomas in the cavernous sinus. Report of three cases. J Neurosurg 2002; 97(5 Suppl):477–480. 114. Kida Y, Kobayashi T, Mori Y. Radiosurgery of cavernous hemangiomas in the cavernous sinus. Surg Neurol 2001; 56(2):117–122. 115. Georg AE, Lunsford LD, Kondziolka D, Flickinger JC, Maitz A. Hemangioblastoma of the posterior fossa. The role of multimodality treatment. Arquivos de Neuro-Psiquiatria 1997; 55(2): 278–286. 116. Pan L, Wang EM, Wang BJ, et al. Gamma knife radiosurgery for hemangioblastomas. Stereotact Funct Neurosurg 1998; 70(Suppl 1):179–186. 117. Jawahar A, Kondziolka D, Garces YI, Flickinger JC, Pollock BE, Lunsford LD. Stereotactic radiosurgery for hemangioblastomas of the brain. Acta Neurochir 2000; 142(6):641–644. 118. Kondziolka D, Lunsford LD, Flickinger JC. The role of radiosurgery in the management of chordoma and chondrosarcoma of the cranial base. Neurosurgery 1991; 29(1):38–45.

127

119. Kondziolka D, Lunsford LD. Stereotactic radiosurgery for squamous cell carcinoma of the nasopharynx. Laryngoscope 1991; 101(5):519–522. 120. Firlik KS, Kondziolka D, Lunsford LD, Janecka IP, Flickinger JC. Radiosurgery for recurrent cranial base cancer arising from the head and neck. Head & Neck 1996; 18(2):160–165. 121. Lee N, Millender LE, Larson DA, et al. Gamma knife radiosurgery for recurrent salivary gland malignancies involving the base of skull. Head & Neck 2003; 25(3):210–216. 122. Maesawa S, Salame C, Flickinger JC, Pirris S, Kondziolka D, Lunsford LD. Clinical outcomes after stereotactic radiosurgery for idiopathic trigeminal neuralgia. J Neurosurg 2001; 94(1):14– 20. 123. Hasegawa T, Kondziolka D, Spiro R, Flickinger JC, Lunsford LD. Repeat radiosurgery for refractory trigeminal neuralgia. Neurosurgery 2002; 50(3):494–500. 124. Niranjan A, Kondziolka D, Baser S, Heyman R, Lunsford LD. Functional outcomes after gamma knife thalamotomy for essential tremor and MS-related tremor. Neurology 2000; 55(3):443– 446. 125. Duma CM. Movement Disorder Radiosurgery. Philadelphia: Lippincott, Williams & Wilkins, Inc., 2003. 126. Ohye C, Shibazaki T, Ishihara J, Zhang J. Evaluation of gamma thalamotomy for parkinsonian and other tremors: survival of neurons adjacent to the thalamic lesion after gamma thalamotomy. J Neurosurg 2000; 93(Suppl 3):120–127. 127. Ohye C, Shibazaki T, Zhang J, Andou Y. Thalamic lesions produced by gamma thalamotomy for movement disorders. J Neurosurg 2002; 97(5 Suppl):600–606. 128. Young RF. Gamma knife radiosurgery as an alternative form of therapy for movement disorders. Arch Neurol 2002; 59(10):1660– 1662; author reply 2–4. 129. Okun MS, Stover NP, Subramanian T, et al. Complications of gamma knife surgery for Parkinson disease. Arch Neurol 2001; 58(12):1995–2002. 130. Friedman DP, Goldman HW, Flanders AE, Gollomp SM, Curran WJ Jr. Stereotactic radiosurgical pallidotomy and thalamotomy with the gamma knife: MR imaging ﬁndings with clinical correlation–preliminary experience. Radiology 1999; 212(1): 143–150. 131. Friehs GM, Ojakangas CL, Schrottner O, Ott E, Pendl G. Radiosurgical lesioning of the caudate nucleus as a treatment for parkinsonism: a preliminary report. Neurol Res 1997; 19(1): 97–103. 132. Keep MF, Mastrofrancesco L, Erdman D, Murphy B, Ashby LS. Gamma knife subthalamotomy for Parkinson disease: the subthalamic nucleus as a new radiosurgical target. Case report. J Neurosurg 2002; 97(5 Suppl):592–599.

1 0

Linear Accelerator Radiosurgery William A. Friedman

Introduction Stereotactic radiosurgery (SRS) is a minimally invasive treatment modality that delivers a large, single dose of radiation to a speciﬁc intracranial target while sparing surrounding tissue. Unlike conventional fractionated radiotherapy, SRS does not maximally exploit the higher radiosensitivity of brain lesions relative to normal brain (therapeutic ratio). Its selective destruction is dependent mainly on sharply focused, high-dose radiation and a steep dose gradient away from the deﬁned target. The biological effect is irreparable cellular damage (probably via DNA strand breaks) and delayed vascular occlusion within the high-dose target volume. Because a therapeutic ratio is not required, traditionally radioresistant lesions can be treated. Because destructive doses are used, however, any normal structure included in the target volume is subject to damage. The basis for SRS was conceived more than 40 years ago by Lars Leksell [1]. He proposed the technique of focusing multiple beams of external radiation on a stereotactically deﬁned intracranial target. The averaging of these intersecting beams results in very high doses of radiation to the target volume but innocuously low doses to non–target tissues along the path of any given beam. His team’s implementation of this concept culminated in the development of the Gamma Knife. The modern Gamma Knife employs 201 ﬁxed cobalt radiation sources in a ﬁxed hemispherical array, such that all 201 photon beams are focused on a single point. The patient is stereotactically positioned in the Gamma Knife so that the intracranial target coincides with the isocenter of radiation. Using variable collimation, beam blocking, and multiple isocenters, the radiation target volume is shaped to conform to the intracranial target. An alternate radiosurgical solution using a linear accelerator (linac) was ﬁrst described in 1984 by Betti et al. [2]. Colombo et al. described such a system in 1985 [3], and linacs have subsequently been modiﬁed in various ways to achieve the precision and accuracy required for radiosurgical applications [4–7]. In 1986, a team composed of neurosurgeons, radiation oncologists, radiation physicists, and computer programmers began development of the University of Florida linac-based radiosurgery system [8]. This system has been used to treat more than 2000 patients at the University of Florida since May 1988 and is in use at multiple sites worldwide. Many other commercial versions of radiosurgical systems are currently available, includ-

ing the BrainLAB system, the Radionics (X-knife) system, the Accuray (CyberKnife) system, and others. Most linac radiosurgical systems rely on the same basic paradigm: A collimated X-ray beam is focused on a stereotactically identiﬁed intracranial target. The gantry of the linac rotates around the patient, producing an arc of radiation focused on the target (Figs. 10-1 and 10-2). The patient couch is then rotated in the horizontal plane and another arc performed. In this manner, multiple non-coplanar arcs of radiation intersect at the target volume and produce a high target dose, with minimal radiation to surrounding brain. This dose concentration method is exactly analogous to the multiple intersecting beams of cobalt radiation in the Gamma Knife. The target dose distribution can be tailored by varying collimator sizes, eliminating undesirable arcs, manipulating arc angles, using multiple isocenters, and differentially weighting the isocenters [9]. In our center, multiple isocenters are used to achieve highly conformal dose distributions, exactly analogous to the Gamma Knife technique (Fig. 10-3). Some linear accelerator systems use an alternative approach that relies upon a computer-driven multileaf collimator to generate nonspherical beam shapes that are conformal to the beam’s-eye view of the tumor. The multileaf collimator can be adjusted statically or dynamically as the linear accelerator rotates. Intensity modulation can be used to achieve dose distributions that are close to those seen with multiple isocenters, and treatment time can be reduced. Achievable dose distributions are similar for linac-based and Gamma Knife systems. With both systems, it is possible to achieve dose distributions that conform closely to the shape of the intracranial target, thus sparing the maximum amount of normal brain. Recent advances in stereotactic imaging and computer technology for dose planning, as well as reﬁnements in radiation delivery systems, have led to improved efﬁcacy, fewer complications, and a remarkable amount of interest in the various applications of SRS. Perhaps of equal importance is the fact that increasing amounts of scientiﬁc evidence have persuaded the majority of the international neurosurgical community that radiosurgery is a viable treatment option for selected patients suffering from a variety of challenging neurosurgical disorders. This chapter will present a brief description of linac radiosurgical technique, followed by a review of the more common applications of stereotactic radiosurgery in the treatment of

129

130

w.a. friedman

FIGURE 10-1. Linear accelerators are the preferred device, worldwide, for conventional radiotherapy. They accelerate electrons to near light speed, then collide them with a heavy metal (like tungsten) in the head of the machine. The collision mainly produces heat, but a small percentage of the energy is converted into highly energetic photons. These photons, because they are electronically produced, are called Xrays. The X-radiation is collimated and focused on the target.

FIGURE 10-3. This choroidal ﬁssure AVM required four 1-cm isocenters to produce a conformal plan. The inner line (70% isodose) is the prescription dose line. The outer line (35% isodose) is half of the prescription dose.

FIGURE 10-2. This diagram shows an add-on device, designed to improve the accuracy of the linear accelerator, in place. The linac arcs around the patient, with its beam always focused on the stereotactically

positioned target. The patient is then moved to a new horizontal (table) position and another arc performed. The result is multiple, noncoplanar arcs of radiation, all converging on the target point.

10.

linear accelerator radiosurgery

intracranial disease (benign tumors, malignant tumors, and arteriovenous malformations).

131

with these difﬁcult tumors continue to be less than optimal [10–12]. A signiﬁcant amount of experience has been accumulated using SRS in the treatment of schwannomas and meningiomas. We will focus on each of these tumor types in turn.

Linac Radiosurgery Technique Although the details of radiosurgical treatment techniques differ somewhat from system to system, the basic paradigm is quite similar everywhere. Following is a detailed description of a typical radiosurgical treatment at the University of Florida. Almost all radiosurgical procedures in adults are performed on an outpatient basis. The patient reports to the neurosurgical clinic the day before treatment for a detailed history and physical, as well as an in-depth review of the treatment options. If radiosurgery is deemed appropriate, the patient is sent to the radiology department for a volumetric magnetic resonance imaging (MRI) scan. A radiosurgical plan can be generated, in advance, using this MRI study. The next morning, the patient arrives at 7:00 a.m. A stereotactic head ring is applied under local anesthesia. No skin shaving or preparation is required. Subsequently, stereotactic computed tomography (CT) scanning is performed. One-millimeter slices are obtained throughout the entire head. The patient is then transported to an outpatient holding area where he and his family have breakfast and relax until the treatment planning process is complete. The stereotactic CT scan and the nonstereotactic volumetric MRI scan are transferred via Ethernet to the treatmentplanning computer. The CT images are quickly processed so that each pixel has an anteroposterior, lateral, and vertical stereotactic coordinate, all related to the head ring previously applied to the patient’s head. Using image fusion software, the nonstereotactic MRI is fused, pixel for pixel, with the stereotactic CT. The “pre-plan” performed the day before is, likewise, fused to the stereotactic CT. Final dosimetry then begins and continues until the neurosurgeon, radiation oncologist, and radiation physicist are satisﬁed that an optimal dose plan has been developed. A variety of options are available for optimizing the dosimetry. The fundamental goal is to deliver a radiation ﬁeld that is precisely conformal to the lesion shape (see Fig. 10-3) while delivering a minimal dose of radiation to all surrounding neural structures. A detailed discussion of dosimetric options is available in Chapter 7. When dose planning is complete, the radiosurgical device is attached to the linac. The patient then is attached to the device and treated. The head ring is removed and, after a short observation period, the patient is discharged. The radiosurgical device is disconnected from the linac, which is then ready for conventional usage. Close clinical and radiologic follow-up is arranged at appropriate intervals depending on the pathology treated and the condition of the patient.

Radiosurgery for Benign Tumors SRS has proved useful for the treatment of a variety of benign intracranial neoplasms. These tumors commonly arise from the skull base, where their dramatic impact on quality of life belies their benign histology and small size. Despite progressive improvement in microsurgical techniques, outcomes for patients

Vestibular Schwannomas Among benign intracranial tumors, vestibular schwannoma (acoustic neuroma) has to date been the most frequent target for stereotactic radiosurgery. This common tumor (representing approximately 10% of all primary brain tumors) is a benign proliferation of Schwann cells arising from the myelin sheath of the vestibular branches of the eighth cranial nerve. These tumors are slightly more common in women, present at an average age of 50 years, and occur bilaterally in patients with neuroﬁbromatosis type 2. Leksell ﬁrst used stereotactic radiosurgery to treat a vestibular schwannoma in 1969 [13]. SRS is a logical alternative treatment modality for this tumor for several reasons. A vestibular schwannoma is typically well demarcated from surrounding tissues on neuroimaging studies. The sharp borders of this noninvasive tumor make it a convenient match for the characteristically steep dose gradient produced at the boundary of a radiosurgical target. This allows the radiosurgeon to minimize radiation of normal tissue. Excellent spatial resolution on gadolinium-enhanced MRI facilitates radiosurgical dose planning. These tumors typically occur in an older population that may be less ﬁt for microsurgical resection under general anesthesia. Finally, the location of these tumors at the skull base in close proximity to multiple critical neurologic structures (i.e., cranial nerves, brain stem) leads to appreciable surgical morbidity and rare mortality even in expert hands. This makes the concept of an effective, less invasive, less morbid alternative treatment that can be performed in a single day under local anesthesia quite attractive. Whether or not radiosurgery ﬁts this description has been extensively debated. Certainly, the role of radiosurgery is limited by its inability to expeditiously relieve mass effect in patients for whom this is necessary. The radiobiology of SRS also requires lower, potentially less effective doses for higher target volumes in order to avoid complications. This limits the use of SRS to the treatment of smaller tumors. Despite these limitations, there is a growing body of literature that substantiates the claim that radiosurgery is a safe and effective alternative therapy for acoustic schwannomas. The published experience using linac-based radiosurgery for the treatment of vestibular schwannomas is relatively limited compared with the Gamma Knife literature. Foote et al. [14] performed an analysis of risk factors associated with radiosurgery for vestibular schwannoma at University of Florida (UF). The aim of this study was to identify factors associated with delayed cranial neuropathy after radiosurgery for vestibular schwannoma (VS) and to determine how such factors may be manipulated to minimize the incidence of radiosurgical complications while maintaining high rates of tumor control. From July 1988 to June 1998, 149 cases of VS were treated using linear accelerator radiosurgery at the University of Florida. In each of these cases, the patient’s tumor and brain stem were contoured in 1-mm slices on the original radiosurgical targeting images. Resulting tumor and brain-stem volumes were coupled

132

w.a. friedman

with the original radiosurgery plans to generate dose-volume histograms. Various tumor dimensions were also measured to estimate the length of cranial nerve that would be irradiated. Patient follow-up data, including evidence of cranial neuropathy and radiographic tumor control, were obtained from a prospectively maintained, computerized database. The authors performed statistical analyses to compare the incidence of posttreatment cranial neuropathies or tumor growth between patient strata deﬁned by risk factors of interest. One hundred thirty-nine of the 149 patients were included in the analysis of complications. The median duration of clinical follow-up for this group was 36 months (range, 18 to 94 months). The tumor control analysis included 133 patients. The median duration of radiology follow-up in this group was 34 months (range, 6 to 94 months). The overall 2-year actuarial incidences of facial and trigeminal neuropathies were 11.8% and 9.5%, respectively. In 41 patients treated before 1994, the incidences of facial and trigeminal neuropathies were both 29%, but in the 108 patients treated since January 1994, these rates declined to 5% and 2%, respectively. An evaluation of multiple risk factor models showed that maximum radiation dose to the brain stem, treatment era (pre-1994 compared with 1994 or later), and prior surgical resection were all simultaneously informative predictors of cranial neuropathy risk. The radiation dose prescribed to the tumor margin could be substituted for the maximum dose to the brain stem with a small loss in predictive strength. The overall radiologic tumor control rate was 93% (59% tumors regressed, 34% remained stable, and 7.5% enlarged), and the 5-year actuarial tumor control rate was 87% (95% conﬁdence interval [CI], 76% to 98%). Based on this study, the authors currently recommend a peripheral dose of 12.5 Gy for almost all acoustics as that dose most likely to yield long-term tumor control without causing cranial neuropathy. Spiegelmann et al. [15, 16] have reported their experience. They reviewed the methods and results of linac radiosurgery in 44 patients with acoustic neuromas who were treated between 1993 and 1997. CT scanning was selected as the stereotactic imaging modality for target deﬁnition. A single, conformally shaped isocenter was used in the treatment of 40 patients; two or three isocenters were used in four patients who harbored very irregular tumors. The radiation dose directed to the tumor border was the only parameter that changed during the study period: In the ﬁrst 24 patients who were treated the dose was 15 to 20 Gy, whereas in the last 20 patients the dose was reduced to 11 to 14 Gy. After a mean follow-up period of 32 months (range, 12 to 60 months), 98% of the tumors were controlled. The actuarial hearing preservation rate was 71%. New transient facial neuropathy developed in 24% of the patients and persisted to a mild degree in 8%. Radiation dose correlated signiﬁcantly with the incidence of cranial neuropathy, particularly in large tumors (≥4 cm3). Several reports on smaller series of patients treated with linac-based radiosurgery for vestibular schwannomas have been published in recent years. Martens et al. reported on 14 patients with at least 1 year of follow-up after radiosurgery on the linac unit in the University Hospital in Ghent, Belgium [17]. A mean marginal dose of 19.4 Gy (range, 16 to 20 Gy) was delivered to the 70% isodose line with a single isocenter. Mean follow-up duration was 19 months (range, 12 to 24 months). During this relatively short follow-up interval, 100% radiographic tumor

control has been achieved (29% regressed, 71% stable, zero enlarged). Rates of delayed facial and trigeminal neuropathy were 21% and 14%, respectively, and two of three facial nerve deﬁcits resolved. Preoperative hearing was preserved 50% of the time. Valentino and Raimondi reported on 23 patients treated with linac radiosurgery in Rome, Italy [18]. Five of these had neuroﬁbromatosis and seven (30%) had undergone previous surgery. Total radiation dose to the tumor margin ranged from 12 to 45 Gy (median, 30 Gy) and was delivered in one to ﬁve sessions. One or two isocenters were used, and mean duration of follow-up was 40 months (range, 24 to 46 months). Results using this less conventional method of multisession radiosurgery were comparable with other radiosurgical techniques. Tumor control was achieved in 96% of patients (38% regressed, 58% stable, 4% enlarged), facial and trigeminal neuropathies each occurred at a rate of 4%, and “hearing was preserved at almost the same level as that prior to radiosurgery in all patients.” The use of linac radiosurgery for acoustics is brieﬂy discussed in reports by Delaney [19] and Barcia [20]. In addition, fractionated stereotactic radiation therapy (SRT) has been used as an alternative management for vestibular schwannomas [1, 5]. This method is proposed as a way of exploiting the precision of stereotactic radiation delivery to minimize dose to normal brain while employing lower fractionated doses in an effort to minimize complications. Thus far, most radiosurgeons feel that optimal results can be achieved with highly conformal singlefraction radiosurgery while sparing the patient the inconvenience of a prolonged treatment course. As of April 2005, the University of Florida experience with vestibular schwannomas comprised 386 patients. The indications for radiosurgery were age >60 (180 cases), failed surgery (81), preference (118), medical inﬁrmity (6). The median treatment volume was 2 cm3. With a median follow-up of 32 months for the entire group, 108 tumors are unchanged, 154 are smaller (Figs. 10-4 and 10-5), and 11 (4%) tumors are larger. Only four

FIGURE 10-4. Pretreatment MRI scan shows left-sided vestibular schwannoma.

10.

linear accelerator radiosurgery

FIGURE 10-5. Four years after treatment, the MRI scan shows the schwannoma of Fig. 10-4 to be much smaller.

(1%) patients have required surgery because of tumor growth after radiosurgery.

Meningiomas Meningiomas are the most common benign primary brain tumor, with an incidence of approximately 7/100,000 in the general population. Surgery has long been thought to be the treatment of choice for symptomatic lesions and is often curative. Many meningiomas, however, occur in locations where attempted surgical cure may be associated with morbidity or mortality, such as the cavernous sinus or petroclival region [21, 22]. In addition, many of these tumors occur in the elderly, where the risks of general anesthesia and surgery are known to be increased. Hence, there is interest in alternative treatments, including radiation therapy and radiosurgery, either as a primary or adjuvant approach. Simpson, in a classic paper, described the relationship between completeness of surgical resection and tumor recurrence [23]. A grade I resection, which is complete tumor removal with excision of the tumor’s dural attachment and involved bone, has a 10% recurrence rate. A grade II resection, complete resection of the tumor and coagulation of its dural attachment, has up to a 20% recurrence rate. Grade III resection is complete tumor removal without dural resection or coagulation. Grade IV resection is subtotal, and grade V resection is simple decompression. Recurrence rates in the grades IV and V groups basically reﬂect the natural history of the tumor, with high rates of recurrence over time. Unfortunately, some common meningioma locations, such as the cavernous sinus or petroclival region, are not readily amenable to a complete dural resection or coagulation strategy because of location and the proximity of vital neural and vascular structures. In addition, relatively high complication rates have been described for meningioma surgery in some locations and in the elderly. Pollock and colleagues recently analyzed 198 patients with meningiomas less than 35 mm in diameter treated with either

133

surgical resection or Gamma Knife radiosurgery [24]. Tumor recurrence was more frequent in the surgical resection group (12% vs. 2%). No statistically signiﬁcant difference was detected in the 3- and 7-year actuarial progression-free survival rate between patients with Simpson grade 1 resections and those who underwent radiosurgery. Progression-free survival rates with radiosurgery were superior to Simpson grades 2, 3, and 4 resections. Complications were lower in the radiosurgery group. Multiple linear accelerator radiosurgical series have been published [25–28]. Hakim and colleagues described the largest such series, and the only one to report actuarial statistics [29]. One hundred twenty-seven patients with 155 meningiomas were treated. Actuarial tumor control for patients with benign tumors was 89.3% at 5 years. Six (4.7%) patients had permanent radiation-induced complications. The University of Florida report on linear accelerator radiosurgery treatment of meningiomas is the largest yet published [30]. Two hundred ten patients were treated from May 1989 through December 2001. All patients had follow-up for a minimum of 2 years, and no patients were lost to follow-up. Actuarial local control for benign tumors was 100% at 1 and 2 years and 96% at 5 years (Figs. 10-6 and 10-7). Actuarial local control for atypical tumors was 100% at 1 year, 92% at 2 years, and 77% at 5 years. Actual control for malignant tumors was 100% at 1 and 2 years but only 19% at 5 years. Permanent radiation-induced complications occurred in 3.8%, all of which involved malignant tumors. These tumor control and treatment morbidity rates compare well with all other published series. We found that reliance on imaging characteristics rather than surgical pathology did not yield a high incidence of missed diagnoses. During the time interval of this study, only two patients were treated as presumed meningiomas and later found to have other diagnoses. One had a dural-based metastasis that was surgically excised when it enlarged. The other had a heman-

FIGURE 10-6. MRI scan shows right cavernous sinus meningioma. The patient presented with a sixth nerve paresis.

134

w.a. friedman

FIGURE 10-7. Three years later, the meningioma of Fig. 10-6 is barely visible. The sixth nerve paresis is completely resolved. We believe that radiosurgery is the treatment of choice for many cavernous sinus meningiomas.

giopericytoma of the lateral cavernous sinus that was surgically excised when it enlarged.

Radiosurgery for Malignant Tumors Malignant tumors are radiobiologically more amenable to fractionated radiotherapy than benign lesions. Malignancies tend to inﬁltrate surrounding brain, resulting in poorly deﬁnable tumor margins. A priori, these two traits of cerebral malignancies would seem to make radiosurgery an unattractive treatment option. Nevertheless, SRS has proved to be a useful weapon in the armamentarium against malignant brain tumors. The most common applications of SRS to malignant tumors are the treatment of cerebral metastases and the delivery of an adjuvant focal radiation “boost” to malignant gliomas.

Cerebral Metastases Metastatic brain tumors are up to 10 times more common than primary brain tumors with an annual incidence of between 80,000 and 150,000 new cases each year [31]. Fifteen percent to 40% of cancer patients will be diagnosed with a brain metastasis during the course of their illness. Once a brain metastasis has been diagnosed, the median life expectancy is less than 1 year; however, in many patients, aggressive treatment of metastatic disease has been shown to restore neurologic function and prevent further neurologic manifestations. Debate exists concerning the optimum treatment for metastatic brain disease. In autopsy series, brain metastases occur in up to 50% of cancer patients [32]. Approximately 30% to 40% present with a solitary metastasis. Brain metastases frequently cause debilitating symptoms that can seriously impact the patient’s quality

of life. With no treatment or steroid therapy alone, survival is limited (1 to 2 months). Whole-brain radiotherapy (WBRT) extends median survival, but the duration of survival is typically low (3 to 4 months). Several randomized trials have suggested that, when possible, surgery followed by WBRT is superior to WBRT alone. Patchell et al. reported a randomized clinical trial involving 46 patients with a single metastasis and well-controlled systemic disease [33]. They found a signiﬁcant improvement in survival (40 weeks vs. 15 weeks) and local recurrences in the CNS (20% vs. 52%) for patients in the surgery plus WBRT arm of the study. Likewise, Noordijk et al. randomized 66 patients and found a signiﬁcant survival advantage (10 vs. 6 months) for the combination therapy arm [34]. In contrast, Mintz et al. studied a group of 84 patients and did not show an advantage of surgery plus radiotherapy over radiotherapy alone [35]. It has been suggested that the inclusion of a higher percentage of patients with active systemic disease and lower performance scores did not allow the beneﬁt of improved local control to affect survival in this series. Haines points out that survival and quality of life are the most important outcomes measures in evaluating a clinical treatment for cancer [36]. Surrogate end points, like local control, are inherently unreliable, especially when the deﬁnition of local control is changed. This applies to a comparison of radiosurgery with surgery for brain metastasis. In surgical series, local control means no visible tumor on follow-up scans. In radiosurgical series, local control means no growth (or sometime minimal growth) on follow-up scans. These end points are unlikely to be equivalent. In addition, comparison of current results to historical controls is fraught with hazard to selection bias. This issue led to erroneous conclusions about the efﬁcacy of brachytherapy for malignant gliomas and to overly optimistic reports regarding the efﬁcacy of intraarterial chemotherapy. Of equal import is the difﬁculty and variability of reporting standards for local control. Few series provide actuarial local control. They simply provide a “raw” number at an arbitrary point in time. Less commonly appreciated is the difﬁculty in documenting local control. Many of these patients die away from the medical center where radiosurgery was performed. It is frequently impossible to determine from family or local physician telephone interview whether the proximate cause of death was loss of local control, new intracranial disease (loss of regional control), or systemic disease. Most radiosurgical series have assumed that, unless an MRI was performed documenting local loss of local control prior to death, local control was maintained. This assumption may lead to a systematic overestimation of local control rates. Sturm [37–39], Black [40, 41], and Adler [42–44] published early reports on linear accelerator radiosurgery for brain metastases. Alexander [41] reported on 248 patients. Median tumor volume was 3 cm3 and median tumor dose was 15 Gy. Median survival was 9.4 months. Actuarial local control was 85% at 1 year and 65% at 2 years. Auchter et al. reported a multi-institutional study of 122 patients [45]. Actuarial 1- and 2-year survivals were 53% and 30%, respectively. Local control was 86%. Many other linac series have been reported [39, 46–53]. As radiosurgery has emerged as a treatment option, clinicians have attempted to deﬁne prognostic factors, which may help to deﬁne patient populations most likely to beneﬁt from

10.

linear accelerator radiosurgery

radiosurgical treatment [54–56]. Multiple factors have been discerned from retrospective analysis and include Karnofsky performance scale score, status of systemic disease, histology, number of metastases, volume of metastases, time interval between the diagnosis of the primary lesion and the metastatic lesion, pattern of enhancement [57, 58], the Radiation Therapy Oncology Group (RTOG) recursive partitioning categories [59], and radiation dose. Recently, the University of Florida published their experience with radiosurgery for brain metastases [60]. Three hundred eighty-three patients were treated. Median survival was 9 months. Melanoma histology and increasing number of metastases predicted poorer survival. Increasing age, somewhat surprisingly, slightly improved survival, possibly because younger patients tended to have more radioresistant histologies. Actuarial local control was 75% (Figs. 10-8 and 10-9). Increasing dose provided better control, and eloquent location was also associated with better control (possibly because eloquent tumors tended to be discovered at a smaller size). Regional control was poorer in melanoma or breast patients and in those with synchronous presentation of brain metastasis and primary tumor. In this retrospective analysis, whole-brain radiotherapy did not improve regional control.

Malignant Gliomas Current conventional treatment for malignant gliomas involves a combination of surgery, radiation, and, often, chemotherapy. The prognosis in these patients remains poor [61]. The majority of recurrences occur within 2 cm of the enhancing lesion as seen on initial imaging. Gross total excision may be associated with prolonged median survival in patients with malignant gliomas. Some studies have shown that other aggressive local therapies, such as interstitial brachytherapy, may favorably impact survival [62–64]. Radiosurgery is another attempt at forestalling local recurrence by aggressive local therapy.

FIGURE 10-8. The patient with known breast carcinoma presented with symptomatic pontine lesions. She was treated with radiosurgery (15 Gy to the 80% isodose line).

135

FIGURE 10-9. Three years later, the site of the lesion of Fig. 10-8 was barely visible.

Malignant gliomas account for approximately 40% of the 17,000 primary brain tumors diagnosed annually in the United States. The prognosis for long-term survival remains poor. More than 80% of recurrences are found within 2 cm of the original tumor site. Many attempts have been made to improve long-term survival by improving local control [65, 66]. Such therapies include aggressive surgical removal, brachytherapy, chemotherapy wafers, and radiosurgery. In this retrospective study, we have attempted to ascertain whether the use of radiosurgical boost, whether given as part of initial tumor therapy or at the time of recurrence, increases survival compared to historical controls. A number of linear accelerator radiosurgery series have been published. Shrieve and colleagues reported on 32 patients receiving interstitial brachytherapy and 86 patients receiving radiosurgical boost [67]. They found similar survival rates between the two groups and recommended radiosurgery because of its outpatient, noninvasive nature. Hall and colleagues reported 35 patients and believed that radiosurgery did confer a survival advantage, with fewer complications than brachytherapy [68]. Buatti et al., at the University of Florida, reported on 11 patients treated with radiosurgical boost [65]. No signiﬁcant survival advantage was found. Likewise, Masciopinto and colleagues [69] reported on 31 patients so treated and found that the “curative value of radiosurgery is signiﬁcantly limited by peripheral recurrence.” Other studies include those of Regine [70], Prisco [71], and Gannett [72]. A recurring theme in all retrospective studies of brain tumor therapies is the question of selection bias inﬂuencing the results of therapy more than the therapy itself. In an attempt to control for selection bias in retrospective treatment trials for malignant gliomas [73], Curran [74] developed the recursive partitioning analysis categories, and Sarkaria and colleagues used this methodology to analyze 115 patients from three institutions treated with linear accelerator radiosurgery [75]. They found that patients treated with radiosurgery had a signiﬁcantly improved 2-year and median survival compared with RTOG

136

w.a. friedman

historical controls. The improvement was seen predominately in the worst prognostic classes (3 to 6 classes). Kondziolka performed a similar analysis on 65 patients who underwent upfront radiosurgery [76]. He also found that patients in RTOG classes 3 to 5 appeared to beneﬁt. At the University of Florida, we have retrospectively reviewed 100 patients with WHO grade III and IV malignant gliomas who received SRS boost therapy for residual or recurrent enhancing disease [77]. The patients in our study were divided into recursive partitioning analysis (RPA) classiﬁcations for comparison with historical controls. Class III and IV patients had median survival times very similar to the historical controls. Class V patients demonstrated an increase in median survival (15.6 vs. 8.9 months) and 2-year survival rate (12.5% vs. 6%) compared with historical controls. Eloquent location correlated with poorer survival. This may be due to the selection of less aggressive therapies for this group of patients. Recurrence at time of radiosurgery was associated with longer survival. Very probably, this reﬂects the fact that patients judged “eligible” for radiosurgery at time of recurrence are already selected for longer survival than the average patient treated up front. However, it remains possible that radiosurgery at time of recurrence is truly more effective than upfront radiosurgery. What about drawbacks of the recursive partitioning technique? The RTOG classes used are broad and do not include all known prognostic variables, most notably tumor size. In addition, important linear variables like age, mental status, and KPS are converted into binary ones. This approach, therefore, is ﬂawed, as are all attempts at retrospective analysis. Irish and colleagues, in an analysis of 101 consecutive malignant glioma patients, have shown that those “eligible” for radiosurgery have a median survival of 23.4 months, compared with 8.6 months for “ineligible” patients [78]. Likewise, Curran found a marked survival advantage in radiosurgery “eligible” versus “ineligible” patients [79]. The only complete solution to the issue of selection bias affecting outcome is a prospective randomized study. Such a study has been performed and the results recently published. RTOG Study 93-05 randomized patients with glioblastoma into two treatment arms [80]. One received postoperative radiosurgery, followed by conventional radiotherapy and BCNU chemotherapy. The other arm received radiotherapy and chemotherapy without radiosurgery. At a median follow-up time of 61 months, the median survival in the radiosurgery group was 13.5 months compared with 13.6 months in the standard treatment arm. There were no signiﬁcant differences in 2or 3-year survival, patterns of failure, or quality of life between the two groups. Notably, RTOG 93-05 did not address the use of radiosurgery for recurrent malignant gliomas.

Arteriovenous Malformations Patient Selection Open surgery is generally favored if an arteriovenous malformation (AVM) is amenable to low-risk resection (e.g., low Spetzler-Martin grade, young healthy patient) or is believed to be at high risk for hemorrhage during the latency period

between radiosurgical treatment and AVM obliteration (e.g., associated aneurysm, prior hemorrhage, large AVM with diffuse morphology, venous outﬂow obstruction). Radiosurgery is favored when the AVM nidus is small (<3 cm) and compact, when surgery is judged to carry a high risk or is refused by the patient, and when the risk of hemorrhage is not believed to be extraordinarily high. Endovascular treatment, although rarely curative alone, may be useful as a preoperative adjunct to either microsurgery or radiosurgery. The history, physical examination, and diagnostic imaging of each patient are evaluated and the various factors outlined above are weighed in combination to determine the best treatment approach for a given case. The decision about optimal AVM treatment is best made by a multidisciplinary team composed of experts in operative, endovascular, and radiosurgical treatment.

Stereotactic Image Acquisition The most problematic aspect of AVM radiosurgery is target identiﬁcation. In some series, targeting error is listed as the most frequent cause of radiosurgical failure [81, 82]. The problem lies with imaging. Although angiography very effectively deﬁnes blood ﬂow (feeding arteries, nidus, and draining veins), it does so in only two dimensions. Using the two-dimensional data from stereotactic angiography to represent the three-dimensional target results in signiﬁcant errors of both overestimation and underestimation of AVM nidus dimensions [83–85]. Underestimation of the nidus size may result in treatment failure, and overestimation results in the inclusion of normal brain within the treatment volume. This can cause radiation damage to normal brain, which—when affecting an eloquent area—may result in a neurologic deﬁcit. To avoid such targeting errors, a true three-dimensional image database is required. Both contrast-enhanced CT and MRI are commonly used for this purpose. Diagnostic (nonstereotactic) angiography is used to characterize the AVM, but because of its inherent inadequacies as a treatment planning database, stereotactic angiography has been largely abandoned at our institution. We use contrast-enhanced, stereotactic CT as a targeting image database for the vast majority of AVMs. Our CT technique employs rapid infusion (1 mL/s) of contrast while scanning through the AVM nidus with 1-mm slices. The head ring is bolted to a bracket at the head of the CT table, ensuring that the head/ring/localizer complex remains immobile during the scan. This technique yields a very clear three-dimensional picture of the nidus. Alternative approaches use MRI/MRA as opposed to CT. Attention to optimal image sequences in both CT and MRI is essential for effective AVM radiosurgical targeting.

Dose Selection Various analyses of AVM radiosurgery outcomes have elucidated an appropriate range of doses for the treatment of AVMs [86–89]. We prefer to deliver a dose of 20 Gy to the periphery of the AVM nidus whenever possible. Larger AVMs, or those in critical locations, may require a lower dose—but this will reduce the chances of complete obliteration.

10.

linear accelerator radiosurgery

FIGURE 10-10. Pretreatment angiogram shows a left parietal AVM. It was treated with radiosurgery (17.5 Gy to the 70% line using three isocenters).

Follow-up Standard follow-up after AVM radiosurgery typically consists of annual clinic visits with MRI/MRA to evaluate the effect of the procedure and monitor for neurologic complications (Figs. 10-10 and 10-11). If the patient’s clinical status changes, he is followed more closely at clinically appropriate intervals. Each patient is scheduled to undergo cerebral angiography at 3 years after radiosurgery, and a deﬁnitive assessment of the success or failure of treatment is made based on the results of angiography. If no ﬂow is observed through the AVM nidus, the patient is pronounced cured and is discharged from followup. If the AVM nidus is incompletely obliterated, appropriate further therapy (most commonly repeat radiosurgery on the day of angiography) is prescribed, and the treatment/follow-up cycle is repeated.

137

analysis of treatment failures in our series in 1998. He found that 26% of the failures were due to targeting error, at least in part. Statistical predictors of failure were increasing AVM size, decreasing treatment dose, and increasing Spetzler-Martin score. Of particular interest were the “cutpoints” that were identiﬁed. There was a dramatic increase in cure rates when the peripheral dose was raised to a least 15 Gy. There was a dramatic decrease in cure rate when AVM size exceeded 10 cm3 (size D). In a more recent analysis, a study was undertaken to determine which factors were statistically predictive of radiographic and clinical outcomes in the radiosurgical treatment of AVMs [89]. The computerized dosimetry and clinical data on 269 patients were reviewed. The AVM nidus was hand contoured on successive enhanced CT slices through the nidus, to allow detailed determination of nidus volume, target miss, normal brain treated, dose conformality, and dose gradient. In addition, a number of patient and treatment factors, including SpetzlerMartin score, presenting symptoms, dose, number of isocenters, radiographic outcome, and clinical outcome were subjected to multivariate analysis. None of the analyzed factors were predictive of permanent radiation-induced complications or of hemorrhage after radiosurgery in this study. Eloquent AVM location and 12 Gy volume correlated with the occurrence of transient radiation-induced complications. Better conformality correlated with a reduced incidence of transient complications. Lower Spetzler-Martin scores, higher doses, and steeper dose gradients correlated with radiographic success. When AVMs are not cured, current practice frequently involves a “retreatment,” usually 3 years after the original treatment. We reviewed the cases of 52 patients who underwent repeat radiosurgery for residual AVM at our institution between December 1991 and June 1998 [90]. In each case, residual arteriovenous shunting persisted beyond 36 months after the initial treatment. The mean interval between the ﬁrst and second treatments was 41 months. Each AVM nidus was measured at

The University of Florida Experience From May 18, 1988 to March 22, 2005, 544 patients with AVMs were treated at the University of Florida. The mean age was 40 (range, 4 to 78 years). The median treatment volume was 7 cm3 (range, 2 to 45.3 cm3). Many patients early in the series were treated with single isocenters (259), but in recent years an effort has been made to produce highly conformal plans by employing multiple isocenters. The median radiation dose to the periphery of the AVM was 1750 cGy and the mean follow-up duration was 31 months. Presenting symptoms included the following: headache/ incidental (188), seizure (227), hemorrhage (179), progressive neurological deﬁcit (23). Spetzler-Martin scores were as follows: I, 29; II, 188; III, 228; IV, 98. AVMs were further delineated into four nidus volume categories: A, <1 cm3; B, 1 to 4 cm3; C, 4 to 10 cm3; D, >10 cm3. Angiographic/MRI cure rates were as follows: A, 92%; B, 79%; C, 64%; and D, 36%. Ellis et al. [81] performed a detailed

FIGURE 10-11. Two years later, the angiogram of the AVM patient of Fig. 10-10 is normal.

138

w.a. friedman

the time of original treatment and again at the time of retreatment, and dosimetric parameters of the two treatments were compared. After retreatment, patients were followed, and their outcomes evaluated, according to our standard post-AVM radiosurgery protocol. Deﬁnitive end points included angiographic cure, radiosurgical failure (documented persistence of AVM ﬂow 3 years after retreatment), and death. The mean original lesion volume was 13.8 cm3 and the mean volume at retreatment was 4.7 cm3, for an average volume reduction of 66% after the initial “failed” treatment. Only two (3.8%) AVMs failed to demonstrate size reduction after primary treatment. The median doses on initial and repeat treatment were 12.5 and 15 Gy, respectively. To date, 25 retreated patients have reached a deﬁnitive end point. These include 15 (60%) angiographically documented cures, 9 (36%) angiographically documented failures, and 1 fatal hemorrhage. A single permanent radiation-induced complication occurred among 52 (1.9%) patients, and 1 patient experienced a transient deﬁcit that resolved with steroid therapy. Two hemorrhages (one fatal) occurred during a total of 130 patient-years at risk, resulting in a 1.5% annual incidence of posttreatment hemorrhage. If one includes retreatments in the analysis of radiosurgical success, the results are as follows: A, 100%; B, 92%; C, 85%; D, 82%. Ten (1.8%) patients sustained a permanent radiationinduced complication. Seventeen (3.1%) had a transient radiation-induced complication. These problems usually resolved within several months of steroid therapy. Most importantly, 42 patients suffered hemorrhages after radiosurgical treatment, and 8 were fatal. Hemorrhage during the latent period after radiosurgery is the major drawback of this procedure. Only surgery at this point can immediately eliminate the risk of hemorrhage in patients with AVMs.

References 1. Leksell L. The stereotaxic method and radiosurgery of the brain. Acta Chir Scand 1951; 102:316–319. 2. Betti OO, Derechinsky VE. Hyperselective encephalic irradiation with a linear accelerator. Acta Neurochir Suppl 1984; 33:385–390. 3. Colombo F, Benedetti A, Pozza F, et al. External stereotactic irradiation by linear accelerator. Neurosurgery 1985; 16:154–160. 4. Hartmann GH, Schlegel W, Sturm V, et al. Cerebral radiation surgery using moving ﬁeld irradiation at a linear accelerator facility. Int J Radiat Oncol Biol Phys 1985; 11:1185–1192. 5. McGinley PH, Butker EK, Crocker IR, Landry JC. A patient rotator for stereotactic radiosurgery. Phys Med Biol 1990; 35:649–657. 6. Podgorsak EB, Olivier A, Pla M, et al. Dynamic stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 1988; 14:115–126. 7. Winston KR, Lutz W. Linear accelerator as a neurosurgical tool for stereotactic radiosurgery. Neurosurgery 1988; 22:454–464. 8. Friedman WA, Bova FJ. The University of Florida radiosurgery system. Surg Neurol 1989; 32:334–342. 9. Friedman WA, Buatti JM, Bova FJ, Mendenhall WM. LINAC Radiosurgery: A Practical Guide. Berlin: Springer-Verlag, 1998. 10. Samii M, Matthies C. Management of 1000 vestibular schwannomas (acoustic neuromas): surgical management and results with an emphasis on complications and how to avoid them. Neurosurgery 1997; 40:11–23. 11. Samii M, Matthies C. Management of 1000 vestibular schwannomas (acoustic neuromas): The facial nerve–preservation and restitution of function. Neurosurgery 1997; 40:684–695.

12. Samii M, Matthies C. Management of 1000 vestibular schwannomas (acoustic neuromas): Hearing function in 1000 tumor resections. Neurosurgery 1997; 40:248–262. 13. Leksell L. A note on the treatment of acoustic tumors. Acta Chir Scand 1971; 137:763–765. 14. Foote KD, Friedman WA, Buatti JM, et al. Analysis of risk factors associated with radiosurgery for vestibular schwannoma. J Neurosurg 2001; 95(3):440–449. 15. Spiegelmann R, Gofman J, Alezra D, Pfeffer R. Radiosurgery for acoustic neurinomas (vestibular schwannomas). Isr Med Assoc J 1999; 1(1):8–13. 16. Spiegelmann R, Lidar Z, Gofman J, et al. Linear accelerator radiosurgery for vestibular schwannoma. J Neurosurg 2001; 94(1): 7–13. 17. Martens F, Verbeke L, Piessens M, Van Vyve M. Stereotactic radiosurgery of •vestibular schwannomas with a linear accelerator. Acta Neurochir 1994; 62(Suppl):88–92. 18. Valentino V, Raimondi AJ. Tumour response and morphological changes of acoustic neurinomas after radiosurgery. Acta Neurochir 1995; 133:157–163. 19. Delaney G, Matheson J, Smee R. Stereotactic radiosurgery: an alternative approach to the management of acoustic neuromas. Med J Austral 1992; 156:440. 20. Barcia Salorio JL, Hernandez G, Ciudad J, et al. Stereotactic radiosurgery in acoustic neurinoma. Acta Neurochir Suppl 1984; 33:373–376. 21. Sekhar LN, Jannetta PJ, Burkhart LE, Janosky JE. Meningiomas involving the clivus: a six-year experience with 41 patients. Neurosurgery 1990; 27:764–781. 22. Sekhar LN, Altschuler EM. Meningiomas of the cavernous sinus. In: Al-Mefty O, ed. Meningiomas. New York: Raven Press, 1991:445–460. 23. Simpson D. The recurrence of intracranial meningiomas after surgical treatment. J Neurol Neurosurg Psychiatry 1957; 20:22–39. 24. Pollock BE, Stafford SL, Utter A, et al. Stereotactic radiosurgery provides equivalent tumor control to Simpson Grade 1 resection for patients with small- to medium-size meningiomas. Int J Radiat Oncol Biol Phys 2003; 55(4):1000–1005. 25. Valentino V, Schinaia G, Raimondi AJ. The results of radiosurgical management of 72 middle fossa meningiomas. Acta Neurochir 1993; 122:60–70. 26. Villavicencio AT, Black PM, Shrieve DC, et al. Linac radiosurgery for skull base meningiomas. Acta Neurochir (Wien) 2001; 143(11):1141–1152. 27. Engenhart R, Kimmig BN, Hover KH, et al. Stereotactic single high dose radiation therapy of benign intracranial meningiomas. Int J Radiat Oncol Biol Phys 1990; 19:1021–1026. 28. Spiegelmann R, Nissim O, Menhel J, et al. Linear accelerator radiosurgery for meningiomas in and around the cavernous sinus. Neurosurgery 2002; 51(6):1373–1380. 29. Hakim R, Alexander III E, Loefﬂer JS, et al. Results of linear accelerator-based radiosurgery for inracranial meningiomas. Neurosurgery 1998; 42:446–454. 30. Friedman WA, Murad G, Bradshaw P, et al. Linear accelerator radiosurgery for meningiomas. J Neurosurg 2005; 103:206–209. 31. Lohr F, Pirzkall A, Hof H, et al. Adjuvant treatment of brain metastases. Semin Surg Oncol 2001; 20:50–56. 32. DeAngelis LM. Brain tumors. N Engl J Med 1990 2001; 344: 114–123. 33. Patchell RA, Tibbs PA, Walsh JW, et al. A randomized trial of surgery in the treatment of single metastases to the brain. N Engl J Med 1990; 322:494–500. 34. Noordijk EM, Vecht CJ, Haaxma-Reiche H, et al. The choice of treatment of single brain metastasis should be made based on extracranial tumor activity and age. Int J Radiat Oncol Biol Phys 1994; 29:711–717.

10.

linear accelerator radiosurgery

35. Mintz AH, Kestle J, Rathbone MP, et al. A randomized trial to assess the efﬁcacy of surgery in addition to radiotherapy in patients with a single cerebral metastasis. Cancer 1996; 78:1470–1476. 36. Haines SJ. Moving targets and ghosts of the past: outcome measurement in brain tumour therapy. J Clin Neurosci 2002; 9:109–112. 37. Sturm V, Kober B, Hover KH, et al. Stereotactic percutaneous single dose irradiation of brain metastases with a linear accelerator. Int J Radiat Oncol Biol Phys 1987; 13:279–282. 38. Sturm V, Kimmig B, Engenhardt R, et al. Radiosurgical treatment of cerebral metastases. J Stereo Func Neurosurg 1991; 57:7–10. 39. Voges J, Treuer H, Erdmann J, et al. LINAC radiosurgery in brain metastases. Acta Neurochir 1994; 62(Suppl):72–76. 40. Black PM. Solitary brain metastases. Radiation, resection, or radiosurgery? Ch 1993; 103:367S–369S. 41. Alexander E, Moriarty TM, Davis RB, et al. Stereotactic radiosurgery for the deﬁnitive, noninvasive treatment of brain metastases. J Nat Cancer Inst 1995; 87:34–40. 42. Adler JR, Cox RS, Kaplan I, Martin DP. Stereotactic radiosurgical treatment of brain metastases. J Neurosurg 1992; 76:444–449. 43. Fuller BG, Kaplan ID, Adler J, et al. Stereotaxic radiosurgery for brain metastases: The importance of adjuvant whole brain irradiation. Int J Radiat Oncol Biol Phys 1992; 23:413–418. 44. Joseph J, Adler JR, Cox RS, Hancock SL. Linear acceleratorbased stereotaxic radisourgery for brain metastases: the inﬂuence of number of lesions on survival. J Clin Oncol 1996; 14:1085–1092. 45. Auchter RM, Lamond JP, Alexander E, et al. A multiinstitutional outcome and prognostic factor analysis of radiosurgery for resectable single brain metastasis. Int J Radiat Oncol Biol Phys 1996; 35:27–35. 46. Becker G, Jeremic B, Engel C, et al. Radiosurgery for brain metastases: the Tuebingen experience. Radiother Oncol 2002; 62: 233–237. 47. Breneman JC, Warnick RE, Albright RE, et al. Stereotactic radiosurgery for the treatment of brain metastases. Cancer 1997; 79: 551–557. 48. Buatti JM, Friedman WA, Bova FJ, Mendenhall WM. Treatment selection factors for stereotactic radiosurgery of intracranial metastases. Int J Radiat Oncol Biol Phys 1995; 32:1161–1166. 49. Caron J-L, Souhami L, Podgordak EB. Dynamic stereotactic radiosurgery in the palliative treatment of cerebral metastatic tumors. J Neuro-Oncol 1992; 12:173–179. 50. Gutin PH, Wilson CB. Radiosurgery for malignant brain tumors. J Clin Oncol 1990; 8:571–573. 51. Mehta MP, Rozental JM, Levin AB, et al. Deﬁning the role of radiosurgery in the management of brain metastases. Int J Radiat Oncol Biol Phys 1992; 24:619–625. 52. Mehta M, Noyes W, Craig B, et al. A cost-effectiveness and costutility analysis of radiosurgery vs. resection for single-brain metastases. Int J Radiat Oncol Biol Phys 1997; 39:445–454. 53. Valentino V. The results of radiosurgical management of 139 single cerebral metastases. Acta Neurochir Suppl 1995; 63:95– 100. 54. Cho KH, Hall WA, Gerbi BJ, et al. Patient selection criteria for the treatment of brain metastases with stereotactic radiosurgery. J Neurooncol 1998; 40:73–86. 55. Fernandez-Vicioso E, Suh JH, Kupelian PA, et al. Analysis of prognostic factors for patients with single brain metastasis treated with stereotactic radiosurgery. Radiat Oncol Invest 1997; 5:31– 37. 56. Maor MH, Dubey P, Tucker SL, et al. Stereotactic radiosurgery for brain metastases: results and prognostic factors. Int J Cancer 2000; 90:157–162. 57. Goodman KA, Sneed PK, McDermott MW, et al. Relationship between pattern of enhancement and local control of brain metas-

58.

59.

60.

61. 62.

63.

64.

65.

66. 67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

139

tases after radiosurgery. Int J Radiat Oncol Biol Phys 2001; 50:139–146. Shiau C-Y, Sneed PK, Shu H-KG, et al. Radiosurgery for brain metastases: Relationship of dose and pattern of enhancement to local control. Int J Radiat Oncol Biol Phys 1997; 37:375–383. Gaspar L, Scott C, Rotman M, et al. Recursive partitioning analysis of prognostic factors in three radiation therapy oncology group (RTOG) brain metastases trials. Int J Radiat Oncol Biol Phys 1997; 37:745–751. Ulm AJ, Friedman WA, Bova FJ, et al. Linear accelerator radiosurgery in the treatment of brain metastases. Neurosurgery 2004; 55:1076–1085. DeAngelis LM. Brain tumors. N Engl J Med 2001; 344(2):114– 123. Bernstein M, Laperriere N, Glen J, Leung P, Thomason C, Landon AE. Brachytherapy for recurrent malignant astrocytoma. Int J Radiat Oncol Biol Phys 1994; 30(5):1213–1217. Chang CN, Chen WC, Wei KC, et al. High-dose-rate stereotactic brachytherapy for patients with newly diagnosed glioblastoma multiformes. J Neurooncol 2003; 61(1):45–55. Prados MD, Gutin PH, Phillips TL, et al. Interstitial brachytherapy for newly diagnosed patients with malignant gliomas: the UCSF experience. Int J Radiat Oncol Biol Phys 1992; 24(4): 593–597. Buatti JM, Friedman WA, Bova FJ, Mendenhall WM. Linac radiosurgery for high-grade gliomas: the University of Florida experience. Int J Radiat Oncol Biol Phys 1995; 32(1):205–210. Friedman WA, Foote KD. Linear accelerator radiosurgery in the management of brain tumours. Ann Med 2000; 32(1):64–80. Shrieve DC, Alexander E III, Wen PY, et al. Comparison of stereotactic radiosurgery and brachytherapy in the treatment of recurrent glioblastoma multiforme. Neurosurgery 1995; 36(2): 275–282; discussion 82–84. Hall WA, Djalilian HR, Sperduto PW, et al. Stereotactic radiosurgery for recurrent malignant gliomas. J Clin Oncol 1995; 13(7):1642–1648. Masciopinto JE, Levin AB, Mehta MP, Rhode BS. Stereotactic radiosurgery for glioblastoma: a ﬁnal report of 31 patients. J Neurosurg 1995; 82(4):530–535. Regine WF, Patchell RA, Strottmann JM, et al. Preliminary report of a phase I study of combined fractionated stereotactic radiosurgery and conventional external beam radiation therapy for unfavorable gliomas. Int J Radiat Oncol Biol Phys 2000; 48(2): 421–426. Prisco FE, Weltman E, de Hanriot RM, Brandt RA. Radiosurgical boost for primary high-grade gliomas. J Neurooncol 2002; 57(2): 151–160. Gannett D, Stea B, Lulu B, et al. Stereotactic radiosurgery as an adjunct to surgery and external beam radiotherapy in the treatment of patients with malignant gliomas. Int J Radiat Oncol Biol Phys 1995; 33(2):461–468. Roberge D, Souhami L. Stereotactic radiosurgery in the management of intracranial gliomas. Technol Cancer Res Treat 2003; 2(2):117–125. Curran WJ Jr, Scott CB, Horton J, et al. Recursive partitioning analysis of prognostic factors in three Radiation Therapy Oncology Group malignant glioma trials. J Natl Cancer Inst 1993; 85(9):704–710. Sarkaria JN, Mehta MP, Loefﬂer JS, et al. Radiosurgery in the initial management of malignant gliomas: survival comparison with the RTOG recursive partitioning analysis. Radiation Therapy Oncology Group. Int J Radiat Oncol Biol Phys 1995; 32(4): 931–941. Kondziolka D, Flickinger JC, Bissonette DJ, et al. Survival beneﬁt of stereotactic radiosurgery for patients with malignant glial neoplasms. Neurosurgery 1997; 41(4):776–783; discussion 83–85.

140

w.a. friedman

77. Ulm AJ, Friedman WA, Bova FJ, et al. Radiosurgery for malignant gliomas: the University of Florida experience. Neurosurgery 2005; 57:512–517. 78. Irish WD, Macdonald DR, Cairncross JG. Measuring bias in uncontrolled brain tumor trials–to randomize or not to randomize? Can J Neurol Sci 1997; 24(4):307–312. 79. Curran WJ Jr, Scott CB, Weinstein AS, et al. Survival comparison of radiosurgery-eligible and -ineligible malignant glioma patients treated with hyperfractionated radiation therapy and carmustine: a report of Radiation Therapy Oncology Group 83-02. J Clin Oncol 1993; 11(5):857–862. 80. Souhami L, Seiferheld W, Brachman D, et al. Randomized comparison of stereotactic radiosurgery followed by conventional radiotherapy with carmustine to conventional radiotherapy with carmustine for patients with glioblastoma multiforme: Report of Radiation Therapy Oncology Group 93-05 protocol. Int J Radiat Oncol Biol Phys 2004; 60(3):853–860. 81. Ellis TL, Friedman WA, Bova FJ, et al. Analysis of treatment failure after radiosurgery for arteriovenous malformations. J Neurosurg 1998; 89(1):104–110. 82. Pollock BE, Flickinger JC, Lunsford LD, et al. Factors associated with successful arteriovenous malformation radiosurgery. Neurosurgery 1998; 42(6):1239–1244; discussion 44–47. 83. Bova FJ, Friedman WA. Stereotactic angiography: an inadequate database for radiosurgery? Int J Radiat Oncol Biol Phys 1991; 20:891–895.

84. Blatt DL, Friedman WA, Bova FJ. Modiﬁcations in radiosurgical treatment planning of arteriovenous malformations based on CT imaging. Neurosurgery 1993; 33:588–596. 85. Spiegelmann R, Friedman WA, Bova FJ. Limitations of angiographic target localization in planning radiosurgical treatment. Neurosurgery 1992; 30:619–624. 86. Flickinger JC, Pollock BE, Kondziolka D, Lunsford LD. A dose-response analysis of arteriovenous malformation obliteration after radiosurgery. Int J Radiat Oncol Biol Phys 1996; 36: 873–879. 87. Karlsson B, Lindquist C, Steiner L. Prediction of obliteration after gamma knife surgery for cerebral arteriovenous malformations. Neurosurgery 1997; 40(3):425–430; discussion 30–31. 88. Pollock BE, Kondziolka D, Lunsford LD, et al. Repeat stereotactic radiosurgery of arteriovenous malformations: factors associated with incomplete obliteration. Neurosurgery 1996; 38(2): 318–324. 89. Friedman WA, Bova FJ, Bollampally S, Bradshaw P. Analysis of factors predictive of success or complications in arteriovenous malformation radiosurgery. Neurosurgery 2003; 52:296– 308. 90. Foote KD, Friedman WA, Ellis TL, et al. Salvage retreatment after failure of radiosurgery in patients with arteriovenous malformations. J Neurosurg 2003; 98:337–341.

1 1

Proton Beam Radiosurgery: Physical Bases and Clinical Experience Georges Noel, Markus Fitzek, Loïc Feuvret, and Jean Louis Habrand

Introduction The introduction of photon therapy into the armamentarium of cancer treatment represented a breakthrough in the early 20th century. Since then, major technical innovations in radiation oncology have been continuously associated with major improvements in local tumor control and patient quality of life. The development of computer technology two decades ago allowed an unprecedented step forward with the design of computer-driven linear accelerators, multileaf collimation, threedimensional treatment planning, and intensity modulation of the beam. In parallel, biological concepts on tumor control probability have been modeled and studied in relation to tumor volume, clonogens sensitivity, biological environment, and so forth. A basic concept remained, however; that is, the probability of local and to some extent distant control are dose-correlated to the primary tumor. This gave the impetus to the exploration of new forms of particles, especially the “biologically” effective ones, like neutrons. In our opinion, disappointing results in this ﬁeld should not discourage, but rather pave avenues to new investigations, dealing with heavy ions for example. An alternative approach is based on purely “ballistically” effective particles, with protons as a paradigm. They have been investigated successfully in multiple dose-escalation trials concerning so-called radioresistant tumors, like ocular melanomas and low-grade sarcomas at the skull base. These results remind us that the simpler the concept, the better! In less-challenging clinical situations, special types of radiation have also been of interest: for many years, radiation oncologists have learned to treat superﬁcial or semideep tumor sites routinely with an electron beam, the optimal and simplest way to spare surrounding organs. Similarly, the management of deep-sited

conditions can be elegantly approached using high-energy protons. Although the technical complexity and cost associated with their production and delivery has frequently been put forward by detractors, it is generally (from a ballistic standpoint) the simplest way to deliver the dose with optimal conformation and with a superior dose-scattering limitation around the target. Thus, the development of proton therapy seems ultimately associated with the need for ionizing radiation that minimizes the risk of long-term side-effects, including carcinogenicity. This might be one of the great public health challenges for the coming decade.

Early History of Proton Radiation Protons are part of the hadron family. This term derives from the old Greek αδρος, which means “strength.” They are actually high-energy particles made up of quarks (elementary particles of the atomic nucleus). In radiology, it has become synonymous with protons or neutrons, as they are basic components of the atomic nucleus, and even with the atomic nuclei themselves. The latter are also known as “light ions”(especially helium, oxygen, and carbon ions) or “heavy ions” (especially neon and argon) [1]. The history of proton therapy began in December 1904 when William Henry Bragg described the peak absorption in the air of alpha particles. In their founding paper (published 1905), Bragg and Kleeman reported the deﬁnite path of the particles, correlated with their initial energy and sharp increase of the ionization density as they approached their range-end [2]. Unlike the rapid development of X-rays in diagnosis and treatment, the medical applications of charged particles did

141

142

g. noel et al.

TABLE 11-1. Hadron therapy centers (except neutrons) and numbers of treated patients (data of January 2005). Center

Country

Particle

Closed centers Uppsala (1) PSI (SIN) Louvain-la-Neuve Dubna (1) Los Alamos Berkeley 184 Berkeley Berkeley Harvard MPRI (1) PMRC (1) TRIUMF

Dates of creation

No. treated patients

Sweden Switzerland Belgium Russia USA USA USA USA USA USA Japan Canada

Protons Pions Protons Protons Pions Protons Helium Heavy ions Protons Protons Protons Pions

1957–1976 1980–1993 1991–1993 1967–1996 1974–1982 1954–1957 1957–1992 1975–1992 1961–2002 1993–1999 1983–2000 1979–1994

73 503 21 124 230 30 2,054 433 9,116 34 700 367

Open centers Western Europe Uppsala (2) PSI (72 MeV) PSI (200 MeV) Nice Orsay Clatterbridge INFN-LNS GSI HMI

Sweden Switzerland Switzerland France France England Italy Germany Germany

Protons Protons Protons Protons Protons Protons Protons Heavy ions Protons

1989 1984 1996 1991 1991 1989 2002 1997 1998

418 4,182 209 2,555 2,805 1,372 82 198 546

Eastern Europe ITEP St. Petersburg Dubna (2)

Russia Russia Russia

Protons Protons Protons

1969 1975 1987

3,785 1,145 296

Asia Chiba PMRC (2) HIMAC NCC HIBMC HIBMC WERC Shizuoka Wanjie, Zibo

Japan Japan Japan Japan Japan Japan Japan Japan China

Protons Protons Heavy ions Protons Protons Heavy ions Protons Protons Protons

1979 2001 1994 1998 2001 2002 2002 2003 2004

145 492 1,796 300 483 30 19 100 1

USA USA USA USA Canada

Protons Protons Protons Protons Protons

1990 1993 2001 1994 1995

9,585 21 973 632 89

South Africa Pions Ions Protons All particles

Protons

1993

468 1,100 4,511 40,801 46,412

North America Loma Linda MPRI (2) NPTC, MGH UCSF-CNL TRIUMF Africa iThemba LABS Total

Source: From Particles. A newsletter of the Particle Therapy Cooperative Group. 2005; 35:10. Available at http://ptcog.web.psi.ch/ptles35.pdf. Used with permission.

11.

143

proton beam radiosurgery

not start until after World War II. Robert Wilson, who had been chief of the cyclotron team during the Manhattan Project, elected to devote (like many other physicists at that time) his future research activities to the beneﬁt of mankind. Soon thereafter (1946) was brought out “radiological use of fastprotons” [3]. This was the ﬁrst proposal for use of proton beam in radiation treatment. He challenged scientists to achieve a uniform dose-distribution whatever the tumor thickness and proposed a solution still valid (i.e., the superimposition of “native” Bragg peaks of different energies and penetration generated by a rotating “ridge ﬁlter” later on described as the “spread out” Bragg peak). This prophetic announcement did not come true until the pioneering experiments in a variety of neurologic disorders by Dr. William Sweet, head of the Neurosurgery Department at Massachusetts General Hospital (MGH), in the early 1960s, then joined by Dr. Raymond Kjellberg, a young Swedish surgeon trained at the Berkeley cyclotron. All in all, it had taken 15 years between the Wilson paper and the ﬁrst patient treated on May 25, 1961! This was a 2year-old girl with a suprasellar tumor (reported in [4]). This group concentrated thereafter on pituitary conditions such as acromegaly, diabetic retinopathy, and eventually in the 1980s on arteriovenous malformations (AVMs) [5]. However, between 1967 and 1973, the future of proton therapy remained unclear with the projected cyclotron shut down. Fortunately, the project was deﬁnitely strengthened upon the arrival of Dr. Herman Suit, recently appointed chief of radiation medicine at MGH. He was joined by Michael Goitein, PhD, in charge of the physics project, and addressed radiobiological issues, in relation to the protons’ relative biologic effectiveness (RBE): this was assessed in 1972, to 1.1, a ﬁgure still in use in most centers worldwide [6]. In February 1974, the ﬁrst patient with a pelvic sarcoma was treated with a fractionated schedule. In parallel, an active ophthalmologic program was initiated in ocular melanomas under the auspices of the Massachusetts Eye and Ear Inﬁrmary (MEEI), which became for many years the cyclotron spearhead. The history has retained the ﬁrst patient’s name, Mr. Mc Kelvey [4]. By January 2005, approximately 40,000 patients worldwide had been treated with protons, almost half at the HCL (Table 11-1) [7].

matter and presents a major straggling effect (deﬁned as “lateral penumbra”) on depth-dose proﬁles. In contrast with indirectly ionizing radiation, charged particles are directly ionizing and exhibit a deﬁnite range in matter. Because of their low masses, electron beams interact at rapid velocity (close to speed of light), with relatively limited and uniform interactions with matter. They also come rapidly to their range-end at the usual energy domain in radiotherapy. When they come to their end, they are strongly submitted to a strong lateral scattering (i.e., large penumbra). Mono energetic proton beams generated in large accelerators such as synchrotrons produce particles of highly uniform range (of the order 1%). This limited but actual heterogeneity is due in part to the nuclear interactions that can affect some of the particles. With current generators, deep beam penetrations (of the order 15 to 30 cm in water equivalent material) can be achieved. As the mass of protons are about 2000 times that of electrons, they evidence considerably higher kinetic energy (at equal velocity) and so the capability for releasing this energy much more considerably as they are slowing down. Furthermore, the projectiles are barely deﬂected during the process (that translates into a sharp lateral penumbra on dose-proﬁles). These properties translate into the so-called Bragg curve (Fig. 11-1): relatively limited interactions in their initial path and limited dose-absorption and a steep rise close to their range-end, followed by a sharp fall-off of the dose. There is an increased linear energy transfer (LET) in the Bragg peak that is in turn linked with increased cell killing and RBE (see below). This phenomenon could potentially have a deleterious effect on normal tissues located at the Bragg peak, although the concomitant abrupt fall-off of the dose seems to minimize considerably its impact. Because of these uncertainties, when planning a proton treatment, it is our practice to avoid abutment of the distal end region to a critical organ surface (especially brain stem).

1.0 proton Bragg peak

0.8

Facilities 0.6

SOBP

DOSE

Approximately 30 centers worldwide have implemented a clinical program dealing with heavy particles. Proton therapy itself approaches 20 centers (Table 11-1).

0.4

Physical Properties Photons as they pass through matter interact with atomic electrons and induce ionization when the energy transfer is superior to the energy that binds them with the nucleus. This corresponds with their main energy transfer process. Because of the statistical nature of the events, and the major deﬂections of interacting particles, the photon beam has no deﬁnite path in

0.2

0.0 0

10

20

30

DEPTH (cm) FIGURE 11-1. Native Bragg peak (in red) and spread-out Bragg peak result of the addition of several native Bragg peaks.

144

g. noel et al.

FIGURE 11-2. (A) Plastic and (B) metallic modulators to modulate, respectively, eye and base of the skull tumors.

When the Bragg peak is commonly “spread out” (in order to cover targets thicker than the “native” Bragg peak width) using a modulator (Fig. 11–2), by the emission of protons of various energies, the narrow, high-LET zone is repositioned many times within the “plateau” and “diluted” among longrange low-LET protons. This dilution does not affect obviously the end of the deeper peak. Another drawback of spreading out the peak is an entrance-dose increase (Fig. 11-1). It should be pointed out that the stopping power of different tissues to protons differs substantially as their densities decrease: corrections need to be done when the beam passes from bones to soft tissues (water equivalent), and above all to air cavities. Technical solutions call for the design of appropriate “compensators,” or the production of “pencil beams” of predeﬁned energies, made possible by the most recent accelerators. We will also mention that the rare nuclear fragmentations that affect the beam at its very end are positron emitters and could in theory be detected using a positron scanning. This property is not as prominent as with heavy ions where it has come to clinical application.

Protons Production and Delivery Systems: Conventional Versus Innovative Heavy charged particles are accelerated in large circular cyclotrons, synchrocyclotrons, and synchrotrons (Fig. 11-3). Although synchrotrons seem the most appropriate to produce protons of high and variable energies, there has recently been a renowned interest for modiﬁed cyclotrons that can achieve excellent performance at relatively low cost. The basic principle of proton production is based on the acceleration in an electromagnetic ﬁeld of a hydrogen plasma (i.e., stripped H nuclei ionized by an electric arc). As mentioned above, the beam can be spread out by changing its energy by (1) combining multiple energies by steps of a few MeV; and (2) interposing upstream of a ﬁxed energy beam speciﬁc rotating “ridge” ﬁlters that alter the beam’s path (Fig. 11-2) [8]. Large accelerators were initially developed in a nuclear physics environment, and so early proton facilities were found at university campuses or nuclear plants. This came

FIGURE 11-3. (A) Magnets of the synchrocyclotron at Orsay (2 yellow rings) and (B) magnet to change the beam direction to serve different rooms.

11.

proton beam radiosurgery

along with simple technology, that is, ﬁxed beam, passive scattering, hand-made compensators, and so forth. Recently, hospital-based facilities have been developed, especially in the United States. This has been made possible because modern accelerators are somewhat more compact than the former ones. Nonetheless, recent equipment goes along with sophisticated beam delivery systems including isocentric gantries that are considerably space-consuming. Thus, the main reason for this geographic transfer is likely elsewhere and lies in the need for patients and staff to beneﬁt from immediate medical support, multidisciplinary management, imaging panel, social support, and so forth. Moreover, part of the indications can be approached using a combination of protons and of photons (which lowers substantially the treatment cost), and younger patients can call for general anesthesia. These services are provided by regular departments of radiation oncology. Two sets of energy beams should be made available to fully cover indications: 50 to 60 MeV for ophthalmologic programs, and 200 to 250 MeV for deep sites. Multiple technical innovations have improved clinical indications and treatment accuracy for the past decade. Isocentric gantry technology is still in progress, but prototypes of remarkable accuracy have been developed despite considerable dimensions (i.e., 6 to 11 m) [9]. Gantries, like in photon therapy, allow isocentric planning and treatment and expand arrangement capabilities of beams (including non-coplanar if combined with couch rotations). In turn, this opens up new potential indications, especially intrathoracic and abdominal malignancies. They also affect positively set-up complexity, duration, and accuracy as well as patient comfort. Proton intensity modulation of the beam is a recent concept that is somewhat different from that developed with photons [10]; although the goal is similar (optimal conformation to the target), the approach is different and based on the beam’s penetration modulation rather than the beam’s ﬂuence modulation. Proton beams in the past have been exclusively generated by passive scattering through lead foils (passive diffusion) and then adapted to the tumor distal shape by compensating depth variations. This was achieved using individualized “boluses” for each beam (compensating also for tissue heterogeneities). Unfortunately, this approach does not compensate for proximal tumor shape and is responsible for unnecessary irradiation of tissues located upstream of the target. This “intrinsic” drawback can only be minimized using a multiple-beams arrangement. Recent innovations have aimed to conform dose tightly in three dimensions: The variable modulation method based on a dynamic multileaf collimator acting in conjunction with stepwise Bragg peaks penetration, in order to trim the target volume in subelements from depth to surface; pencil beam scanning eliminates the need for individualized collimators and compensators [11], but requires tight control processes of the dose. Any altered proton ﬂuence, or treatment interruption, or patient’s displacement can seriously affect dose delivery. This is true for raster scanning, an approach in which the beam is deﬂected by two dipole magnets, and more importantly in spot (i.e., point by point) scanning, probably the “ultimate” conformation in radiation oncology.

145

Radiobiological Considerations RBE measurements constitute an integral part of the calibration of clinical hadron beams and contribute, together with dosimetry and, possibly, microdosimetry, to specify the radiation quality. Today, the worldwide reference biological system is the “intestinal crypt regeneration in mice.” RBE multiple biological investigations were conducted at the Harvard cyclotron from the 1960s to the early 1980s that explored mainly brain tolerance in mammals to single fractions administered with different beam apertures. Most of the current knowledge on single-dose radiosurgery (both with photons and protons) has emerged from these data. The rodent intestinal crypt regeneration tests allowed early assessment of the RBE value for rapidly proliferating tissues. It was assessed to a mean of 1.1 [12], although it was likely higher in the peak region (or in the deeper peak as far as the spread-out Bragg peak): 1.4 to 1.5 as mentioned previously. This led to the deﬁnition of the CGE (cobalt Gray equivalent) as the proton dose “unit” by the Boston group. One CGE corresponds with the physical dose times the estimated 1.1 mean RBE. Some experiments conducted elsewhere suggested RBE values ranging between 0.9 and 1.25 [12, 13]. Some lower values can be attributed to the use of orthovoltage equipment as reference radiation instead of cobalt-60. Microdosimetric investigations have recently been conducted that conﬁrm and reﬁne animal models. If RBE estimates can substantially affect dose-reports, physical dose measurements take their own toll: this was evidenced in a recent multicentric intercomparison in which doseunderestimates up to 17% were found for those dealing with Faraday cups against ionization chambers [14]. This led, for example, the Boston group to move ongoing prescribed tumordoses and tolerance-doses to the critical organs up and down by 10%, respectively.

Patient Setup Patient setup is largely conditioned by the treatment room conﬁguration. The procedure implemented at the Orsay center will be brieﬂy discussed in this section and in the following one, which is based on a ﬁxed horizontal beamline. The patient can be either lying on his back or seated. For immobilization purposes, a thick custom-made thermoplastic mask is used. This rather “conventional” approach is offset by a stereotactic alignment that includes the implantation under local anesthesia of four to ﬁve radiopaque ﬁducial markers in the outer skull (made of gold or titanium). Orthogonal X-ray ﬁlms are compared with digitally reconstructed radiographs (DRRs) before each daily session to check alignment and repeated for each ﬁeld. The ﬁducial markers allow appropriate translation and angular corrections using an original computer program (Rotaplus) developed at the Orsay Proton Therapy Center (CPO). Proton treatment time averages 25 minutes with 20 minutes for patient setup and 5 minutes for irradiation itself. Appropriate corrections are made with high precision, using a robotic chair or a couch. Setup accuracy has been checked to less than 1 mm in the x, y, and z directions and less than 1° rotation (Fig. 11-4).

146

g. noel et al.

< 2 mm x 0.8

FIGURE 11-4. Patient setup. Gold ﬁducial marker (A). Fiducial makers in outer bone (B1) or in skull (B2). Patient postioning in two different robotic chairs (C).

Patient Treatment Planning and Simulation The process starts with the acquisition of 3-mm-thick slices from a contrast-enhanced computed tomography (CT) scan and 1.5-mm-thick slices from a contrast-enhanced magnetic res-

onance imaging (MRI) scan, both performed with the patient in a supine position. Target volume and organs at risk beneﬁt from matched or fused imagings, combining both modalities. Threedimensional treatment plans, including dose-volume histograms (DVHs), are generated using a “homemade” software (ISIS 3D

11.

proton beam radiosurgery

147

π patchesεσ

FIGURE 11-5. Beams in patch technique. Example: Distal penumbra of left posterior (A) and a right lateral-posterior (B) beams are in contact with the lateral penumbra of a right lateral beam (C) allowing delivery of a tumor dose (blue) without irradiating the brain stem– spinal cord junction (D).

Treatment Planning System; Technology Diffusion Company). This treatment planning system (TPS) has the advantage of combining photon, electron, and proton dose-distributions. A unique advantage of proton beams arrangement should be mentioned here: the ability to design “patched” ﬁelds, in which the lateral penumbra of a beam is matched with the distal one of a

second (and sometimes a third) beam. This leads to remarkable “bended” isodoses around critical structures such as brain stem or spinal cord (Fig. 11-5). Individualized modulators and compensators are eventually designed by a computer-driven milling machine for each proton beam. Beam shaping is performed using Cerrobend blocks or electro-cut brass blocks (Fig. 11-6).

Without compensator

FIGURE 11-6. Collimator (A) to limit irradiation in two dimensions (2D irradiation) and compensator (B) to compensate indepth the irradiation (third dimension). Collimator was used for a superior ﬁeld and shielded optic nerves and chiasm.

With compensator

148

g. noel et al.

Clinical Applications of Proton Beam Therapy

Eye Tumors

Proton therapy has proved highly valuable in the management of slow-growing, radioresistant tumors that abut radiosensitive structures. In adults, ocular melanomas and base of the skull chordomas and chondrosarcomas typically represent such a challenge and have enjoyed remarkable improvements both in their outcome and in radiation-related toxicity. Proton therapy should also be considered seriously in the therapeutic armamentarium of patients with exquisite sensitivity to radiation (i.e., primarily young patients). Promising experiences have been brought out in pediatric conditions located at the skull base and brain [15–24]. Past and ongoing phase I and II studies could pave new avenues in the treatment of adult gliomas [15–17, 25, 26], intracranial AVMs [27–30], medulloblastomas [31, 32], head and neck malignancies [28, 33, 34], especially nasopharyngeal carcinomas [35–37], esophagus [38, 39], prostate [40, 41], rectum [42, 43], gynecological carcinomas [44–46], soft tissues sarcomas [47], lung carcinomas [48], and hepatocarcinomas [49].

Uveal melanomas have been treated with protons since 1975 at HCL, followed by a remarkable program at the Paul Scherrer Institute (PSI) in Switzerland. Total dose ranged between 60 and 70 CGE in four to ﬁve consecutive fractions [50]. The ﬁrst retrospective analyses evidenced an increased 3-year overall survival for the groups treated with protons over those with enucleation [51–58] (Table 11-2). A deﬁnitive retrospective analysis of MGH-MEEI patients showed in 1488 cases a 5-year local control rate of 96%, along with useful vision preservation in 65% and 39% whether the tumor was small or large, respectively [52]. Egger et al. reported a 10-year local control rate of 94.8% in the Swiss series. In multivariate analysis, prognostic factors for local control were gender, tumor size, margin close to the ciliary body, and extent of safety margin around the GTV [53]. Courdi et al. reported similar ﬁgures in his series of 538 patients treated in Nice. Five-year speciﬁc local control and survival rates were 89% and 86.3%, respectively. The metastatic rate was 8%. Fifteen enucleations were performed for tumor progression, 12 for

TABLE 11-2. Ocular melanoma (relevant series only). Authors

Location

No. of cases

Type of study

Dose/fractionation

Results

Munzenrider et al. [52]

Choroid

1488

P

70 CGE in 4 Fr

Choroid

2435

P

60–70 CGE in 4 Fr

Courdi et al. [54]

Choroid

538

P

57.2 CGE in 4 Fr

Lumbroso et al. [55]

Choroid

1062

P

60 CGE in 4 Fr

Damato et al. [56]

Choroid

349

P

58.4 CGE in 4 Fr

Damato et al. [57]

Iris

88

P

58.4 CGE in 4 Fr

Gragoudas et al. [58]

Choroid

Rand.

70 CGE vs. 50 CGE To decrease radiation-induced complication

Median follow-up 59 months Large tumors Relapse: 2.4% 5- and 10-year FMS: 68%, 60% Useful vision: 39% Small tumors Relapse: 0.5% 5- and 10-year FMS: 86%, 79% Useful vision: 65% Median follow-up 40 months 5- and 10-year LC: 95.8%, 94.8% 10-year OS: 72.6% Useful vision: 50% LC: 89% 5-year DFS: 86% Mets: 8% Useful vision: 50% Median follow-up 38 months LC: 97% 5-year DFS: 78% Mets: 15% Useful vision: 47% Median follow-up 37.2 months 5-year LC: 96.5 % 5- and 8-year FMS: 90%, 83.9% 5-year useful vision: 79.1% Median follow-up 32.4 months 4-year LC: 96.7 % 4-year cataract rate: 63% Dose reduction did not limit loss of visual acuity. Local tumor recurrence and mets death rates = both groups

Egger et al. [53]

188

DFS, disease-free survival rate; FMS, free-metastasis survival; Fr, fraction(s); LC, local control rate(s); OS, overall survival rate; P, prospective-retrospective series; Rand, randomized.

11.

proton beam radiosurgery

major complications, and 6 for both [54]. The Orsay group with 1062 patients and a median 32 months follow-up reported a 5-year local control and overall survival rate of 97% and 78%, respectively. Metastases were detected in 15%. Prognostic factors for survival were location of the tumor, diameter, thickness, and tumor volume [55]. Approximately 50% of patients, with a pretreatment useful vision, retained it after proton beam. A total of 7.9% of enucleations were performed for tumor relapse or for complications. All published series evidence neovascular glaucoma (45% of the enucleation) as the leading cause for enucleation when tumor is controlled [52–55]. In order to decrease the complication rate, a randomized trial comparing two radiation dose levels (i.e., 50 vs. 70 CGE) in ﬁve fractions was investigated in 188 MEEI patients. There was no difference in terms of local control, overall and disease-free survival, complication rate, and preservation of a useful vision. It has also been suggested to lower dose per fraction while keeping total dose constant in order to decrease the complication rate without affecting the excellent local control [58]. Recently, Liverpool-Clatterbridge published results of the ﬁrst large series of iris melanoma. Local 4-year control rate was high, 86.7%, and one of the most remarkable ﬁndings of this study was the minimal nature of the complications after treatment (mainly cataract), which was eminently treatable [57]. Because of the development of stereotactic fractionated irradiation, a comparative theoretical dosimetric study was performed at Paul Scherrer Institute. A ﬁxed proton horizontal beam and intensity-modulated spot scanning proton therapy (IMPT), with multiple non-coplanar beam arrangements, was compared with linear accelerator–based stereotactic radiotherapy using a static or a dynamic microleaf collimator and intensity-modulated radiotherapy. From imaging of a patient without eye melanoma (with a brain metastasis) were deﬁned several cases of tumor location. The results suggested that use of fractionated stereotactic radiotherapy compared with protons provided similar levels of dose conformation. Tumor dose inhomogeneity was always increased with photon planning. Furthermore, to dose all the contralateral organs at risk was completely eliminated with proton planning [59]. We will also mention ocular hemangiomas, a benign condition located at the posterior eye pole that can affect vision seriously, as a potential indication for low-dose proton beam [60]. Interestingly, recently Tsina et al. reported results of treatment of 76 eyes with ocular metastasis in 63 patients. Treatment delivered 28 CGE in two fractions. This treatment did not need implantation of ﬁducial markers. Proton beam irradiation is a useful therapeutic approach for uveal metastases; it allows retention of the globe, achieves a high probability of local tumor control, and helps to avoid pain and visual loss [61]. Proton beam has become the gold standard for primary ocular malignancies in adults. Its current indications cover posterior and equatorial sites and also large anterior ones (i.e., >5 mm thickness). Few indications remain for radioactive plaques (mainly small anterior tumors) and enucleation (massive ocular involvement) [62].

149

CNS Malignancies Base of the Skull and Upper Cervical Spine Chordomas and Chondrosarcomas Chordomas and low-grade chondrosarcomas of the skull base and upper cervical spine (see Table 11-3) [63–70] represent a paradigm for three-dimensional conformal proton therapy for many reasons: (1) both are radioresistant, that is, poorly controlled with the conventional doses administered in the literature. Surgery plus conventional radiotherapy (i.e., with doses below 60 Gy) yield a 5-year local control rate of 17% to 33% [71–73] and an overall survival rate of 33% to 90% [63, 74–77]. (2) Both are rarely totally resectable, because they are located in sites of difﬁcult surgical access: sphenoid body, petrous bone, clivus, spinal canal. This makes in turn escalated doses suitable. (3) Both lie along and even abut sensitive anatomic structures whose injury could be lethal or highly damaging such as optic nerves, chiasm, brain stem, and spinal cord. (4) Both are located in bony areas that are clearly visible on imaging (especially CT scan) and so easily delineated on three-dimensional simulation and accurately positioned for treatment using bony landmarks. The limited number of new cases per year was also compatible with the time-consuming process that was made necessary for patient setup. The Bragg peak with its sharp lateral and distal dose fall-off proves theoretically of major interest in such challenging situations, and this led to early clinical experiments in Boston in the late 1970s. The impact of dose escalation was clearly suggested through nonrandomized studies that ranged between 55.8 CGE and 83 CGE [23, 63–65, 78–82] delivered with protons (alone or in combination with photons). Local control was switched to 59% to 100%, according to various prognosticators: the strongest one was the pathologic type with chondrosarcomas faring better than chordomas: 59% to 78% for chordomas versus 78% to 100% for chondrosarcomas [23, 64, 65, 78, 80, 83]. Also of prognostic value (but less convincingly) were evidenced: tumor volume, local extension, and ﬁnally gender in chordomas alone (females having a less favorable outcome) [84]. However, this last prognostic factor remains unclear [85]. Interestingly, in the Boston historical series of 141 patients, on 26 local relapses (18%), only 6 of 26 were found after doses ≥70 CGE and 15 of 26 after doses below that level. The main reason for underdosage was the compliance with a dose-constraint due to the proximity of an organ at risk [83]. These ﬁndings led the Proton Radiation Oncology Group (PROG) to explore in a randomized fashion two dose-levels (PROG 25– 86): (1) 66.6 versus 72 CGE in the low-risk group (i.e., all chondrosarcomas and male chordomas), and (2) 72 versus 79 CGE in the high-risk group (i.e., all cervical sites and female chordomas). Results are still pending. In a similar population, Noel et al. showed the importance of quality of radiation, in terms of dose-uniformity within the GTV, a ﬁnding also related to underdosed areas [23]. Two other ﬁndings from our experience should also be mentioned: the delay between treatment and failure that rarely exceeds 3 years. The possibility in few cases is of regional failures that can take the form of tumor seeding within nodal drainage, and/or surgical route, which is highly typical of chordomas.

150

g. noel et al.

TABLE 11-3. Chordoma (CH) and chondrosarcoma (CS) of the base of the skull and upper cervical spine. Type of study

Dose/fractionation/aim

Results

68

P

69 CGE, 1.8 CGE, 80% protons

CH CS

58

P

71 CGE (65–79)

Munzenrider and Liebsch [65]

CH CS

621

P

67 CGE (66–83)

Noel et al. [22, 23, 66]

CH CS

67

P

67 CGE (60–70), 1/3 protons

Igaki et al. [67]

CH

13

P

72 CGE (63–95)

Weber et al. [68]

CH CS

29

P

68–74 CGE

Baumert et al. [69]

Brain

7

TP

Lomax et al. [70]

Miscellaneous

9

TP

Large range of total dose but same dose for photon and proton dosimetry. Study of the distribution of dose of three types of irradiation Ph, IMRT, Pr

Median follow-up 34 months 5-year LC: 82% 10-year LC: 58% Median follow-up 33 months CH 5-year LC: 59% CH5-year OS: 79% CS 5-year LC: 75% CS 5-year OS: 100% Median follow-up 41 months CH 5-, 10-year LC: 73%, 54% CH 5-, 10-year OS: 80%, 54% CS 5-, 10-year LC: 98%, 94% CS 5-, 10-year LC: 91%, 88% Median follow-up 29–31 months CH 4-year LC: 53.8% CH 5-year OS: 80.5% CS 3-year LC: 85% CS 4-year OS: 75% Median follow-up 69.3 months CH 5-year LC: 46% CH 5-year OS: 66;7% Median follow-up 29 months CH 3-year LC: 87.5% CH 3-year OS: 90% CS 3-year LC: 100% CS 3-year OS: 93.8% Mean CI Photons Protons 1.5 (1.15–2.03) 1.2 (1.05–1.38)

Authors

Tumors

Austin-Seymour et al. [78]

CH CS

Hug et al. [104]

No. of cases

healthy tissues irradiated with Pr vs. Ph/ IMRT. Tumor coverage with Pr vs. Ph and equal to IMRT.

CI, conformity index; IMRT, intensity-modulated radiotherapy (using photons); LC, local control rate; OS, overall survival rate; P, prospective-retrospective series; Ph, photons; Pr, protons; TP, theoretical publication.

Lower spinal/paraspinal conditions proved extraordinarily difﬁcult to manage using the ﬁxed beamlines only available at that time and are generally excluded from these studies [86]. Nonetheless, Hug et al. reported on a limited series of 20 patients an outcome close to the rest of the population: 56% and 100% 5-year local control in chordomas and chondrosarcomas, respectively. Total dose ranged from 55 to 82 CGE. Five of 16 chordomas and 0 of 4 chondrosarcomas failed. There was a tendency for improved local control as far as patients treated upfront with radiation rather than at the time of relapse, patients who underwent total/subtotal removal, and those who received a dose above 77 CGE [87]. Salvage therapy after proton therapy has been considered as purely symptomatic in the absence of demonstrated chemosensitivity and of possibility of reirradiation within 5 years. The introduction of recent biological agents (Glivec) could pave new avenues in these situations as well as less advanced presentations [88].

Meningiomas The indication for proton therapy is deﬁnitely more controversial in this tumor type. The main reason is that no clear-cut

dose-response relationship has been evidenced so far (Table 11-4) [89–94]. The initial study by Austin-Seymour et al. in 13 patients was a mixture of benign, malignant, and atypical variants. Median dose was 59.4 CGE (range, 54 to 71.6) and followup 26 months. Local control was 100% [90]. Gudjonsson et al. described results of a hypofractionated regime in 19 patients, delivering 26 CGE in 6 fractions. With a median follow-up of 36 months, local control was again 100% [89]. Miralbell et al. reported on 11 patients treated with photons and protons after incomplete surgery. With a median follow-up of 53 months, no relapse was observed against 6 of 25 patients with comparable tumors irradiated with photons alone [91]. Later on, the HCL published the results of 46 patients with benign meningiomas. Median dose delivered in tumor volume was 59 CGE. Median follow-up was 53 months. Five- and 10-year local control rates were 100% and 88%, respectively, and overall survival rates were 93% and 77%, respectively [24]. The same group published the results in 31 malignant or atypical meningiomas irradiated either conventionally with photons alone or with an escalated dose by a combination of photons and protons. Fiveyear local control rates were 17% and 80%, respectively [93]. Seventeen patients treated at the CPO with a combination of photons and protons to a slightly escalated dose were recently

11.

151

proton beam radiosurgery

TABLE 11-4. Meningiomas. Authors

Tumors

Gudjonsson et al. [89]

Benign, atypical, malignant meningiomas Benign, atypical, malignant meningiomas Benign, atypical, malignant meningiomas Benign, atypical, malignant meningiomas Benign meningiomas

Austin-Seymour et al. [90] Miralbell et al. [91]

Noel et al. [92]

Wenkel et al. [24]

Hug et al. [93]

Noel et al. [94]

No. of cases

Benign, atypical, malignant meningiomas Benign meningiomas

Type of study

Dose/fractionation/aim

Results

19

P

36 CGE in 4 Fr

Median follow-up 36 months No tumoral progression

13

P

59.4 CGE, 1.8 CGE/Fr

Follow-up ≥26 months No tumoral progression

11

P

Combination of photon-protons 54–69 CGE 1.8 CGE/Fr

Median follow-up 53 months No tumoral progression

17

P

Combination of photon-protons 61 CGE 1.8 CGE/Fr

46

P

59 CGE (53–74)

31

P

50–72 CGE

51

P

60.6 CGE Clinical improvement

Median follow-up 37 months 4-year LC: 82.4% 5-year OS: 88.9% Median follow-up 53 months 5- and 10-year LC: 100% and 88% 5- and 10-year OS: 93% and 77% 5-year LC: Photons: 17% Protons: 80% ocular symptoms: 68.8% other symptoms: 67%

Fr, fraction(s); LC, local control rate; OS, overall survival rate; P, prospective/retrospective.

analyzed. Patients were irradiated after surgery or at the time of relapse after surgery. Median follow-up was 37 months. Four-year local control and 5-year overall survival were 87.5% and 88.9%, respectively. The median disease-free interval was increased by 24 months for the patients treated after relapse, a ﬁgure similar to the disease-free interval between surgery and relapse [92, 95]. Our group also brought out the outcome of 51 patients with purely benign meningiomas. The main ﬁnding was the improvement in functional outcome (67% cases) and especially ocular preservation (68.8%) [94].

Gliomas The survival of patients diagnosed with a glioblastoma multiform is dismal (Table 11-5). It has been demonstrated that the majority of patients will fail locally (i.e., within 2 cm around the macroscopic tumor extension). The positive impact of dose-

escalation has been shown using brachytherapy or radiosurgical boosts with photons. Blomquist and Carlsson suggested a proton-based strategy in grades III to IV gliomas: (1) surgical removal of the bulky tumor, (2) high-precision, high-dose proton beam fractionated irradiation in a limited volume encompassing the area at risk plus minimal margin around [96]. The 90-Gy dose level has been investigated through multiple studies: Tatsuzaki et al. conducted dosimetric investigations showing that at this level using protons, none of the brain stem received 60 Gy/CGE compared with 5 cm3 with photons, and that the volume of “nontarget” brain receiving >70 CGE was almost doubled by photons (175 cm3 and 94 cm3, respectively). The reverse side of the coin was a superior dose to the skin delivered with protons alone (63 vs. 45 CGE) [25]. Baumert et al. compared the dose-distribution between modulated-intensity protons and photons using multileaf collimator in seven cases of brain tumors. The conformity index (CI) was better

TABLE 11-5. Brain gliomas. Authors

Tumors

Tatsuzaki et al. [25]

Glioblastoma

No. of cases

1

Type of study

Dose/fractionation/aim

Results

TP

60 Gy/CGE + 30 Gy/CGE (boost) Comparison of normal irradiated volume

20

Phase II

68,2 CGE, 1.8/Fr, grade II 79.7 CGE, 1.8/Fr, grade III Dose escalation

23

Phase II

90 CGE, 1.8/Fr, 2 Fr/day, at least 33% of the total dose with Pr Dose escalation

Photons Protons Normal brain 175 cm3 94 cm3 Brain stem 5 cm3 0 cm3 Max. dose chiasm 60 Gy 60 CGE Max. dose skin 45 Gy 63 CGE Grade III: median survival 29 Grade II not reached 5-year OS grade II: 71% 5-year OS grade III: 23% No increase in LC rate or time of relapse Median survival: 20 months 1-, 2-, 3-year OS: 78%, 34%, 18% Survival: 5–11 months compared with photons

Fitzek et al. [17]

Grade II–III gliomas

Fitzek et al. [16]

Glioblastoma

LC, local control rate; Fr, fraction(s); Pr, protons; OS, overall survival rate; TP, theoretical publication.

152

g. noel et al.

for protons than for photons for complex or concave lesions, or when the target volume was close to critical structures. In other cases, the CI was comparable for both modalities. The number of beams was lower with protons than with photons, which suggests a decrease in the integral dose to the brain [69]. Fitzek et al. reported clinical outcome of a phase II trial that enrolled 23 patients irradiated for a glioblastoma. This institutional study explored a dose of 90 CGE delivered predominately with protons and accelerated fractionation. The median survival time was 20 months, with 4 patients being alive 22 to 60 months after diagnosis. Analysis according to the RTOG prognostic criteria or MRC indices showed a 5- to 11-month increase in median survival time compared with patients treated conventionally in the literature. Radiation necrosis was evidenced in 7 of 29 patients, and survival was signiﬁcantly shortened in patients treated at the time of failure (p = 0.01). Tumor relapse occurred most commonly in areas that received doses >70 CGE; conversely, only one case failed in the 90-CGE volume. The authors concluded that future investigations should aim to cover more widely areas at risk to 90 CGE, although in practice this is rarely achievable due to the risk of radiation necrosis [16]. Another phase II trial was reported by the same group in 20 patients with less aggressive gliomas (grades II to III of the Daumas-Duport classiﬁcation). Prescribed dose were 68.2 CGE in grade II and 79.7 CGE in grade III. Five-year overall survival (OS) rates were 71% and 23%, respectively. The authors concluded that tumor recurrence was neither prevented nor noticeably delayed in these patients compared with the published series on photons [17].

Arteriovenous Malformations Single-dose proton therapy at MGH in the early 1960s gave the impetus to radiosurgical programs worldwide (Table 11-6). A single dose of 10 to 50 CGE was delivered according to the tumor size and radiobiological data sets as mentioned above. Obliteration rate was 20% and complication rate 3% [5]. In an updated series of 95 patients, Amin-Hanjani et al. reported a 17% (before irradiation) to 9% (after irradiation) reduction of the annual hemorrhage rate, but the complication rate was substantial: 26.5%, including 16.3% permanent neurologic deﬁcits and 3% death [27]. Seifert et al. reported a 16% obliteration rate for German patients treated with protons in the United States. They observed clinical improvement in 44%, stability in 27%, and worsening in 29% cases. An unexpected ﬁnding was that therapy was less effective (and so not recommended) for

lesions >3 cm [97], which contradicts dosimetric considerations that plead for the superiority of protons in lesions >4 cm [98]. Recently, Vernimmen et al. reported the experience of a South Africa proton center regarding 64 patients treated for predominately large intracranial AVMs. Irradiation was delivered according to a hypofractionated schedule and dose ranged between 18.4 and 22 single-fraction equivalent CGE. Obliterations were observed in 67% of the lesions with a volume inferior to 14 cm3 and 43% in those with volume superior to 14 cm3. Grade IV complications were reported in 3% of the patients [30].

Vestibular Schwannoma This is another typical case for monofractionated radiosurgery. Weber et al. reported the results of 88 patients with vestibular schwannomas that were treated at HCL with proton beam stereotactic radiosurgery. Two to four convergent ﬁxed beams of 160-MeV protons were applied. The median cross section and target volume were 16 mm and 1.4 cm3, respectively. Previous surgical resection had been made possible in 15 (17%) patients. Facial and trigeminal nerves functions were normal in 79 (89.8%) patients. Eight (9%) patients had good or excellent hearing, and 13 (15%) patients a useful hearing. A median dose of 12 CGE was prescribed to the 70% to 108% isodose lines. Median follow-up was 38.7 months. The actuarial 2- and 5-year tumor control rates were 95.3% and 93.6%, respectively. The actuarial 5-year cumulative radiologic reduction rate was 94.7%. Of the 21 patients (24%) with functional hearing, 7 retained “serviceable” hearing ability. Actuarial 5-year normal facial and trigeminal nerve function preservation rates were 91.1% and 89.4%, respectively. Univariate analysis revealed that prescribed dose (p = 0.005), maximum dose (p = 0.006), and the inhomogeneity index (p = 0.03) were associated with a signiﬁcant risk of long-term facial neuropathy [99, 100]. These results compare favorably with other published radiosurgical series. In summary, current knowledge on proton therapy in skull base and probably spinal canal low-grade sarcomas make proton therapy highly suitable in a majority of patients both in terms of oncologic outcome and quality of life. The place of surgical resection remains crucial because it can greatly improve ballistics to critical structures, and possibly by reducing tumor burden. Randomized studies (and possibly meta-analyses) might conﬁrm these ﬁndings as the number of patients submitted to this approach is expanding. The introduction of modern technologies especially isocentric gantries should make possible

TABLE 11-6. Arteriovenous malformations. Authors

Tumors

No. of cases

Type of study

Dose/fractionation/aim

Results

Kjellberg et al. [5]

AVM

75

P

10.5–50 CGE, 1 Fr

Amin-Hanjani et al. [27] Vernimmen et al. [30]

AVM

95

P

AVM

64

P

Median max. dose: 18.3 CGE (8–36.7), 1 Fr 18.38–22.05 SFEGyE

Complete occlusion rate: 20% Partial occlusion rate: 56% Death: 2 cases Reduction hemorrhage risk: 17% to 9% Complication: 26.6% Median follow-up: 62 months Obliteration rate for lesion >14 cm3: 67% Obliteration rate for lesion >14 cm3: 43% Complication: 3%

Fr, fraction(s); P, prospective; SFEGyE, single-fraction equivalent Gy equivalent.

11.

proton beam radiosurgery

the management of a variety of spinal/paraspinal malignancies in the future. PT remains controversial in meningiomas although it represents an elegant option for sparing normal tissues. In high-grade gliomas, PT has been disappointing despite substantial tumor dose increase. Future directions should explore combined chemoradiation and light ions.

Pediatric Tumors Despite the prominent role held by chemotherapy in this age group, radiotherapy still plays a crucial role in most solid tumors especially those located in the brain (approximately 20% of cases). There is a considerable amount of data indicating that radiotherapy even at low dose can induce severe sideeffects especially in young children (Table 11-7) [101–106]. For example in the CNS, it can affect cognitive function and pituitary-driven growth at >20 Gy in whole brain, sellar area, auditory function at >30 Gy in cochlea, and so forth. Another major concern comes from irradiation of bones in the prepubertal age, with impairment of growth plates at >15 Gy. Carcinogenicity has also represented a serious threat, especially in the usual combined chemo-radiation approaches. A salient feature of proton dose-distribution that can be of considerable interest in this age group is the absolute reduction of the integral dose (mainly due to the lack of “exit” beam) that minimizes areas receiving low and moderate doses around the target. On the other hand, there are practical serious limitations that have made this approach still almost conﬁdential in this indication, to mention a few: (1) The rarity of pediatric oncology centers (correlated with the rarity of the disease itself) that are restricted to a few places not necessarily close to a particle center. This comes along with the unique expertise required from radiation oncologists, physicists, and technologists involved. (2) The difﬁculty for performing extensive and uncomfortable daily setups without the help of deep sedation and even general anesthesia in the youngest patients. There is obviously a need for careful patient selection through preliminary dosimetric investigations in order for example to deﬁne the merits of different technological approaches like conformal photons or IMRT. In a comparative study between protons and conformal photons, in optical pathway gliomas, Fuss et al. showed that the CI (ratio of GTV to non-GTV encompassed in the 95% isodose) was better with protons than with photons. They also showed that doses delivered with protons in normal optic nerve, chiasm, pituitary gland, and temporal lobes were respectively reduced by 47%, 11%, 13%, and 39% compared with those delivered with photons [101]. For tumors located in the posterior fossa, Lin et al. showed that cochlea received 25% of the proton dose versus 75% of photons. Furthermore, 40% of temporal lobes were fully spared from protons, whereas 90% were exposed to photons to a minimum 30% of the total dose [31]. Based on theoretical models, Miralbell et al. reported a potential 10% drop in the predictable risk of IQ decline in medulloblastoma treated with CNS irradiation by the age of 4 years, using protons compared with photons. NTCP (normal tissue complication probabilities) values were also lowered by protons but to a modest extent when compared with highly conformal photons [107]. The same authors compared both techniques in cervical irradiation to 27 Gy at the age of 2 to 3

153

years. As far as cervical spine, the volume receiving ≥50% dose was ﬁvefold using photons versus of protons (i.e., 100% vs. 20%). As far as heart, the difference was even more striking (≥45% dose, photons; 100%, protons; 0% volume). Low doses to thyroid, liver, and gonads were similarly reduced. The authors’ estimates of “height sparing effect” by protons was of the order 10 cm [32]. It is interesting to mention that most authors acknowledge that the beneﬁt of protons increases with tumor size, a datum of special interest in brain malignancies [26, 70, 98]. Similarly, reduction of the carcinogenic risks is of primary concern in this age group and can be expected from protons (see above) [102]. The Swiss group developed mathematical algorithms based on radioprotection estimates to quantify these risks [108]. Carcinogenic risk between “rival” techniques (i.e., conformal photons/protons, and photons/ protons IMRT) were appraised: in parameningeal rhabdomyosarcomas, the “protective” effect of protons was ≥2-fold, and in medulloblastoma, 8- to 15-fold [109]. Elegant beam arrangements are made possible with protons as mentioned previously. Sophisticated patch techniques make sparing of abutting critical organs feasible. Impressive sparing of highly sensitive structures such as growing plates has been brought out in retinoblastomas by Krengli [102], orbital rhabdomyosarcomas, and lumbar neuroblastomas by Hug [104]. Recent theoretical study including cases of retinoblastoma, medulloblastoma, and pelvic sarcoma cases concluded that protons delivered superior target dose coverage and sparing of normal structure. In pelvic sarcoma studied in this series, none of the ovaries received dose superior to 2 Gy; furthermore, as expected, proton lowdose volume is greatly inferior to that obtained with IMRT. These volumes receiving low dose of irradiation have been suspected to be the site of radiation-induced secondary cancer [103]. Clinical series on the use of protons in children are still scarce. The Boston group reported on 18 children aged 4 to 18 years with a skull base chordoma/chondrosarcoma. With a median 72 months follow-up, 5-year OS and RFS were 68% and 63%, respectively. The complication rate was limited to one temporal lobe necrosis [64]. The preliminary Orsay experience was reported by Noel et al. on 17 children (median age, 12 years), with skull base sarcomas as the main indication, who received combined photon-proton irradiation (approximately 50%–50% of the dose). With a median 27 months follow-up (range, 3 to 81), 3-year local control was 91.7% and 1-, 2-, and 3-year OS 93.3%, 83%, and 83%, respectively. One child failed in-ﬁeld and one at the margin. No late side-effect was reported although follow-up is deﬁnitely short [18, 106]. McAllister et al. at Loma Linda University Medical Center (LLUMC) reported on 28 children treated for grades 2 to 4 glioma. At the time of analysis, three were dead from disease, one was alive with tumor progression, and the others were with NED. Complication rate was low [105]. Hug et al. reported on 27 children treated by protons for low-grade malignancies. Six relapsed, four died, and the others were with NED and complicationfree. A subgroup of six children with optic glioma had visual preservation or improvement [21]. Giant cell tumors seen in the pediatric age are regarded as benign conditions, although they can behave aggressively. In this situation, high-dose proton therapy is warranted (just as in the adult sarcomas) and can induce prolonged remissions [110].

154

g. noel et al.

TABLE 11-7. Pediatric cancers. No. of cases

Type of study

Authors

Tumors

Dose/fractionation/aim

Results

Lin et al. [31]

Posterior fossa

9

TP

54 Gy/CGE Comparison 3D Ph and Pr

Fuss et al. [101]

Optic glioma

7

TP

50.4–50 Gy/CGE Dose distribution comparison for Pr, 3D conformal Ph, standard pH (std)

Krengli et al. [102]

Retinoblastoma

1

TP

Lee et al. [103]

Retinoblastoma

3

TP

46 CGE in GTV, 40 CGE in CTV Proton beam arrangements for various intraocular tumor locations Comparison of different technique of Ph including IMRT and proton

% prescribed Pr PH dose in cochlea 25% 75% Pr: 40% temporal lobes excluded irradiated volume PH: 90% temporal lobes received ≥31% of prescribed dose CI 3D Ph Std Ph Pr 2.9 7.3 2.3 Reduction of dose/Pr 3D Ph Std Ph Contralateral optic 47% 77% nerve 11% 16% Chiasm 13% 16% Pituitary 39% 54% Temporal lobes PT risks K2 + cosmetic / functional sequelae

Mean % volume/ technique 5 Gy orbit 20 Gy optic nerve 25 Gy cochlea 10 Gy pituitary 10 Gy thyroid 10 Gy lung 10 Gy kidney 15 Gy heart 5 Gy ovary 20 Gy vertebra 30 Gy bowel

3D RT 25 53 64 91 24 15 18 2 100 20 11

IMRT 69 55 33 81 100 14 15 59 29 29 12

Medulloblastoma

3

Pelvic sarcoma

3

Hug et al. [104]

Neuroblastoma

1

P

34.2 CGE Dose distribution

50% ipsilateral kidney <16 CGE 50% contralateral kidney <1 CGE 50% liver <2.6 CGE 20% liver ≥10 CGE Spinal cord <3 CGE

Benk et al. [64]

CH-CS

18

P

Follow-up ≥72 months 5-year LC: 78% 5-year OS: 68%

Krengli et al. [102]

Various ocular tumor

69 CGE, 1.8 CGE /Fr, 80% protons Combination photons and protons (60–80 CGE) 46 CGE in GTV, 40 CGE in CTV

Hug et al. [21]

Low-grade glioma

27

P

Mean dose 55.2 CGE (50.4–63) 1.8 CGE/Fr

McAllister et al. [105]

Miscellaneous

28

P

Noel et al. [18, 106]

Miscellaneous

17

P

Pr only: median: 54 CGE (40–70) Mixed Pr and pH: Ph: median 36 Gy (18–45) Pr: median 18 CGE (13–32) Mixed Pr and pH: Ph: median 40 Gy (24–54) Pr: median 20 CGE (9–31)

Protons 10 29 6 21 7 2 2 0 0 9 5

Pr: K2 + cosmetic outcome and functional sequelae Median follow-up 39 months OS: 85% LC: 78% 4 hypopituitarisms 1 asymptomatic brain necrosis 4 relapses Complications: 2 seizures 1 hormonal deﬁcit 1 cataract Median follow-up: 27 months (3–81) 1 in site + 1 marginal relapse 3-year CL: 91.7% 1-, 2-, 3-year OS: 93.3%, 83%, 83%

LC, local control rate; Fr, fraction(s); CI, conformity index; CTV, clinical target volume; GTV, gross tumor volume; K2, secondary cancer; Ph, photons; Pr, protons; OS, overall survival rate; std, standard; P, prospective/retrospective series; TP, theoretical publication.

11.

155

proton beam radiosurgery

TABLE 11-8. Pelvic cancers. Authors

Tumors

Type of study

Dose/fractionation/aim

Results

Isacsson et al. [42]

Inoperable rectum

No. of cases

6

TP

46 Gy/CGE Photons 4 b; protons 3 b Photons 4 b + protons 3 b

Tatsuzaki et al. [43]

Rectum

1

TP

50 Gy/CGE ± 5 Gy/CGE (boost) Comparison photons 2, 3, 4 b Protons 1 b, conformal or not

Cella et al. [111]

Prostate

1

TP

81 CGE • Ph 6 MeV 18 conformal (6 b) • Pr 214 MeV (2 b) • IMRT 15 MeV (5 b) • IMPT 177–200 MeV (5 b)

Schulte et al. [41, 112]

Localized prostate stage T1-T2b

911

P155

Pr alone or combining with Ph 74–75 CGE

Shipley et al. [113]

Prostate, T3–4, Nx, N0–2

202

Rand.

Ph: 50.4 Gy + 16.8 Gy Pr: 50.4 Gy + 25.2 CGE, 1.8 Gy/CGE/Fr Comparison dose escalation

Reduction of the irradiated digestive tract, femoral head and of bladder volume with proton, increased TCP from 8 to 14% if NTCP = 5% Proton treatment allows higher doses to be delivered to the tumor, with a probable increase in TCP, or a reduction in NTCP • Better homogeneity with proton • Low or mean dose volume decreased with proton • Improvement of NTCP of the rectum (<5%, grade 3) IMPT or IMRT 5-year biological survival: 82% No grade 3–4 complication Grade 2 rectal complication rate: 3.5% Grade 2 bladder complication rate: 5.4%. Median follow-up: 61 months Improvement of local control for the most undifferentiated tumors with Pr

b, beams; Fr, fraction(s); IMPT, intensity-modulated protontherapy; IMRT, intensity-modulated radiotherapy; LC, local control rate; NTCP, normal tissue complication probability; OS, overall survival rate; Ph, photons; Pr, protons; Rand., randomized; TCP, tumor control probability; TP, theoretical publication.

In summary, pediatric conditions represent a major challenge to radiation oncologists, mainly due to the risk for serious long-term side-effects. This challenge could possibly be overcome by a more systematic use of proton therapy especially in young children as suggested by preliminary studies. The pediatric community is certainly waiting anxiously for deﬁnitive clinical evaluations.

Pelvic Cancers Protons are attractive in this location because there is a number of relatively sensitive structures such as bladder, rectum, and femoral heads to be spared. On the other hand, tumors are generally located deeply and require beam energy >160 MeV, which is not widely available. (Table 11-8) [111–113]. If so, protons can be offered with good chances to achieve improved conformation, tumor homogeneity, and decreased integral dose as evidenced by the dosimetric intercomparisons by Cella et al. To a dose of 81 Gy, only intensity-modulation photons and protons could comply with acute rectal toxicity [111]. It has also been estimated that doses could be escalated by 20% this way, with parallel normal organs volume reduction up to 60%, if an isocentric gantry was available [44–46]. In case of mixed beams, Tatsuzaki et al. evidenced a possible relationship between beneﬁt and proportion of photons and protons [43]. Similarly in inoperable rectal carcinoma, Isacsson et al. estimated that tumor control probability was increased by 8% with mixed photon-protons and 14% with protons alone, for a ﬁxed 5% complication risk [42]. Most clinical studies have dealt with prostate carcinomas managed both at late and early stages: in Boston, between 1982

and 1992, T3-T4, Nx, N0-2, and M0 were elected to receive 50.4 Gy with photons in a box technique followed in a random fashion by a conformal boost of 25.2 CGE with protons or 16.8 Gy with photons. Total dose was slightly superior using protons rather than photons (75.6 CGE vs. 67.2 Gy). Among 202 patients, 8-year local control after proton and photon treatments were 77% and 60%, respectively (p = 0.089) with similar OS and DFS. A subgroup analysis in 57 poorly differentiated (Gleason 4 or 5) tumors evidenced an 8-year local control rate of 84% with protons versus 19% with photons (p = 0.0014). Grades 1 to 2 postirradiation rectal bleeding (i.e., not requiring surgery or hospitalization), and correlated with telangiectasia, affected 34% and 16% patients treated with protons and photons, respectively (p = 0.013) [113, 114]. This increased toxicity in the proton arm was related to a superior total dose but also with a relatively simple technique for boosting based on a single perineal ﬁeld. Oppositely, LLUMC explored early stages. On 911 patients managed with protons between 1991 and 1996, 5-year “biological” (i.e., normal PSA) local control was 85% [41, 112]. In summary, there is a considerable potential for pelvic tumor sites (and related ones like sacral malignancies) for centers with the most advanced equipment. The place of proton therapy in prostate carcinoma remains highly controversial especially in early stages, until comparative studies with alternative techniques (Brachytherapy, surgery, etc.) are contemplated.

Head and Neck Tumors There is again a substantial body of data based on dosimetric intercomparisons but few clinical reports. (Table 11-9) [115, 116]. In a comparative study testing current photon and proton

156

g. noel et al.

TABLE 11-9. Head and neck tumors. Authors

Tumors

Type of study

Dose/fractionation/aim

Results

Miralbell et al. [28]

Cancer of the maxillary sinus

1

TP

64 Gy/CGE • Classic Ph • Conformal 3D Ph • Ph-Pr

Slater et al. [34]

Cancer tonsillar T2 and T3

2

TP

T2: 70 Gy/CGE + 5 Gy/CGE boost; Ph 2 b; Pr 1 b T3: 70 Gy + 5 Gy boost; Ph 2 b; 70 CGE + 15 CGE boost Pr 1 b

• homogeneity with mixed and 3D technique • Dose escalation in tumoral volume without dose increase in organs risk T2: dose in contralateral parotid gland + mandible T3 same result with conformity

Slater et al. [33]

Oropharyngeal carcinoma stages II–IV Advanced head and neck cancer Orbital or periorbital tumor

29

P

50.4 Gy Ph 28 Fr of 1.8 Gy + 25.5 CGE Pr as boost in 17 Fr

6

TP

4

TP

54 Gy/CGE PH-electrons; photons 3D: IMRT Passive Pr or spot scanning 30–50.4 Gy/CGE Modulated intensity with photons or protons

Brown et al. [35] Noel et al. [37]

Nasopharyngeal carcinoma Nasopharyngeal carcinoma

2

TP

70 Gy/CGE; 3D Ph and Pr

5

TP

Lin et al. [36]

Relapsed nasopharyngeal carcinoma

70 Gy/CGE; 3D Ph; Ph + Pr boost Dose distribution in tumoral volume and organs at risk 62.8 CGE (59.4–70.2); 1.8 to 2 CGE/Fr

Cozzi et al. [115] Miralbell et al. [116]

No. of cases

12

P

Median follow-up: 28 months 2- and 5-year LC: 96% and 88% 2- and 5-year DFS: 81% and 65% • homogeneity with protons • dose spinal cord and parotid with protons NTCP (mean) Pr Ph Ipsilateral lens 0.30% 1.63% Ipsilateral choroid 1.03% 1.10% Ipsilateral parotid 0% 0.10% Contralateral lens 0.30% 1.40% Mean dose 5 Gy in the tumor with protons and organs at risk CI: 3.1 for mixed Ph-Pr and 5.7 for Ph only 78% of the 68 criteria studied in favor of Ph-Pr combination Median follow-up duration: 23.7 months 2-year LC: 50% 2-year DFS: 50% 2-year OS: 50%

b, beams; Fr, fraction(s); CI, conformity index; DFS, disease-free survival; IMRT, intensity-modulated radiotherapy; LC, local control rate; CL, rate of local control; NTCP, normal tissue complication probability; OS, overall survival; Ph, photons; Pr, protons; TCP, tumor control probability; TP, theoretical publication.

techniques (including passive diffusion, raster and spot scanning; see above) on ﬁve different treatment plans by Cozzi et al., main organs (i.e., spinal cord and parotid gland) shielding was always superior using protons. Dose homogeneity in the target was also improved but to a modest extent in comparison with conformal photons [117]. Miralbell et al. made a similar evaluation in maxillary sinus with similar conclusions but pointed out the difﬁculty evaluating the actual proton dose within air cavities [28]. There is also a study by Slater et al. in tonsillar primary, evidencing the potential improved sparing of salivary glands and mandible [34]. Recently, the same group presented results about 29 patients treated with accelerated irradiation including proton boost for an oropharyngeal tumor. Two- and 5-year disease-free survival were 81% and 65%, respectively. The dose increase, up to 75.9 CGE, was particularly well tolerated with only three cases of grade 3 complication [33]. Nasopharyngeal carcinoma represents a paradigm as there is evidence in the literature on the positive role of doseescalation. Brown et al. [35] exempliﬁed two cases. As far as clinical studies, protons have been investigated in recurrent nasopharyngeal carcinomas. It has been acknowledged since the initial reports by C.C.Wang and others that reirradiation was feasible but at the price of increased toxicity in extensive failures at the skull base. Lin et al. reported on 16 such cases. At initiation of therapy, 20 patients had symptoms consistent with intracranial involvement. Median dose at reirradiation was

62.8 CGE (range, 59.4 to 70.2), and cumulated with the previous one, 134.6 CGE (range, 110 to 148). With a median 23.7 months follow-up (range, 4 to 47), 2-year local/regional DFS and OS were both 50%. The authors pointed out that chances for prolonged remission were correlated with target dose homogeneity. Despite the high cumulative dose, only two severe complications were observed, due to a good sparing of the organs at risk the second time [36]. In summary, proton therapy has not gained wide acceptance in head and neck carcinomas for multiple reasons, such as the current interest for alternative techniques that are more readily accessible (especially photon IMRT), the difﬁculty of modeling with current biophysical algorithms, and considerable tissue heterogeneities, especially in the paranasal area. These limitations could be overcome in the future with expected technological and biophysical improvements. It will certainly remain debatable to put protons forward for an entire treatment program, including lymphatic drainage coverage.

Bronchial Cancer and Esophagus Bush et al. explored stages I to III lung cancers in 37 patients (Table 11-10). The group was divided in two according to whether patients had good or poor cardiopulmonary function. The ﬁrst group received 73.8 Gy by combined photons and

11.

157

proton beam radiosurgery

TABLE 11-10. Thoracic and esophagus cancers. Authors

Tumors

Type of study

Dose/fractionation/aim

Results

Bush et al. [48]

Stage I and III lung cancer

37

P

Tsunemoto et al. [39]

Stage T1–4 esophagus cancer Stage T1–4 esophagus cancer

16

P

Good cardiac function Ph 45 Gy + Pr 28.8 CGE; 1.8 Gy/CGE/Fr Bad cardiac function Pr 51 CGE, 5.1 CGE/Fr 80–88 CGE

46

P

Median duration of 14-month monitoring 2-year LC: 87% 2-year DFS: 63% 2-year OS: 31% Complete response in 13 patients LC T1: 83% T2–4: 29% OS Overall: 34% T1: 55%, T2–4: 13% DFSS Overall: 67% T1: 95% T2–4: 33%

Sugahara et al. [38]

No. of cases

Ph-Pr combination* Median dose Ph: 48 Gy Median dose Ph: 31.7 CGE Total dose range: 69.1–87.4 CGE Pr only Total dose range: 75–89.5 Gy

DFS, disease-free survival rate; DFSS, disease-free speciﬁc survival rate; Fr, fraction(s); LC, local control rate; OS, overall survival rate; Ph, photons; Pr, protons; P, prospective. *RBE = 1 in this study.

protons (45 Gy + 28.8 CGE) to the mediastinum and GTV + cone down to the primary only. The second group received exclusive protons to the GTV only, up to 51 CGE with accelerated regime (10 weekly fractions over 2 weeks). With a median follow-up of 14 months, local control was 87%. Two-year OS and DFS were 63% and 31%, respectively [48]. These results compared favorably with those of photons in the literature. Tsunemoto et al. managed 16 esophageal carcinomas with doses escalated up to 88 CGE. Thirteen enjoyed complete response [39]. Saguhara reported on 46 cases treated with mixed beams (40) or protons alone (6). Respective median doses were 76 and 82 CGE. Eighteen patients failed either locoregionally (16) or distantly (2). Five-year local control was 83% in T1 and 29% in T2-4. Five-year OS and DFS in T1-T2-T4 were 34%, 55%, 13% and 67%, 95%, 33%, respectively. These results suggest that proton beam therapy can be a reasonable option at least in early presentations [38]. In summary, no ﬁrm conclusion can be drawn from such limited and heterogeneous series, and comparative studies with alternative therapies are missing.

Toxicity Despite the relatively few long-term side-effects reported, a precious database has been accumulated especially by the Boston group with remarkable accuracy on dose, volume, and risk estimates. It provides highly valuable pieces of information on tolerance of normal organs to radiation (especially in the brain) that might expand our knowledge beyond the “particles” community and feed, for example, NTCP models in threedimensional conformal therapy. As far as temporal lobes, Santoni et al. conducted a study in 96 chordoma/chondrosarcoma patients. The 2- and 5-year complication rates (i.e., clinical and/or radiologic symptoms of radionecrosis) were 8% and 13%, respectively, after doses of

66 to 72 CGE. Surprisingly, male gender showed up as an independent risk factor [117] (we mentioned above the adverse role of female gender in failures in the same population). Glosser et al. focused on neuropsychological outcome in 17 patients who received radiation up to 66 CGE for skull base malignancies. They did not observe any early or late cognitive side-effect related to temporal lobe necrosis among NED patients; however half had psychomotor speed impairments and few developed minor transient symptoms like depression or anxiety [118]. Brain-stem tolerance was studied by Debus et al. in the same population. They reported 17 of 367 (4.5%) injuries. The 10-year complication rate was 12% and the mean free interval 10 months (90% within 3 years). Risk of injury was correlated with the number of previous surgical procedures, dose >60 CGE, and association with diabetes mellitus [119]. In benign meningiomas, Wenkel et al. found 1 of 46 (2%) BS necrosis, which is consistent with the lower dose administered (i.e., 59 CGE) [24]. The authors mentioned noncompliance to recommended doseconstraints as a major risk factor. Spinal cord tolerance was studied by Marucci et al. Thirteen minor and four severe injuries out of 85 patients with cervical spine malignancies were found (15% and 4.4%, respectively). There was no dose-effect relationship in the range explored (≤55 to 58 CGE cord center and ≤67 to 70 CGE cord surface). The only predictor of toxicity was the number of previous surgeries [120]. Radiation-related hypothalamic-pituitary endocrine damage was reported in the Boston series by Munzenrider et al. in 40% of cases [65]. From Slater et al., patients deteriorate between 14 and 45 months postirradiation and the risk is dosedependent: 50% at the 67.6 CGE level [121]. In the Orsay population, risk of symptomatic toxicity has been set to 8 of 64 (12.5%) but with a somewhat shorter follow-up [92]. Neurovisual impairments represent undoubtedly a major concern in escalated-dose skull base protocols (along with BS toxicity). This is due to their dramatic impact on quality of life

158

g. noel et al.

for patients with generally extended life expectancy. It strongly inﬂuences the approach of our group in the management of skull base sarcomas, as patients with tumor abutting the visual pathway are generally offered a limited reoperation in order to remove this extension by at least 1 or 2 mm. Given the sharp proton lateral penumbra (15%/mm), this means a 5- to 10-CGE drop toward the critical structure. Risk estimate has been appraised by Habrand et al. to ∼10% at 55 CGE and ≥20% at >60 CGE [122]. In Wenkel’s series of meningiomas, 4 of 46 (9%) experienced neurovisual symptoms [24]. Kim et al. demonstrated that the “patch” technique (see above) could be at risk just as usual photon beam matching if ﬁeld abutments concerned nerves or chiasm [123]. Several series point out the predisposing role of underlying vasculopathies related to diabetes mellitus or HBP [92, 122]. Permanent hearing loss is a symptom frequently related to cochlear injury [65, 93, 110, 124]. Schoenthaler et al. showed that in two of three affected patients, dose had been >62.7 CGE [110]. The rest of cranial nerves has been found relatively “radioresistant” as evidenced by Munzenrider and Liebsch [65]. In the detailed Urie et al. analysis in 27 patients, based on anatomic nerve reconstructions in three dimensions, 17 symptomatic injuries (in 5 patients) of 594 structures were found (3%), with a mean 74 months interval. Using logistic regression, they estimated the risk for “common” cranial nerves to 1% at 60 CGE and 5% at 70 CGE [124]. Using the model based on radioprotection data mentioned above, Schneider et al. came up with a twofold decrease in the carcinogenic risk for adults irradiated for Hodgkin disease when protons replaced photons [109]. Carcinogenicity in the pediatric age has been discussed above. All previous data concern adult patients treated with conventional fractionation. We are still missing age-based information in children. In summary, a remarkably low complication rate has been reported despite the considerable dose generally administered. This makes proton therapy highly attractive in most tumor sites, especially in children.

Socioeconomic Aspects Initiating a proton program is not as painful as expected when a “physics machine” becomes available for therapy. For example, the Orsay project has been conducted so far on an existing synchrocyclotron, kindly offered by the French Scientiﬁc Research Authority (CNRS). Adaptations for two treatment rooms were approximately *600,000, and yearly running cost (including beam’s hour, and staff salaries) approximately *2 million. With approximately 250 ocular and 50 intracranial malignancies managed per year, patients have been charged approximately *1000 per session, a ﬁgure still 10-fold that of conventional radiation (but less compared with sophisticated three-dimensional photons). The treatment cost has been actually substantially reduced since generally one-half to two-thirds of treatment is performed using conformal photons. Fully operational new equipment is more costly by 2 orders of magnitude. Goitein et al. have simulated the relative cost of state-of-the-art proton and photon facilities (i.e., with isocentric gantries and

intensity modulation capabilities) [8]. The following costs were attributed to protons: two-gantry facility, *62.5 million; and operating cost per fraction, *1025. As far as photons, they were respectively *16.8 million and *425. The authors estimated that relative treatment costs would be of the order 2.4 initially and down to 1.7 in the long run. Cost-beneﬁt evaluations are also emerging that suggest a ﬁnancial beneﬁt in using protons: this has been shown in pediatric medulloblastoma, when rehabilitation of children disabled by photon irradiation was taken into account [125]. New technologies that aim to produce protons at a lower cost are also in progress, like particle beams generated by ultraintense laser pulses [126].

Conclusion There is a respectable body of evidence showing the ballistic superiority of protons over conformal photons currently. This might be conﬁrmed in future investigations comparing intensity modulation with both particles. From a clinical standpoint, current successes in intracranial and ocular malignancies do not preclude the need for future controlled trials. In this context, the place of new particles, especially light ions, is emerging and should also be deﬁned.

References 1. Orecchia R, Zurlo A, Loasses A, et al. Particle beam therapy (hadron therapy): basis for interest and clinical experience. Eur J Cancer 1998; 34:459–468. 2. Bragg WH, Kleeman R. On the alpha particles of radium, and their loss of range in passing through various atoms and molecules. Philos Mag 1905; S6:318–340. 3. Wilson RR. Radiological use of fast proton. Radiology 1946; 47: 487–491. 4. Wilson R. A Brief History of the Harvard University Cyclotrons. Cambridge, MA: Harvard University Department of Physics, 2004. 5. Kjellberg RN, Hanamura T, Davis KR, et al. Bragg-peak protonbeam therapy for arteriovenous malformations of the brain. N Engl J Med 1983; 309:269–274. 6. Urano M, Verhey LJ, Goitein M, et al. Relative biological effectiveness of modulated proton beams in various murine tissues. Int J Radiat Oncol Biol Phys 1984; 10:509–514. 7. Brown A, Suit H. The centery of the discovery of the Bragg peak. Int J Radiat Oncol Biol Phys 2004; 73:268. 8. Goitein G, Lomax A, Pedroni ES. Treating cancer with protons. Physics Today 2002; 45–55. 9. Moyers MF, Lesyna W. Isocenter characteristics of an external ring proton gantry. Int J Radiat Oncol Biol Phys 2004; 60:1622– 1630. 10. Tourovsky A, Lomax AJ, Schneider U, Pedroni E. Monte Carlo dose calculations for spot scanned proton therapy. Phys Med.Biol 2005; 50:971–981. 11. Chu WT, Ludewigt BA, Renner TR. Instrumentation for treatment of cancer using proton and light-ion beams. Rev Sci Instrum 1993; 64:2055–2122. 12. Paganetti H, Niemierko A, Ancukiewicz M, et al. Relative biological effectiveness (RBE) values for proton beam therapy. Int J Radiat Oncol Biol Phys 2002; 53:407–421. 13. Molokanov AG, Spurny F. Biological weighted effective dose in 205 MeV clinical proton beam. Phys Med Biol 2005; 50:281– 287.

11.

proton beam radiosurgery

14. Delacroix S, Bridier A, Mazal A, et al. Proton dosimetry comparison involving ionometry and calorimetry. Int J Radiat Oncol Biol Phys 1997; 37:711–718. 15. Archambeau JO, Slater JD, Slater JM, Tangeman R. Role for proton beam irradiation in treatment of pediatric CNS malignancies. Int J Radiat Oncol Biol Phys 1992; 22:287–294. 16. Fitzek MM, Thornton AF, Rabinov JD, et al. Accelerated fractionated proton/photon irradiation to 90 cobalt gray equivalent for glioblastoma multiforme: results of a phase II prospective trial. J Neurosurg 1999; 91:251–260. 17. Fitzek MM, Thornton AF, Harsh G, et al. Dose-escalation with proton/photon irradiation for Daumas-Duport lower-grade glioma: results of an institutional phase I/II trial. Int J Radiat Oncol Biol Phys 2001; 51:131–137. 18. Habrand JL, Mammar H, Ferrand R, et al. Proton beam therapy (PT) in the management of CNS tumors in childhood. Strahlenther Onkol 1999; 175(Suppl 2):91–94. 19. Hug EB, Slater JD. Proton radiation therapy for pediatric malignancies: status report. Strahlenther Onkol 1999; 175(Suppl 2): 89–91. 20. Hug EB, Muenter MW, Adams JA, et al. 3-D-conformal radiation therapy for pediatric giant cell tumors of the skull base. Strahlenther Onkol 2002; 178:239–244. 21. Hug EB, Muenter MW, Archambeau JO, et al. Conformal proton radiation therapy for pediatric low-grade astrocytomas. Strahlenther Onkol 2002; 178:10–17. 22. Noel G, Habrand IL, Jauffret, de Crevoisier. Radiation therapy for chordoma and chondrosarcoma of the base of the skull and the cervical spine: prognostic factors and patterns of failure. Strahlenther Onkol 2003; 179:241–248. 23. Noel G, Feuvret L, Ferrand R, et al. Radiotherapeutic factors in the management of cervical-basal chordomas and chondrosarcomas. Neurosurgery 2004; 55:1252–1260. 24. Wenkel E, Thornton AF, Finkelstein D, et al. Benign meningioma: partially resected, biopsied, and recurrent intracranial tumors treated with combined proton and photon radiotherapy. Int J Radiat Oncol Biol Phys 2000; 48:1363–1370. 25. Tatsuzaki H, Urie MM, Linggood R. Comparative treatment planning: proton vs. x-ray beams against glioblastoma multiforme. Int J Radiat Oncol Biol Phys 1992; 22:265–273. 26. Wambersie A, Gregroire V, Brucher JM. Potential clinical gain of proton (and heavy ion) beams for brain tumors in children. Int J Radiat Oncol Biol Phys 1992; 22:275–286. 27. Amin-Hanjani S, Ogilvy CS, Candia GJ, et al. Stereotactic radiosurgery for cavernous malformations: Kjellberg’s experience with proton beam therapy in 98 cases at the Harvard Cyclotron. Neurosurgery 1998; 42:1229–1236. 28. Miralbell R, Crowell C, Suit HD. Potential improvement of three dimension treatment planning and proton therapy in the outcome of maxillary sinus cancer. Int J Radiat Oncol Biol Phys 1992; 22: 305–310. 29. Steinberg GK, Fabrikant JI, Marks MP, et al. Stereotactic heavycharged-particle Bragg-peak radiation for intracranial arteriovenous malformations. N Engl J Med 1990; 323:96–101. 30. Vernimmen FJ, Slabbert JP, Wilson JA, et al. Stereotactic proton beam therapy for intracranial arteriovenous malformations. Int J Radiat Oncol Biol Phys 2005; 62:44–52. 31. Lin R, Hug EB, Schaefer RA, et al. Conformal proton radiation therapy of the posterior fossa: a study comparing protons with three-dimensional planned photons in limiting dose to auditory structures. Int J Radiat Oncol Biol Phys 2000; 48:1219– 1226. 32. Miralbell R, Lomax A, Russo M. Potential role of proton therapy in the treatment of pediatric medulloblastoma/primitive neuroectodermal tumors: spinal theca irradiation. Int J Radiat Oncol Biol Phys 1997; 38:805–811.

159

33. Slater JD, Yonemoto LT, Mantik DW, et al. Proton radiation for treatment of cancer of the oropharynx: Early experience at Loma Linda University Medical Center using a concomitant boost technique. Int J Radiat Oncol Biol Phys 2005; 62:494–500. 34. Slater JM, Slater JD, Archambeau JO. Carcinoma of the tonsillar region: potential for use of proton beam therapy. Int J Radiat Oncol Biol Phys 1992; 22:311–319. 35. Brown AP, Urie MM, Chisin R, Suit HD. Proton therapy for carcinoma of the nasopharynx: a study in comparative treatment planning. Int J Radiat Oncol Biol Phys 1989; 16:1607–1614. 36. Lin R, Slater JD, Yonemoto LT, et al. Nasopharyngeal carcinoma: repeat treatment with conformal proton therapy–dose-volume histogram analysis. Radiology 1999; 213:489–494. 37. Noel G, Boisserie G, Dessard-Diana B, et al. Comparaison par les histogrammes dose-volume de deux techniques d’irradiation conformationnelle des cancers du nasopharynx classés T4N0M0: l’une par photons et protons et l’autre par photons seuls. Cancer Radiother 2002; 6:337–348. 38. Sugahara S, Tokuuye K, Okumura T, et al. Clinical results of proton beam therapy for cancer of the esophagus. Int J Radiat Oncol Biol Phys 2005; 61:76–84. 39. Tsunemoto H, Ishikawa T, Morita S, et al. Indications of particle radiation therapy in the treatment of carcinoma of the esophagus. Int J Radiat Oncol Biol Phys 1992; 22:321–324. 40. Sandler HM, McShan DL, Lichter AS. Potential improvement in the results of irradiation for prostate carcinoma using improved dose distribution. Int J Radiat Oncol Biol Phys 1992; 22:361– 367. 41. Schulte RW, Slater JD, Rossi CJ Jr, Slater JM. Value and perspectives of proton radiation therapy for limited stage prostate cancer. Strahlenther Onkol 2000; 176:3–8. 42. Isacsson U, Montelius A, Jung B, Glimelius B. Comparative treatment planning between proton and X-ray therapy in locally advanced rectal cancer. Radiother Oncol 1996; 41:263–272. 43. Tatsuzaki H, Urie MM, Willett CG. 3-D comparative study of proton vs. x-ray radiation therapy for rectal cancer. Int J Radiat Oncol Biol Phys 1992; 22:369–374. 44. Levin CV. Potential for gain in the use of proton beam boost to the para-aortic lymph nodes in carcinoma of the cervix. Int J Radiat Oncol Biol Phys 1992; 22:355–359. 45. Slater JD, Slater JM, Wahlen S. The potential for proton beam therapy in locally advanced carcinoma of the cervix. Int J Radiat Oncol Biol Phys 1992; 22:343–347. 46. Smit BM. Prospects for proton therapy in carcinoma of the cervix. Int J Radiat Oncol Biol Phys 1992; 22:349–353. 47. Gademann G, Wannenmacher M. Charged particle therapy to pediatric tumors of the retroperitoneal region: a possible indication. Int J Radiat Oncol Biol Phys 1992; 22:375–381. 48. Bush DA, Slater JD, Bonnet R, et al. Proton-beam radiotherapy for early-stage lung cancer. Chest 1999; 116:1313–1319. 49. Kawashima M, Furuse J, Nishio T, et al. Phase II study of radiotherapy employing proton beam for hepatocellular carcinoma. J Clin Oncol 2005; 23:1839–1846. 50. Gragoudas ES, Marie LA. Uveal melanoma: proton beam irradiation. Ophthalmol Clin North Am 2005; 18:111–118. 51. Seddon JM, Gragoudas ES, Albert DM, et al. Comparison of survival rates for patients with uveal melanoma after treatment with proton beam irradiation or enucleation. Am J Ophthalmol 1985; 99:282–290. 52. Munzenrider J, Gragoudas E, Walsh S. conservative treatment of uveal melanoma: survival, local control, eye retention, and vision in patients with intermediate and large tumor after proton beam therapy. Int J Radiat Oncol Biol Phys 1994; 30(Suppl 1):233. 53. Egger E, Schalenbourg A, Zografos L, et al. Maximizing local tumor control and survival after proton beam radiotherapy of uveal melanoma. Int J Radiat Oncol Biol Phys 2001;51:138–147.

160

g. noel et al.

54. Courdi A, Caujolle JP, Grange JD, et al. Results of proton therapy of uveal melanomas treated in Nice. Int J Radiat Oncol Biol Phys 1999; 45:5–11. 55. Lumbroso L, Levy C, Plancher C, et al. [Results of proton beam irradiation for treatment of choroidal melanoma]. J Fr Ophtalmol 2002; 25:290–297. 56. Damato B, Kacperek A, Chopra M, et al. Proton beam radiotherapy of choroidal melanoma: the Liverpool-Clatterbridge experience. Int J Radiat Oncol Biol Phys 2005; 62:1405–1411. 57. Damato B, Kacperek A, Chopra M, et al. Proton beam radiotherapy of iris melanoma. Int J Radiat Oncol Biol Phys 2005; 63: 109–115. 58. Gragoudas ES, Lane AM, Regan S, et al. A randomized controlled trial of varying radiation doses in the treatment of choroidal melanoma. Arch Ophthalmol 2000; 118:773–778. 59. Weber DC, Bogner J, Verwey J, et al. Proton beam radiotherapy versus fractionated stereotactic radiotherapy for uveal melanomas: a comparative study. Int J Radiat Oncol Biol Phys 2005; 63: 373–384. 60. Frau E, Rumen F, Noel G,. Low-dose proton beam therapy for circumscribed choroidal hemangiomas. Arch Ophthalmol 2004; 122:1471–1475. 61. Tsina EK, Lane AM, Zacks DN, et al. Treatment of metastatic tumors of the choroid with proton beam irradiation. Ophthalmology 2005; 112:337–343. 62. Desjardins L, Levy C, D’hermies F, et al. [Initial results of proton therapy in choroidal melanoma at the d’Orsey Center for Proton Therapy; the ﬁrst 464 cases]. Cancer Radiother 1997; 1:222–226. 63. Saito K, Unni KK, Wollan PC, Lund BA. Chondrosarcoma of the jaw and facial bones. Cancer 1995; 76:1550–1558. 64. Benk V, Liebsch NJ, Munzenrider JE, et al. Base of skull and cervical spine chordomas in children treated by high- dose irradiation. Int J Radiat Oncol Biol Phys 1995; 31:577–581. 65. Munzenrider JE, Liebsch NJ. Proton therapy for tumors of the skull base. Strahlenther Onkol 1999; 175(Suppl 2):57–63. 66. Noel G, Feuvret L, Calugaru V, et al. Chordomas of the base of the skull and upper cervical spine. One hundred patients irradiated by a 3D conformal technique combining photon and proton beams. Acta Oncol 2005; 44(7):700–708. 67. Igaki H, Tokuuye K, Okumura T, et al. Clinical results of proton beam therapy for skull base chordoma. Int J Radiat Oncol Biol Phys 2004; 60:1120–1126. 68. Weber DC, Rutz HP, Pedroni E, et al. Results of spot scanning proton radiation therapy for chordoma and chondrosarcoma of the skull base: the Paul Scherrer institut experience. Int J Radiat Oncol Biol Phys 2005; 63:401–409. 69. Baumert BG, Lomax AJ, Miltchev V, Davis JB. A comparison of dose distributions of proton and photon beams in stereotactic conformal radiotherapy of brain lesions. Int J Radiat Oncol Biol Phys 2001; 49:1439–1449. 70. Lomax AJ, Bortfeld T, Goitein G, et al. A treatment planning inter-comparison of proton and intensity modulated photon radiotherapy. Radiother Oncol 1999; 51:257–271. 71. Fuller DB, Bloom JG. Radiotherapy for chordoma. Int J Radiat Oncol Biol Phys 1988; 15:331–339. 72. Keisch ME, Garcia DM, Shibuya RB. Retrospective long-term follow-up analysis in 21 patients with chordomas of various sites treated at a single institution. J Neurosurg 1991; 75:374–377. 73. Romero J, Cardenes H, la Torre A, et al. Chordoma: results of radiation therapy in eighteen patients. Radiother Oncol 1993; 29:27–32. 74. Forsyth PA, Cascino TL, Shaw EG, et al. Intracranial chordomas: a clinicopathological and prognostic study of 51 cases. J Neurosurg 1993; 78:741–747. 75. Gay E, Sekhar LN, Rubinstein E, et al. Chordomas and chondrosarcomas of the cranial base: results and follow- up of 60 patients. Neurosurgery 1995; 36:887–896.

76. Mark RJ, Tran LM, Sercarz J, et al. Chondrosarcoma of the head and neck. The UCLA experience, 1955–1988. Am J Clin Oncol 1993; 16:232–237. 77. Ruark DS, Schlehaider UK, Shah JP. Chondrosarcomas of the head and neck. World J Surg 1992; 16:1010–1015. 78. Austin-Seymour M, Munzenrider J, Goitein M, et al. Fractionated proton radiation therapy of chordoma and low-grade chondrosarcoma of the base of the skull. J Neurosurg 1989; 70:13–17. 79. Hug EB, Loredo LN, Slater JD, et al. Proton radiation therapy for chordomas and chondrosarcomas of the skull base. J Neurosurg 1999; 91:432–439. 80. Hug EB, Slater JD. Proton radiation therapy for chordomas and chondrosarcomas of the skull base. Neurosurg Clin N Am 2000; 11:627–638. 81. Noel G, Habrand JL, Mammar H, et al. Combination of photon and proton radiation therapy for chordomas and chondrosarcomas of the skull base: the Centre de Protontherapie D’Orsay experience. Int J Radiat Oncol Biol Phys 2001; 51:392–398. 82. Suit HD, Goitein M, Munzenrider J, et al. Deﬁnitive radiation therapy for chordoma and chondrosarcoma of base of skull and cervical spine. J Neurosurg 1982; 56:377–385. 83. Austin JP, Urie MM, Cardenosa G, Munzenrider JE. Probable causes of recurrence in patients with chordoma and chondrosarcoma of the base of skull and cervical spine. Int J Radiat Oncol Biol Phys 1993; 25:439–444. 84. Muthukumar N, Kondziolka D, Lunsford LD, Flickinger JC. Stereotactic radiosurgery for chordoma and chondrosarcoma: further experiences. Int J Radiat Oncol Biol Phys 1998; 41:387–392. 85. Halperin EC. Why is female sex an independent predictor of shortened overall survival after proton/photon radiation therapy for skull base chordomas? Int J Radiat Oncol Biol Phys 1997; 38:225–230. 86. Terahara A, Niemierko A, Goitein M, et al. Analysis of the relationship between tumor dose inhomogeneity and local control in patients with skull base chordoma. Int J Radiat Oncol Biol Phys 1999; 45:351–358. 87. Hug EB, Fitzek MM, Liebsch NJ, Munzenrider JE. Locally challenging osteo- and chondrogenic tumors of the axial skeleton: results of combined proton and photon radiation therapy using three-dimensional treatment planning. Int J Radiat Oncol Biol Phys 1995; 31:467–476. 88. Casali PG, Messina A, Stacchiotti S, et al. Imatinib mesylate in chordoma. Cancer 2004; 101:2086–2097. 89. Gudjonsson O, Blomquist E, Nyberg G, et al. Stereotactic irradiation of skull base meningiomas with high energy protons. Acta Neurochir (Wien) 1999; 141:933–940. 90. Austin-Seymour M, Munzenrider J, Linggood R, et al. Fractionated proton radiation therapy of cranial and intracranial tumors. Am J Clin Oncol 1990; 13:327–330. 91. Miralbell R, Linggood RM, De La MS, et al. The role of radiotherapy in the treatment of subtotally resected benign meningiomas. J Neurooncol 1992; 13:157–164. 92. Noel G, Mammar H, Ferrand R, et al. Résultats de l’irradiation par association de photons et de protons de 47 chordomes et chondrosarcomes et de 17 méningiomes. Expérience du Centre de Protonthérapie d’Orsay. Rev Neurol (Paris) 2002; 158:497–509. 93. Hug EB, DeVries A, Thornton AF, et al. Management of atypical and malignant meningiomas: role of high-dose, 3D-conformal radiation therapy. J Neurooncol 2000; 48:151–160. 94. Noel G, Bollet MA, Calugaru V, et al. Functional outcome of patients with benign meningioma treated by 3D conformal irradiation with a combination of photons and protons. Int J Radiat Oncol Biol Phys 2005; 62:1412–1422. 95. Noel G, Habrand IL, Mammar H, et al. Highly conformal therapy using proton component in the management of meningiomas: preliminary experience of the centre de protonthérapie d’Orsay. Strahlenther Onkol 2002; 178:480–485.

11.

proton beam radiosurgery

96. Blomquist E, Carlsson J. Strategy for planned radiotherapy of malignant gliomas: postoperative treatment with combinations of high dose proton irradiation and tumor seeking radionuclides. Int J Radiat Oncol Biol Phys 1992; 22:259–263. 97. Seifert V, Stolke D, Mehdorn HM, Hoffmann B. Clinical and radiological evaluation of long-term results of stereotactic proton beam radiosurgery in patients with cerebral arteriovenous malformations. J Neurosurg 1994; 81:683–689. 98. Verhey LJ, Smith V, Serago CF. Comparison of radiosurgery treatment modalities based on physical dose distributions. Int J Radiat Oncol Biol Phys 1998; 40:497–505. 99. Harsh GR, Thornton AF, Chapman PH, et al. Proton beam stereotactic radiosurgery of vestibular schwannomas. Int J Radiat Oncol Biol Phys 2002; 54:35–44. 100. Weber DC, Chan AW, Bussiere MR, et al. Proton beam radiosurgery for vestibular schwannoma: tumor control and cranial nerve toxicity. Neurosurgery 2003; 53:577–586. 101. Fuss M, Hug EB, Schaefer RA, et al. Proton radiation therapy (PRT) for pediatric optic pathway gliomas: comparison with 3D planned conventional photons and a standard photon technique. Int J Radiat Oncol Biol Phys 1999; 45:1117–1126. 102. Krengli M, Hug EB, Adams JA, et al. Proton radiation therapy for retinoblastoma: Comparison of various intraocular tumor locations and beam arrangements. Int J Radiat Oncol Biol Phys 2005; 61:583–593. 103. Lee CT, Bilton SD, Famiglietti RM, et al. Treatment planning with protons for pediatric retinoblastoma, and pelvic sarcoma: how do protons compare with other conformal techniques? Int J Radiat Oncol Biol Phys 2005; 63:362–372. 104. Hug EB, Nevinny-Stickel M, Fuss M, et al. Conformal proton radiation treatment for retroperitoneal neuroblastoma: introduction of a novel technique. Med Pediatr Oncol 2001; 37:36–41. 105. McAllister B, Archambeau JO, Nguyen MC, et al. Proton therapy for pediatric cranial tumors: preliminary report on treatment and disease-related morbidities. Int J Radiat Oncol Biol Phys 1997; 39:455–460. 106. Noel G, Habrand JL, Helfre S, et al. Proton beam therapy in the management of central nervous tumors in childhood: the preliminary experience of the Centre de Protonthérapie d’Orsay. Med Pediatr Oncol 2003; 40:309–315. 107. Miralbell R, Lomax A, Bortfeld T, et al. Potential role of proton therapy in the treatment of pediatric medulloblastoma/primitive neuroectodermal tumors: reduction of the supratentorial target volume. Int J Radiat Oncol Biol Phys 1997; 38:477–484. 108. ICRP. Recommendations of the International Commission on Radiobiological Protection. Publication 60. Annals of the ICRP. Oxford: Pergamon Press 1991; 21:1–3. 109. Miralbell R, Lomax A, Cella L, Schneider U. Potential reduction of the incidence of radiation-induced second cancers by using proton beams in the treatment of pediatric tumors. Int J Radiat Oncol Biol Phys 2002; 54:824–829. 110. Schoenthaler R, Fullerton B, Maas A, et al. Relationship between dose to auditory pathways and audiological outcomes in skull-base tumor patients receiving high-dose proton-photon radiotherapy. Int J Radiat Oncol Biol Phys 1996; 36:291.

161

111. Cella L, Lomax A, Miralbell R. Potential role of intensity modulated proton beams in prostate cancer radiotherapy. Int J Radiat Oncol Biol Phys 2001; 49:217–223. 112. Slater JD, Rossi CJ Jr, Yonemoto LT, et al. Conformal proton therapy for early-stage prostate cancer. Urology 1999; 53:978–984. 113. Shipley WU, Verhey LJ, Munzenrider JE, et al. Advanced prostate cancer: the results of a randomized comparative trial of high dose irradiation boosting with conformal protons compared with conventional dose irradiation using photons alone. Int J Radiat Oncol Biol Phys 1995; 32:3–12. 114. Benk VA, Adams JA, Shipley WU, et al. Late rectal bleeding following combined X-ray and proton high dose irradiation for patients with stages T3-T4 prostate carcinoma. Int J Radiat Oncol Biol Phys 1993; 26:551–557. 115. Cozzi L, Fogliata A, Lomax A, Bolsi A. A treatment planning comparison of 3D conformal therapy, intensity modulated photon therapy and proton therapy for treatment of advanced head and neck tumours. Radiother Oncol 2001; 61:287–297. 116. Miralbell R, Cella L, Weber D, Lomax A. Optimizing radiotherapy of orbital and paraorbital tumors: intensity- modulated X-ray beams vs. intensity-modulated proton beams. Int J Radiat Oncol Biol Phys 2000; 47:1111–1119. 117. Santoni R, Liebsch N, Finkelstein DM, et al. Temporal lobe (TL) damage following surgery and high-dose photon and proton irradiation in 96 patients affected by chordomas and chondrosarcomas of the base of the skull. Int J Radiat Oncol Biol Phys 1998; 41: 59–68. 118. Glosser G, McManus P, Munzenrider J, et al. Neuropsychological function in adults after high dose fractionated radiation therapy of skull base tumors. Int J Radiat Oncol Biol Phys 1997; 38:231– 239. 119. Debus J, Hug EB, Liebsch NJ, et al. Brainstem tolerance to conformal radiotherapy of skull base tumors. Int J Radiat Oncol Biol Phys 1997; 39:967–975. 120. Marucci L, Niemierko A, Liebsch N, et al. Spinal cord tolerance to high dose fractionated 3D conformal proton-photon irradiation as evaluated by equivalent uniform dose and dose volume histogram analysis. Radiother Oncol 2002; 64(Suppl 1):S55. 121. Slater JD, Austin-Seymour M, Munzenrider J, et al. Endocrine function following high dose proton therapy for tumors of the upper clivus. Int J Radiat Oncol Biol Phys 1988; 15:607–611. 122. Habrand JL, Austin-Seymour M, Birnbaum S, et al. Neurovisual outcome following proton radiation therapy. Int J Radiat Oncol Biol Phys 1989; 16:1601–1606. 123. Kim J, Munzenrider JE, Maas A, et al. Optic neuropathy following combined proton and photon radiotherapy for base of skull tumors. Int J Radiat Oncol Biol Phys 1997; 39:272. 124. Urie MM, Fullerton B, Tatsuzaki H, et al. A dose response analysis of injury to cranial nerves and/or nuclei following proton beam radiation therapy. Int J Radiat Oncol Biol Phys 1992; 23:27–39. 125. Lundkvist J, Ekman M, Ericsson SR, et al. Cost-effectiveness of proton radiation in the treatment of childhood medulloblastoma. Cancer 2005; 103:793–801. 126. Malka V, Fritzler S, Lefebvre E, et al. Practicability of protontherapy using compact laser systems. Med Phys 2004; 31:1587–1592.

1 2

Robotics and Radiosurgery Cesare Giorgi and Antonio Cossu

Robotic Basics and History of Use in Medicine and Surgery Similar to surgery, radiosurgery is rapidly becoming a ﬁeld of application of robotics. Surgeons are confronted with the limits of their dexterity, endurance, and ability to process large amounts of data while operating. Radiation oncologists are able to plan treatments using large amounts of morphologic and functional data and start using robots to deliver the dose with higher precision. The introduction of automation in these ﬁelds of medicine has different origins and has occurred at a different pace. Surgery has entered the Information Age with the advent of modern diagnostic capabilities and computer power. The parallel development of endoscope, operative microscope, microinstruments, and navigators has downsized the operative ﬁeld and minimized surgical exposure. This has resulted in reduction of morbidity-mortality and shortening of hospital stay. The challenge now consists of the development of humaninstrument interfaces, both motor and sensory, to expand surgical possibilities beyond human capabilities. Radiosurgery faces less demanding tasks consisting of the development of devices that achieve high dose conformity and homogeneity in more complex disease geometries and in possibly moving organs.

History The term robot is less than a century old and was introduced by playwright Karel Capeck in R.U.R. (Rossum’s Universal Robots). Robota is a Czech word for “slave”; in Capeck’s play, living and intelligent working machines built to free humans from work [1]. The concept is at least as old as our culture; Aristotle, two dozen centuries ago, wrote in The Politics, “There is only one condition in which we can imagine managers not needing subordinates, and masters not needing slaves. This condition would be that each [inanimate] instrument could do its own work. . . . as if a shuttle should weave of itself, and a plectrum should do its own harp playing” [2]. In the following centuries, skill and imagination have left increasingly complex testimony of the concept of an automaton, not only in Western but also in Islamic culture, where in the 13th century Ibn Ismail Ibn al Razzaz al-Jazari published his

Al-jami bain al-lim wal-amal al-naﬁ ﬁ sinat’at al-hiyal (“Treatise on the theory and practice of the mechanical arts”) [3]. Western countries expressed mechanical “creatures” working with increasingly complicated clockwork mechanisms from the 14th century throughout the 17th century, to reach maximal expression with the mechanical wonders that ﬂourished at the end the 18th century, when the Swiss inventors Pierre and Henri-Louis Jacquet-Droz created their Automatic Scribe, which could write messages up to 40 characters long, and a robotic woman playing the piano [2]. At the turning of the following century, automation exited the role of exotic curiosity and entered the realm of useful devices. Joseph Jacquard invented a programmable loom, operated by punch cards, and went to mass production [4]. Toward the end of that century, Seward Babbitt created the ﬁrst robot, consisting of a crane with a gripper to remove ingots from a furnace, the ﬁrst machine designed to substitute for human work in a hostile environment [5]. Contemporarily, Nikola Tesla manufactured wireless controlled vehicles, and coined the term teleautomatics for his study of robotics. [6]. In the 20th century, the term robotics became popular after publication of Isaac Asimov’s “Runaround” story, which introduced his Laws of Robotics. These laws express the concepts that robots must obey human instructions and protect themselves but never cause harm to human beings directly or through inaction [1]. These fundamentals, more than six decades old, will always stand at the base of any design, particularly of machinery interacting with human life. The introduction of computers had an astounding impact on robot technology in the following decades, starting in 1948 with Norbert Wiener’s concept of cybernetics; communication and control in electronic, mechanical, and biological systems [7]. Programming of robots was the ﬁrst step, accomplished by George Davol in 1946 by means of a magnetic process recorder, and in 1954 with a computer [8]. Robots made their appearance on a production line of an automobile factory in 1963. The next leap was made by Shakey (so-called because of its jerky motion), the ﬁrst robot with vision, bump detectors at the base, a TV camera, and triangulating range ﬁnder capable of interaction with the surroundings. It was the ﬁrst mobile robot that could claim to reason about its actions [9]. Soon this achievement allowed for eye-hand coordination in assembly

163

164

c. giorgi and a. cossu

robots. In 1974, touch and pressure sensors were added to perform small parts assembly. The past three decades saw robots becoming widely represented components in the ﬁelds of industry, research, and even entertainment.

Robots in Medicine and Surgery: Neurosurgery Robots made their appearance in medicine in the mid-1980s, and the ﬁrst application was neurosurgical [10]. Stereotaxy was a ﬁeld where computational ability overcame manual skills. It was the time when neurophysiology was the dominant guideline to surgery. The advent of computers and the introduction of digital three-dimensional (3D) neuroanatomic images prompted pioneer researchers to use robots to help surgeons calculate the trajectory of stereotactic probes and hold instruments along that trajectory. Targets and trajectories were calculated from a multimodal set of data fast and error-free and transferred to the surgical stereotactic space to assist in performing functional procedures or biopsies. Neuromate, a contemporary 6 degrees of freedom (6DOF) articulated arm, evolved from original work published by Benabid et al. in Grenoble in 1987 [11]. The Neuromate performs all calculations involved and is able to direct the probe to target, by registering itself with respect to the stereotactic space. The Minerva, a Cartesian robot, was developed in Switzerland in 1994 [12] to perform biopsies under computed tomography (CT) control. Soon after, a similar device, completely nonmagnetic, was manufactured in Japan to perform biopsies of cerebral lesions within the gantry of a magnetic resonance imaging (MRI) instrument [13]. The concept of “controlling a probe position in stereotactic space” was extended to the robotic operative microscope, where the “probe” was the optical axis. The Zeiss MKM, Elekta Surgiscope, and the Möller-Wedel scope mounted on a Staubli Rx90 robot [14–16] were all devices that allowed the surgeon to “navigate” within the brain with the conﬁdence provided by the display of the position of his line of sight within the 3D reconstruction of the brain anatomy, based on CT or MR data. The movement of those robots was passive, controlled by the surgeon’s hand. Diffusion of these instruments within the neurosurgical community after the curiosity aroused at the presentation of each of them has not followed the expected trend: computer-assisted surgery (CAS), or image-guided surgery, has won the leadership. The obvious advantage of extracting accurate 3D morphologic data from neuroanatomic images and displaying the position of surgical instruments as in a virtual rendering of the surgical ﬁeld has rapidly gained broad consensus even among the most conservative neurosurgeons. With CAS, operative skills remain human, but the surgeon’s conﬁdence is enhanced by the added perception of complete control of the operative scenario. Minimally invasive procedures are facilitated, even in the absence of obvious anatomic landmarks. Distortions of preoperative imaging (swelling and brain shift) during surgery have to be taken into account. Intraoperative imaging techniques (US, CT, or MRI) are being evaluated in order to extend the beneﬁts of image-guided methods throughout the procedure. Following this road, with continuous perfection of CAS and consequent further miniaturization of

surgery, even in neurosurgery eventually the limits of human dexterity and stamina will be unveiled.

Robot Development in Other Fields of Surgery Other ﬁelds of medicine witnessed robots being developed to the clinical stage in the early 1990s. Davies described a device for soft tissue removal that developed into Prorobot, a system for transurethral removal of prostatic tissue, used in clinical practice [17]. Shortly after, Robodoc became the ﬁrst industrial production robot for orthopedics (hip replacement). It was an example of transfer of technology from industry to medicine. Bony structures were handled with a precision far exceeding that obtained with traditional surgery [18]. The most successful ﬁeld for robot development in surgery has been that of telemanipulators, used in endoscopic surgery. The Automated Robotic System for Optimal Position (AESOP; Computer Motion) was the ﬁrst robotic arm to position and hold an endoscopic camera to reach commercial diffusion and to clear the U.S. FDA as a surgical robot, in 1996 [19]. Two similar products appeared in the same years, the ZEUS (Computer Motion) and the da Vinci (Intuitive Surgical) telemanipulators, which have been used since then worldwide in a number of thoracic and abdominal endoscopic procedures with the largest bulk of literature regarding cardiac surgery for coronary artery bypass, mitral valve repair, and atrial septal surgery [20, 21]. Both systems consist of a surgical interface and a controller that relay action performed by the surgeon to the robotic arms, equipped with custom-designed endoscopic instruments. Visual perception of the operative ﬁeld is achieved through an endoscope mounted on a third arm, which is also controlled by the surgeon. In spite of all technical efforts though, natural visual 3D perception cannot be matched by camera pictures, and true force feedback equal to that obtained by natural human contact with tissue is still far from being effectively simulated. Still, the advantages of surgery, performed with these robots, have not been clearly demonstrated in experimental and clinical settings. It is true that hand tremor is ﬁltered and movements are scaled down. Compared with non–robotic laparoscopy, hand-eye coordination is maintained and hand movements are natural. Shear forces are neutralized, and operator’s postural distress is minimized. These facts lead us to say that the level of perfection reached by existing robotic devices allows them to be deﬁned as dexterity enhancers, far from being able to replace human operators. The deﬁnition of robots, “a mechatronic device interacting with its environment under remote or programmed control” [22], applies to the case of surgical robots, but the “interaction with the environment” is their true limit.

Robot Types Different robot types are currently used in the medical ﬁeld (Table 12-1). Selective compliance arm for robotic assembly (SCARA) and articulated robots, usually derived from industrial applications, are typically used in surgery. The reason for this choice is evident in the requirements of such procedures, where the robot is used to precisely handle low-weight

12.

robotics and radiosurgery

TABLE 12-1 Robot types in medical applications. Name

Type

Application

Freedom

Parallel

CyberKnife

Articulated

Robodoc Caspar da Vinci

SCARA Articulated Four articulated arms

Radiotherapy/radiosurgery: patient positioning Radiotherapy/radiosurgery: linac positioning Orthopedic surgery Orthopedic surgery General purpose: heart, thoracic, urologic

instrumentations. In terms of kinematics, they are serial robots, the end-effector is moved thanks to a series of joints, each responsible for a degree of freedom. SCARA and articulated robots characteristically handle large working volume. On the contrary, compared with other robots, they have poor performance in terms of payload. In radiotherapy, robots can be used to move the linear accelerator (linac), move the patient, or to conform the radiation beam. In the existing scenario, the robot-moving linac (CyberKnife) uses the 6 degrees of freedom both to shape the beam and to perform patient position tracking. Other devices tend to use a combination of the multileaf collimators to shape the beam and the robotic couch to position the patient and to track organ movement. Theoretically, the multileaf collimator alone could shape the beam and track the position, but the ﬁeld of view is ﬁxed and there are too few degrees of freedom. Speed of the blade is also too high for the current technology.

Moving Radiation Source This philosophy has been followed in developing the CyberKnife device, where a 6DOF robotic arm allows total freedom in positioning the radiation source. The choice of an articulated robot, combined with the necessary restrictions in terms of robot weight and size, results in a limitation of the source energy.

Robotic Couches with Existing Linacs The chosen technology in terms of patient positioning is the parallel robot, where the motion results from the simultaneous

165

movement of several actuators (typically six, one for each DOF). Compared with the articulated robot, this kind of kinematics allows high payload and high degree of accuracy at the cost of a reduced working volume.

Conformal Micro-Multileaf This kind of robotic device is mainly made by several actuators moving each leaf during the treatment in order to precisely conform the radiation to the shape of the lesion. Advantages and disadvantages of the different solutions will be examined in the next chapters. Table 12-2 contains a brief classiﬁcation of different robot kinematics.

Robotics and Image Guidance Radiation therapy, similar to surgery, has evolved with the progress of imaging technologies and computers. Conformal planning and delivery of maximal dose to the tumor, while sparing surrounding structures, has become more and more efﬁcient. The introduction of intensity-modulated radiation therapy (IMRT) in 1992 has made it possible to tailor the distribution of the dose according to the often non-homogeneous lesion grading, complex geometry, and the presence of surrounding critical structures. MRI, including perfusion-diffusion, spectroscopy and diffusion tensor imaging, SPECT and PET all contribute to the deﬁnition of planning target volume (PTV). Patient positioning has become extremely critical; methods and devices currently in use may soon be considered inadequate to deliver such sophisticated treatment plans with acceptable precision. Body respiratory movements and natural internal organ motion, particularly for thoracic and upper abdominal lesions, represent an additional challenge. They need to be accurately described and tracked in order to extend to moving targets the advantages of contemporary treatment planning techniques. Image-guided radiation therapy (IGRT) captures images of the body immediately prior to treatment delivery. Images can be produced by ultrasound, ﬂuoroscopy, X-rays, or cone beam tomography; they can be matched with treatment plan images, in order to detect and correct movements of the organs, or compensate for displacement of the target due to respiratory

TABLE 12-2. Kinematics of most common robot types. Type of robot

Kinematics

Cartesian

Positioning is done in the workspace with prismatic joints. This conﬁguration is useful when a large workspace must be covered or when consistent accuracy is expected from the robot. It has a revolute motion about a base, a prismatic joint for height, and a prismatic joint for radius. This robot is well suited to round workspaces. A robot whose axes form a polar coordinate system. A robot that has two parallel rotary joints to provide compliance in a plane. This robot conforms to cylindrical coordinates, but the radius and rotation is obtained by a two planar links with revolute joints. A robot whose arm has at least three rotary joints. The robot uses three revolute joints for positioning. Generally, the work volume is spherical. This robot most resembles the human arm, with a waist, shoulder, elbow, and wrist. A parallel robot is one whose arms (primary axes) have at least three concurrent prismatic joints or both prismatic and rotary joints.

Cylindrical Polar SCARA Articulated Parallel

166

c. giorgi and a. cossu

movements or other physiologic conditions. This increases the conﬁdence of radiotherapists to give a higher dose to the tumor and reduce the side effects due to irradiation of surrounding critical structures.

Ultrasound Imaging Devices like the MOMOS Radiation Oncology Batcam, an echographic probe used to locate precisely the prostate, use ultrasound to image the target precisely to improve treatment accuracy. Prior to daily treatment section, the physician obtains ultrasound images that are registered in the treatment space, and corrections in the patient positioning are obtained by means of optical tracking. Variations in daily patient setup are promptly detected; bladder and rectum are visualized and target is shown in actual position. Comparison with CT and MRI data set obtained at the time of treatment planning provides x, y, and z alignment offsets, corresponding with a new couch position. Sensitive organs can be spared, resulting in reduced discomfort and complications, and radiation treatment increases efﬁcacy through dose escalation to the target with an improved local control. Drawbacks of this system are poor imaging quality of ultrasound in presence of organs containing air or covered by thick bony structures. Also, the procedure for image acquisition involves instruments that are not integrated in the normal treatment scenario, with consequent signiﬁcant lengthening of the treatment time.

X-ray Imaging On-Board Imager, a feature of Varian linacs, consists of an Xray 150-kV tube and an amorphous silicon ﬂat-panel X-ray detector facing it on the other side of the patient, both mounted on articulated motorized arms. It generates diagnostic quality projection images of the patient at low energy, as opposed to the megavolt images generated by the former technology of electronic portal imaging (EPI). Advantages of the new device consist of the possibility of high-quality image acquisition, optimal acquisition geometry with respect to the target, speed, and lower dose to the patient. Images are compared with the treatment planning images, and appropriate patient positioning corrections are made. Two-dimensional (2D) projection imaging is adequate when the target lies close to bony structures and its position is unaffected by respiratory or other physiologic movements or by gravity. This condition typically applies to the brain, where the skull can serve as a stereotactic frame, and, to a lesser extent, to the contents of the spinal canal and the paravertebral tissues. To image soft tissue, metal markers can be implanted and then imaged prior to treatment with biplanar X-rays. For this solution, kV imaging provides better resolution than MV and allows for the use of markers as small as 1.2 mm diameter. Still, 2D imaging does not provide sufﬁcient information regarding movement of the target. The solution has been offered by Varian and Elekta, who have equipped their latest accelerators with CT quality imaging, obtained with the “cone beam” technology. The conical beam of the accelerator is used to acquire several hundreds of images of the patient in a single 360° rotation. The result is a 3D image of soft tissue, but when the tumor to be treated lies in the lungs or in organs adjacent

to the diaphragm, the fourth dimension (time) has to be visualized. Asking the patient to hold the breath during treatment at a speciﬁc cycle (deep inspiration), teaching him to breathe only when treatment is stopped, or using physical restraints for the chest can be very demanding, especially for patients that have limited respiratory capacity. A different solution has been formulated, based on the description of the movement of the chest or abdomen during breathing, obtained with optical tracking of markers applied to the skin and observed by infrared cameras. The waveform of the patient’s breathing pattern is synchronized with the CT image acquisition, so that gating of the radiation beam can be automatically performed by the cameras during treatment when they detect the position of the markers that corresponds with the position of the tumor in the ﬁeld chosen for treatment.

Other Imaging Techniques Yet another approach to image-guided radiotherapy is that of helical tomotherapy (HT) combining a rotating intensity-modulated fan beam with integrated CT imaging. The elegant concept of imaging the body during treatment was ﬁrst described by Carol [23] but successfully exploited only later by Mackie et al. [24] after the introduction of spiral CT. The result is deﬁned as adaptive radiotherapy highly integrated adaptive radiotherapy (Hi Art). Signiﬁcant clinical follow-up documenting beneﬁts related to the introduction of these novel technologies is not yet available, but the combination of IMRT and IGRT promises improved tumor control and superior normal tissue sparing. All this is achievable at the expense of instrument cost, complication of the procedure, and treatment time. A favorable element of this equation is the enhancement of automation in patient positioning, which is the basis for the introduction of robotics in radiotherapy.

Applications of Robotics to Radiotherapy The evolution of morphologic and functional diagnostic imaging and the availability of computer power have made possible the design of complex treatment plans, highly conformal to the target and capable of sparing adjacent organs, even in cases of complex geometry like concave targets surrounding and overlapping organs at risk. Translating the treatment plan into reality with continuously improving accuracy is where a large portion of research in radiotherapy is aimed.

Radiosurgery: The Gamma Knife The principles of radiosurgery—single session, highly conformal dose delivery—are the most demanding in terms of accurate targeting. Radiosurgery was introduced in the 1950s for treatment of central nervous system lesions. It has been very efﬁciently accomplished with the Gamma Knife, a hemispheric array of cobalt-60 sources collimated to the center. Imaging of the target and positioning of the patient is obtained with reference to a frame, which is ﬁrmly applied to the patient’s skull. The ﬁxed spatial relationship between the frame and the target structure, identiﬁed by means of “localizers” applied to the frame during

12.

robotics and radiosurgery

image acquisition, translates into precise targeting of the lesion once the frame is positioned for treatment, and the target is placed at the center of the collimated sources. When the shape of the target is not spherical, the treatment is obtained with a cluster of spherical lesions that “cover” the irregularly shaped volume to be treated. The Gamma Knife is a concept that has limitations in its use: only cranial-cervical lesions with a single dose can be treated, the dose distribution to the target is less homogeneous than that obtained with other methods, but still it represents the gold standard, to which all other technologies are confronted. In recent years Cartesian robot has been incorporated in the latest version of this equipment, with the aim of positioning the frame during treatment so that each “shot” or spherical radiation dose can be automatically and precisely directed in sequence to the target. Lately, the Gamma Knife “Perfexion” has been introduced. It represents the ultimate evolution of this elegant technique. Beside signiﬁcant room added for cranial and cervical targets, a clever robotic delivery of the does by means of mechanical alignment of Co60 sources with different size collimators, combined with robotic movements of the couch head holder system, allows for improvement of does conformity with signiﬁcant shortening of treatment time. Meanwhile, other techniques have been introduced that take advantage of the enormous development of computers and imaging technology that has occurred in the past couple of decades. Physicists, radiation oncologists, and neurosurgeons having witnessed the tremendous achievements of radiosurgery, obtained with the original technique, have expanded the neurosurgical concept of stereotaxy to ﬁnd new applications for treatment of other districts of the body. This concept has also deeply modiﬁed the way fractionated radiotherapy is delivered.

Linac and CyberKnife, Single and Multiple Stereotactic Fractions Linac radiosurgery was introduced in the early 1980s. The elegant concept of multiple non-coplanar arc therapy has been complemented in the past decade with the introduction of micromultileaf collimators. Whether they are built into the linac head or added to it as an accessory, these multiaxial robots allow for highly conformed and uniform dose to target and can modulate the intensity of the dose according to treatment requirements. Dosimetric performance of some of these devices, particularly of the ones that claim better design characteristics (double focusing and interdigitation of blade travel, thinner leaves, dynamic movements, minimal transmission and leakage between the leaves), is very high, inferior only to protons [25]. Charged particle beams dosimetry is in fact superior to any other photon therapy method and can also be very effectively shaped by electromagnetic steering. Access to this treatment modality is out of reach for the large majority of radiation oncologists, being only available at present at very few centers worldwide. About a decade ago, CyberKnife was introduced in clinical practice: it was the ﬁrst device for precision radiotherapy to use an industrial robot to direct a photon beam generated by a small on-board linac to the target. Detailed description of this innovative device is dealt with in another chapter of this book; here we will only describe and comment on some of its features, which include unobstructed access to the entire body; high

167

mechanical precision; and innovative image-guided control loop, with target tracking capability. This is a device that truly represents the evolution of the Gamma Knife concept, and similarly it has been developed by a neurosurgeon [26]. It resembles the method: covering the target with high conformity, using a non-isocentric array of beams (rather then multiple treatment “shots”). The similarity ends here: the system is open and allows for the unobstructed treatment of the whole body. The source is a lightweight 6-MeV linac, with circular secondary collimators of various diameters, mounted on an industrial robot (KUKA Roboter GmbH). An X-ray imaging system is part of the system control. The treatment planning is similar to other inverse planning systems: identiﬁcation of the target and critical structures surrounding it from pretreatment morphologic and functional data set, prescription of the dose to the target, and description of constraints. A CT acquisition, among other appropriate studies, is mandatory, to generate pre- and intratreatment digitally reconstructed radiographs (DRRs), necessary for patient positioning and subsequent tracking. Planning of dose delivery is performed in an unique fashion: the process chooses a number of “nodes” laying on a sphere some 80 cm around the target volume. Guided by the prescription dose and the constraints for critical structures, the system chooses beam directions and weights for each node, to reach optimal conformity to the prescribed dose distribution. This “non-isocentric” technique is very effective in designing highly conformal plans. “Wraparound” doses for critical structures, like the spinal cord in the treatment of spinal metastases, can be obtained. A unique feature of the system is represented by the image guide loop during treatment. Two ﬂat-panel, amorphous silicon X-ray cameras are used for patient positioning and treatment tracking. The systems generates a sequence of DRRs from the pretreatment CT study that are matched with the couple of orthogonal X ray images acquired during treatment. Changes in the position of the target during treatment are disclosed with the acquisition of orthogonal projection images. The new target position is compared with the position at the planning phase, and the beam directions are corrected, accordingly. This step is repeated at each treatment node.

Frameless Head and Spine Radiosurgery The introduction of image-guided radiotherapy is certainly a major improvement compared with the simple immobilization techniques that have been used so far in the treatment of the body. The frameless solution introduced by the CyberKnife for the treatment of cerebral lesions appeals to patients and eliminates the need for a surgical act. This solution is not applicable to the Gamma Knife, an exquisitely frame-based device, but can be promptly adopted in linac sites, equipped with ﬂat screen panel and optical tracking. A similar evolution has been observed in neurosurgery. The need for stereotactic assistance in neurosurgical procedures, where location and dimensions of the lesion required it, prompted the introduction of the stereotactic frame guidance. Soon it became evident that obstruction and access limitation was a major limit of the method.

168

c. giorgi and a. cossu

Frameless solutions for surgery were anticipated in the early 1990s [27]. Large diffusion of “neuronavigation” assisted neurosurgical procedures has followed: optical tracking or magnetic ﬁeld patient’s registration allow for unobstructed access to the operative ﬁeld. This is evidently a legitimate choice, justiﬁed by incompatibility between frame-based localization and open neurosurgery and represents an evolution compared with the previously adopted frame-based technique. It needs to be pointed out, though, that when a procedure requires high precision in target localization (like deep brain stimulating electrodes in the subthalamus), no surgeon risks the potential error introduced by frameless methods. One could argue that ﬁducials glued to the skin can shift slightly and that biplanar X-ray registration is more accurate, but even in oldfashioned “functional” operating room sites, where biplanar teleradiography hardware is in use, no stereotactic procedure would take place without the reassuring presence of a frame. Naturally, if a patient is asked whether he prefers to be treated with the inconvenience of a frame or by means of a “soft” ﬁxation, he will prefer the latter, but he would certainly have doubts, if he would be plainly informed that frameless radiosurgery involves computational power, extra labor and irradiation, to match what is readily obtained with a frame. Frame positioning, in experienced hands, has never been an issue for the patient and will not take more than minutes to be done. Frameless radiosurgery of the spine, on the contrary, is without doubt perfectly adequate, and is the only sound method available, and in the past years an increasing number of body localizers devoted to body radiosurgery positioning have appeared on the market.

reconﬁguration of the beam can be performed. An organ position is best described by imaging of the organ itself. Determining the position of adequate bony landmarks or of metal ﬁducials implanted in the organ is not always feasible in the ﬁrst case, or is invasive with all related consequences in the second. Fiducial markers, if implanted in sufﬁcient number, can be automatically detected and the rototranslation of the target can be solved so rapidly that real-time ﬂuoroscopy can be used. Fluoroscopy, even performed with the best possible efﬁciency, delivers a skin dose of 2 cGy per minute, which for prolonged treatment as in hypofractionation or radiosurgery is unacceptable. A proposed solution is correlation between different data: intermittent imaging of the organ and continuous measures of movement of chest or abdomen or spirometry that is used to interpolate target position between radiographic images. This works satisfactorily for some tumor locations but not for the majority of lung tumors, where motion is too complex. Proposed solutions that develop mathematical model to predict breathing motion or empirical approaches with adaptive ﬁlters are being pursued, but still the timing of intermittent respiratory image acquisition should be well below one-half second. The issue of the dose necessary for imaging of the target remains a critical one: the accumulation of radiographic dose during treatment should not be greater than the therapy beam leakage and scatter dose. For CyberKnife, it has been calculated that the radiographic dose is smaller than the leakage dose only for treatment of the head and spine [28]. Imaging of target affected by respiratory movements is still an unsolved issue for any technology.

Other Methods for Directing the Beam Tracking Targets Being able to correctly position a patient for treatment and to compensate for movements during treatment has a number of advantages: it becomes possible to explore different treatment schemes that are difﬁcult to perform with traditional loose ﬁxation. Fractionated treatment can reduce or eliminate the need for tumor motion margin. This is particularly evident for the treatment of moving organs but can reduce the difference between target deﬁnition in radiosurgery and in fractionated radiotherapy, thanks to the ability to determine the target position, with accuracy very close to that of a stereotactic frame. Once a patient is correctly positioned, random and cyclic movements have to be taken into account. Treatment of the CNS and spine only deals with random motion both in time and direction of the target, but these can become signiﬁcant as the time for fraction increases, as in radiosurgery. The solution adopted by CyberKnife, assuming random movements of the patient occurring once every 2 minutes, sets the time for imaging interval to 1 minute. Researches state that this solution will yield misdirection of less than 1% of the dose to more than 1 mm off-target, and conclude that this ﬁgure is sufﬁcient to maintain radiosurgical standard [28]. Unfortunately, quite often patients that undergo radiosurgery are not very cooperative: fast random motion can occur, and this tracking interval could be insufﬁcient to correct the beam direction. Tracking an organ with cyclic movements (respiration and heartbeat) involves determining the actual position of the organ and predicting its future position, so that new alignment and

Determining variations in target position requires the possibility to redirect the beam. Apart from the CyberKnife, other methods have been described, namely moving the patient with a robotic couch or shifting the aperture of a multileaf collimator. Patient positioning, motion control, and tracking of the target are required to deliver the prescribed dose to the lesion with minimal irradiation of the surrounding tissue. Generation of images during treatment and robotic beam or patient positioning during single or multiple treatment fractions are the instruments currently under development. As a result, single-session treatments of head and spine lesions without the constraint of a rigid ﬁxation are already feasible, and similar stereotactic delivery methods can be applied to small lesions throughout the body, in single or multiple sessions. Robotization of the delivery of the dose by means of a robot-mounted linac is one possible solution, another one considers the automatic positioning of the couch (Table 12-3). Again, compensating for changes in patient position or random intrafraction movements is a task that can be effectively solved by different methods, but tracking moving targets increases complexity for the reasons illustrated above. Studies are being conducted regarding shifting the aperture of a multileaf collimator [29]. In the case of a conventional linac mounted unit, the alignment of the beam can only be maintained in the plane of the treatment ﬁeld. This satisﬁes the stochastic category of motion, but in case of targets rotating out of plane, it could be very difﬁcult to maintain treatment conformality, mainly due to limited leaf speed.

12.

robotics and radiosurgery

TABLE 12-3. Feature comparison between whole body radiotherapy robotic devices. Feature

CyberKnife

Robotic couch

Patient’s motion discomfort Mechanical precision Dose conformity Dose homogeneity Body access IGRT

None

Possible

High

High

Very high Poor Easy 2D low detail, X-ray quality Labor intensive Yes

Very high Very high (Tomotherapy) Easy 3D high detail, CT quality

Custom

Regular

Ready

Feasible

Limited by design

Unlimited

Safety Linac energy and weight limitations Site dimension and shielding Frameless neuro-SRS Future system development

Standard No

Robotic Couch Target positioning by means of a robotic couch presents several advantages, mainly because a robotic couch can be ﬁtted to any linac. Combined with optical tracking and IGRT, superior image acquisition solutions are complemented by a very effective tool for positioning the patient and compensating for target movement. Fast-moving and rotating targets, limited patient compliance to continuous motion, and the presence of inertia could pose serious limits to this solution alone and require the combination of the multi-leaf collimator (MLC) beam shifting. Out of the several solutions available on the market, it is noteworthy to mention the latest evolution: it has been designed by 3Dline Medical Systems, taking into account previous experience with similar parallel architectures. It can substitute existing couches, but it can also be mounted on top of one, without limiting the use of the linac for more “traditional” procedures. It has 6 degrees of freedom; linear excursions cover 80 mm; angular orientation span 4° on the three axes, in any position of the working volume. It is a “lineapode” architecture that uses six rods to obtain all translational and angular movements. All actuators are mounted on the base of the structure, thus reducing to 200 mm the height of the device at rest and reducing the inertia to a minimum.

Glossary CAS Computer-aided surgery. Initially, CAS meant a technology of surgical simulation using three-dimensional organ models and reconstructed medical imaging by computer graphics technique. In Japan, Prof. Takeyoshi Dohi and Prof. Masakazu Tsuzuki at the University of Tokyo were the ﬁrst who used this word for their research. In other countries, the term “computer-assisted surgery” is commonly used.

169

CT Computed tomography. Originally known as computed axial tomography (CAT or CAT scan) and body section roentgenography, it is a medical imaging method employing tomography where digital geometry processing is used to generate a three-dimensional image of the internals of an object from a large series of two-dimensional X-ray images taken around a single axis of rotation. DOF Degrees of freedom. The set of independent displacements that specify completely the displaced or deformed position of the body or system. In robotics, DOF is often used to describe the number of directions in which a robot can pivot or move a joint. EPR Electronic portal image IGRT Image-guided radiation therapy IMRT Intensity-modulated radiation therapy Laparoscopy Laparoscopic surgery, also called keyhole surgery (when natural body openings are not used), Band-Aid surgery, or minimally invasive surgery. MIS Minimally invasive surgery MRI Magnetic resonance imaging, formerly referred to as magnetic resonance tomography (MRT) or nuclear magnetic resonance (NMR), is a method used to visualize the inside of living organisms. It is primarily used to demonstrate pathologic or other physiologic alterations of living tissues and is a commonly used form of medical imaging. PET Positron emission tomography is a nuclear medicine medical imaging technique that produces a three-dimensional image or map of functional processes in the body. PTV Planning target volume Radiosurgery A medical procedure that allows noninvasive brain surgery (i.e., without actually opening the skull) by means of directed beams of ionizing radiation. It is a relatively recent technique (1951), which is used to destroy, by means of a precise dosage of radiation, intracranial tumors and other lesions that could be otherwise inaccessible or inadequate for open surgery. Radiotherapy The medical use of ionizing radiation as part of cancer treatment to control malignant cells (also called radiation therapy). US Ultrasonography (medical sonography) is a useful ultrasound-based diagnostic medical imaging technique used to visualize the fetus, muscles, tendons, and many internal organs; their size, structure, and any pathologic lesions.

References 1. Robotics Research Group, University of Texas at Austin. Learn More History. Available at http://www.robotics.utexas.edu/rrg/ learn_more/history. 2. Malone R. The Robot Book. New York: Push Pin Press, 1978. 3. Al-Hassani STS. Foundation for Science Technology and Civilisation. Al-Jazari—The Mechanical Genius. Available at http://www. muslimheritage.com/topics/default.cfm?ArticleID=188. 4. Spartacus Educational. Joseph Jacquard. Available at http:/www. spartacus.schoolnet.co.uk/SCjacquard.htm. 5. II Robotics. Timeline—Back to the Beginnings. Available at http:/ www.iirobotics.com/webpages/robothistory.php. 6. Tesla Society. Available at ieee.org/web/aboutus/history_center/ biography/tesla.html. 7. Vallée R. Norbert Weiner. International Society for Systems Sciences. Available at http://www.isss.org/lumwiener.htm.

170

c. giorgi and a. cossu

8. Robot Hall of Fame. The 2003 Inductees: Unimate. Carnegie Mellon University. Available at http://www.robothalloffame.org/ unimate.html. 9. SRI International. Shakey the Robot. SRI International. Available at http://www.sri.com/about/timeline/shakey.html. 10. Kwoh YS, Hou J, Jonckheere EA, et al. A robot with improved absolute positioning accuracy for CT guided stereotactic brain surgery. IEEE Trans Biomed Eng 1988; 35:153–160. 11. Benabid AL, Cinquin P, Lavallée S, et al. Computer driven robot for stereotactic surgery connected to CT scan and magnetic resonance imaging: technological design and preliminary results. Appl Neurophysiol 1987; 50:153–154. 12. Fankhauser H, Glauser D, Flury P, et al. Robot for CT-guided stereotactic neurosurgery. Stereotact Funct Neurosurg 1994; 63:93–98. 13. Masamune K, Kobayashi E, Masutani Y, et al. Development of an MRI-compatible needle insertion manipulator for stereotactic neurosurgery. J Image Guid Surg 1995; 1:242–248. 14. Steinmeier R, Rachinger J, Kaus M. Factors inﬂuencing the application accuracy of neuronavigation systems. Stereotact Funct Neurosurg. 2000; 75:4. 15. Bonev I. Delta Parallel Robot—The Story of Success. ParalleMIC—The Parallel Mechanisms Information Center. Available at http://www.parallemic.org/Reviews/Review002.html. 16. Giorgi C, Sala R, Riva D, et al. Robotics in child neurosurgery. Child’s Nerv Syst 2000; 16:832–834. 17. Ng WS, Davies BL, Hibberd RD, et al. A ﬁrst hand experience in transurethral resection of the prostate. IEEE Med Biol Soc Mag 1993; 120–125. 18. Bargar WL, Bauer A, Borner M. Primary and revision total hip replacement using the Robodoc system. Clin Orthop 1998; 354: 82–91.

19. AESOP, Computer Motion Robot. Available at http://www. timeforce.com/Medical_Robotics/Medical_Robotics_Companies/ computermotionproﬁle.html. 20. Damiano RJ Jr, Reichenspurner H, Ducko CT. Robotically assisted endoscopic coronary artery bypass grafting: current state of the art. Adv Card Surg 2000; 12:37–57. 21. Mohr FW, Falk V, Diegeler A, et al. Computer-enhanced coronary artery bypass surgery. J Thorac Cardiovasc Surg 1999; 117: 1212–1214. 22. Hootman R. Homebrewed Robots. Available at www.virtuar. com/click/2005/robonexus/. 23. Carol MP. Peacock: a system for planning and rotational delivery of intensity modulated ﬁelds. Int J Imag Sys Technol 1995; 6:56–61. 24. Mackie TR, Balog J, Ruchala K, et al. Tomotherapy. Semin Radiat Oncol 1999; 9:108–117. 25. Clivio A, Bolsi, Cozzi L, et al. Advanced radiotherapy techniques applied to brain tumours. A comparative study. Presented at 8th Biennial ESTRO Meeting on Physics and Radiation Technology for Clinical Radiotherapy, Lisbon, Spain, September, 2005. 26. Adler JR, Chang SD, Murphy MJ. The CyberKnife: A frameless robotic system for radiosurgery. Stereotact Funct Neurosurg 1997; 69:124–128. 27. Giorgi C. Imaging techniques and computers. Stereotact Funct Neurosurg 1994; 63:8–15. 28. Murphy MJ. Tracking moving organs in real time. Sem Radiat Oncol 2004; 14:91–100. 29. Jiang S, Zygmansky P, Kung J. Gated motion adaptive therapy (GMAT): modiﬁcation of IMRT MLC leaf sequence to compensate for tumor motion. Med Phys 2002; 29:1347–1348.

1 3

CyberKnife Radiosurgery John R. Adler Jr., Alexander Muacevic, and Pantaleo Romanelli

Image-Guided Robotic Radiosurgery

CyberKnife Technology

Conceptually, the lineage of CyberKnife (Accuray Inc., Sunnyvale, CA) technology derives from the clinical principles that underlie stereotactic radiosurgery. This minimally invasive procedure involves the precise delivery of large doses of ionizing radiation to destroy well-deﬁned targets without injuring the surrounding and intervening healthy tissue. This objective is achieved using large numbers of narrow beams that emanate from a wide array of directions and intersect (and therefore accumulate) within the volume selected for ablation. The cumulative dose that can be administered this way overwhelms any capacity for cellular repair, thereby typically ensuring tissue destruction. Until recently, radiosurgery was conﬁned to the brain and skull base, and for almost 30 years now, the Gamma Knife (Elekta AB, Stockholm, Sweden) system has been the standard instrument for the neurosurgical application of radiosurgery. However, modiﬁed linear accelerators (linacs) also emerged over the years to become well-accepted radiosurgical technologies. Gamma Knife and conventional linacs both require the application of an invasive frame on the patient’s head to achieve the desired stereotactic accuracy of ±1 mm. The frameless targeting system of the CyberKnife represents a radical departure from this approach. Furthermore, the combination of imagedguided targeting with robotic technology is enabling the original scope of radiosurgery to be dramatically expanded. During CyberKnife radiosurgery, real-time intraoperative imaging is used instead of a stereotactic frame to establish the tumor position with reference to skeletal anatomy. In several clinical circumstances, the combination of image guidance and robotics offers a material advantage over more conventional approaches to radiosurgery. Primary among these beneﬁts is the fact that the beam can track lesion motion in any direction throughout the body. Recently, a new tracking method based on the correlation between external (chest) and internal (lesion) motion has made it possible to also follow targets that move with respiration while the treatment beam is on. These developments have extended what was once a solely intracranial application to targets throughout the chest and abdomen. In this chapter, we describe the CyberKnife and some of its more unique clinical indications.

The CyberKnife is composed of a lightweight and compact high-energy X-ray source (6-MeV linac, dose rate 6 Gy/min) coupled to a robotic arm capable of moving with 6 degrees (6D) of freedom (Kuka GmbH, Augsburg, Germany) (Fig. 13-1). During treatment, the manipulator is capable of aiming 1200 beam directions (1600 with the newest G4 model) toward the lesion being treated. The robot is spatially calibrated to a computerized localization system consisting of two X-ray generators that are ﬁxed on the ceiling to enable orthogonal images of the target region. Images are recorded on ﬂoor-mounted silicon detectors that generate high-resolution digital images. X-rays are registered to digitally reconstructed radiographs (DRRs) derived from the planning computed tomography (CT) scan (Fig. 13-2), and deviations of the target region are corrected automatically during initial patient setup on the ﬁve-axis patient couch. During treatment, X-rays are frequently acquired and used to automatically compensate for patient movements within a range of 10 mm by adjusting the direction of the treatment beam. The targeting accuracy of this system design has been repeatedly demonstrated to be submillimetric [1, 2]. At each position of the robot, beams can be directed toward different areas of the target region. This design enables the treating surgeon to select from a large array of non-isocentric, non-coplanar beams during the process of constructing a treatment plan and thereby create dose distributions that conform to even irregularly shaped lesion volumes (Fig. 13-3). In contrast, more conventional radiosurgical devices construct spherical dose distributions around a discrete isocenter.

CyberKnife Spinal Radiosurgery Image-guided robotic radiosurgery can also be a useful tool for ablating a broad spectrum of spinal lesions. It is worth emphasizing that in terms of dose conformality and targeting accuracy, CyberKnife spinal radiosurgery compares favorably with standard frame-based intracranial radiosurgery [2]. The initial method for targeting the spine during CyberKnife radiosurgery required the percutaneous implantation of ﬁducials (typically, three to four 4 × 2 mm stainless steel screws placed in vertebrae adjacent to the target lesion). By comparing the positions of

171

172

j.r. adler jr. et al.

FIGURE 13-1. The main components of the CyberKnife system. (Image used with permission from Accuray Incorporated.)

FIGURE 13-2. A library of digitally reconstructed radiographs (DRRs) is generated from the treatment planning CT. Each of these images, which approximate an oblique projection, emulate a unique pose of the patient’s anatomy.

FIGURE 13-3. (a) With standard radiosurgery dispersed isocentrical beams, all intersect a common region (light gray). Because multiple spherical volumes are needed to cover nonround lesions, the resulting

dose distribution tends to be inhomogeneous. (b) Non-isocentric beams from various directions. Beams do not all cross in a single point and dose need not be inhomogeneous.

13.

cyberknife radiosurgery

ﬁducials in pretreatment DRRs with their location in intratreatment radiographs, it is possible to calculate a rigid-body 6D (3 × translation, 3 × rotation) offset and adjust the treatment beam accordingly. To compensate for patient motion during treatment, the imaging-and-alignment process is repeated between beams. The clinically relevant accuracy of the above approach was investigated by Yu and co-workers [2]. In this study, measurements were performed at three different CyberKnife facilities using head and torso phantoms loaded with packs of radiochromic ﬁlm. The absolute displacement of the dose contours from that deﬁned during treatment planning was reported to be submillimetric; the average offset was 0.3 ± 0.1 mm. Fiducial tracking error was below 0.3 mm for radial translations smaller than 14 mm and less than 0.7 mm for rotations up to 4.5 degrees.

Fiducial-Free Tracking More recently, ﬁducial-free tracking has been developed for CyberKnife spinal radiosurgery (Xsight; Accuray, Sunnyvale, CA). This technology enables the tracking of spinal lesions based on anatomic landmarks instead of surgically implanted ﬁducials. Similar in many ways to the head-tracking algorithm, Xsight automatically references radiographically visible skeletal structures. The fact that much of the spine is a nonrigid body introduces an added dimension of complexity to the underlying algorithm. Nevertheless, the approach has proved remarkably robust and accurate. Recent investigation demonstrates that the targeting accuracy of Xsight compares favorably with the published precision of ﬁducial-based localization [3]. The main beneﬁts of a ﬁducial-free system are (1) nonrigid deformation is accounted for, potentially improving treatment accuracy in the many cases where patient pose changes have occurred, and (2) marker insertion is not required. The latter precludes any associated risks and dramatically increases convenience for both patient and clinician. A further, yet more minor, advantage of skeletal-based localization is that the spatial ﬁdelity of targeting is maintained in the event of ﬁducial migration, the risk of which is admittedly small with spinal screws.

173

CyberKnife Body Radiosurgery The CyberKnife can also be used to ablate soft tissue targets within the major body cavities. As a prelude to such radiosurgery, a minimally invasive procedure is required to percutaneously implant several (at least 3) small (0.8 × 4 mm) gold seeds in close proximity to the treated lesion. These markers provide a radiopaque frame of reference for radiosurgical targeting. However, intrathoracic and intraabdominal organs move throughout treatment, especially because of diaphragm motion [4–9]. In an attempt to compensate, conventional radiotherapy systems may utilize respiratory gating (turning on the treatment beam when the moving tumor is in range) or breathing restriction (using frames that apply abdominal pressure) to limit tumor movement. The accuracy that can be achieved with either approach is still less than ideal and a substantial margin (1 cm or more) is commonly included around the lesion being irradiated. In contrast with standard gantry-based systems, the robotic arm of the CyberKnife can move the linac in any direction. This capability makes it feasible to compensate for complex breathing motion in real time. The Synchrony motion-tracking system was developed to accomplish this objective. This technology senses in real time three-dimensional respiratory motion by means of infrared (IR) LEDs attached to the patient’s chest or abdomen and a pair of IR cameras. The absolute location of the tumor at a given point in time is determined from the position of percutaneously implanted gold ﬁducials within the CyberKnife’s orthogonal X-ray imaging system (Fig. 13-4). The Synchrony system develops a continuously updated model that relates the position of the implanted gold markers to the moving skin surface of the patient. The correlation model built by Synchrony begins with a series of radiographs taken after patient setup and prior to the start of treatment. The positions of the ﬁducials in the digital radiographs are correlated with the time-stamped positions of the IR emitters at the moment of X-ray acquisition (Fig. 13-5). As the patient breathes, software computes a correlation model that is used to dynamically adjust the aim of the linac. In the process, the position of the treatment beam is continuously adapted to the location of the tumor. Periodically during

FIGURE 13-4. A CT image is displayed that was obtained during the process of implanting a ﬁducial into a right-sided lung metastasis from a tonsillar carcinoma. The needle with an ejected 5-mm gold seed at the outer tumor margin is shown.

174

j.r. adler jr. et al.

such inverse planning proves to be of greatest advantage for the most irregular target volumes. After the radiation oncologist, often working with a surgeon, speciﬁes the dose to a tumor and surrounding critical structures, the treatment planning system determines directions and activation durations for each non-isocentric, non-coplanar beam (selected from a universe of approximately 1200 beam directions) such that the primary speciﬁcations are met. Next, an optimization method is used to compute beam weights such that initial dose constraints are fulﬁlled. Typically, the ﬁnal plan utilizes more than 100 beam directions in a single treatment. Recently, planning has been extended to the case of moving targets [12, 13]. However, while the target may move during treatment, critical structures in the vicinity may not move with the same speed or direction. The relative motion of organs with respect to one another requires highly accurate planning. The temporal changes over the entire respiration cycle are represented in a motion model (obtained from either fourdimensional CT or by deforming an inhalation image stack into an exhalation image stack). This motion model is used in fourdimensional planning. It has been shown recently [14] that the linear programming approach used by the CyberKnife can be extended to four-dimensional inverse planning. FIGURE 13-5. This schematic illustrates several aspects of dynamic respiratory compensation with the Synchrony system. In this situation, the movement of a lung tumor is determined in real time by correlating the location of internal gold seeds (not visible) with external markers (IR emitters). Tumor position is then fed back to the robot, which adjusts accordingly for tumor motion.

treatment, radiographs are obtained and the correlation model is updated. Movements of the IR emitters that violate the correlation model (as when a patient’s breathing pattern changes) turn the treatment beam off and initiate the model-building procedure again.

CyberKnife Treatment Planning Radiation planning with the CyberKnife is generally quite different from the standard isocentrically arrayed radiosurgical system (e.g., the Gamma Knife). Although treatment plans are always developed on top of CT image data, other imaging modalities (e.g., MRI, fMRI, PET) are readily supported by means of image fusion. Like other radiosurgical devices, the CyberKnife uses secondary collimators to shape cylindrical beams with diameters ranging from 5 to 60 mm. However, during the planning phase of standard radiosurgical systems, beam directions and beam weights are iteratively deﬁned by the radiation oncologist, physicist, and/or dosimetrist, often with a surgeon, until a suitable distribution of radiation dose is achieved. Although an isocentric treatment mode for spherically symmetric lesions is available with the CyberKnife, most planning is done using an inverse approach. For more complex and irregularly shaped tumors, robotic radiosurgery provides an enormous range of possibilities for placing beams and beam arrays. In designing plans for these targets, the goal of the CyberKnife planning algorithm is to ﬁnd a scheme of beams that returns a maximally conformal distribution for a given shape of target [10, 11]. Practically speaking,

Intracranial Lesions Most conditions treated with standard frame-based radiosurgery are readily treatable with the frameless CyberKnife using the same basic radiosurgical principles. However, the combination of real-time image guidance and robotic treatment delivery can offer distinct advantages over conventional frame-based radiosurgery under some clinical circumstances. Because of its robotic nature, the CyberKnife enables an increase in both the range of beam trajectories and in some cases the volume of space over which they are distributed. Clinically speaking, this design provides greater homogeneity, when desirable, and enhanced dose conformality for irregularly shaped targets. Even more importantly, frameless stereotaxy makes it practical to both incorporate hypofractionation into a radiosurgical context and, for the ﬁrst time, perform extracranial radiosurgery with true stereotactic precision [2]. The CyberKnife can be used to treat all the common intracranial conditions currently treated radiosurgically, such as intracranial tumors, arteriovenous malformations, and functional indications such as trigeminal neuralgia. However, the use of hypofractionation makes it feasible to treat many larger tumors or lesions adherent to especially radiation-sensitive brain structures such as the anterior visual pathways. Moreover, CyberKnife radiosurgical ablation is readily extended to lesions that originate beneath or extend through the skull base and involve the hypopharynx, such as tumors of the foramen magnum or nasopharyngeal carcinoma. Although it is beyond the scope of this chapter to describe all potential intracranial indications, here we highlight a few of them.

Perioptic Tumors Hypofractionation is especially useful for treating lesions immediately adjacent to the optic pathways or lesions involving other

13.

cyberknife radiosurgery

cranial nerves. High rates of tumor control and preservation of visual function after multisession robotic radiosurgery have been reported [15–17]. In these retrospective studies, patients with meningioma, pituitary adenoma, and craniopharyngioma, all within 2 mm of the optic apparatus, were treated with CyberKnife radiosurgery delivered in two to ﬁve sessions to a mean cumulative marginal dose of 20.3 Gy. After 4 years of follow-up, this multisession approach resulted in very high levels of local tumor control (>95%) and was nearly universally safe; one patient with a history of conventional radiation therapy and three courses of perioptic radiosurgery suffered a radiationinduced optic nerve injury. This experience is remarkable for the high incidence of visual function preservation and tumor control in what was otherwise a particularly challenging group of patients; nearly all had a history of either surgical resection (sometimes multiple) and/or conventional radiotherapy.

175

Spinal Indications In the ﬁrst reported use of image-guided robotics to perform spinal radiosurgery, Ryu and co-workers demonstrated the safety and short-term efﬁcacy for a variety of neoplastic and vascular lesions [21]. Surgical implantation of ﬁducials into adjacent vertebral segments was necessary for tracking the ablated spinal lesion. There were no adverse events related to the implantation of ﬁducials. Given the appreciable uncertainty about the spinal cord’s tolerance to radiosurgery, hypofractionation, using two to ﬁve fractions, was utilized in nearly all cases. Subsequent to the above investigation, Gerstzen and coworkers found in their clinical series of 125 patients that singlefraction spinal radiosurgery was safe and effective under most circumstances [22]. A mean tumor dose of 14 Gy as prescribed to the 80% isodose was used (Fig. 13-6). Most patients’ lesions

Vestibular Schwannomas Small- to moderate-sized vestibular schwannomas can be treated with single or fractionated radiosurgical regimes. Since 1999, the CyberKnife at Stanford University has been used to treat more than 350 patients with vestibular schwannoma, delivering 18 to 21 Gy in three sessions separated by 24 hours. To date, only three patients within this cohort has shown evidence of tumor progression, and in no case was treatment-related trigeminal or facial nerve dysfunction observed. Among those patients with Gardner-Robertson grade I or II hearing preoperatively, 74% retained these hearing levels after an average 4 years of follow-up. Furthermore, there were no cases of total hearing loss [18].

Trigeminal Neuralgia Radiosurgical rhizotomy, most commonly performed with the Gamma Knife, is well-established in the management of trigeminal neuralgia. However, after more than a decade of experience, treatment latencies and the overall response rate continue to be less than ideal. With the goal of overcoming these limitations, the standard Gamma Knife trigeminal rhizotomy has been modiﬁed by using the capacity of the CyberKnife to deliver non-isocentric plans. Rapid onset of pain relief after CyberKnife treatment was ﬁrst reported in a small series of patients by Romanelli and co-workers [19]. A more recent and larger multi-institutional study conﬁrmed these results [20]. In this later study, a prescribed dose of 60 to 70 Gy was delivered to a 6- to 8-mm length of the retrogasserian region of the trigeminal nerve. The median latency to pain relief was only 7 days. Initial pain control was ranked as excellent in 88% of patients, whereas three patients reported no pain relief and two experienced only a moderate reduction of pain. Although pain relief appeared durable in 78% of this cohort, half of the patients in this series eventually developed facial numbness. Because a clear relationship was observed between the length of the trigeminal nerve treated and the onset of numbness, a gradual dose and volume de-escalation was subsequently conducted. The current parameters used for trigeminal rhizotomy at Stanford include a 6-mm length of nerve and dose prescriptions of 60 Gy marginal and 75 Gy Dmax.

FIGURE 13-6. Examples of CyberKnife treatment planning for a T12 spinal metastasis from a renal cell carcinoma. Isodose distribution is overlaid on both axial (a) and coronal (b) MRI scans. In this case, the selected treatment dose was 18 Gy prescribed to the 70% isodose. Note the steep dose gradient adjacent to the spinal cord.

176

j.r. adler jr. et al.

were metastatic (108 cases), and 59% of them had been treated previously with conventional irradiation. The authors reported an improvement in pain scores in 74 of 79 patients with axial and radicular pain prior to radiosurgery. No acute radiation toxicity or new neurologic deﬁcits occurred during the median 18-month follow-up. Meanwhile, Degen et al. [23] treated 51 patients with 72 spinal tumors (58 metastatic and 14 primary) with multiple-session CyberKnife radiosurgery. After a mean follow-up of 1 year, there were signiﬁcant and durable reductions in pain scores as well as a maintenance of physical and mental quality of life measures. Side effects were mild and selflimiting. Because by nature the spinal cord is fragile, poorly vascularized, and highly sensitive to radiation damage, the ﬁndings of both of the above studies are remarkable. The beneﬁts are even more impressive when one considers that a signiﬁcant percentage of patients in these series had undergone previous conventional irradiation. Furthermore, the relative safety of this procedure at these institutions reinforces claims about the overall system accuracy of robotic radiosurgery. In management of brain tumors, radiosurgery is often used as one component of a multimodality approach. The same philosophy can also be applied to spinal lesions. As an example, CyberKnife radiosurgery has been combined with kyphoplasty to address pathologic compression fractures [24]. This is a new treatment paradigm for metastatic spinal tumors. By coupling two minimally invasive procedures, the risk of major surgical complications is lessened, treatment costs are lowered, and quality of life is improved. Even more integrated radiosurgical approaches are likely to emerge in the future that will further change the classic surgical management of many spinal lesions. One of the more important nonneoplastic applications of CyberKnife spinal radiosurgery is the treatment of intramedullary spinal cord arteriovenous malformations (SCAVMs). Alternative therapies for this speciﬁc group of AVMs, which include embolization and surgery, have only a limited role and are almost never curative. Because there are so few viable treatment options for most cases of SCAVM, spinal radiosurgery could prove a welcome new therapeutic tool. With this in mind, 21 patients with intramedullary spinal cord AVMs (11 cervical, 7 thoracic, 3 lumbar) were treated with the CyberKnife between 1997 and 2006 as part of a gradual dose escalation study at Stanford University. Preliminary ﬁndings from this experience have been reported [25]. However, for the entire series of patients, radiosurgery was delivered in one to ﬁve sessions to an average AVM volume of 1.8 cm3 using an average marginal dose of 19.5 Gy. Patients were followed with annual MRI and angiography every 3 years. Average follow-up now exceeds 2.5 years. Among the six patients studied with posttreatment angiography, AVM obliteration was partial in four and complete in two. Signiﬁcant AVM obliteration has been observed on MRI in nearly every case that was more than 1 year from radiosurgery; AVM involution appears complete in three cases and conﬁrmatory angiography is pending. Not surprisingly, the radiologic outcome to date suggests that more aggressive radiosurgical regimens (generally using fewer fractions) correlate with a higher rate of AVM obliteration. No patients suffered post-SRS hemorrhage or any signiﬁcant neurologic deterioration attributable to SRS. Despite the still-evolving radiosurgical experience with SCAVM, it seems safe to say that the CyberKnife now offers an important new treatment option for these challenging lesions.

Intrathoracic and Intraabdominal Lesions The past half decade has witnessed an explosion of interest in using ever more precise irradiation to treat lesions of the chest and abdomen. All these procedures combine noninvasive external immobilization and targeting with conventional medical linacs [26–31]. Meanwhile, a review of the preliminary stereotactic radiotherapy experience treating liver malignancies has been recently provided by Fuss and Thomas [32]. Image-guided robotic radiosurgery adds a powerful and versatile tool to this ﬁeld. In fact, because of Synchrony’s unique capabilities for tracking and correcting for respiratory motion, the CyberKnife may eventually have its biggest clinical impact in treating lesions of the chest, abdomen, and pelvis. Consistent with this idea, multiple clinical trials are now under way worldwide to assess the utility of CyberKnife ablation of lung tumors. It is important to reiterate that tumor targeting and tracking with Synchrony requires the implantation of radiopaque ﬁducials near the lesion [33, 34]. Gold seeds are preferred because they are inert, readily inserted, and easy to image with the CyberKnife’s imaging module. The feasibility of CyberKnife radiosurgical ablation for pulmonary lesions was ﬁrst investigated in 23 patients with biopsyproven lung tumors (15 primary and 8 metastatic lesions) by Whyte and co-workers [35] in a pilot study. After CT-guided percutaneous ﬁducial placement, each patient was treated with 15 Gy in a single fraction. Although radiosurgery itself was well tolerated, several patients experienced complications as a result of the implantation of ﬁducials. After limited postsurgical follow-up, radiographic progression was found only in two patients; the follow-up period ranged from 1 to 26 months (mean, 7 months). Because several local failures were subsequently observed, the radiosurgical dose for lung cancer has been gradually escalated. Le et al. [36] conducted a phase I dose-escalation study to assess the effects of single-fraction CyberKnife radiosurgery in patients with inoperable T1–2N0 non–small cell lung cancers (NSCLCs) or solitary lung metastases. Doses ranging from 15 to 30 Gy were tested in 32 patients. At 1 year, doses greater than 20 Gy prevented local progression of NSCLC in 91% of patients, whereas doses less than or equal to 20 Gy resulted in only a 54% freedom from local progression. However, higher doses (25 to 30 Gy) were also associated with serious complications in patients with prior radiotherapy and larger midline tumors. NSCLCs responded better than metastatic tumors. The authors concluded that single-fraction CyberKnife was feasible and effective, but in selected cases, single fractions of 25 Gy or more may be unacceptably toxic. These results are consistent with those emerging during the past decade showing that high-dose stereotactic radiotherapy can be effective for NSCLC, but they also suggest that a hypofractionated approach may be required to minimize complications. The treatment of pancreatic adenocarcinomas with CyberKnife radiosurgery was ﬁrst reported by Romanelli et al. [19]. Twelve patients were treated with 15, 20, or 25 Gy delivered as a single fraction. The treatment was well tolerated, values of the pancreatic cancer marker CA 19–9 were decreased in most patients, and all patients with pain prior to treatment experienced improvement within days. These preliminary results were expanded upon by Koong and associates [37], who observed local control of tumor growth with a single dose of

13.

cyberknife radiosurgery

25 Gy, without serious gastrointestinal toxicity. In a later paper [38], CyberKnife radiosurgery was delivered as a boost after conventional intensity-modulated radiotherapy (IMRT; 45 Gy delivered in 1.8-Gy fractions) and concurrent chemotherapy (5-ﬂuorouracil or capecitabine). Although high rates of tumor control were achieved (15 of 16 patients were free from local progression until death), the incidence of complications increased signiﬁcantly above that observed with radiosurgery alone. In addition, there was no improvement in survival. This ﬁnding encouraged the authors to subsequently forego IMRT in such patients in favor of radiosurgery alone, with or without adjuvant chemotherapy. Although CyberKnife radiosurgery has proved highly effective in achieving local control and palliating symptoms, survival continues to be largely dictated by the frequent occurrence of regional metastases. Nevertheless, control of pancreatic tumors itself is an impressive accomplishment; if such an effect could be combined with more effective systemic therapies, noninvasive radiosurgical ablation holds out the promise of prolonging survival. Although preliminary, the above results with abdominal and thoracic tumors have been encouraging. Nevertheless, much more evidence is needed to establish a deﬁnite role for robotic radiosurgery in managing nonneurologic disorders. In this regard, multiple clinical trials are under way worldwide that seek to discover if CyberKnife radiosurgery can substantially impact the overall clinical outcome for a broad range of extracranial indications.

Future Directions Stated simply, the CyberKnife was developed to enable the safe, accurate, and effective application of radiosurgery throughout the body. Future enhancements in the CyberKnife will be largely dictated by the experiences of ongoing clinical studies as well as the creativity of medical professionals in radiosurgery. Nevertheless, certain improvements seem likely. For example, the radiation output of the CyberKnife linac is likely to increase. In addition to shortening the length of treatment, this development should enable progressively smaller collimators to create ever more conformal treatment volumes. Given the relative youth of the CyberKnife concept, one can readily envision how a number of similar subsystems will be optimized with time. Fiducial targeting of lesions within the major body cavities is both robust and extremely accurate. However, the implantation process can be technically demanding, adds to the complexity of the overall radiosurgical procedure, and is modestly invasive. In the case of lung treatment, pneumothorax is not uncommon. Although the challenges are considerable, a technology for targeting intraabdominal and thoracic lesions without ﬁducials will be a clear improvement in robotic radiosurgery. Recent research suggests that motion compensation without implanted gold markers may be clinically practical [39]. Such technology would considerably simplify treatment protocols and enhance patient comfort and safety. The proposed method for ﬁducial-less targeting extends the current CyberKnife X-ray image correlation targeting system by also incorporating breathing motion into the pretreatment DRR library. Using four-dimensional imaging and image registration, several pretreatment CT scans are acquired at different stages of the respiratory cycle. A large array of DRRs are

177

then computed from each CT volume. In effect, each of these synthetic images incorporates both patient and respiratory movement. Similar to the current process, throughout radiosurgery live X-ray images are compared with the precalculated DRRs. However, the new comparison group would include synthetic X-rays that also reﬂect a range of respiratory states. In performing two-dimensional–two-dimensional (2D-2D) registrations, the best matching DRR is identiﬁed, thereby enabling the current phase of the breathing cycle and the location of the tumor to be determined. With dynamic motion compensation (i.e., Synchrony), the treatment beams move synchronously with the target, and the beam strikes the target region approximately as planned. However, some tumors deform or rotate with respiration, and tissue surrounding the tumor may exhibit a different motion pattern than the tumor itself. Information on the relative motion of organs with respect to treatment beams (during the actual radiosurgery) can be incorporated into the planning process, thus further improving the overall precision and safety of the treatment. This could be of particular beneﬁt when treating lesions close to the spinal cord where the precise dose tolerance can be critical. Current robotic systems use only cylindrical collimators. However, multileaf collimators in combination with robotic systems could potentially further reduce treatment time and improve treatment conformality. Given the ﬂexibility with which the CyberKnife can conﬁgure treatment beams, a relatively simple multileaf collimator might be utilized instead of the more standard microleafed device used in conventional radiation therapy. On the other hand, the same improvement in time efﬁciency might also be achieved by further increasing linac output, which unlike multileaf collimation would not sacriﬁce beam penumbra. Whether taken alone or in aggregate, these technical improvements may in turn usher in new clinical radiosurgical applications that have yet to be envisioned. Although future technological developments have the potential to improve the process and clinical outcome of robotic radiosurgery, we are equally excited by the prospects for improving our basic radiobiological understanding of large-fraction irradiation. A number of important questions remain unanswered. The biggest of these continues to be what are the optimal doses and fractionation schemes for speciﬁc tumor entities, particularly within the thoracic and abdominal cavities. Furthermore, the combination of radiosurgery with new systemic immunotherapies and chemotherapies, which attack microscopic malignancies, and may in fact be radiation sensitizers, are likely to play a vital role for improving future clinical outcomes.

Conclusion Radiosurgery is in the midst of a technological revolution. The recent introduction of image guidance and robotic delivery has dramatically expanded the scope of this ﬁeld. As inevitable improvements in neuroimaging and computer technology emerge over the coming years, they will serve as an impetus for further improvements in robotic radiosurgical technology. This perspective stems in large part from the inherent, and somewhat unique, ﬂexibility of the CyberKnife’s basic design. As the full extent of this vision is realized, the concept of radiosurgical ablation will continue to expand into new anatomic

178

j.r. adler jr. et al.

regions and disorders. This evolution will also require that surgeons of nearly all stripes and radiation oncologists reexamine some of the basic tenets encompassed within their respective disciplines.

References 1. Chang SD, Main W, Martin DP, et al. An analysis of the accuracy of the CyberKnife: a robotic frameless stereotactic radiosurgical system. Neurosurgery 2003; 52:140–146; discussion 146–147. 2. Yu C, Main W, Taylor D, et al. An anthropomorphic phantom study of the accuracy of CyberKnife spinal radiosurgery. Neurosurgery 2004; 55:1138–1149. 3. Ho AK, Fu D, Cotrutz C, et al. A study of the accuracy of CyberKnife spinal radiosurgery using skeletal structure tracking. Neurosurgery 2007; 60(2 Suppl 1):ONS147–156. 4. Gierga DP, Chen GT, Kung JH, et al. Quantiﬁcation of respiration-induced abdominal tumor motion and its impact on IMRT dose distributions. Int J Radiat Oncol Biol Phys 2004; 58: 1584–1595. 5. Kaus MR, Netsch T, Kabus S, et al. Estimation of organ motion from 4D CT for 4D radiation therapy planning of lung cancer. Presented at Medical Image Computing and Computer-Assisted Intervention—MICCAI 2004, 7th International Conference, Saint-Malo, France, September 26–29, 2004. 6. Langen KM, Jones DT. Organ motion and its management. Int J Radiat Oncol Biol Phys 2001; 50:265–278. 7. Mageras GS, Pevsner A, Yorke ED, et al. Measurement of lung tumor motion using respiration-correlated CT. Int J Radiat Oncol Biol Phys 2004; 60:933–941. 8. Plathow C, Ley S, Fink C, et al. Analysis of intrathoracic tumor mobility during whole breathing cycle by dynamic MRI. Int J Radiat Oncol Biol Phys 2004; 59:952–959. 9. Shirato H, Seppenwoolde Y, Kitamura K, et al. Intrafractional tumor motion: lung and liver. Semin Radiat Oncol 2004; 14:10–18. 10. Webb S. Conformal intensity-modulated radiotherapy (IMRT) delivered by robotic linac—testing IMRT to the limit? Phys Med Biol 1999; 44:1639–1654. 11. Webb S. Conformal intensity-modulated radiotherapy (IMRT) delivered by robotic linac—conformality versus efﬁciency of dose delivery. Phys Med Biol 2000; 45:1715–1730. 12. Li JG, Xing L. Inverse planning incorporating organ motion. Med Phys 2000; 27:1573–1578. 13. Unkelbach J, Oelfke U. Incorporating organ movements in inverse planning: assessing dose uncertainties by Bayesian inference. Phys Med Biol 2005; 50:121–139. 14. Schlaefer A, Fisseler J, Dieterich S, et al. Feasibility of fourdimensional conformal planning for robotic radiosurgery. Med Phys 2005; 32:3786–3792. 15. Adler JR Jr, Gibbs IC, Puataweepong P, Chang SD. Visual ﬁeld preservation after multisession CyberKnife radiosurgery for perioptic lesions. Neurosurgery 2006; 59(2):244–254. 16. Mehta VK, Lee QT, Chang SD, et al. Image guided stereotactic radiosurgery for lesions in proximity to the anterior visual pathways: a preliminary report. Technol Cancer Res Treat 2002; 1:173–180. 17. Pham CJ, Chang SD, Gibbs IC, et al. Preliminary visual ﬁeld preservation after staged CyberKnife radiosurgery for perioptic lesions. Neurosurgery 2004; 54:799–810; discussion 810–812. 18. Chang SD, Gibbs IC, Sakamoto GT, et al. Staged stereotactic irradiation for acoustic neuroma. Neurosurgery 2005; 56:1254– 1261; discussion 1261–1253. 19. Romanelli P, Heit G, Chang SD, et al. CyberKnife radiosurgery for trigeminal neuralgia. Stereotact Funct Neurosurg 2003; 81: 105–109.

20. Lim M, Villavicencio AT, Burneikiene S, et al. CyberKnife radiosurgery for idiopathic trigeminal neuralgia. Neurosurg Focus 2005; 18:E9. 21. Ryu S, Fang Yin F, Rock J, et al. Image-guided and intensitymodulated radiosurgery for patients with spinal metastasis. Cancer 2003; 97:2013–2018. 22. Gerszten PC, Ozhasoglu C, Burton SA, et al. CyberKnife frameless stereotactic radiosurgery for spinal lesions: clinical experience in 125 cases. Neurosurgery 2004; 55:89–98; discussion 98–99. 23. Degen JW, Gagnon GJ, Voyadzis JM, et al. CyberKnife stereotactic radiosurgical treatment of spinal tumors for pain control and quality of life. J Neurosurg Spine 2005; 2:540–549. 24. Gerszten PC, Germanwala A, Burton SA, et al. Combination kyphoplasty and spinal radiosurgery: a new treatment paradigm for pathological fractures. J Neurosurg Spine 2005; 3:296–301. 25. Sinclair J, Chang SD, Gibbs IC, Adler JR Jr. Multisession CyberKnife radiosurgery for intramedullary spinal cord arteriovenous malformations. Neurosurgery 2006; 58:1081–1089; discussion 1081–1089. 26. Bilsky MH, Yamada Y, Yenice KM, et al. Intensity-modulated stereotactic radiotherapy of paraspinal tumors: a preliminary report. Neurosurgery 2004; 54:823–830; discussion 830–821. 27. Herfarth KK, Debus J, Lohr F, et al. Stereotactic single-dose radiation therapy of liver tumors: results of a phase I/II trial. J Clin Oncol 2001; 19:164–170. 28. Shiu AS, Chang EL, Ye JS, et al. Near simultaneous computed tomography image-guided stereotactic spinal radiotherapy: an emerging paradigm for achieving true stereotaxy. Int J Radiat Oncol Biol Phys 2003; 57:605–613. 29. Timmerman R, Papiez L, McGarry R, et al. Extracranial stereotactic radioablation: results of a phase I study in medically inoperable stage I non-small cell lung cancer. Chest 2003; 124: 1946–1955. 30. Uematsu M, Shioda A, Suda A, et al. Computed tomographyguided frameless stereotactic radiotherapy for stage I non-small cell lung cancer: a 5-year experience. Int J Radiat Oncol Biol Phys 2001; 51:666–670. 31. Yenice KM, Lovelock DM, Hunt MA, et al. CT image-guided intensity-modulated therapy for paraspinal tumors using stereotactic immobilization. Int J Radiat Oncol Biol Phys 2003; 55: 583–593. 32. Fuss M, Thomas CR Jr. Stereotactic body radiation therapy: an ablative treatment option for primary and secondary liver tumors. Ann Surg Oncol 2004; 11:130–138. 33. Schweikard A, Glosser G, Bodduluri M, et al. Robotic motion compensation for respiratory movement during radiosurgery. Comput Aided Surg 2000; 5:263–277. 34. Schweikard A, Shiomi H, Adler J. Respiration tracking in radiosurgery. Med Phys 2004; 31:2738–2741. 35. Whyte RI, Crownover R, Murphy MJ, et al. Stereotactic radiosurgery for lung tumors: preliminary report of a phase I trial. Ann Thorac Surg 2003; 75:1097–1101. 36. Le QT, Loo BW, Ho A, et al. Results of a phase I dose-escalation study using single-fraction stereotactic radiotherapy for lung tumors. J Thorac Oncol. 2006 Oct; 1(8):802–809. 37. Koong AC, Le QT, Ho A, et al. Phase I study of stereotactic radiosurgery in patients with locally advanced pancreatic cancer. Int J Radiat Oncol Biol Phys 2004; 58:1017–1021. 38. Koong AC, Christofferson E, Le QT, et al. Phase II study to assess the efﬁcacy of conventionally fractionated radiotherapy followed by a stereotactic radiosurgery boost in patients with locally advanced pancreatic cancer. Int J Radiat Oncol Biol Phys 2005; 63:320–323. 39. Schweikard A, Shiomi H, Adler JR. Respiration tracking in radiosurgery without ﬁducials. Int J Med Robotics Comput Assist Surg 2005; 1:19–27.

PA R T IV

Treatment of Disease Types

179

1 4

Brain Metastases John H. Suh, Gene H. Barnett, and William F. Regine

Introduction Brain metastases are a signiﬁcant cause of morbidity and mortality that affects an estimated 20% to 30% of cancer patients [1]. The annual incidence in the United States has been estimated to be as high as 170,000 cases [2]. Brain metastases most commonly originate from tumors of the lung, breast, renal cells, and colon and from melanoma. In some cases, the primary is unknown. Multiple lesions are seen more frequently with melanoma and lung cancer, whereas single lesions are more common with breast, colon, and renal cell cancer. An estimated 30% to 40% of patients with a brain metastasis present with a single lesion [3]. Despite improvements in imaging, surgical technique, and radiation delivery, the prognosis for these patients remains dismal. Patients who receive no treatment have a median survival of only 1 month. Those who are treated with steroids survive a median of 1 to 2 months, and those who undergo whole-brain radiation therapy survive for a median of 4 to 7 months after therapy [4–7]. As a result, the development of brain metastasis is one of the most feared complications of cancer patients, and it represents a therapeutic challenge to neurosurgeons, radiation oncologists, medical oncologists, and neuro-oncologists. Historically, the treatment of brain metastasis has included whole-brain radiation therapy (WBRT), which was ﬁrst reported in the 1950s and early 1960s [8, 9]. Response to WBRT is best among patients with small cell lung cancer, lymphoma, and germ cell tumors. The Radiation Therapy Oncology Group (RTOG) conducted numerous trials from 1971 through 1993 to investigate various fractionation schemes and doses of WBRT, which are listed in Table 14-1 [10–18]. In addition, the RTOG has investigated the use of hyperfractionation and radiation sensitizers such as bromodeoxyuridine and found no improvement in overall survival. Based on these studies, the use of 3000 cGy in 10 fractions became a popular fractionation scheme to consider for patients with brain metastases. Although neurologic symptoms improved in the majority of patients, local control was low resulting in neurologic death in 25% to 54% of patients [10]. Given the poor prognosis for patients with brain metastases, alternative strategies to improve outcomes have been explored. Stereotactic radiosurgery (SRS) is a technique that

delivers a single, high dose of ionizing radiation using stereotactically directed narrow beams to small intracranial targets while sparing the surrounding brain tissue [19]. Since Sturm’s initial report of 12 lesions treated on a modiﬁed linear accelerator, a number of papers have corroborated the beneﬁt of SRS in newly diagnosed and recurrent brain metastases [20]. Currently, brain metastases represent the most common indication for SRS. Although SRS has become an important treatment option, its use has also sparked controversy about the appropriate treatment for patients with brain metastases. This chapter will review the prognostic factors, surgical results, rationale for SRS, results of surgery compared with SRS, the institutional results of SRS with WBRT and SRS alone, the results of completed phase II/III clinical trials, complications of SRS, and future direction of SRS for brain metastases.

Prognostic Factors The most commonly used prognostic scale for patients with brain metastasis is the RTOG recursive partitioning analysis (RPA) [21]. This scale divides patients into three classes based on 1200 consecutive patients enrolled in three RTOG trials from 1979 to 1993. The vast majority of these patients had unresectable and/or multiple metastases but received standard doses of WBRT. Table 14-2 lists the various classes and their components. The most important factors include extracranial metastases, patient age, Karnofsky performance status, and control of primary tumor. A study from Lagerwaald and colleagues reviewed 1292 patients with brain metastases [22]. In this Dutch study, lung cancer was the most common primary disease (56%). Median survival was 3.4 months, and the 1-year and 2-year survivals were 12% and 4%, respectively. The most important factors were treatment modality, performance status, extracranial disease burden, and response to steroid treatment.

Surgery Given the low local control rates associated with WBRT alone, surgical removal of tumors—particularly single or symptomatic lesions—was explored in hopes of improving local control and

181

182

j.h. suh et al.

TABLE 14-1. Prospective Radiation Therapy Oncology Group (RTOG) brain metastases studies (1971–1993). Protocol

Year

No. patients

Total dose/no. fractions

Median survival

RTOG 6901 [10]

1971–1973

RTOG 7361 [11]

1973–1976

RTOG 6901 [12] RTOG 7361 [12] Ultrarapid RTOG 7606 [13] Favorable patients RTOG 8528 [14]

1971–1973 1973–1976

233 217 233 227 447 228 227 26 33

30 Gy/10 Fx/2 weeks 30 Gy/15 Fx/3 weeks 40 Gy/15 Fx/3 weeks 40 Gy/20 Fx/4 weeks 20 Gy/5 Fx/1 week 30 Gy/10 Fx/2 weeks 40 Gy/15 Fx/3 weeks 10 Gy/1 Fx/1 day 12 Gy/2 Fx/2 days

21 weeks 18 weeks 18 weeks 16 weeks 15 weeks 15 weeks 18 weeks 15 weeks 13 weeks

RTOG 9104 [15]

1991–1995

RTOG 7916 [16] Misonidazole

1979–1983

130 125 30 53 44 36 213 216 193 200 196 190

18 weeks 17 weeks 4.8 months 5.4 months 7.2 months 8.2 months 4.5 months 4.5 months 4.5 months 4.1 months 3.1 months 3.9 months

RTOG 8905 [17] BrdU

1989–1993

30 Gy/10 Fx/2 weeks 50 Gy/20 Fx/4 weeks 48 Gy/1.6 Gy bid 54.5 Gy/1.6 Gy bid 64 Gy/1.6 Gy bid 70.4 Gy/1.6 Gy bid 30 Gy/10 Fx 54.4 Gy/1.6 Gy bid 30 Gy/10 Fx/2 weeks 5 Gy/6 Fx/3 weeks 30 Gy/10 Fx + Miso 5 Gy/6 Fx + Miso 37.5 Gy/15 Fx/3 weeks 37.5 Gy/15 + BrdU

1976–1979 1986–1989

36 34

6.1 months 4.3 months

Fx, fraction; bid, twice daily; Miso, misonidazole; BrdU, bromodeoxyuridine. Source: Adapted from Sneed PK, Larson DA, Wara WM. Neurosurg Clin N Am 1996; 7:505–515.

TABLE 14-2. RTOG RPA classes for brain metastases. Factors

Median survival (months)

Class I

Age <65 and KPS ≥70 and Controlled primary tumor and Extracranial metastases

7.1

Class II

KPS ≥70 and Age ≥65 and/or Uncontrolled primary and/or Extracranial metastases

4.2

Class III

KPS ≤60

2.3

Source: Gaspar L, Scott C, Rotman M, et al. Recursive partitioning analysis (RPA) of prognostic factors in three Radiation Therapy Oncology Group (RTOG) brain metastases trials. Int J Radiat Oncol Biol Phys 1997; 37:745–751.

survival. Surgery can also establish or conﬁrm the diagnosis, especially for patients without primary tumor conﬁrmation. In addition, surgery can provide immediate and effective palliation of symptomatic mass effect. Three randomized trials comparing WBRT with surgery followed by WBRT for patients with single metastasis have been completed and are summarized in Table 14-3 [23–25]. These trials were based on the premise that improved local control of a single brain metastasis would result in improved survival. Patchell’s study demonstrated that surgery and WBRT improved survival (40 vs. 15 weeks, p < 0.01), local control (80% vs. 48%, p < 0.02), and functional independence (38 vs. 8 weeks, p < 0.005) compared with biopsy and WBRT [23]. Noordijk’s study included 63 patients with CT-conﬁrmed single metastasis who were randomized to surgery and WBRT or WBRT alone [24]. Surgery and WBRT improved survival and functional independence more than WBRT alone (10 vs. 6 months, p = 0.04 and 7.5 vs. 3.5 months, p = 0.06, respectively). Patients with

TABLE 14-3. Phase III trials of WBRT versus surgery and WBRT. Author

Year

Surgery

WBRT Dose (cGy)/no. fractions/ no. weeks

N

MS (weeks)

KPS ≥70 (weeks)

CNS death

Local control

Patchell [23]

1990

Noordijk [24]

1994

Mintz [25]

1996

Yes No Yes No Yes No

3600 / 12 / 2.5 weeks 3600 / 12 / 2.5 weeks 4000 / 20 / 2 weeks 4000 / 20 / 2 weeks 3000 / 10 / 2 weeks 3000 / 10 / 2 weeks

25 23 32 31 41 43

40 15 43 26 22 25

38 8 33 15 N/A* N/A*

29% 52% 35% 33% 14% 28%

80% 48% N/A N/A N/A N/A

N, number of patients; MS, median survival; N/A, not applicable; N/A*, the mean proportion of days that the patients had KPS ≥70 was not different.

14.

brain metastases

active systemic disease did not beneﬁt from surgery. The third trial, which was from Mintz, did not demonstrate a survival beneﬁt for patients undergoing surgery (6.3 vs. 5.6 months, p = 0.24) [25], which was most likely a result of two factors: (1) the study consisted of a large percentage of patients with active systemic disease and (2) patients with lower baseline Karnofsky performance status (KPS) were included.

Rationale for Using SRS for Brain Metastases Brain metastases have features that make them ideal targets for SRS. The majority of these tumors are pseudospherical in shape, are located in the gray-white junction, have a maximum tumor diameter of less than 4 cm, and, unlike primary gliomas, are noninﬁltrative. These features also allow for accurate target delineation, planning, and treatment delivery. In addition, a single large fraction of radiation appears to have an equal effect in all tumor types, even among the “radioresistant” tumors such as renal cell carcinoma and melanoma [26–28] (Case Study 14-1). The use of SRS alone or as an adjunct to WBRT is based on the premise that improved local control will reduce morbidity, improve quality life, and prolong survival. Nieder reviewed the computed tomography (CT) scans of 332 patients with brain metastases to evaluate local control and time to local progression based on dose [29]. A biologically effective dose (BED) using the linear quadratic model with an alpha/beta of 10 was derived for each patient. The partial response rates signiﬁcantly improved for patients with higher BED suggesting that higher doses were needed to improve local control. As mentioned, two

183

Case Study 14-1 This is a 73-year-old male with a history of renal cell carcinoma s/p right nephrectomy 10 years ago. He presented with left-sided weakness. MRI was obtained, which revealed a 1.7-cm right frontal lesion (Fig. 14-1a, b). Staging workup was negative. He underwent Gamma Knife radiosurgery to the right frontal lesion (Fig. 14-1c): 2400 cGy was prescribed to the 50% local isodose line, which covered 100% of the target. The plan utilized 8 shots using an 8-mm helmet without plugs. Tumor volume = 2.0 cm3. Maximum dose = 4800 cGy. Maximum diameter = 1.9 cm. MD/PD = 2.00. PIV/TV = 1.70. MRI 3 months after Gamma Knife radiosurgery (Fig. 14-1d, e) shows the patient’s left-sided weakness improved.

phase III trials comparing WBRT versus surgical resection and WBRT demonstrated that improved local tumor control of a single metastasis led to improved survival. Given the potential advantages of SRS, interest developed in using SRS rather than surgery to improve local control of brain metastasis. Normal brain tolerance to radiation therapy is related to total dose, fractional dose, elapsed time of radiation delivery, volume of normal brain irradiated, and the amount of and interval from exposure of prior radiation therapy [30]. From a radiobiologic standpoint, brain metastases are considered category IV targets, which are targets with early-responding tissue

FIGURE 14-1. MRI of male with a history of renal cell carcinoma. See Case Study 14-1.

184

j.h. suh et al.

surrounded by late-responding normal tissue [31]. Because the tumor cells are within the target, one would expect a high therapeutic index for these targets given the steep dose gradient associated with SRS. Unlike surgery, SRS has the potential to sterilize cells that may not be contained within the high-dose region.

Surgery Versus SRS Based on results of surgical resection for single metastasis, investigators became interested in evaluating the use of SRS for brain metastases. This interest was based on the potential advantages of SRS, which include outpatient delivery, elimination of general anesthesia, minimal risk for bleeding or infection, lower costs, short convalescence time, and lower risk for complications. In addition, SRS is not limited by location, number of lesions, comorbidity, or coagulopathy. For these various reasons, SRS has become a viable alternative to surgery given the potential cost savings and quality-of-life beneﬁts [32, 33]. A multi-institutional retrospective review from the University of Wisconsin at Madison, University of Florida at Gainesville, and Joint Center for Radiation Therapy included 122 patients with newly diagnosed, potentially resectable single brain metastases treated by SRS and WBRT [34]. The local control rate was 86%. The median survival was 56 weeks with 25% of deaths related to brain tumor progression. Based on results of this retrospective review, the investigators concluded that SRS could be used in place of surgery for patients with brain metastasis. Another study from the M.D. Anderson Cancer Center compared 13 patients treated by SRS with 62 patients treated by surgery [35]. These patients were retrospectively matched for important prognostic factors. The median survival was 7.5 months for the radiosurgery group and 16.4 months for the surgery group. The local recurrence for SRS was 21% versus 8% for surgery. Based on these results, they concluded surgery should be the preferred treatment for surgically accessible lesions. A study from the Mayo Clinic compared the outcomes of radiosurgery versus surgery [36]. This retrospective review included patients with tumors that would have allowed for either type of treatment. No difference in 1-year survival was noted between the groups. A retrospective study from Muacevic compared surgery and radiotherapy with Gamma Knife radiosurgery for patients with a single tumor 3.5 cm or less in diameter [37]. The 1-year survival rates (53% vs. 43%, p = 0.19), 1-year local control rates (75% vs. 83%, p = 0.49), and 1-year neurologic death rates (37% vs. 39%, p = 0.8) for the surgical group and SRS group, respectively, were not statistically different. Because the previously described studies are all retrospective, patient selection bias was probably the key contributor to the mixed results and controversy regarding the beneﬁts of surgery versus SRS. The Joint Center for Radiation Therapy attempted to enroll patients onto a phase III trial comparing surgery to SRS in the 1990s. After 3 years, only six patients were enrolled because of patient or physician preference [38]. M.D. Anderson Cancer Center is currently performing a randomized

trial of surgery versus radiosurgery for patients with a single brain metastasis who are considered eligible for either treatment. Primary end points are survival and local control. This study was opened in 1997. For patients with resectable tumors larger than 3.5 cm or those without histologic conﬁrmation of a primary tumor, most would agree that surgical resection should be considered rather than radiosurgery. Surgical removal of the tumor offers faster improvement in neurologic function and minimizes the use of chronic steroid use. Prolonged use of steroids can result in a number of problems including diabetes, proximal muscle weakness, peripheral edema, psychosis, susceptibility to infections, and gastrointestinal perforation [39]. Surgery is also more effective in alleviating mass effect.

Treatment of Recurrent Metastatic Disease One of the initial reports of SRS for brain metastases consisted of patients with recurrent brain metastases after WBRT or surgery. Loefﬂer reported the Joint Center for Radiation Therapy (JCRT) retrospective results of 18 patients with recurrent metastases after previous surgery, WBRT, or both [40]. The majority of patients improved neurologically with local control being achieved in all patients (Case Study 14-2). A follow-up report from Alexander consisting of mostly patients with recurrent brain metastases reported tumor control rates of 85% at 1 year and 65% at 2 years, although the rate was lower for recurrent lesions [41]. Median survival was 9 months after SRS. Another series from the University of Cincinnati reported the results of 84 patients with 1 to 6 lesions recurrent after WBRT [42]. Median survival was 43 weeks from SRS. Median time to local failure was 35 weeks for all lesions and 52 weeks

Case Study 14-2 This is a 57-year-old female with history of a T3N1M0, stage III adenocarcinoma of the esophagus s/p radiation therapy (3000 cGy/15 fractions, 150 cGy/fraction twice daily) to the distal esophagus with concurrent chemotherapy s/p esophageal resection followed by an additional 3000 cGy/15 fractions, 150 cGy/fraction twice daily to the tumor bed with concurrent chemotherapy. She did well for 10 months until she presented with a right-sided seizure. MRI revealed a 2-cm left frontoparietal lesion (Fig. 14-2a, b). She underwent WBRT (3750 cGy/15 fractions using 250 cGy) followed by Gamma Knife radiosurgery within 1 week of completing WBRT (Fig. 14-2c). Treatment planning information: 1800 cGy was prescribed to the 50% isodose line, which covered 100% of the target. The plan utilized 6 shots using an 18-mm helmet without plugs. Tumor volume = 5.3 cm3. Maximum dose = 36 Gy. Maximum diameter = 2.5 cm. MD/PD = 2.00. PIV/TV = 1.64. Film 3 months after Gamma Knife radiosurgery (Fig. 14-2d, e); she had excellent response to WBRT and Gamma Knife radiosurgery.

14.

brain metastases

185

FIGURE 14-2. MRI of female with a history of adenocarcinoma of the esophagus. See Case Study 14-2.

for lesions treated with greater than or equal to 18 Gy. A study from Noël of 54 consecutive patients reported 1- and 2-year local control rates of 91.3% and 84% and 1- and 2-year brain control rates of 65% and 57%, respectively [43]. Unfortunately, no prospective study has evaluated the use of SRS with other treatment options such as surgery, conformal radiation therapy, or chemotherapy for patients with recurrent brain metastases after WBRT.

Radiosurgery with WBRT Based on the results of SRS for recurrent brain metastasis, the use of SRS with WBRT was investigated for newly diagnosed patients with the hypothesis that the combination of WBRT with SRS would improve local and regional control of brain metastases. A number of the institutional studies have demonstrated high local control rates with the addition of SRS to WBRT [41, 44–52]. These local control rates range from 63% to 97% versus 42% to 87% with use of SRS alone and are summarized in Table 14-4 [26, 50, 52–54].

TABLE 14-4. Retrospective studies of SRS versus SRS and WBRT (local tumor control).

Flickinger [26] Chidel [50] Shehata [52]* Pirzkall [53] Sneed [54]

SRS alone (%)

SRS+WBRT (%)

p value

47 52 87 72 42

82 80 97 86 63

0.008 0.034 0.0001 0.13 0.008

*For metastases less than or equal to 2 cm in diameter.

Shehata evaluated the optimal SRS dose and inﬂuence of WBRT on tumor control among 160 patients with 468 recurrent and newly diagnosed metastases ≤2 cm in diameter treated between October 1992 and May 2001 [52]. On multivariate analysis, the most important factor for local tumor control was the addition of WBRT to SRS (97% vs. 87% for those who did not undergo WBRT; p = 0.001). For patients undergoing WBRT, the use of doses >20 Gy resulted in higher-grade 3 and 4 neurotoxicity (5.9% vs. 1.9% for those receiving <20 Gy) and did not result in better local control. Thus, the authors recommended using 20 Gy for metastases ≤2 cm in diameter with planned combined WBRT. A multi-institutional retrospective study from 10 institutions analyzed outcomes for 502 patients with newly diagnosed brain metastases treated by SRS and WBRT. Patients were stratiﬁed by the RTOG recursive partitioning analysis [55]. For all RPA classes, survival improved for the SRS and WBRT patients compared with WBRT-alone patients. Median survival was 16, 10, and 8 months for the SRS and WBRT patients versus 7.1, 4.2, and 2.3 months for the WBRT patients, respectively. Three prospective trials have been completed comparing the use of SRS with WBRT versus WBRT alone. The ﬁrst trial from the University of Pittsburgh trial enrolled 27 patients with two to four brain metastases measuring less than 2.5 cm in maximum diameter [56]. The study was stopped at an interim analysis at 60% because a signiﬁcant difference in the 1-year local failure rate was noted (8% in the SRS arm vs. 100% in the WBRT arm). The median time to local failure was 6 months for WBRT versus 36 months for the WBRT and SRS arm. Median survival was not signiﬁcantly different for the SRS and WBRT versus WBRT arm (11 months vs. 7.5 months,

186

j.h. suh et al.

respectively). The small sample size as well as possible selection and statistical bias may have distorted the results. The Brown University trial enrolled 96 patients onto a three-armed, prospective randomized trial of Gamma Knife radiosurgery, WBRT, or both [57]. Fifty-one patients underwent surgical resection of large symptomatic lesions prior to randomization, which was not evenly distributed among the treatment arms. No difference in overall survival was noted among the three study arms (7, 5, and 9 months for the SRS alone, WBRT plus SRS, and WBRT arms, respectively). The local control rates were 87%, 91%, and 62%, respectively, for the three arms. The risk for developing new brain lesions was higher for the patients who did not receive WBRT (43% vs. 19% and 23% for the WBRT groups). Unfortunately, the study had major treatment bias given the unequal distribution of patients undergoing surgical resection in the three arms. The RTOG recently reported on a large prospective trial (RTOG 9508) of 333 patients with one to three newly diagnosed brain metastases on magnetic resonance imaging (MRI), each ≤4 cm in diameter, randomized to WBRT (3750 cGy/15 fractions) and SRS versus WBRT alone [58]. Patients had a minimum KPS of 70 and were excluded if the metastases were located in the brain stem or within 1 cm from the optic apparatus. Survival signiﬁcantly improved for the patients with a single brain metastasis treated with SRS plus WBRT versus WBRT (6.5 months vs. 4.9 months, p = 0.04) despite the fact that 19% of patients failed to receive the intended treatment. On multivariate analysis, RPA class (1 vs. 2) and tumor type (squamous or non–small cell vs. other) had a statistically signiﬁcant effect on survival. Overall survival did not improve for the patients with two to three lesions although KPS at 6 months had stabilized or improved with the addition of SRS plus WBRT versus WBRT alone (43% vs. 27%, respectively; p = 0.03). Steroid use was also lower in the SRS and WBRT arm. Local control was improved at 1 year (82% vs. 71%; p = 0.01). Using a Cox model, the risk of developing local recurrence was 43% greater with WBRT alone (p = 0.0021). The type of SRS unit did not inﬂuence results. Based on the results of RTOG 9508, a phase III trial (RTOG 0320) is under way for NSCLC patients with one to three brain metastases. This trial will randomize patients to SRS plus WBRT, SRS plus WBRT and geﬁtinib, or SRS plus WBRT and temozolomide.

and Women’s Hospital of surgically staged IIIA patients with non–small cell lung cancer (NSCLC) found that the most common site of recurrence was the brain, in particular for patients with residual nodal disease after neoadjuvant therapy and nonsquamous histology who had a risk of 53% at 3 years [64]. The RTOG is currently performing a phase III trial (RTOG 0212) for patients with locally advanced NSCLC; patients are randomized to PCI (3000 cGy/15 fractions) versus observation. Several prospective trials have been performed to evaluate whether patients with brain metastases have neurocognitive deﬁcits prior to the initiation of treatment. A recent prospective phase III trial evaluated patients with brain metastasis with monthly neurocognitive testing consisting of memory, ﬁne motor speed, executive function, and global neurocognitive testing [65]. Table 14-5 lists the various neurocognitive testing that was performed for patients enrolled on this study. This trial demonstrated that 90.5% of patients had impairment of one or more neurocognitive tests at baseline and 42.4% of patients had impairment in four or more tests. Another study from RTOG demonstrated that tumor progression resulted in a signiﬁcant decline in the mini-mental status exam at 3 months that was much more pronounced for the patients enrolled on the phase III trial of 30 Gy/10 fractions versus 54.4 Gy/30 fractions using 1.6 Gy twice daily [66]. Chang reported on a prospective trial of neurocognitive testing for patients with 1 to 3 newly diagnosed brain metastases to determine if neurocognitive function was spared with SRS alone [67]. Eighteen of the 27 enrolled patients had evaluable neurocognitive function data. At baseline, 66% of patients had some degree of impairment that was related to total brain metastases volume. After SRS, acute and subacute improvements in executive function and processing speed occurred, but learning and memory skills declined. Seven to 9 months after SRS, there was a suggestion of global decline of neurocognitive function. Recently, the RTOG reported on the feasibility of performing ﬁve neurocognitive measures and administering a qualityof-life instrument in patients with brain metastases [68]. The overall compliance rate for administration and completion of these measures and instrument prior to treatment, at the completion of WBRT, and 1 month after WBRT was ≥95%, ≥84%, and ≥70%, respectively. Based on the encouraging results of

Radiosurgery Alone Despite evidence supporting the use of WBRT for patients with single brain metastasis, the use of SRS alone for treatment of brain metastases has gained popularity and has become a contentious issue. This has been mostly driven by concerns regarding quality-of-life issues and the potential side effects of WBRT, especially cognitive effects for long-term survivors [59–60]. The commonly referenced paper by DeAngelis and colleagues used fractionation schemes (400 to 600 cGy/fraction in 10/12 patients) that are used for patients with a poor prognosis; with clinical dementia limited to only those patients treated with >3 Gy per fraction [59]. In addition, some believe that patients can be effectively managed by repeat SRS rather than WBRT. Several trials of prophylactic irradiation (PCI) have demonstrated a decrease in the development of brain metastases but no survival advantage [61–63]. A recent study from Brigham

TABLE 14-5. Neurocognitive testing used for phase III brain metastases trial. Test

Neurocognitive domain

Hopkins Verbal Learning Test (recall) Hopkins Verbal Learning Test (recognition) Hopkins Verbal Learning Test (delay) Trailmaking A Trailmaking B Controlled oral word association Grooved pegboard dominant hand Grooved pegboard nondominant hand

Memory Memory Memory Executive Executive Executive Fine motor Fine motor

Source: Umsawasdi T, Valdivieso M, Chen TT, et al. Role of elective brain irradiation during combined chemoradiotherapy for limited disease non-small cell lung cancer. J Neurooncol 1984; 2:253–259.

14.

brain metastases

this study, and others of similar design, current and future developing brain tumor/metastases studies have and will more routinely incorporate baseline and follow-up neuropsychometric testing as an inherent part of study design and patient outcome evaluation. This is particularly important as it is becoming increasingly evident that neurologic/neurocognitive decline seen in cancer patients is multifactorial and not due to the effects of radiation therapy alone. This has been demonstrated in recent publications implicating the potential signiﬁcant impact of other treatment modalities such as chemotherapy and/or surgery on neurologic/neurocognitive function [69–73]. In addition to the potential toxicity concerns of WBRT, a randomized trial from Patchell that showed no survival beneﬁt to WBRT is often quoted as a reason why WBRT should not be given. This phase III randomized trial of 95 patients compared surgery alone with surgery followed by WBRT for patients with a single brain metastasis [74]. The addition of WBRT to surgery decreased the chance for neurologic death (14% vs. 44%, p = 0.003), decreased local recurrence (10% vs. 46%, p < 0.01), and decreased tumor recurrence anywhere in the brain (XX% vs. 70%, p < 0.001). The majority of patients (61%) received WBRT at time of recurrence resulting in a large crossover of the observation arm to WBRT. Survival was not different, although the study’s end point was local control and thus it was not powered to demonstrate a survival advantage [75]. It is also important to note that patients who undergo SRS alone may be at higher risk for morbidity associated with brain tumor recurrence. A retrospective study from the University of Kentucky reviewed 36 patients with newly diagnosed brain metastases—22 had a single lesion [76]. These patients were treated with SRS alone. Seventeen of the 36 (44%) patients developed recurrent brain metastases. Of the 17 patients, 12 (71%) were symptomatic and 10 (59%) had neurologic deﬁcits.

Large Institutional/Multi-Institutional Results of SRS Alone Sneed reported on a UCSF retrospective study of outcomes of patients with one to four metastases treated by SRS alone compared with WBRT plus SRS [54]. Physician preference and referral patterns inﬂuenced the use of upfront WBRT. The patients had similar median survivals (11.1 months for SRS alone vs. 11.3 months for WBRT plus SRS) and local tumor control rates (71% vs. 79%). The incidence of distant brain metastasis was, however, signiﬁcantly higher in the SRS-only group versus WBRT and SRS (72% vs. 31%). Despite the higher rate of distant brain metastases, these metastases were controlled with salvage therapies including WBRT, partial brain radiotherapy, SRS, and surgery. The authors concluded that survival was not compromised by the omission of WBRT. Another retrospective paper from Hasegawa reviewed 172 patients with brain metastases (3.5 cm or less in diameter) managed by SRS alone [77]. One hundred twenty-one patients had follow-up imaging with 80% of patients having solitary lesions. The median survival was 8 months, and the local tumor control rate was 87%. At 2 years, the local control rate, distant brain control, and total intracranial control rates were 75%, 41%, and 27%, respectively. Tumor volume signiﬁcantly

187

predicted for local control (p = 0.02) with tumor volumes of at least 4 cm3 having local control rate of 49% at 1 and 2 years. Based on these results, the authors advocated avoiding WBRT for selected patients with one or two tumors with good control of their primary cancer, better KPS (90 or higher), and younger age (<60 years old). One study prospectively evaluated the role of SRS alone for patients with up to three brain metastases measuring less than 3 cm in maximum diameter [78]. The study consisted of 101 patients with a total of 155 lesions. At 1 year, the local control, distant brain freedom from progression, and overall brain freedom from progression rates were 91%, 53%, and 51%, respectively. The best predictor for improved overall brain freedom from progression was the presence of a single lesion and a greater than 2-year interval from primary diagnosis to diagnosis of brain metastases. The overall survival stratiﬁed by the RTOG RPA class was 13.4, 9.3, and 1.5 months for class I, II, and III, respectively (p < 0.0001). Median survival was 7.6 months. These results suggested that SRS alone can provide good local control but that marked distant failure can be expected. Another study from Pirzkall reviewed their experience of 236 patients with 311 brain metastases who underwent SRS alone (158 patients) or SRS with WBRT (73 patients). No difference was noted in terms of survival or local control. The SRS with WBRT group had a trend toward better survival compared with SRS alone (15.4 vs. 8.3 months; p = 0.08) for the patients with no evidence of extracranial disease [73]. The local control and distant brain control rates were worse with SRS alone. A multi-institutional retrospective review from 10 institutions analyzed the results of 569 patients with newly diagnosed single or multiple brain metastases treated initially with SRS alone versus SRS and WBRT [79]. For all RPA classes, the survival was comparable between the two treatment groups (RPA class I, 14 vs. 15 months; RPA class II, 8 vs. 7 months; RPA class III, 5 vs. 6 months) suggesting that upfront WBRT does not improve survival better than SRS alone.

Phase II/III Trials of SRS with or Without WBRT The Eastern Cooperative Oncology Group (ECOG) phase II feasibility trial of SRS alone for renal cell carcinoma, sarcoma, and melanoma patients with one to three brain metastases was presented at the 2004 ASCO meeting [80]. The objective of the study was to determine the radiographic and neurologic progression rates at 3, 6, and 9 months. The trial enrolled 36 patients; 32 were eligible by entry criteria. The SRS dose was based on tumor size (24, 18, or 15 Gy). The median survival was 8.3 months. Progression occurred in 41% of the cases. The authors concluded that omission of WBRT was associated with a high brain failure rate and should be approached judiciously. The interim report of the Japanese Radiation Oncology Study Group’s (JROSG) phase III study of SRS versus SRS plus WBRT (3000 cGy/10 fractions) was presented at the 2004 ASCO meeting [81]. This trial randomized patients with one to four brain metastases to SRS alone (61 patients) or SRS plus WBRT (59 patients). The primary end point of the study was survival with the secondary points including cause of death, freedom from new brain metastases, KPS preservation rate, local tumor control, and late radiation morbidity. No signiﬁcant

188

j.h. suh et al.

difference in survival, cause of death, or preservation of neurologic or systemic function was noted. At 12 months, the local control was 86% for the WBRT and SRS group versus 70% for the SRS alone group (p = 0.019). The freedom from new brain metastasis was also signiﬁcantly better at 6 months for the WBRT and SRS group (82%) compared with the SRS alone group (49%) (p = 0.003). The American College of Surgical Oncology Group (ACOSOG) initiated a phase III trial (ACOSOG Z0300) in February 2002, which randomized patients with one to three brain metastases to SRS versus SRS plus WBRT (3000 cGy in 12 fractions). The primary end point of the study was to determine if overall survival was equal or better for patients undergoing SRS alone than for patients undergoing SRS and WBRT. Secondary end points included evaluation of local failure, quality of life, functional independence, neurocognitive status, and toxicities associated with each treatment arm. Patients were stratiﬁed by age (≥60 vs. <60), extracranial disease (control ≤3 months vs. >3 months), and number of lesions (1 vs. >1). The goal was to accrue 480 patients. The study is temporarily suspended secondary to poor accrual.

Complications Associated with SRS Acute Complications Acute complications during the ﬁrst week after SRS are uncommon. Some of the complications include headache after frame removal, infection of the pin sites, nausea and vomiting, seizures, transient worsening of preexisting neurologic conditions, and fatigue. The incidence of severe headaches after frame removal is low [51]. Nausea/vomiting can be minimized if the dose to the area postrema is kept below 375 cGy [39]. The risk for seizures has been reported to range from 2% to 6% [41, 42, 51]. The risk for seizures is higher for patients with cortical lesions and for those with history of seizures. For these patients, anticonvulsants should be therapeutic or considered, and the steroid dose should be increased.

Subacute Complications These complications occur within the ﬁrst 6 months after SRS. Alopecia has been reported in 5.6% of patients [41]. These patients had superﬁcial tumors that resulted in a dose of 4.4 Gy to the scalp. Neurologic deterioration can occur in some patients, which is usually treated with steroids.

Chronic Complications Radiation necrosis represents the most serious chronic complication. In general, the risk of radiation necrosis increases with higher doses, prior radiation therapy, and larger tumor volumes. This entity can be difﬁcult to distinguish from tumor recurrence on MRI and may require the use of surgery, positron emission tomography (PET), and/or magnetic resonance spectroscopy (MRS). RTOG 9005 was a phase I/II trial to determine the maximum tolerated radiosurgery dose for patients with recurrent primary brain tumors or brain metastases treated previously with

TABLE 14-6. RTOG CNS toxicity criteria used for RTOG 9005. Grade 1 Grade 2 Grade 3 Grade 4

Grade 5

Mild neurologic symptoms; No medication required Moderate neurologic symptoms; Outpatient medication required (e.g., steroids) Severe neurologic symptoms; Outpatient or inpatient medication required Life-threatening neurologic symptoms (e.g., uncontrolled seizure, paralysis, coma); includes clinically and radiographically suspected radiation necrosis and histologically proven radiation necrosis at time of operation Death

Source: Hasegawa T, Kondziolka D, Flickinger JC, et al. Brain metastases treated with radiosurgery alone: an alternative to whole brain radiotherapy? Neurosurgery 2003; 52:1318–1326.

fractionated radiation therapy [82]. In this trial, the maximum tolerated doses were inversely correlated with the maximum tumor diameter. The doses were 24 Gy for a tumor ≤20 mm in diameter, 18 Gy for a tumor 21 to 30 mm in diameter, and 15 Gy for a tumor 31 to 40 mm in diameter. Of note, investigators were reluctant to escalate the dose for tumors ≤20 mm in diameter even though the maximum tolerated dose was not reached. The rate of radiation necrosis was 5%, 8%, 9%, and 11% at 6, 12, 18, and 24 months after SRS, respectively. Other factors that predicted for grades 3 to 5 neurotoxicity were tumor dose and KPS. Table 14-6 lists the RTOG CNS toxicity criteria used for RTOG 9005. The incidence of complications was also associated with tumor dose and KPS. The study also reported on physics and quality control assessments (MD/PD ratio, a measure of dose homogeneity, and PIV/TV ratio, a measure of conformity of the treated volume relative to target volume) that are useful for all tumors treated by SRS [83]. Valery reported on 377 patients with 760 lesions treated with linear accelerator–based SRS. Seven patients had severe complications including nine patients who developed radiation necrosis. The median tumor volume was 4.9 cm3. The median prescribed tumor dose was 15.6 Gy. The only factor that inﬂuenced the risk for radiation necrosis was the conformality index [84]. The general management strategy for radiation necrosis is to decrease the edema and necrosis. Usually, high doses of steroids are used to minimize neurologic deterioration. If the patient becomes symptomatic despite steroids, surgical resection can be considered. In some cases, hyperbaric oxygen has been used for patients who are poor surgical candidates, have multiple areas of radiation necrosis, or have a surgically inaccessible lesion. In a retrospective review of 40 patients, 90% reported subjective improvement and 80% had objective neurologic improvement [85]. Steroids were discontinued or decreased in two thirds of patients. A randomized trial is under way at the University of Cincinnati comparing surgery with hyperbaric oxygen.

Future Directions Based on the results of RTOG 9508, a randomized trial, RTOG 0525, is ongoing for NSCLC patients with one to three brain metastases. The primary end point of the study is survival.

14.

brain metastases

Another phase II study (PCYC-0224) is evaluating the use of motexaﬁn gadolinium, a radiation sensitizer, with WBRT (37.5 Gy in 15 fractions) with SRS for patients with one to four metastases. Others are incorporating newer imaging modalities such as MRS and PET to aid in planning or to deliver higher doses to certain regions of the target [86–87].

Conclusion Stereotactic radiosurgery has an important role in the management of brain metastases. Retrospective studies have shown the beneﬁts of SRS as sole treatment, as an adjunct to WBRT, and as salvage treatment after WBRT in some patients. Retrospective studies suggest that patients who undergo SRS or surgery have comparable outcomes. The results of RTOG 9508 provide level 1 evidence of a survival beneﬁt of SRS and WBRT compared with WBRT alone for patients with a single lesion. For patients with two to three lesions, the use of SRS and WBRT can be considered based on performance status, extent and activity of extracranial disease, and steroid use. The omission of WBRT has been driven by the concerns of the potential risks of WBRT and apparent lack of survival beneﬁt of WBRT rather than evidence from prospective clinical studies. The potential side effects of WBRT need to be balanced with the risk for neurologic and neurocognitive decline of uncontrolled brain metastases and the additional cost of more frequent scans. Recent phase III trials have shown that many patients with brain metastases have neurocognitive deﬁcits prior to WBRT. As the prognosis improves for cancer patients, the challenge of improving survival while limiting acute and long-term side effects will continue to make the management of brain metastases controversial. We encourage the participation of patients with brain metastases in clinical trials to improve outcomes and to help answer many important questions about management of this very common disease.

References 1. Wen PY, Loefﬂer JS. Management of brain metastasis. Oncology (Huntingt) 1999; 13(7):941–954. 2. Greenberg H, Chandler WF, Sandler HM. Brain metastases. In Greenberg H, Chandler WF, Sandler HM, eds. Brain Tumors. Oxford: Oxford University Press, 1999:299–317. 3. Lohr F, Pirzkall A, Hof H, et al. Adjuvant treatment of brain metastases. Semin Surg Oncol 2001; 20:50–56. 4. Zimm S, Wampler GL, Stablein D, et al. Intracerebral metastases in solid-tumor patients: natural history and results of treatment. Cancer 1981; 48(2):384–394. 5. Posner JB. Intracranial metastases. Neurologic Complications of Cancer, vol. 45. Philadelphia: FA Davis Company, 1995:77–110. 6. Nussbaum ES, Djalilian HR, Cho KH, et al. Brain metastasis: histology, multiplicity, surgery and survival. Cancer 1996; 78: 1781–1788. 7. Hazra T, Mullins GM, Lott S. Management of cerebral metastases from bronchogenic carcinoma. Johns Hopkins Med J 1972; 130: 377–383. 8. Chao JH, Phillips R, Nickson JJ. Roentgen-ray therapy of cerebral metastases. Cancer 1954; 7:682–689. 9. Chu FCH, Hilaris BB. Value of radiation therapy on the management of intracranial metastases. Cancer 1961; 4:577–581. 10. Borgelt B, Gelber R, Kramer S, et al. The palliation of brain metastasis: ﬁnal results of the ﬁrst two studies by the Radiation

11.

12.

13.

14.

15.

16.

17.

18. 19. 20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

189

Therapy Oncology Group. Int J Radiat Oncol Biol Phys 1980; 6:1–9. Gelber RD, Larson M, Borgelt BB, et al. Equivalence of radiation schedules for the palliative treatment of brain metastases in patients with favorable prognosis. Cancer 1981; 48:1749–1753. Borgelt B, Gelber R, Larson M, et al. Ultra-rapid high-dose irradiation schedules for the palliation of brain metastasis: ﬁnal result of the ﬁrst two studies by the Radiation Therapy Oncology Group. Int J Radiat Oncol Biol Phys 1981; 7:1633–1638. Kurtz JM, Gelber R, Brady LW, et al. The palliation of brain metastasis in a favorable patient population: a randomized clinical trial by the Radiation Therapy Oncology Group. Int J Radiat Oncol Biol Phys 1981; 7:891–895. Sause WT, Scott C, Krisch R, et al. Phase I/I1 trial of accelerated fractionation in brain metastasis RTOG 85-28. Int J Radiat Oncol Biol Phys 1993; 26:653–657. Murray KJ, Scott C, Greenberg HM, et al. A randomized phase III study of accelerated hyperfractionated versus standard in patients with unresected brain metastasis: a report of the Radiation Therapy Oncology Group (RTOG) 9104. Int J Radiat Oncol Biol Phys 1997; 39:571–574. Komarnicky LT, Philips TL, Martz K, et al. A randomized phase III protocol for the evaluation of misonidazole combined with radiation in the treatment of patients with brain metastases (RTOG 7916). Int J Radiat Oncol Biol Phys 1991; 20:53–58. Phillips TL, Scott CB, Leibel S et al. Results of a randomized comparison of radiotherapy and bromodeoxyuridine with radiotherapy alone for brain metastasis: a report of RTOG trial 89-05. Int J Radiat Oncol Biol Phys 1995; 33:339–348. Sneed PK, Larson DA, Wara WM. Radiotherapy for cerebral metastases. Neurosurg Clin N Am 1996; 7:505–515. Leksell L. The stereotactic method and radiosurgery of the brain. Acta Chir Scand 1951; 102:316–319. Sturm V, Kober B, Hover KH, et al. Stereotactic percutaneous single dose irradiation of brain metastases with a linear accelerator. Int J Radiat Oncol Biol Phys 1987; 13:279–282. Gaspar L, Scott C, Rotman M, et al. Recursive partitioning analysis (RPA) of prognostic factors in three Radiation Therapy Oncology Group (RTOG) brain metastases trials. Int J Radiat Oncol Biol Phys 1997; 37:745–751. Lagerwaald FJ, Levendag PC, Novak PJ, et al. Identiﬁcation of prognostic factors in patients with brain metastases: a review of 1292 patients. Int J Radiat Oncol Biol Phys 1999, 43:795–801. Patchell RA, Tibbs PA, Walsh JW, et al. A randomized trial of surgery in the treatment of single metastases to the brain. N Engl J Med 1990; 322:494–500. Noordijk EM, Vecht CJ, Haaxma-Reiche H, et al. The choice of treatment of single brain metastasis should be based on extracranial tumor activity and age. Int J Radiat Oncol Biol Phys 1994; 29:711–717. Mintz AH, Kestle J, Rathbone MP, et al. A randomized trial to assess the efﬁcacy of surgery in addition to radiotherapy in patients with a single cerebral metastasis. Cancer 1996; 78:1470–1476. Flickinger JC, Kondziolka D, Lunsford LD, et al. A multi-institutional experience with stereotactic radiosurgery for solitary brain metastasis. Int J Radiat Oncol Biol Phys 1994; 28:797–802. Yu C, Chen JCT, Apuzzo MLJ, et al. Metastatic melanoma to the brain: prognostic factors after gamma knife radiosurgery. Int J Radiat Oncol Biol Phys 2002; 52:1277–1287. Goyal LK, Suh JH, Reddy CA, et al. The role of whole brain radiotherapy and stereotactic radiosurgery on brain metastasis from renal cell carcinoma. Int J Radiat Oncol Biol Phys 2000; 47:1007–1012. Nieder C, Nestle U, Walter K, et al. Dose/effect relationships for brain metastases. J Cancer Res Clin Oncol 1998; 1214:346– 350.

190

j.h. suh et al.

30. Leibel SA, Sheline GE. Radiation therapy for neoplasms of the brain. J Neurosurg 1987; 66:1–22. 31. Larson DA, Flickinger JC, Loefﬂer JS, et al. The radiobiology of radiosurgery. Int J Radiat Oncol Biol Phys 1993; 25:557–561. 32. Mehta M, Noyes W, Craig B, et al. A cost-effectiveness and cost-utility analysis of radiosurgery vs. resection for single brain metastases. Int J Radiat Oncol Biol Phys 1997; 39:445–454. 33. Sperduto PW, Hall WA. Radiosurgery, cost-effectiveness, gold standards, the scientiﬁc method, cavalier cowboys, and the cost of hope. Int J Radiat Oncol Biol Phys 1996; 36:511–513. 34. Auchter RM, Lamond JP, Alexander E, et al. A multiinstitutional outcome and prognostic factor analysis of radiosurgery for resectable single brain metastasis. Int J Radiat Oncol Biol Phys 1996; 35:27–35. 35. Bindal AK, Bindal RK, Hess KR, et al. Surgery versus radiosurgery in the treatment of brain metastases. J Neurosurg 1996; 84:748–754. 36. O’Neill BP, Iturria NJ, Link MJ, et al. A comparison of surgical resection and stereotactic radiosurgery in the treatment of solitary brain metastases. Int J Radiat Oncol Biol Phys 2003; 55:1169– 1176. 37. Muacevic A, Kreth FW, Horstmann GA, et al. Surgery and radiotherapy compared with gamma knife radiosurgery in the treatment of solitary cerebral metastases of small diameter. J Neurosurg 1999; 91:35–43. 38. Loefﬂer JS, Shrieve DC. What is appropriate therapy for a patient with a single brain metastasis? Int J Radiat Oncol Biol Phys 1994; 29:915–917. 39. Weissman DE. Glucocorticoid treatment of brain metastases and epidural spinal cord compression: a review. J Clin Oncol 1988; 6:543–551. 40. Loefﬂer JS, Kooy HM, Wen PW, et al. The treatment of recurrent brain metastasis with stereotactic radiosurgery. J Clin Oncol 1990; 8:576–582. 41. Alexander E 3rd, Moriarty TM, Davis RB, et al. Stereotactic radiosurgery for the deﬁnitive, noninvasive treatment of brain metastasis. J Natl Cancer Inst 1995; 87:34–40. 42. Breneman JC, Warnick RE, Albright RE Jr, et al. Stereotactic radiosurgery for the treatment of brain metastases: results of a single institution series. Cancer 1997; 79:551–557. 43. Noël G, Proudhom MA, Valery CA, et al. Radiosurgery for reirradiation of brain metastases: results in 54 patients. Radiother Oncol 2001; 60:61–67. 44. Kihlstrom L, Karlsson B, Lindquist C. Gamma knife surgery for cerebral metastases. Implications for survival based on 16 years experience. Stereotact Funct Neurosurg 1993; 61:45–50. 45. Valentino V. The results of radiosurgical management of 139 single cerebral metastases. Acta Neurochir (Suppl) 1995; 63:95– 100. 46. Shiau CY, Sneed PK, Shu HK, et al. Radiosurgery for brain metastases: relationship of dose and pattern of enhancement to local control. Int J Radiat Oncol Biol Phys 1997; 37:375–383. 47. Matsuo T, Shibata S, Yasunaga A, et al. Dose optimization and indication of Linac radiosurgery for brain metastases. Int J Radiat Oncol Biol Phys 1999; 45:931–939. 48. Schoeggl A, Kitz K, Ertl A, et al. Prognostic factor analysis for multiple brain metastases after gamma knife radiosurgery: results in 97 patients. J Neurooncol 1999; 42:168–175. 49. Simonova G, Liscak R, Novotny J. Solitary brain metastases treated with the Leksell gamma knife: prognostic factors for patients. Radiother Oncol 2000; 57:207–213. 50. Chidel MA, Suh JH, Reddy CA, et al. Application of recursive partitioning analysis and evaluation of the use of whole brain radiation among patients treated with stereotactic radiosurgery for newly diagnosed brain metastases. Int J Radiat Oncol Biol Phys 2000; 47:993–999.

51. Joseph J, Adler JR, Cox RS, et al. Linear accelerator-based stereotaxic radiosurgery for brain metastases: the inﬂuence of number of lesions on survival. J Clin Oncol 1996; 14:1085–1092. 52. Shehata MK, Young B, Reid B, et al. Stereotactic radiosurgery of 468 brain metastases < or = 2 cm: implications for SRS dose and whole brain radiation therapy. Int J Radiat Oncol Biol Phys 2004; 59:87–93. 53. Pirzkall A, Debus J, Lohr F, et al. Radiosurgery alone or in combination with whole-brain radiotherapy for brain metastases. J Clin Oncol 1998; 16:3563–3569. 54. Sneed PK, Lamborn KR, Forstner JM, et al. Radiosurgery for brain metastases: is whole brain radiotherapy necessary? Int J Radiat Oncol Biol Phys 1999; 43:549–558. 55. Sanghavi SN, Miranpuri SS, Chappell R, et al. Radiosurgery for patients with brain metastases: a multi-institutional analysis, stratiﬁed by the RTOG recursive partitioning analysis method. Int J Radiat Oncol Biol Phys 2001; 51:426–434. 56. Kondziolka D, Patel A, Lunsford LD, et al. Stereotactic radiosurgery plus whole brain radiotherapy versus radiotherapy alone for patients with multiple brain metastases. Int J Radiat Oncol Biol Phys 1999; 45:427–434. 57. Chougule P, Burton-Williams M, Saris S, et al. Randomized treatment of brain metastasis with Gamma Knife radiosurgery, whole brain radiotherapy or both. Int J Radiat Oncol Biol Phys 2000; 48(Suppl):114. 58. Andrews DW, Scott CB, Sperduto PW, et al. Whole brain radiation therapy with or without stereotactic radiosurgery boost for patients with one to three brain metastases: phase III results of the RTOG 9508 randomised trial. Lancet 2004; 363: 1622–1672. 59. DeAngelis LM, Posner LB. Radiation-induced dementia in patients cured brain metastases. Neurology 1989; 39:789–796. 60. So NK, O’Neill BP, Frytak S, et al. Delayed leukoencephalopathy in survivors with small cell lung cancer. Neurology 1987; 37:1198– 1201. 61. Cox JD, Stanley K, Petrovich Z, et al. Cranial irradiation in cancer of the lung of all cell types. JAMA 1981; 245:469–472. 62. Russell AH, Pajak TE, Selim HM, et al. Prophylactic cranial irradiation for lung cancer patients at high risk for development of cerebral metastasis: results of a prospective randomized trial conducted by the Radiation Therapy Oncology Group. Int J Radiat Oncol Biol Phys 1991; 21:637–643. 63. Umsawasdi T, Valdivieso M, Chen TT, et al. Role of elective brain irradiation during combined chemoradiotherapy for limited disease non-small cell lung cancer. J Neurooncol 1984; 2:253– 259. 64. Mamon HJ, Yeap BY, Reblando J, et al. High risk of brain metastases in surgically staged IIIA non-small cell lung cancer patients treated with surgery, chemotherapy, and radiation. J Clin Oncol 2005; 23:1530–1537. 65. Meyers CA, Smith JA, Bezjak A, et al. Neurocognitive function and progression in patients with brain metastases treated with whole-brain radiation and motexaﬁn gadolinium: results of a randomized phase III trial. J Clin Oncol 2004; 22:157–165. 66. Regine WF, Scott C, Murray K, et al. Neurocognitive outcome in brain metastases patients treated with accelerated-fractionation vs. accelerated hyperfractionation radiotherapy: an analysis from the Radiation Therapy Oncology Group 91-04. Int J Radiat Oncol Biol Phys 2001; 51:711–717. 67. Chang EL, Wefel JS, Meyers CA, et al. Prospective neurocognitive evaluation of patients with 1 to 3 newly diagnosed brain metastases treated with stereotactic radiosurgery alone. Int J Radiat Oncol Biol Phys 2004; 60:S164–S165. 68. Regine WF, Schmitt FA, Scott CB, et al. Feasibility of neurocognitive outcome evaluations in patients with brain metastases in a multi-institutional cooperative group setting: results of Radiation

14.

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

brain metastases

Therapy Oncology Group trial BR-0018. Int J Radiat Oncol Biol Phys 2004; 58:1346–1352. Sonderkaer S, Schmiegelow M, Carstensen H, et al. The long-term neurologic outcome of childhood brain tumors treated by surgery alone. J Clin Oncol 2003; 21:1347–1351. Tchen N, Juffs HG, Downie FP, et al. Cognitive function, fatigue, and menopausal symptoms in women receiving adjuvant chemotherapy for breast cancer. J Clin Oncol 2003; 21:4175– 4183. Schagen SB, van Dam FSAM, Muller MJ, et al. Cognitive deﬁcits after postoperative adjuvant chemotherapy for breast cancer. Cancer 1999; 85:640–650. van Dam FSAM, Schagen SB, Muller MJ, et al. Impairment of cognitive function in women receiving adjuvant treatment for high risk breast cancer: high-dose versus standard dose chemotherapy. J Natl Cancer Inst 1998; 90:210–218. Brezden CB, Phillips KA, Abdolell M, et al. Qaeda to function and breast cancer patients receiving adjuvant chemotherapy. J Clin Oncol 2000; 18:2695–2701. Patchell RA, Tibbs PA, Regine WF, et al. Postoperative radiotherapy in the treatment of single metastases to the brain: a randomized trial. JAMA 1998; 280:1485–1489. Patchell RA, Regine WF. The rationale for adjuvant whole brain radiation therapy with radiosurgery in the treatment of single brain metastases. Technol Ca Res Treat 2003; 2:111–115. Regine WF, Huhn JL, Patchell RA, et al. Risk of symptomatic brain tumor recurrence and neurologic deﬁcit after radiosurgery alone in patients with newly diagnosed brain metastases: results and implications. Int J Radiat Oncol Biol Phys 2002; 52:333– 338. Hasegawa T, Kondziolka D, Flickinger JC, et al. Brain metastases treated with radiosurgery alone: an alternative to whole brain radiotherapy? Neurosurgery 2003; 52:1318–1326. Lutterbach J, Cyron D, Henne K, et al. Radiosurgery followed by planned observation in patients with one to three brain metastases. Neurosurgery 2003; 52:1066–1074.

191

79. Sneed PK, Suh JH, Goesch SJ, et al. A multi-institutional review of radiosurgery alone versus radiosurgery with whole brain radiotherapy as the initial management of brain metastases. Int J Radiat Oncol Biol Phys 2002; 53:519–526. 80. Manon RR, Oneill A, Mehta M, et al. Phase II trial of radiosurgery (RS) for 1-3 newly diagnosed brain metastases from renal cell, melanoma, and sarcoma: An Eastern Cooperative Oncology Group Study (E6397). Proceedings of the 40th ASCO meeting [abstract no. 1507]. J Clin Oncol 2004. 81. Aoyama H, Shirato H, Nakagawa K, et al. Interim report of the JROSG99-1 multi-institutional randomized trial, comparing radiosurgery alone versus radiosurgery plus whole brain irradiation for 1-4 brain metastases. Proceedings of the 40th ASCO meeting [abstract no. 1506]. J Clin Oncol 2004. 82. Shaw E, Scott C, Souhami L, et al. Single dose radiosurgical treatment of recurrent previously irradiated primary brain tumors and brain metastases: ﬁnal report of RTOG protocol 90-05. Int J Radiat Oncol Biol Phys 2000; 47:291–298. 83. Shaw E, Scott C, Souhami L, et al. Radiosurgery for the treatment of previously irradiated recurrent primary brain tumors and brain metastases: initial report of the Radiation Therapy Oncology Group protocol 90-05. Int J Radiat Oncol Biol Phys 1996; 34: 647–654. 84. Valery CA, Cornu P, Noël G, et al. Predictive factors of radiation necrosis after radiosurgery for cerebral metastases. Stereotact Funct Neurosurg 2003; 81:115–119. 85. Warnick RE, Darakchiev BJ, Breneman JC. Stereotactic radiosurgery for patient with solid brain metastases: current status. J Neuro-oncol 2004; 69:125–137. 86. Ma L, Chin LS, DiBiase SJ, et al. Concomitant boost of stratiﬁed target area with gamma knife radiosurgery: a treatment planning study. Am J Clin Oncol 2003; 26:e100–e105. 87. Levivier M, Massager N, Wikler D, et al. Use of stereotactic PET images in dosimetry planning of radiosurgery for brain tumors: clinical experience and proposed classiﬁcation. J Nucl Med 2004; 45:1146–1154.

1 5

Metastatic Brain Tumors: Surgery Perspective Raymond Sawaya and David M. Wildrick

Introduction Metastatic brain tumors occur more frequently than other intracranial neoplasms and are a serious complication of systemic cancer [1]. Their annual incidence exceeds 100,000 [2, 3] and is on the rise as patients live longer from improved treatments for extracranial cancer [4, 5]. Thus, patients with brain metastases constitute a signiﬁcant fraction of the case load in the neurosurgery services at many oncologic care centers. Early in the 1900s, the results of surgical treatment of single brain metastases were disappointing [6], with patients seldom showing a median survival time of longer than 6 months [7]. Although this was longer than the 4- to 6-week survival time typical of patients with untreated brain metastases, the surgical morbidity was high (15% to 50%) [8], and whole-brain radiation therapy (WBRT) alone was frequently favored for management of patients with brain metastases. Since then, the Radiation Therapy Oncology Group (RTOG) has demonstrated that treatment of brain metastasis patients with WBRT alone can extend their median survival time to 16 weeks [9–11]. In addition, two independent, prospective randomized studies showed that surgery plus WBRT offered a survival advantage superior to WBRT alone in such patients [12, 13]. Since Leksell demonstrated the utility of focused beams of high-energy X-rays in the ablation of intracranial tumors [14, 15], the technique of stereotactic radiosurgery (SRS) has continued to evolve. During the 1990s, SRS was increasingly used in the management of brain metastases, with some investigators recently claiming that SRS alone produces results similar to those obtained by combining SRS and WBRT, or potentially even equivalent to the outcome with surgical resection [16–20]. With this in mind and because of the ease of use of SRS and the perception that it costs less than conventional surgery, some have even suggested that most metastatic brain tumors should be managed exclusively with SRS. This controversial notion has already led to treatment of some series of brain metastasis patients with SRS alone followed by observation [18, 19]. At The University of Texas M. D. Anderson Cancer Center (“M. D. Anderson”), SRS is regarded as a specialized tool to be used judiciously, as a given patient’s situation dictates, rather than as a generalized treatment modality for brain metastases. We continue to see surgical resection as playing the central role in the management of patients with a limited number of brain

metastases [4, 21–24]. In this chapter, we present our view of the relative roles of surgery and SRS in terms of issues concerning patient selection, treatment outcome, cost-effectiveness of the treatment, and the patients’ quality of life. These differences are summarized in Table 15-1 [25–27].

Patient Selection When a patient presents with a brain metastasis and symptoms of mass effect, there is seldom a dispute that the lesion should be removed surgically. Similarly, when a patient with a brain metastasis presents in too poor a medical condition to be a surgical candidate (or declines surgery), SRS represents a logical treatment modality. Other patients with single brain metastases can be sorted into three groups. The ﬁrst group has relatively large lesions (exceeding 3 cm in maximum diameter) that can only be effectively removed by surgical resection. Treatment of these tumors with SRS is not effective because the radiation dose must be reduced as the tumor size increases to prevent damage to surrounding brain tissue. This relationship was clearly demonstrated by Mehta et al. [25], who showed that with SRS, the rate of complete response (CR; total disappearance of the magnetic resonance [MR] image of the lesion after SRS) falls off dramatically with tumor volume such that if a tumor 2 cm3 in volume has a 50% CR rate, an 8- to 9-cm3 lesion shows about a 20% CR rate (Fig. 15-1). Patients in the second group have very small lesions (less than 10 mm in maximum diameter) that are not surgically accessible and are located deep within the brain. In this situation, SRS provides an effective alternative to resection, superior to WBRT, which was the only treatment previously available. The third group of patients are those who have metastases that are less than 3 cm in maximum diameter and are surgically accessible. Whether surgery or SRS is the best treatment for these lesions is the subject of much current debate. The 3-cm upper size limit referred to above is probably too high for adequate SRS treatment. At M. D. Anderson, a recent study of 153 brain metastases from melanoma treated with SRS [28] showed that the 1-year local control rate of smaller tumors with a maximum diameter of no more than 1.5 cm (volume = 2 cm3) was superior to that of larger lesions (75.2% and 42.3%, respectively; p < 0.05). Moreover, another

193

194

r. sawaya and d.m. wildrick

TABLE 15-1. Differences between surgery and stereotactic radiosurgery in the treatment of brain metastasis.

Patient selection Tissue diagnosis Lesion size Surgical candidate Treatment outcome Tumor status Local control Local recurrence rate Median survival time Complications Presentation Major neurological Mortality (30-day)†

Surgical resection

Stereotactic radiosurgery

Conﬁrms the lesion is a tumor Large, ≥1.5 cm*; especially if there is mass effect Yes

Cannot conﬁrm the lesion is a tumor Small, ≤2.0 cm*; signiﬁcant mass effect absent No (or patient declines surgery)

Removed (≥94% on average) [4] 85% for >40 months (up to 5 years) [23] 8% [23] to 12% [4] 10 [13] to 16.4 [23] months

Not removed 85% at 12 months; 65% at 24 months [25] 30% [26] to 47% [27] 7.5 [23] to 14 [16] months

Usually immediate

Frequently delayed; necrosis may necessitate surgical resection 25% in eloquent brain (RTOG grade 3) [24] 1.8%

Cost-effectiveness

7% in eloquent brain; 6% overall [4] <2% Based on 3-day hospital stay

Quality of life Relief from mass effect Steroid use Follow-up visits Patient debilitation

Immediate Tapered off over 2–4 weeks Few Not debilitated at home

Based on 1-day outpatient stay; includes no costs of maintenance steroids, follow-up ofﬁce visits, or follow-up MRI Delayed Can last for months; may cause dependence Many Debilitated at home

*Maximum diameter. †

Occurring within 30 days after the procedure.

recent study at M. D. Anderson used SRS to treat brain metastases from different primary tumor types [26] and found a 1year actuarial local control rate of 86% for lesions that were 1 cm or less in maximum diameter (0.5 cm3) but only a 56% local control rate for lesions larger than this (p = 0.0016). Furthermore, a slightly earlier study by a radiosurgery group in Pittsburgh [29] included only brain metastases that were no

100

GTR

Treatment Outcome

80

Survival

Percent

60

40 CR 20

0 0

larger than 2.5 cm in maximum diameter and found that treatment with SRS plus WBRT was superior to WBRT-alone. It is important to note that control rates similar to surgery for lesions as large as 2 cm (volume ∼5 cm3) can be achieved with the addition of fractionated WBRT [30]. Collectively, these ﬁndings clearly indicate that for effective use of SRS alone (i.e., without the addition of WBRT), a cutoff point in brain metastasis diameter that is lower than 3 cm is warranted.

2

4

6

8

10

12

14

16

18

20

Tumor Size (cc) FIGURE 15-1. Tumor size versus response to radiosurgery and surgery for brain metastases. Comparison of a representative radiosurgery complete response (CR) curve [25] with a surgery gross-total resection (GTR) curve (average percentage, 94% from Table 15-2), showing that CR rate with radiosurgery is tumor size dependent, whereas GTR rate with surgery is not. (From Sawaya R. Surgical treatment of brain metastases. Clin Neurosurg 1999; 45:41–47. Used with permission.)

The controversy remains as to whether surgery or SRS is more appropriate and effective to treat patients in the group with single brain metastases that are surgically accessible and range between 1 and 3 cm in maximum diameter. Because no one has yet been able to complete a prospective randomized study comparing treatment of such patients with surgical resection and SRS, the question of which of the two is more effective lacks a deﬁnitive answer. In lieu of this, many retrospective studies have been performed comparing surgical resection and SRS for brain metastasis treatment. Among the more carefully constructed studies of this type are those of Auchter et al. [16], Bindal et al. [23], and Cho et al. [31]. Auchter and coworkers [16] performed a multiinstitutional retrospective outcome and prognostic factor analysis of patients with single cerebral metastases who were treated with SRS plus WBRT. From their database of 533 patients with brain metastases treated with SRS and WBRT, they selected 122 patients who fulﬁlled the criteria for surgical resection established in the prospective randomized trial of Patchell et al.

15.

195

metastatic brain tumors: surgery perspective

[13], including having a single brain metastasis that was surgically resectable (but no urgent need for surgery), no prior radiotherapy or surgical treatment, independent functional status (a Karnofsky Performance Scale [KPS] score >70), and a non– radiosensitive tumor. They then compared the outcome of these 122 patients with that of the patients in the prospective randomized surgical series of Patchell et al. [13] and Noordijk et al. [12]. The actuarial median survival time after treatment with SRS plus WBRT was 56 weeks compared with 40 weeks [13] and 43 weeks [12], respectively, after surgery plus WBRT. The median duration of functional independence was 44 weeks after SRS and WBRT [16] compared with 38 weeks [13] and 33 weeks [12], respectively, after surgery and WBRT. These results indicated to Auchter et al. [16] that SRS plus WBRT produced outcomes for patients with single brain metastases that were comparable with, if not better than, surgery plus WBRT. Cho and coworkers [31] retrospectively reviewed a series of 225 single brain metastases in patients treated with either WBRT alone, SRS plus WBRT, or surgery and WBRT. There was a similar distribution of prognostic factors such as age, sex, KPS score, and metastasis location in these three groups except for extracranial cancer, which was 14% higher in the group treated with SRS and WBRT. The actuarial median survival times for patients treated with SRS and WBRT (9.8 months) and surgery plus WBRT (10.5 months) were nearly identical, but both were signiﬁcantly better than for those treated with WBRT alone (3.8 months). Other more recent retrospective studies have also labeled resection and SRS as equally effective in the treatment of small to moderately sized brain metastases [32]. At M. D. Anderson, Bindal and coworkers [23] also retrospectively compared surgical resection and SRS treatment of brain metastases. Thirty-one patients treated with SRS were followed prospectively, and 62 patients treated with conventional surgery were retrospectively matched to those in the SRS group. Patients were matched according to primary tumor histology, extent of systemic disease, preoperative KPS score, time to brain metastasis, number of brain metastases, and patient age and sex. WBRT treatment was similar in both groups, and patient eligibility criteria for SRS were the same as for surgery. Lesions treated by SRS were limited to those <3 cm in maximum diameter, and 81% of them were deemed surgically resectable. There was a statistically signiﬁcant survival advantage in the surgically treated group (16.4 months median survival) relative to the group treated with SRS (7.5 months median survival) (p = 0.0009 by multivariate analysis). In contrast to the conclusions of Auchter et al. [16] and Cho et al. [31], Bindal’s group [23] concluded that surgery was superior to SRS in clinically similar patients in terms of survival, local recurrence, and morbidity. Thus, determination of whether treatment of brain metastases with SRS is better than, equivalent to, or worse than surgical resection must await a phase III prospective trial comparing SRS with conventional surgery for the treatment of single brain metastases.

Tumor Local Control and Recurrence When brain metastases are removed surgically, local control is produced by total removal of the tumor and elimination of surrounding edema (Fig. 15-2). At M. D. Anderson, gross-total

FIGURE 15-2. (A) Preoperative and (B) postoperative gadolinium contrast-enhanced MR images of a brain metastasis in a patient with non–small cell lung cancer. In each panel, the left and right images are T1- and T2-weighted, respectively. (From Sawaya R. Surgical treatment of brain metastases. Clin Neurosurg 1999; 45:41–47. Used with permission.)

resection (equivalent to complete response in radiosurgery) of metastatic tumors is achieved in 90% to 96% of instances, in both eloquent and noneloquent brain regions (Table 15-2). With surgery, local control (absence of recurrent tumor growth after gross-total resection) typically persists at an 85% level at 1 year and remains there for an interval ranging from 40 months to 5 years (Fig. 15-3). TABLE 15-2. Gross-total resections performed for metastatic tumors in different brain regions (gross-total resection is equivalent to a complete response in radiosurgery). Metastasis location (Tumor functional grade*)

No. of patients

Noneloquent (I) Near-eloquent (II) Eloquent (III) Total

79 61 54 194

GTR (%)

75 (95) 55 (90) 52 (96) Average, 94

GTR, gross-total resection. *Grade according to functional location. Source: Adapted from Sawaya R. Surgical treatment of brain metastases. Clin Neurosurg 1999; 45:41–47. Used with permission.

Percent Free From Local Recurrence

196

r. sawaya and d.m. wildrick

100 MDACC Surgery 80 BRIGHAM Radiosurgery 60 40 MDACC Radiosurgery 20 0 0

10

20

30

40

Follow-up Time (months) FIGURE 15-3. Freedom from local recurrence (local control) of brain metastases with radiosurgery and with surgery. Curve showing time from radiosurgical treatment to local failure (BRIGHAM Radiosurgery) in 42 patients treated at Brigham and Women’s hospital [25] superimposed on curves showing time from surgical resection (MDACC Surgery) to local recurrence in 62 patients and time to local failure in 31 patients radiosurgically treated (MDACC Radiosurgery) at The University of Texas M. D. Anderson Cancer Center [23]. (From Sawaya R. Surgical treatment of brain metastases. Clin Neurosurg 1999; 45: 41–47. Used with permission.)

In contrast, when brain metastases are treated with SRS, it is common for the tumor to remain visible on computed tomography (CT) or MR images. Local tumor control with SRS is less stringently deﬁned and, in addition to tumor regression includes lesions remaining unchanged in size as well as those undergoing no more than a 25% increase in size/volume above the baseline measurement. This makes it difﬁcult and misleading to compare values published for local control of brain metastases by surgical resection and SRS. In some patients who have a brain metastasis that is stable on MR images or increases in size by less than 25%, surgery may be required to relieve neurologic deﬁcits resulting from mass effect and persistent edema (Fig. 15-4) [33].

After surgical resection of brain metastases at M. D. Anderson, we have observed local tumor recurrence at a rate ranging from about 8% to 12% [4, 23], which is consistent with the observations of others. In patients treated for brain metastases with SRS, similar 1-year local control failure rates (6% to 15%) are often reported [16, 18, 19, 34–38]. These numbers reported for SRS are ﬂawed not only because of the less stringent deﬁnition of local control with SRS but also because tumor recurrence in brain metastasis patients is a function of survival time, and many radiosurgical series of these patients report such short survival intervals (8 months on average) that failure of local tumor control cannot be reliably determined. Treatment of brain metastasis patients with SRS at Brigham and Women’s Hospital [25] gave an 85% local control rate at 1 year, but this dropped steadily to 65% after 2 years (Fig. 15-3). More recent series of brain metastasis patients treated with SRS have presented overall recurrence values (30% to 47%) [26, 27] that are much higher than those typical of surgically treated patients. These numbers may be more realistic, reﬂecting a trend toward increased follow-up monitoring of patients, with more routine neuroimaging [28].

Complications Complications arising from surgical resection of brain metastases are usually evident immediately. Deleterious effects of SRS treatment are frequently delayed and may be missed in follow-up visits, leading to an underreporting of complications. A recent study using SRS alone to treat brain metastases showed a complication rate of only 8% [19]; nevertheless, 70% of the complications were acute and included increases in seizures and worsening of preexisting neurologic symptoms. Similarly, in a study of recurrence of brain metastasis after SRS, Regine et al. [27] observed that recurrence was symptomatic in 71% of patients and was associated with a neurologic deﬁcit in 59%. With the advent of modern neurosurgical techniques, especially intraoperative stereotaxy and ultrasonography, few metastatic brain tumors are surgically inaccessible. In a study of 194 patients at M. D. Anderson [4], the 30-day morbidity rate for any major neurologic deﬁcit after brain metastasis resection was 6% (Table 15-3). Even for resections in eloquent brain regions, this value only rose to 7%. The common perception that SRS is safer to use than conventional surgery to treat brain metastases within or near eloquent brain areas may not be warranted. We recently performed a retrospective study of neurologic complications in patients

TABLE 15-3. Major neurologic complications of surgery for metastases in different brain regions. Metastasis location

FIGURE 15-4. CT scan of a patient with non–small cell lung cancer who had a 3-cm brain metastasis in the right anterior parietal lobe (left) that was treated with radiosurgery. Three months later, the tumor shrank by 23%, but the edema and the patient’s symptoms were unchanged (right). The patient subsequently underwent a craniotomy and total resection of the mass. The results of the surgery are shown in Fig. 15-2. (From Sawaya R. Surgical treatment of brain metastases. Clin Neurosurg 1999; 45:41–47. Used with permission.)

Noneloquent Near-eloquent Eloquent Average major neurologic deﬁcit rate

No. patients*

No. complications (%)

79 61 54

1 (1) 6 (10) 4 (7) (6)

*Total = 194 patients. Source: Adapted from Sawaya R. Surgical treatment of brain metastases. Clin Neurosurg 1999; 45:41–47. Used with permission.

15.

197

metastatic brain tumors: surgery perspective 100

A.Pre

B.Post 80

% of Patients

60

28≤ 26 24 22 20 18 16 14 12 10 8

6

4

2

40

20 0 0 2

4

6

8 10 12 14 16 18 20 22 24 26 ≥28

Time (Weeks) FIGURE 15-5. Graphs depicting the duration (in weeks) of steroid facilitated tapering off of steroid use within 2 to 4 weeks. (From Vecil treatment from the time of conventional resection of the index lesion GG, Suki D, Maldaun MV, et al. Resection of brain metastases pre(Time 0). The duration of continuous treatment is shown both before viously treated with stereotactic radiosurgery. J Neurosurg 2005; 102(2): (A) and after (B) conventional resection. Note that surgical resection 209–215. Used with permission.)

treated with SRS for brain metastases at M. D. Anderson and found that such complications (including radiation necrosis, seizures, and neurologic symptom development or worsening) were highest for tumors in eloquent brain, reaching 25% for complications of RTOG grade 3 and above [24]. At M. D. Anderson, the current mortality rate for surgical resection of brain metastases is less than 2%. Although it has been claimed that SRS poses little risk of permanent neurologic morbidity in brain metastasis treatment, mortalities have occurred. In the study by Bindal et al. [23], spontaneous intratumoral hemorrhage occurred in 3 of 31 patients treated with SRS, causing the death of one. Kondziolka and coworkers [29] compared brain metastasis treatment with SRS plus WBRT versus WBRT alone and reported a 1.8% mortality rate within 4 weeks after SRS. Thus, it appears that mortality rates for SRS and surgical resection of brain metastases are similar. Another potential source of complications when treating brain metastases with SRS stems from its inability to provide a tissue biopsy for pathologic diagnosis. Because 5% to 11% of patients with a prior history of cancer and a brain lesion that appears consistent with a metastasis may actually have nonmetastatic disease (primary brain tumor, abscess, multiple sclerosis plaque, or a hemorrhage) [13, 39], SRS treatment of such lesions can lead to an adverse outcome. With surgical resection of brain metastases, it is easy to histologically conﬁrm that the lesion is metastatic cancer, and this situation does not arise.

Cost Effectiveness SRS is frequently promoted as a lower-cost alternative to conventional surgery in the management of brain metastases. Two publications have emerged using retrospective analyses to compare relative costs of SRS and surgery in attempt to bolster this position [40, 41]. Unfortunately, these cost analyses have relied on surgical expenses and data on length of hospital stay from publications on patient series that were already more than a decade old, such as that of Patchell et al. [13], and have not taken into account the drastic cost-cutting measures adopted by

most hospitals and surgeons around the country during the past decade. Moreover, these cost-comparison studies [40, 41] make the assumption that SRS and surgery are “equiefﬁcacious” in the management of patients with cancer metastatic to the brain. They also assign inferior median survival values to studies employing surgical resection plus WBRT for brain metastasis treatment relative to SRS, thereby inﬂating the average cost of surgery. Because there have been no prospective randomized studies comparing SRS and surgery with respect to patient outcomes, there is no accurate statistical basis for the presumed “equiefﬁcaciousness” or the survival numbers put forth. To accurately reﬂect the total costs of SRS and surgical resection for brain metastases, the expenses involved in patient follow-up must be included. Typically, with SRS, only the 1-day cost of treatment is considered in cost-effectiveness studies, with follow-up being ignored. At M. D. Anderson, follow-up costs after surgery for brain metastases have been severely reduced because most patients are now discharged from the hospital after 3 days. Follow-up expenses after SRS are more signiﬁcant than for surgery because many routine magnetic resonance imaging (MRI) studies and ofﬁce visits are required to monitor the delayed effects of the radiation. These delayed effects require extended administration of medications such as dexamethasone, which becomes costly. In our recent study of patients undergoing resection of brain metastases after failure of SRS to control their lesions [33], after their SRS treatment and prior to surgery, 40% of 53 patients required continuous steroid administration for 12 weeks, which is consistent with steroid dependency and is not a trivial expense. After surgery, more than 95% of these patients were able to cease steroid usage within 2 to 4 weeks (Fig. 15-5).

Quality of Life Surgical resection of a brain metastasis usually provides immediate relief of neurologic symptoms produced by mass effect of the tumor and edema surrounding it. Thus, the patient can be

198

r. sawaya and d.m. wildrick

on his feet that much sooner, and few follow-up visits are required. With SRS treatment of a brain metastasis, radiation damage to the tumor frequently takes time to develop, and the therapeutic effect with consequent resolution of neurologic symptoms may be delayed for days or weeks. Multiple followup visits are required to monitor the tumor’s response to SRS with MRI or CT. For symptom relief, the patient must receive continuous steroid administration that can last for months, while remaining debilitated at home.

Conclusion The debate continues whether surgical resection or SRS is better in managing tumors metastatic to the brain. Settling the issue of which modality affords patients the longer median survival time awaits the outcome of a prospective randomized trial comparing the two. Both methods have clear beneﬁts. Surgery has the advantages of immediate resolution of mass effect, procurement of tissue for pathologic diagnosis, and lack of the risk of radiation necrosis [42]. SRS carries a decreased risk of hemorrhage and infection, has no risk of tumor seeding, and is less invasive, potentially less costly, and requires a shorter hospital stay than standard craniotomy. Disadvantages of SRS include potential radiation necrosis or exacerbation of peritumoral edema and a requirement for long-term steroid administration [43, 44]. In our experience at M. D. Anderson, we believe that the size and location of the metastases along with the patient’s clinical presentation can be used to recommend one modality or the other. We almost always favor surgery in the case of brain metastases larger than 3 cm in maximum diameter, whereas deeply located lesions smaller than 1 to 1.5 cm in maximum diameter are usually treated with SRS. If the lesions lend themselves to treatment by either method, we let the patient’s symptoms be the guide; lesions that produce symptoms are more frequently treated surgically, and lesions that do not can be treated with SRS. Of course, this approach may be modiﬁed depending upon the patient’s systemic cancer status or medical condition. Patients who cannot tolerate surgery, who have progressive systemic disease, or who are expected to live for less than 3 months are treated with SRS. The best management strategy for patients with brain metastases should involve the complementary use of surgery, SRS, and WBRT. Acknowledgment. We thank the Barbara Falik Research Fund for helping make this work possible.

References 1. Posner JB, Chernik NL. Intracranial metastases from systemic cancer. Adv Neurol 1978; 19:579–592. 2. Tsukada Y, Fouad A, Pickren JW, Lane WW. Central nervous system metastasis from breast carcinoma. Autopsy study. Cancer 1983; 52(12):2349–2354. 3. Sawaya R, Bindal RK, Lang FF, Abi-Said D. Metastatic brain tumors. In: Kaye AH, Laws ER, eds. Brain Tumors: An Encyclopedic Approach, 2nd ed. Edinburgh: Churchill Livingstone, 2001:999–1026. 4. Sawaya R. Surgical treatment of brain metastases. Clin Neurosurg 1999; 45:41–47.

5. Young RF. Radiosurgery for the treatment of brain metastases. Semin Surg Oncol 1998; 14(1):70–78. 6. Grant FC. Concerning intracranial malignant metastases: their frequency and the value of surgery in their treatment. Ann Surg 1926; 84:635–646. 7. Cairncross JG, Posner JB. The management of brain metastases. In: Walker MD, ed. Oncology of the Nervous System. Boston: Martinus Nijhof, 1983:341–377. 8. Horwitz NH, Rizzoli HV. Postoperative complications of intracranial neurological surgery. Baltimore: Williams & Wilkins, 1982. 9. Borgelt B, Gelber R, Kramer S, et al. The palliation of brain metastases: ﬁnal results of the ﬁrst two studies by the Radiation Therapy Oncology Group. Int J Radiat Oncol Biol Phys 1980; 6(1):1–9. 10. Diener-West M, Dobbins TW, Phillips TL, Nelson DF. Identiﬁcation of an optimal subgroup for treatment evaluation of patients with brain metastases using RTOG study 7916. Int J Radiat Oncol Biol Phys 1989; 16(3):669–673. 11. Sneed PK, Larson DA, Wara WM. Radiotherapy for cerebral metastases. Neurosurg Clin N Am 1996; 7(3):505–515. 12. Noordijk EM, Vecht CJ, Haaxma-Reiche H, et al. The choice of treatment of single brain metastasis should be based on extracranial tumor activity and age. Int J Radiat Oncol Biol Phys 1994; 29(4):711–717. 13. Patchell RA, Tibbs PA, Walsh JW, et al. A randomized trial of surgery in the treatment of single metastases to the brain. N Engl J Med 1990; 322(8):494–500. 14. Leksell L. The stereotaxic method and radiosurgery of the brain. Acta Chir Scand 1951; 102:316. 15. Leksell L. Stereotactic radiosurgery. J Neurol Neurosurg Psychiatry 1983; 46(9):797–803. 16. Auchter RM, Lamond JP, Alexander E, et al. A multiinstitutional outcome and prognostic factor analysis of radiosurgery for resectable single brain metastasis. Int J Radiat Oncol Biol Phys 1996; 35(1):27–35. 17. Flickinger JC, Kondziolka D. Radiosurgery instead of resection for solitary brain metastasis: the gold standard redeﬁned [editorial] [see comments]. Int J Radiat Oncol Biol Phys 1996; 35(1):185–186. 18. Hasegawa T, Kondziolka D, Flickinger JC, et al. Brain metastases treated with radiosurgery alone: an alternative to whole brain radiotherapy? Neurosurgery 2003; 52(6):1318–1326; discussion 1326. 19. Lutterbach J, Cyron D, Henne K, Ostertag CB. Radiosurgery followed by planned observation in patients with one to three brain metastases. Neurosurgery 2003; 52(5):1066–1073; discussion 1073–1074. 20. Sneed PK, Suh JH, Goetsch SJ, et al. A multi-institutional review of radiosurgery alone vs. radiosurgery with whole brain radiotherapy as the initial management of brain metastases. Int J Radiat Oncol Biol Phys 2002; 53(3):519–526. 21. Bindal RK, Sawaya R, Leavens ME, Lee JJ. Surgical treatment of multiple brain metastases. J Neurosurg 1993; 79(2):210–216. 22. Bindal RK, Sawaya R, Leavens ME, et al. Reoperation for recurrent metastatic brain tumors. J Neurosurg 1995; 83(4):600–604. 23. Bindal AK, Bindal RK, Hess KR, et al. Surgery versus radiosurgery in the treatment of brain metastasis. J Neurosurg 1996; 84(5):748–754. 24. Dare AO, Sawaya R. Part II: Surgery versus radiosurgery for brain metastasis: surgical advantages and radiosurgical myths. Clin Neurosurg 2004; 51:255–263. 25. Mehta MP, Rozental JM, Levin AB, et al. Deﬁning the role of radiosurgery in the management of brain metastases. Int J Radiat Oncol Biol Phys 1992; 24(4):619–625. 26. Chang EL, Hassenbusch SJ 3rd, Shiu AS, et al. The role of tumor size in the radiosurgical management of patients with ambiguous brain metastases. Neurosurgery 2003; 53(2):272–280.

15.

metastatic brain tumors: surgery perspective

27. Regine WF, Huhn JL, Patchell RA, et al. Risk of symptomatic brain tumor recurrence and neurologic deﬁcit after radiosurgery alone in patients with newly diagnosed brain metastases: results and implications. Int J Radiat Oncol Biol Phys 2002; 52(2):333– 338. 28. Selek U, Chang EL, Hassenbusch SJ 3rd, et al. Stereotactic radiosurgical treatment in 103 patients for 153 cerebral melanoma metastases. Int J Radiat Oncol Biol Phys 2004; 59(4):1097– 1106. 29. Kondziolka D, Patel A, Lunsford LD, Kassam A, Flickinger JC. Stereotactic radiosurgery plus whole brain radiotherapy versus radiotherapy alone for patients with multiple brain metastases. Int J Radiat Oncol Biol Phys 1999; 45(2):427–434. 30. Shehata MK, Young B, Reid B, et al. Stereotatic radiosurgery of 468 brain metastases ≤ 2 cm: implications for SRS dose and whole brain radiation therapy. Int J Radiat Oncol Biol Phys 2004; 59(1): 87–93. 31. Cho KH, Hall WA, Lee AK, et al. Stereotactic radiosurgery for patients with single brain metastasis. J Radiosurg 1998; 1(2):79– 85. 32. O’Neill BP, Iturria NJ, Link MJ, et al. A comparison of surgical resection and stereotactic radiosurgery in the treatment of solitary brain metastases. Int J Radiat Oncol Biol Phys 2003; 55(5):1169– 1176. 33. Vecil GG, Suki D, Maldaun MV, et al. Resection of brain metastases previously treated with stereotactic radiosurgery. J Neurosurg 2005; 102(2):209–215. 34. Alexander E 3rd, Moriarty TM, Davis RB, et al. Stereotactic radiosurgery for the deﬁnitive, noninvasive treatment of brain metastases. J Natl Cancer Inst 1995; 87(1):34–40. 35. Flickinger JC, Kondziolka D, Lunsford LD, et al. A multiinstitutional experience with stereotactic radiosurgery for solitary

36.

37.

38.

39.

40.

41.

42.

43.

44.

199

brain metastasis. Int J Radiat Oncol Biol Phys 1994; 28(4):797– 802. Kihlstrom L, Karlsson B, Lindquist C, et al. Gamma knife surgery for cerebral metastasis. Acta Neurochir Suppl (Wien) 1991; 52:87–89. Muacevic A, Kreth FW, Horstmann GA, et al. Surgery and radiotherapy compared with gamma knife radiosurgery in the treatment of solitary cerebral metastases of small diameter. J Neurosurg 1999; 91(1):35–43. Petrovich Z, Yu C, Giannotta SL, et al. Survival and pattern of failure in brain metastasis treated with stereotactic gamma knife radiosurgery. J Neurosurg 2002; 97(5 Suppl):499–506. Voorhies RM, Sundaresan N, Thaler HT. The single supratentorial lesion. An evaluation of preoperative diagnostic tests. J Neurosurg 1980; 53(3):364–368. Mehta M, Noyes W, Craig B, et al. A cost-effectiveness and costutility analysis of radiosurgery vs. resection for single-brain metastases. Int J Radiat Oncol Biol Phys 1997; 39(2):445–454. Rutigliano MJ, Lunsford LD, Kondziolka D, et al. The cost effectiveness of stereotactic radiosurgery versus surgical resection in the treatment of solitary metastatic brain tumors. Neurosurgery 1995; 37(3):445–453; discussion 453–455. Loefﬂer JS, Alexander E 3rd. Radiosurgery for the treatment of intracranial metastases. In: Alexander E 3rd, Loefﬂer JS, Lunsford LD, eds. Stereotactic Radiosurgery. New York: McGraw-Hill, 1993:197–206. Sturm V, Kimmig B, Engenhardt R, et al. Radiosurgical treatment of cerebral metastases. Method, indications and results. Stereotact Funct Neurosurg 1991; 57(1–2):7–10. Kondziolka D, Lunsford LD. Brain metastases. In: Apuzzo MLJ, ed. Brain Surgery: Complication Avoidance and Management. New York: Churchill Livingstone, 1993:615–641.

1 6

Brain Metastases: Whole-Brain Radiation Therapy Perspective Roy A. Patchell and William F. Regine

Introduction Whole-brain radiation therapy (WBRT) is the mainstay of treatment for most patients with intracerebral metastases. It is useful for most newly diagnosed tumors and sometimes also as salvage therapy for recurrent tumors.

Efﬁcacy of WBRT Given the general consensus that WBRT is an effective treatment for brain metastases, it is surprising to note that there have been no randomized trials comparing WBRT with best supportive care or no treatment [1]. The evidence for the effectiveness of WBRT is derived from retrospective studies, nonrandomized prospective trials, and commonsense reasoning. Perhaps the best support for the general efﬁcacy of WBRT comes from comparing the outcomes of patients after WBRT treatment with the natural history of untreated brain metastases. Data from large retrospective studies [1–3] have shown that WBRT increases the median survival time to 3 to 6 months and that more than half of patients treated with WBRT die ultimately of progressive systemic cancer and not as a direct result of brain metastases. This contrasts with the natural history of untreated patients, which is that patients live for a median of 1 to 2 months and virtually all die as result of their brain lesions [4–7]. There is little doubt that WBRT does control or improve symptoms, at least temporarily, in the majority of patients with brain metastases. That WBRT has been accepted as effective should not be a serious problem when we remember that the efﬁcacy of most of the treatments used in medicine has not been established by the use of randomized trials.

Unsettled Issues Although there is general agreement on efﬁcacy, no consensus exists on the optimum WBRT dose and schedule for the treatment of brain metastases. The best available data on the effect of dose and schedule for the treatment of brain metastases comes from several large-scale multi-institutional trials conducted by the Radiation Therapy Oncology Group (RTOG) [1,

8–11]. These studies have shown no signiﬁcant difference in the frequency and duration of response for total radiation doses ranging from 2000 cGy over 1 week to 5000 cGy over 4 weeks. Regimens of 1000 cGy in a single dose or 1200 cGy in two doses were less effective and are no longer in use. Typical radiation treatment schedules for brain metastases consist of short courses (7 to 15 days) of WBRT with relatively high doses per fraction (150 to 400 cGy per day) and total doses in the range of 3000 to 5000 cGy. These schedules minimize the duration of treatment while still delivering adequate amounts of radiation to the tumor.

Attempts to Improve WBRT Different fractionation and dosing schemes have been tried in order to improve the effectiveness of WBRT with varying results. Epstein et al. [12] reported a phase I/II dose-escalation study of hyperfractionated radiotherapy using total doses of 48, 54.4, 64, and 70.4 Cy. No increased toxicity was identiﬁed and the three highest dose arms had a statistically signiﬁcant improvement in median survival over the lowest dose arm. This study suggested a dose-dependent effect with the use of hyperfractionated radiation in unresected, single brain metastasis. Unfortunately, a large randomized RTOG trial, studying both single and multiple brain metastases, failed to show improvement in overall survival using a hyperfractionated schedule of 54.5 Gy in 34 fractions compared with a standard regimen of 30 Gy in 10 fractions [13]. Another attempt to improve the efﬁcacy of WBRT was the addition of a boost dose to the tumor. However, a retrospective study [14] has found that increased conventional focal irradiation to the tumor, or “boost” dosing, did not increase survival or time to neurologic recurrence when compared with WBRT-alone. Radiation cell sensitizing agents have also been used in an attempt to increase tumor cell death. The rationale was based on the observation that hypoxic cells (often found centrally in a tumor) are more resistant to the effects of ionizing radiation. Agents such as misonidazole have the potential to increase cell sensitivity to irradiation. In the past, none of the radiation cell sensitizers has been shown to provide any additional beneﬁt

201

202

r.a. patchell and w.f. regine

over conventional radiotherapy [1, 15, 16]. However, two recent trials involving motexaﬁn gadolinium in lung cancer [17] and RSR13 in breast cancer [18] have shown guardedly promising results, and further investigation of these agents is needed to determine if they are effective.

Complications of WBRT WBRT has complications. Almost all patients experience a temporary loss of hair; hair usually returns 6 to 12 months after completing therapy. Also, in the short-term, patients may have a transient worsening of neurologic symptoms while receiving therapy. Many physicians believe that maintaining patients on steroids during radiotherapy will minimize radiation complications, although conclusive proof of this has not been forthcoming. The long-term side effects of radiotherapy are usually not a signiﬁcant issue in the treatment of patients with brain metastases because of the relatively short survival time of these patients. The frequency of serious long-term complications is unknown. One often quoted retrospective study by DeAngelis et al. [19] suggests that as many as 11% of long-term survivors (>12 months) of brain metastases treated with WBRT develop dementia. However, virtually all of the patients in that sample who developed dementia had been treated with atypically large radiation fractionation schedules. The patients treated with fraction sizes less than 3.0 Gy per day did not develop clinically apparent dementia. Thus, the actual frequency of radiationrelated dementia when using convention fractionation schedules is not known but is certainly less than 11%. In any event, the frequency of long-term neuropsychological side effects of WBRT in adult brain metastases patients appears to have been overestimated and seems to be within the acceptable range when modern fractionation schemes are employed.

WBRT in the Treatment of Multiple Brain Metastases Because the majority of patients have multiple metastases, the efﬁcacy data for WBRT comes in a large part from treatment of patients with multiple brain metastases. There is little doubt that WBRT is effective for multiple metastases, and it is used routinely. The main controversy regarding the treatment of multiple metastases involves the use of radiosurgery (either with or without WBRT). To date, there have been three randomized trials [20–22] assessing the efﬁcacy of radiosurgery in the treatment of multiple metastases. The ﬁrst randomized trial was reported by Kondziolka et al. [20]. In that study, 27 patients with multiple brain metastases were randomized to treatment with WBRTalone or WBRT plus a radiosurgery boost. The study was stopped early because the authors claimed to have found a large difference in the recurrence rates in favor of radiosurgery. Unfortunately, the study used nonstandard end points to measure recurrence. The investigators used any change in measurement of the lesion rather than the more usual 25% increase in diameter. No attempt was made to control for steroid use, radiation changes, or other factors that might produce small ﬂuctuations in lesion size on magnetic resonance imaging

(MRI). Also, a study with only 27 patients in it lacked the statistical power to support any meaningful conclusion, regardless of p values. As a result, this study was uninterpretable. A second study reported in abstract form by Chougule et al. [21] randomized patients with one to three brain metastases to treatment with radiosurgery-alone, radiosurgery plus WBRT, or WBRT-alone. The study had 109 patients. There was no statistically signiﬁcant difference in survival among the three treatment arms. Median survival times for the radiosurgery, radiosurgery plus WBRT, and the WBRT-alone treated groups were 7, 5, and 9 months, respectively. Local control rates in the brain were also not signiﬁcantly different. This trial suffered from several methodological problems. The most serious error was that 51 of the patients had had surgery for at least one symptomatic brain metastasis prior to entry into the study. No attempt was made to stratify for previous surgery or to otherwise ensure that surgical patients were equally distributed among the treatment groups. The inclusion of the surgical patients effectively made this a six-arm trial (the original three subdivided again into surgically treated patients and nonsurgically treated patients), and therefore, the size of this trial was not large enough to support a meaningful analysis. Also, because surgery is in all probability an effective therapy for brain metastases, the nonrandom distribution of surgically treated patients among the treatment arms substantially weakened the trial. Therefore, this study, although ostensibly negative, is really uninterpretable. A third study was reported by Andrews et al. [22] This study (RTOG 9508) contained 333 evaluable patients with one to three brain metastases who were randomized to treatment with either WBRT (37.5 Gy) plus radiosurgery or WBRT (37.5 Gy) alone. The primary end point was survival. Overall, there was no signiﬁcant difference in survival between the two treatment groups (median, 6.5 months for radiosurgery plus WBRT and 5.7 months for WBRT-alone, p = 0.1356). There was no survival beneﬁt from radiosurgery in patients with multiple metastases (median, 5.8 months for radiosurgery plus WBRT and 6.7 months for WBRT-alone, p = 0.9776). (However, for patients with single metastases, there was a signiﬁcant survival advantage favoring radiosurgery, median 6.5 months vs. 4.9 months, p = 0.0393). Lower posttreatment Karnofsky scores and steroid dependence were more common in the WBRT-alone group. Multiple subgroup analyses were made and a beneﬁt for radiosurgery plus WBRT was found in several subgroups that included patients with single and multiple metastases. These subgroups were RPA class 1 patients, patients with metastases size equal to or larger than 2 cm, and lung cancer patients with squamous cell histology. However, these subset analyses were not prespeciﬁed, and the p values needed for signiﬁcance should have been adjusted for inﬂation of type I error. When this was done, none of these subgroup analyses showed a positive beneﬁt for radiosurgery [23]. So, for multiple brain metastases, this was a completely negative trial with regard to the major end points prevention of death due to neurologic causes and overall survival. Radiosurgery has been put to the test in the treatment of multiple metastases and has not been established as effective. Therefore, based on the best available evidence, WBRT-alone is the treatment of choice for most patients with multiple brain metastases (and the word “multiple” in this context means more than one).

16.

brain metastases: whole-brain radiation therapy perspective

WBRT in the Treatment of Single Brain Metastases For the treatment of single brain metastases, randomized trials have established the superiority of focal treatment (either conventional surgery or radiosurgery) plus WBRT over treatment with WBRT-alone. Therefore, good-prognosis patients with single brain metastases should be treated with upfront surgery or radiosurgery. However, with the establishment of the efﬁcacy of focal treatments for single brain metastases, a new controversy has arisen as to whether adjuvant WBRT is really necessary after a “complete resection” or “successful” treatment with radiosurgery. Adjuvant WBRT is thought to be of beneﬁt because there may be residual disease in the tumor bed or at distant microscopic sites in the brain. However, brain metastases tend to be discrete masses that are theoretically capable of being removed totally or destroyed, and so WBRT may not be necessary after “successful” focal therapy. There are several reasons for eliminating WBRT. First, WBRT has adverse, long-term neuropsychological side effects. Second, there are also the costs and time commitment of the patient that must be considered. And ﬁnally, there is the possibility that WBRT may simply not be needed at all. It is theoretically possible to remove single brain metastases by surgery totally or to control them with radiosurgery. Furthermore, neuroimaging has improved, and it may now be possible to detect reliably additional metastases that may be present and treat these with additional focal therapy. If these last two statements are true, then there would be little justiﬁcation for adjuvant WBRT. On the other hand, compelling reasons exist for giving adjuvant WBRT. As a practical matter, it is probably impossible to remove completely all metastases with conventional surgery, and radiosurgery does not completely control the tumors. In addition, neuroimaging may not have reached the point yet where we can be absolutely certain that all metastases are being detected, and therefore some type of additional treatment may be needed. Also, although WBRT does have side effects, these side effects may not be as severe or as common as was previously thought. Furthermore, most patients with brain metastases have relatively limited overall survival times, and so the really serious long-term side effects are usually not an issue in their care. Two randomized trials [24, 25] have addressed the question of adjuvant WBRT in conjunction with focal treatment. A study published in 1999 by Patchell et al. [24] examined the effect of WBRT in conjunction with conventional surgery. In that study, 95 patients who had single brain metastases that were completely surgically resected were randomized to treatment with postoperative WBRT (50.4 Gy) or to observation with no further treatment of the brain metastasis (until recurrence). Recurrence of tumor anywhere in the brain was less frequent in the radiotherapy group than in the observation group (18% vs. 70%, p < 0.001). Postoperative radiotherapy prevented brain recurrence at the site of the original metastasis (10% vs. 46%, p < 0.001) and at other sites in the brain (14% vs. 37%, p < 0.01). As a result, patients in the radiotherapy group were less likely to die of neurologic causes than patients in the observation group (6 of 43 who died [14%] vs. 17 of 39 [44%]; p = 0.003). There was no signiﬁcant difference between the two

203

groups in overall length of survival or the length of time that patients remained functionally independent. The effect of WBRT in association with radiosurgery was examined in a randomized trial conducted by the Japanese Radiation Oncology Study Group and reported in abstract form by Aoyama et al. [25] In that study, 132 patients with one to four brain metastases were randomized to treatment WBRT plus radiosurgery or with radiosurgery-alone. The WBRT total dose was 30 Gy. The radiosurgery dose was 18 to 25 Gy at the periphery of the lesion in the radiosurgery-alone group and was reduced by 30% in the WBRT plus radiosurgery group. The median survival time was 7.5 months in the WBRT plus radiosurgery group and 8.0 months in the radiosurgery-alone group, p = 0.42. The 12-month brain metastases recurrence rates were signiﬁcantly (p < 0.001) different (47% in the WBRT plus radiosurgery group and 76% in the radiosurgery-alone group). Death due to neurologic causes and neurologic functioning were not signiﬁcantly different between the two groups. Despite the fact that both of the randomized trials [24, 25] showed clearly that WBRT prevented recurrences, these studies have actually provoked controversy rather than settling the issue. Results of these trials have been used as reasons both to give and not to give adjuvant WBRT. The justiﬁcation for not giving WBRT holds that because no survival difference was found in either of the trials, WBRT really adds nothing to the treatment. This argument fails on several counts. The Patchell study [24] used tumor recurrence as the primary end point and was not designed either to show a difference in survival or to rule one out. There was actually an increase of 11% in survival time in the WBRT group when compared with the observation group. The relative risk of improved survival with WBRT was 1.1. However, this was not a statistically signiﬁcant difference. Because there was a statistically signiﬁcant reduction in death due to neurologic causes, ultimately adjuvant WBRT might have had some positive impact on overall survival time. The estimated sample size required to detect a signiﬁcant difference of 11% in overall survival with adequate power would have been 1005 patients per group or 2010 patients total. For practical reasons, the study could not be designed to have this large of a sample size and, therefore, was not designed to detect moderate differences in survival, even one as large as 11%. There is an even stronger reason for discounting the apparent lack of efﬁcacy of postoperative WBRT with regard to length of survival in the Patchell trial [24]. Recurrence of tumor in the brain was the primary end point of that randomized trial, and this end point was the only truly direct measure of the effects of adjuvant WBRT. Up until recurrence of tumor, the two treatment groups were distinct, and the patients in each had received the treatment assigned by randomization. However, at recurrence, no speciﬁc treatment was mandated by the study design, and as a result, patients received a variety of additional treatments. There was an extremely large crossover of the observation group to WBRT. Of the 32 patients in observation group who developed recurrent brain metastases, 28 patients got WBRT. Overall, that means that 61% (28 patients of 46 total) in the “no WBRT” observation arm were, in fact, treated ultimately with WBRT. For the purposes of length of survival and functional independence, the study was virtually a comparison of surgery plus immediate WBRT versus surgery plus delayed

204

r.a. patchell and w.f. regine

WBRT. This substantially diluted the effect of WBRT given immediately postoperatively because WBRT probably improved the length of survival and functional independence in the observation group. Therefore, for all the reasons given above, the Patchell study did not “prove” that WBRT had no positive effect on survival. The Aoyama trial [25] also failed to ﬁnd any difference in survival with the addition of WBRT. This study differed in design from the Patchell study and was devised with survival as the primary end point; it was originally powered to be able to detect a 30% or greater difference with 89 patients per treatment arm. (Given that the Patchell study [24] and numerous retrospective studies had failed to ﬁnd a difference in survival even close to 30%, the Aoyama study [25] should have been powered as an equivalence trial [if survival was going to be used as the primary end point].) The study was stopped after an interim analysis at 132 patients indicated that the study was not going to show a statistically signiﬁcant difference in survival, even if all of the 178 patients originally projected to be in the trial had been randomized. Therefore, any negative conclusion based on the original power calculation, that is, that there was no more than a 30% difference in survival (or any other reasonable difference), cannot be supported. This study also had an 11% crossover that further eroded the strength of any conclusions about survival. In addition, about half of the patients in both treatment arms had multiple brain metastases. The study by Andrews et al. [22] failed to establish the efﬁcacy of radiosurgery in the treatment of multiple metastases, and so the inclusion of these patients is problematic and further reduces the number of valid patients on which to base a conclusion about survival beneﬁt. Thus, like the Patchell trial [24] before it, this Aoyama study [25] was unable to show that there was no difference in survival. Therefore, arguments based on the supposed lack of efﬁcacy of (immediate) postoperative WBRT on survival are based on a misunderstanding of the design and limits of the randomized trials. In addition, the Aoyama study demonstrated that omitting WBRT does not produce any difference in either gross neurologic or neurocognitive functioning. From this information, Aoyama et al. [25] and the Journal of the American Medical Association (JAMA) editorial writer [26] concluded that the addition of WBRT is not necessary and can be safely omitted in the treatment of most patients with brain metastases. However, even if one takes the data presented in the paper at face value, it is possible to draw exactly the opposite conclusion. As stated by the authors and the editorial writer, the main reason for not giving WBRT is to avoid the long-term neurotoxic effects of WBRT. Yet, this study found no difference in neurologic functioning, neurocognitive functioning, gross radiation-induced side effects, or survival times between the two groups. In fact, deterioration in neurologic function attributable to progression of brain metastases was observed in 59% of patients in the WBRT group and 86% in the SRS-alone group (p = 0.05) indicating a signiﬁcantly higher rate of neurologic deterioration as a consequence of tumor progression in patients when WBRT is withheld. Thus, at very least, WBRT appears to signiﬁcantly reduce the recurrence of brain metastases without demonstrable neurotoxicity. Therefore, the trial by Aoyama et al. [25] seems to support strongly the use of WBRT

as upfront treatment in the management of most patients with brain metastases. The most forceful argument in favor of adjuvant WBRT involves an examination of the effects of not giving WBRT. Patients who do not receive adjuvant WBRT suffer substantially more recurrent brain metastases than patients who are treated with WBRT. As previously noted, the harmful side effects of WBRT appear to have been overestimated in the past and are probably in the acceptable range. Unfortunately, the same cannot be said of the side effects of recurrence of brain metastases. Several studies [27, 28] have demonstrated that the recurrence of brain metastases has a negative effect on the neurocognitive functioning of patients. A study by Regine et al. [27] found that in 36 patients with brain metastases treated with SRS alone, 47% had recurrence of brain metastases and 71% of the recurrences were symptomatic. Signiﬁcantly, 59% of the patients with recurrent tumors had associated neurologic deﬁcits and 17% were unable to undergo salvage brain therapy because of their overall poor general status associated with brain tumor recurrence. These ﬁndings are now substantiated by the level 1 evidence provided by the Aoyama phase III trial where deterioration in neurologic function attributable to progression of brain metastases was observed in 59% of patients in the WBRT group versus 86% in the SRS-alone group (p = 0.05); indicating a signiﬁcantly higher rate of neurologic deterioration as a consequence of tumor progression in patients when WBRT is withheld [25]. Another study by Regine et al. [28] showed that, at 3 months after treatment, patients treated for brain metastases with WBRT had greater negative changes in their mini-mental status examinations with uncontrolled brain tumors than they did with controlled brain tumors (−6.3 points versus −0.5 points, p = 0.02). Also relevant (but perhaps somewhat farther aﬁeld) was a study by Taylor et al. [29] showing that, in patients with primary brain tumors at 12 months after treatment, changes in mini-mental status examinations were worse in patients with uncontrolled tumors (−2.42 points) than in patients with controlled tumors (+0.076 points) (p = 0.0046). All of the patients in this study had received large total doses of conventional radiation therapy. These studies all strongly suggest that uncontrolled brain tumors result in a substantial decrease in mental performance and that this reduction far outweighs any decrement seen with cranial radiation therapy. Therefore, the side effects of recurrent tumors are worse than the side effects of preventive treatment. This is an extremely strong argument for the use of adjuvant WBRT in association with focal therapy.

WBRT for Recurrent Brain Metastases Brain metastases often recur, and CNS progression may be accompanied by systemic tumor progression and a decline in functional status. In general, the same types of treatment used for newly diagnosed brain metastases are also available for recurrent tumors. However, the type of previous therapy may limit the therapeutic options available at recurrence, and the development of radioresistance is not uncommon.

16.

brain metastases: whole-brain radiation therapy perspective

If patients have not already had WBRT, they should be treated with it on recurrence. However, often patients with recurrences have already been treated with WBRT, and this limits the amount of subsequent radiation that can be given safely. The amount of additional radiation that can be offered is usually in the range of 1500 to 2500 cGy, a total dose usually inadequate to control tumor growth. Several retrospective studies [30–34] have attempted to asses the efﬁcacy of salvage WBRT. It is difﬁcult to assess efﬁcacy from these reports. Rates of improvement ranged from 27% to 70%; however, the range of duration of response was fairly uniform and was 2.5 to 3 months. The median survival ranged from 1.8 to 4.0 months. Relatively few long-term complications were reported; however, because the median survival is quite short, most patients did not live long enough to develop the long-term complications of radiation. The problem with the interpretation of these studies is that they often used different end-point measurements for improvement and had heterogeneous patient populations. Some included patients with poor performance status and extensive disease and others selected out favorable subgroups for radiation. One recommendation based on a retrospective study [30] is to restrict reirradiation to patients who showed an initial favorable response to radiotherapy, had a longer disease-free interval, and who remain in good general condition when the cerebral recurrence develops. However, even in this favorable subgroup, only 42% of patients showed symptomatic improvement, and the median survival after reirradiation was 5 months. Despite such relatively poor results, additional radiotherapy is frequently one of the few treatment options for patients with recurrent disease.

References 1. Berk L. An overview of radiotherapy trials for the treatment of brain metastases. Oncology 1995; 9:1205–1219. 2. Posner JB. Management of brain metastases. Rev Neurol 1992; 148:477–487. 3. Cairncross JG, Posner JB. The management of brain metastases. In: Walker MD, ed. Oncology of the Nervous System. Boston: Martinus Nijhoff, 1983:341–377. 4. Horton J, Baxter DH, Olson KB. The management of metastases to the brain by irradiation and corticosteroids. Am J Roentgenol Radium Ther Nucl Med 1971; 111:334–335. 5. Markesbery WR, Brooks WH, Gupta GD, et al. Treatment for patients with cerebral metastases. Arch Neurol 1978; 35:754–756. 6. Ruderman NB, Hall TC. Use of glucocorticoids in the palliative treatment of metastatic brain tumors. Cancer 1965; 18:298–306. 7. Chang DB, Yang PC, Luh KT, et al. Late survival of non-small cell lung cancer patients with brain metastases. Chest 1992; 101: 1293–1297. 8. Borgelt B, Gelber R, Kramer S, et al. The palliation of brain metastases: ﬁnal results of the ﬁrst two studies by the Radiation Therapy Oncology Group. Int J Radiat Oncol Biol Phys 1980; 6: 1–9. 9. Gelber RD, Larson M, Borgelt BB, et al. Equivalence of radiation schedules for the palliative treatment of brain metastases in patients with favorable prognosis. Cancer 1981; 48:1749–1753. 10. Kurtz JM, Gelber RD, Brady LW, et al. The palliation of brain metastases in a favorable patient population: a randomized clinical trial by the Radiation Therapy Oncology Group. Int J Radiat Oncol Biol Phys 1981; 7:891–895.

205

11. Deiner-West M, Dobbins TW, Phillips TL, et al. Identiﬁcation of an optimal subgroup for treatment evaluation of patients with brain metastases using RTOG study 7916. Int J Radiat Oncol Biol Phys 1989; 16:669–673. 12. Epstein BE, Scott CB, Sause WT, et al. Improved survival duration in patients with unresected solitary brain metastasis using accelerated hyperfractionated radiation therapy at total doses of 54.4 gray and greater. Results of the Radiation Therapy Oncology Group 85–28. Cancer 1993; 7:1362–1367. 13. Murray KJ, Scott C, Greenberg H, et al. A randomized phase III study of accelerated hyperfractionated versus standard radiation in patients with unresected brain metastases: a report of the Radiation Therapy Oncology Group (RTOG) 9104. Int J Radiat Oncol Biol Phys 1997; 39:571–574. 14. Hoskins PJ, Crow J, Ford HT. The inﬂuence of extent and local management on the outcome of radiotherapy for brain metastases. Int J Radiat Oncol Biol Phys 1990; 19:111–115. 15. Eyre HJ, Ohlsen JD, Frank J, et al. Randomized trial of radiotherapy versus radiotherapy plus metronidazole for the treatment metastatic cancer to the brain. J Neuro-Oncology 1984; 2:325–330. 16. Komarnicky LT, Phillips TL, Martz K, et al. A randomized phase III protocol for the evaluation of misonidazole combined with radiation in the treatment of patients with brain metastases (RTOG-7916). Int J Radiat Oncol Biol Phys 1991; 20:53–58. 17. Mehta MP, Rodrigus P, Terhaard CH, et al. Survival and neurologic outcomes in a randomized trial of motexaﬁn gadolinium and whole-brain radiation therapy in brain metastases. J Clin Oncol 2003; 21:2529–2536. 18. Suh JH, Stea B, Nabid A, et al. Phase III study of efaproxiral as an adjunct to whole brain radiation therapy for brain metastases. J Clin Oncol 2006; 24:106–114. 19. DeAngelis LM, Mandell LR, Thaler HT, et al. The role of postoperative radiotherapy after resection of single brain metastases. Neurosurgery 1989; 24:798–805. 20. Kondziolka D, Patel A, Lunsford LD, Kassam A, Flickinger JC. Stereotactic radiosurgery plus whole brain radiotherapy versus radiotherapy alone for patients with multiple brain metastases. Int J Radiat Oncol Biol Phys 1999; 45:427–434. 21. Chougule PB, Burton-Williams M, Saris S, et al. Randomized treatment of brain metastases with gamma knife radiosurgery, whole brain radiotherapy or both. Int J Radiat Oncol Biol Phys 2000; 48:114. 22. Andrews DW, Scott CB, Sperduto PW, et al. Whole brain radiation therapy with and without stereotactic radiosurgery boost for patients with one to three brain metastases: phase III results of the RTOG 9508 randomized trial. Lancet 2004; 363:1665– 1672. 23. Elveen T, Andrews DW. Summary of RTOG 95-01 phase III randomized trial of whole brain radiation with and without stereotactic radiosurgery boost, including presentation of a clinical case study. Am J Oncol Rev 2004; 3:592–600. 24. Patchell RA, Tibbs PA, Regine WF, et al. Postoperative radiotherapy in the treatment of single metastases to the brain: a randomized trial. JAMA 1998; 280:1485–1489. 25. Aoyama H, Shirato H, Tago M, et al. Stereotactic radiosurgery plus whole-brain radiation therapy vs Stereotactic radiosurgery alone for the treatment of brain metastases. JAMA 2006; 295: 2483–2491. 26. Raizer J. Radiosurgery and whole-brain radiation therapy for brain metastases. JAMA 2006; 295:2535–2536. 27. Regine WF, Scott C, Murray K, Curran W. Neurocognitive outcome in brain metastases patients treated with acceleratedfraction vs. accelerated-hyperfractionated radiotherapy: an analysis from Radiation Therapy Oncology Group Study 91-04. Int J Radiat Oncol Biol Phys 2001; 51:711–717.

206

r.a. patchell and w.f. regine

28. Regine WF, Huhn JL, Patchell RA, et al. Risk of symptomatic brain tumor recurrence and neurologic deﬁcit after radiosurgery al in patients with newly diagnosed brain metastases: results and implications. Int J Radiat Oncol Biol Phys 2002; 52:333–338. 29. Taylor BV, Buckner JC, Cascino TL, et al. Effects of radiation and chemotherapy on cognitive function in patients with highgrade gliomas. J Clin Oncol 1998; 16:2195–2201. 30. Hocht S, Wiegel T, Hinkelbein W. Reirradiation for recurrent brain metastases. Controversies in Neuro-Oncology. Front Radiat Oncol Basal 1999; 33:327–331.

31. Hazuka MB, Kinzie JJ. Brain metastases: results and effects of reirradiation. Int J Radiat Oncol Biol Phys 1988; 15:433–437. 32. Kurup P, Reddy S, Hendrickson FR. Results of re-irradiation for cerebral metastases. Cancer 1980; 46:2587–2589. 33. Wong WW, Schild SE, Sawyer TE, et al. Analysis of outcome in patients reirradiated for brain metastases. Int J Rad Oncol Biol Phys 1996; 34:585–590. 34. Cooper JS, Steinfeld AD, Lerch IA. Cerebral metastases: value of reirradiation in selected patients. Radiology 1990; 174:883– 885.

1 7

High-Grade Gliomas David Roberge and Luis Souhami

Introduction Primary brain tumors are classiﬁed according to their predominant cell type. Glial neoplasms are the most common primary intracranial malignancies and are classiﬁed as astrocytic tumors, oligodendroglial tumors, and ependymal tumors. Astrocytomas are the most common variant of glioma, and most adult astrocytomas are of high grade. Of these, glioblastoma multiforme (GBM), or astrocytoma grade IV, represents the most common subtype [1], making this the most common malignant tumor in adults. Despite the best contemporary use of surgery, radiation, and chemotherapy, long-term survival of patients with GBM has been distinctly uncommon. In unselected series, the 5-year survival has been as low as 2% [2, 3]. Anaplastic astrocytoma (AA), or astrocytoma grade III, represents less than 20% of malignant gliomas and is associated with a more favorable outcome [1, 4]. For the scope of this chapter, only high-grade astrocytic tumors will be discussed. High-grade astrocytic tumors are inﬁltrating tumors that can rapidly enlarge, resulting in various signs and symptoms, such as focal or generalized seizures, headache, visual disturbances, speech disturbances, changes in mental status, and motor or sensory deﬁcits. Malignant glial tumors are thought to evolve from an accumulation of multiple genetic aberrations in normal precursor cells. In a stepwise fashion, this accumulation of deleterious genetic alterations may lead to transformation to a low-grade glioma and to the subsequent aggressive phenotype associated with high-grade tumors. Several molecular studies have generated multiple markers linked with malignant gliomas, including chromosomal deletion, addition, mutation, and gene ampliﬁcation, which may have important clinical implications [5]. Both GBM and AA can be of two types based on their clinical presentation. Primary high-grade astrocytomas occur de novo and are associated with short duration of symptoms and worse prognosis, and secondary high-grade gliomas often occur in patients with a previous low-grade astrocytoma, suggesting a different pathogenesis [6]. High-grade glioma (HGG) is a local disease. Distant spread is a rare event, and 90% of recurrences are located within 2 cm of the original enhancing lesion [7–10]. Overall, the prognosis of HGG is related to tumor grade, performance status, age, and treatment. Five other prognostic factors were

incorporated in a 1993 prognostic scheme based on recursive partitioning analysis (RPA) of the individual patient data from three Radiation Therapy Oncology Group (RTOG) trials [4]. Using several prognostic variables (Table 17-1), patients were divided into six classes with median survivals ranging from 4.7 to 58.6 months. The RPA classiﬁcation is now frequently used in the comparison of treatment results from different series. Despite intermittent waves of enthusiasm regarding various treatment modalities, the survival of patients with HGG has not increased substantially from 1950 to 2000. In this chapter, we describe and contextualize the use of stereotactic radiosurgery (SRS) and fractionated stereotactic radiotherapy (F-SRT) in the management of primary and recurrent malignant gliomas.

Historical Perspective The poor prognosis of HGG is known for many years. In 1926, Bailey and Cushing, in reference to what they at the time called “spongioblastoma multiforme,” made the following observation: “Operative procedures, howsoever radical [block extirpations repeated on signs of recurrence; saturation with X-rays or radium emanations after wide decompression with or without surgical interference with the tumor], have apparently done little more than to prolong life, save vision, and alleviate headache for an average of a few months. . . . Whether deep Roentgenization ever does more than hold the growth temporarily in check is problematical” [11]. By the time Elvidge, Penﬁeld, and Cone published their “McGill series” in 1937, the name “glioblastoma multiforme” had become widely accepted, and neurosurgery was still faced with the same “difﬁcult human problem” posed by these patients [12]. Thirteen years after the publication of Elvidge and colleagues [12], a case history was published as part of a paper reviewing 70 glioblastoma multiforme patients treated at the Monteﬁore Hospital [13] (Case Study 17-1). The described case represented the longest surviving patient in the literature of the day [13]. Prognostic factors for

207

208

d. roberge and l. souhami

TABLE 17-1. RTOG recursive partitioning analysis classes. Class

Deﬁnition

Median survival (months)

Two-year survival (%)

I

Age <50, AA, normal mental status

58.6

76

II

Age ≥50, KPS 70–100, AA, ≥3 months from ﬁrst symptoms to treatment

37.4

68

III

Age <50, AA, abnormal mental status or Age <50, GBM, KPS 90–100

17.9

35

IV

Age <50, GBM, KPS <90 or Age ≥50, KPS 70–100, AA, ≥3 months from ﬁrst symptoms to treatment or Age >50, GBM, surgical resection, good neurologic function

11.1

15

V

Age ≥50, KPS 70–100, GBM, either surgical resection and poor neurologic function or biopsy only followed by ≥54.4 Gy EBRT or Age ≥50, KPS <70, normal mental status

8.9

6

VI

Age ≤50, KPS <70, abnormal mental status Age ≥50, KPS 70–100, GBM, biopsy only, <54.4 Gy EBRT

4.6

4

Source: Curran WJ Jr, Scott CB, Horton J, et al. Recursive partitioning analysis of prognostic factors in three Radiation Therapy Oncology Group malignant glioma trials. J Natl Cancer Inst 1993; 85:704–710.

the Monteﬁore series were reviewed: “An analysis of factors in survival revealed that operation had no signiﬁcant effect on longevity and irradiation a slight effect, if any. The longest survivals occurred in patients with onset at 24 to 42 years of age. However, age at onset was not a constant prognostic factor. The sex of the patient, the location of the neoplasm, or the histologic appearance gave no indication as to longevity. A signiﬁcant lengthening of survival was found in those patients whose initial symptom was a motor seizure.” If patients dying within 1 week of surgery were excluded, the median overall survival was 13.1 months with postoperative irradiation and 12

Case Study 17-1 Case 1. S.S. Autopsy no. 9588. “In 1929, at the age of 37, this right-handed white male had a convulsive seizure lasting a few minutes. . . . In 1931, following an attack, he noticed loss of sensation on the right side of the body and a right hemiparesis. In 1933, a partial motor aphasia developed. Dr. Paul C. Bucy operated on him in October 1933. A ﬁrm area measuring 2 cm in diameter was found on the left side in the gyrus just anterior to the precentral gyrus. . . . An attempt at complete removal was not made because of the location. A small piece was taken for study, . . . The tissue was diagnosed by Dr. Percival Bailey as a malignant glioma, probably a glioblastoma multiforme. . . . Following the operation, the patient was given a course of roentgen therapy for a total of 6068 r. . . . In September 1937, he was struck by an automobile, and was unconscious for 24 hours. Sometimes thereafter his seizures returned. He gradually lost power on the right side, and the aphasia became worse. In October 1939, he was seen by Dr. Leo Davidoff. . . . An encaphalogram was interpreted as showing left-sided cerebral atrophy without evidence of regrowth. Nothing further was done.

months without. It must however be noted that for all cases of prolonged survival reported in this paper, the patient had received irradiation. As demonstrated by the previous series, irradiation had been in common use long before randomized trials demonstrating its efﬁcacy. In 1947, Bush and Christensen wrote: “While it is very difﬁcult to judge of the effect of radiation in these cases, it is our impression that this therapy is of deﬁnite value. A halfhearted attempt of giving every other patient Roentgen therapy was never carried through as we felt ourselves unable to deprive the particular patient of what we felt was the best chance” [14].

He entered Monteﬁore Hospital in August 1939 at the age of 49. . . . there were now astereognosis in the right hand, a partial motor aphasia, and some blurring of the right disc margin. . . . During his stay in the hospital, both focal motor and generalized seizures were noted. Irradiation was not deemed advisable, and he was discharged to be followed in the outpatient clinic. He had difﬁculties in expressing himself. Because of this, he insisted on being re-admitted in July 1941. At this time it was felt that the paretic side was more spastic, and that the motor aphasia now had a sensory component. . . . He was given a course of roentgen-therapy amounting to 3000 r. with slight improvement. Thereafter he continued to have right-sided seizures until July 1943, when he rapidly became worse and died. . . . Autopsy. A broncho-pneumonia was found to be the immediate cause of death. The brain weighed 1420 gm. . . . There was a hemorrhagic tumor nodule, measuring 3 × 2 cm. in the left 3rd frontal convolution. . . . The tumor was a glioblastoma multiforme. There were some spongioblastic portions, but there were, in addition, extensive necrosis, pseudopallisading around focal necrotic zones, thrombosis of vessels and endothelial proliferation, perivascular lymphocytes, numerous mitotic ﬁgures, and pleomorphism.”

17.

high-grade gliomas

It was only in the early 1980s that both the Brain Tumor Cooperative Group (BTCG) and the Scandinavian Glioblastoma Study Group (SGSG) published results of randomized trials of adjuvant irradiation [15, 16]. These trials conﬁrmed a large and signiﬁcant increase in median survival after the administration of postoperative irradiation compared with surgery alone. In the Scandinavian trial, 45 Gy of whole-brain irradiation increased the median survival from 5.2 months to 10.8 months. In BTCG study 6901, median survival increased from 3.2 months to 8.1 months after 50 to 60 Gy of whole-brain radiotherapy. A randomized trial by the Medical Research Council [17] conﬁrmed an improvement in median survival from 9 to 12 months when 60 Gy was compared with 45 Gy. This trial suggested a possible dose-control relationship. For HGG, 60 Gy has been considered the standard postoperative radiotherapy dose by most investigators. Initially as part of the same trials looking at adjuvant irradiation, a large number of randomized trials have been conducted testing the use of concurrent and/or adjuvant chemotherapy agents and radiation sensitizers in the management of HGG [15, 18–22]. In 1993, a meta-analysis by Fine et al. summarized data from 16 randomized trials involving more than 3000 patients [23]. Optimistic physicians saw a large (52%) difference in 2-year survival; pessimistic physicians saw a relatively small difference in median survival. A subsequent meta-analysis from the Glioma Meta-analysis Trialists group conﬁrmed a 1.5-month increase in median survival and failed to demonstrate any beneﬁt of multidrug regimens over singleagent nitrosourea compounds [24]. Adjuvant BCNU chemotherapy was considered standard of care by most although it was used variably in clinical practice. Recently, in a joint European Organization for Research and Treatment of Cancer (EORTC) and National Cancer Institute of Canada (NCIC) randomized trial [25], concomitant and adjuvant temozolomide signiﬁcantly increased the median survival of patients diagnosed with glioblastoma multiforme from 12.1 months to 14.6 months and more than doubled the 2-year survival (26.5% vs. 10.4%). It is of interest to note that the control arm (radiotherapy-alone) had the same median survival as reported in the 1949 series from the Monteﬁore Hospital, highlighting how little progress had been made in the past 50 years. Based on this EORTC/NCIC trial, the current standard of care for GBM patients with a performance status of 0 to 1 (on subgroup analysis, there was no beneﬁt for patients with an ECOG performance status of 2) is external beam radiotherapy to 60 Gy and temozolomide. Although patients with AA histology were not included in the trial, this regimen will likely also become the de facto standard for these patients as well.

Rationale for Stereotactic Radiosurgery Inﬁltrating high-grade glial neoplasms would appear to be poor candidates for the stereotactic application of single-fraction irradiation. These tumors are hypoxic, acute-responding [26], and admixed with normal tissue [27, 28]. Despite these biological roadblocks, stereotactic radiation has been pursued in the management of HGG. The use of stereotactic radiosurgery

209

(SRS; deﬁned here as a radiotherapy technique characterized by accurate delivery of high doses of radiation in a single session to small intracranial targets in such a way that the dose fall-off outside the target volume is very sharp) has been based on the pattern of failure of this disease, the dose-response data from external beam radiation, and early data from interstitial brachytherapy trials [29, 30]. As mentioned previously, 90% of patients recur within 2 cm of the contrast-enhancing lesions–-this despite the fact that tumor cells can be found pathologically at larger distances, often following the peritumoral edema. Furthermore, multicentric or metastatic disease is rare [10]. Moreover, in an analysis of the Brain Tumor Study Group data, a dose-response relationship has been shown for doses of 50 to 60 Gy [31]. In this data set, the median survival increases from 28 weeks at 50 Gy to 42 weeks at the 60 Gy level. Signiﬁcant improvement in median survival was also observed in the randomized British trial comparing 45 Gy to 60 Gy [17]. Thus, all of these characteristics made SRS an attractive option to be used as a focal boost in selected patients with HGG. In the early 1990s, phase I/II data suggested that interstitial brachytherapy improved local control and survival in selected primary and recurrent HGG patients. At University of California San Francisco (UCSF), the median survival of the ﬁrst 18 patients treated with interstitial brachytherapy for recurrent GBM was 52 weeks with 2 patients surviving more than 5 years [32]. In a subsequent North Carolina Oncology Group (NCOG) study [30], 107 patients with HGG were enrolled in a program of brachytherapy added to EBRT and adjuvant procarbazine, lomustine and vincristine (PCV) chemotherapy. In the 63 evaluable patients who were actually implanted, the median survivals were 157 weeks and 88 weeks, respectively, for patients with grade III astrocytoma and GBM. Brachytherapy was associated with a high risk of radionecrosis, and it was believed that SRS might offer the focal dose escalation beneﬁts of implants with lessened toxicity. Unfortunately, concerns about selection bias [33], as a possible explanation for these improved median survivals after brachytherapy, appear to be have been conﬁrmed by two subsequent negative phase III trials of brachytherapy by the BTCG [34] and the Princess Margaret Hospital [35] groups. In these randomized studies, the added use of a focal brachytherapy boost did not lead to an improvement in survival in patients harboring a HGG. In the Toronto trial, the median survival was 13.2 months on the standard arm and 13.8 months on the brachytherapy arm, and in the BTCG study it was 58.8 weeks on the standard arm and 64.1 weeks on the brachytherapy arm. The cumulative proportion surviving between the two treatment groups was not statistically signiﬁcantly different in either study. In 1985, Columbo et al. reported on the ﬁrst patients treated in Vinceza, Italy, with a new technique for linear accelerator (linac)-based radiosurgery using non-coplanar arcs [36]. Only 6 of 22 patients had adequate follow-up. Of these, one patient was treated for a 3.5-cm grade III astrocytoma. Two doses of 20 Gy were delivered during separate procedures. This patient worsened within 2 months and was reoperated. After this early experience, several institutional reports have been published on the use of SRS boost in the management, at presentation or at time of recurrence, of patients with HGG. The noninvasive

210

d. roberge and l. souhami

FIGURE 17-1. Publications on the radiosurgical treatment of HGG.

nature of SRS coupled with the promising early results of interstitial brachytherapy led to the inﬂated enthusiasm for the technique in patients with HGG and a growing number of publications became available over the years (Fig. 17-1). Most of these reports were published before the results of the randomized trials of brachytherapy and SRS were available. Single-fraction stereotactic radiation has predominately been used in one of two scenarios: in primary lesions as a supplement to wider volume external-beam radiation therapy (EBRT) and in recurrent lesions as a single modality. With the advent of noninvasive immobilization devices, the use of fractionated stereotactic radiation therapy (F-SRT) has become more common. Initially, F-SRT was explored to increase the therapeutic ratio in previously irradiated patients; it has since also been explored as a boost in the treatment of newly diagnosed HGG patients.

Treatment Planning Although each case must be approached on an individual basis, the following general guidelines represent a reasonable approach to the delivery of stereotactic therapy for HGG.

possible rapid tumor progression. Completely resected tumors are generally not considered for SRS boosts. The maximum clinically tolerated doses for the treatment of HGG has been derived from the RTOG 90-05 phase I study and is tumor size dependent. In the study by Shaw et al. [37], single doses of 24, 18, and 15 Gy were found to be the maximum tolerated doses for tumor diameters of ≤20 mm, 21 to 30 mm, and 31 to 40 mm, respectively. These doses are usually prescribed to an isodose surface between 50% and 90%. Caution should be exercised when treating brain-stem lesions and lesions within 10 mm of the optic chiasm, as these lesions were not included in the RTOG study and the maximum tolerated radiosurgery dose has not been clearly established for these structures. For recurrent disease, similar principles apply. It is a general rule that only lesions smaller than 4 cm are considered for SRS. In these cases, the GTV is the enhancing lesion on a gadolinium-enhanced T1 MRI (or contrast-enhanced CT scan). On a case-by-case basis, a clinical target volume (CTV) of a few millimeters (∼2 to 5 mm) can be added. For rigid immobilization systems, no additional margin is added for the PTV. Table 17-2 contrasts commonly used treatment parameters for stereotactic irradiation in HGG.

F-SRT For both primary and recurrent lesions, larger volumes can be considered for F-SRT than SRS. Depending on the planned dose of F-SRT and the use (concurrent or prior) of EBRT, volumes of up to approximately 100 cm3 can be considered for treatment. Margins will depend on the volume to be treated, proximity of critical structures, prior treatment, and immobilization device used. A reasonable schema would have the GTV equal the enhancing lesion on a gadolinium-enhanced T1 MRI (or contrast-enhanced CT scan), no CTV margin, and a PTV of 2 to 3 mm (corresponding with the reproducibility of a common three-ply thermoplastic mask-based immobilization system [38, 39]). Table 17-3 [40–47] contains various published fractionation schemes. Depending on the number of isocenters, dose would be prescribed to the 50% to 90% isodose surface after assessing the plan with regard to volume irradiated, subjective isodose distribution, conformity, homogeneity, and dose to critical structures. Currently, there is no evidence to suggest that higher MDPD (maximum dose/prescription dose) ratios should be favored.

SRS In the primary treatment of HGG, SRS is used as a boost to EBRT. The goal of radiosurgery is to inﬂict precise damage to tissue within the target volume, in this case proliferating glial cells. Thus, only tumors with a limited diameter (at presentation or postoperatively) should be considered for SRS. The RTOG considered a diameter of 40 mm the maximum diameter allowed for patient entry into the SRS trials. The target volume (PTV) is the tumor (as viewed on computed tomography [CT] or magnetic resonance imaging [MRI]) without margins. If MRI registration is used, a contrast-enhanced CT scan is obtained on the day of the radiosurgery to prevent mistargeting because of

TABLE 17-2. Comparison of common parameters for stereotactic irradiation modalities. Parameter

SRS

F-SRT

Tumor size

≤40 mm 15 to 24 Gy Single 50% to 80% None Rigid

≤60 mm Variable Multiple 80% to 90% 2 to 5 mm Relocatable

Total dose Fractions Prescription isodose Margins Stereotactic frame

17.

211

high-grade gliomas

TABLE 17-3. Results of F-SRT for primary and recurrent high-grade gliomas. Lederman [43]‡

Regine [44, 45]

Baumert [46]

Cho [47]

Staten Island

Kentucky

Zurich

Minnesota

2000 88 recurrent

2000 8 primary 1 recurrent GBM 6 AA 2 LGG 1

2003 17 primary

2004 10 primary

GBM 15 AA 2

GBM 10

Author

Glass [40]*

Shepherd [41]

Hudes [42]

Institution

Temple

Date of publication Number of patients

1997 20 recurrent

Royal Marsden 1997 33 recurrent

Thomas Jefferson 1999 20 recurrent

Histology

GBM 13 AA 7

AA 29 AO 3

GBM 19 AA 1

GBM

Median age (range)

44.5 (6–73)

37 (19–55)†

52 (26–77)

56 (21–82)

≥18

51 (25–64.8)

57 (17–80)

KPS

—

80 (60–100)†

80 (60–100)

70 (50–90)

≥60

70 (60–90)

Time from Diagnosis (for recurrent lesions) Median follow-up (months) Median tumor volume (range) Median peripheral dose (range) Median survival (months) 1-year survival (%) 2-year survival (%) Median prescription isodose (range) EBRT

—

29†

3.1 (0.8–45.5) from EBRT

N/A

(5/12 WHO 0/1, 7 WHO 2) N/A

N/A

—

—

>12

15†

25 (9–50)

—

14.3 (1.76–122)

24 (3–93)†

12.66

32.7 (1.5–150.3)

7.4 (1.5–27.2)†

42/7 (—)

35/7 (20/4–50/10) AA 11

30/10 (21/7–35/10) 10.5

24/4 (18/4–36/4) 7

14–28/2–4

— — 90 (80–90)

20 — 89 (80–95)

17 2 90 (80–90)

56 44 50

All prior EBRT (45–60) 6†

All prior EBRT 60 (44–72) 25 (for progression) No GR III

99%

12.7 — — ≥70 All prior EBRT?

Reoperation (%)

25

Toxicity

15% necrosis

45% at 24 months†

8.5

6.5

34 (4–70) 15 20/5 2 10/2 20

27.5/11 (20–35/8–14) 15.9 67 — 88 (75–90)

59.4/33

77 42 Prescribed at isocenter 60

60 (50.4–60)

12

57†

11/17

20

—

4/15 GR IV†

6% necrosis

10% necrosis

11

*With cis-platinum. †Of a larger group including other histologies. ‡With Taxol.

Linac Versus Gamma Knife HGGs are large, inﬁltrating tumors where submillimeter accuracy in delivery should not be an issue. Because of the necrotic and presumed radioresistant core of high-grade tumors, there might be a theoretical advantage to more inhomogeneous plans. As Gamma Knife (GK) treatments commonly use multiple isocenters and are prescribed to lower isodose surfaces, they might offer a theoretical advantage over single-isocenter plans (if this were ever clinically veriﬁed, linac centers could then choose to prescribe treatments to lower isodose surfaces). This hypothesis was tested by Larson and colleagues at UCSF in a phase II trial [48]. In this trial, prescriptions were made to isodoses as low as 25%. The results, however, appear no different than would be expected with conventional treatment (15 weeks median survival for patients with recurrent grade IV tumors).

A review of the single-institution series will reveal a mix of GK and linac-based treatments. The two largest series are from Pittsburgh [49] and Boston [50]–-prototypical GK and linacbased programs. Not unexpectedly, there is no difference in result between these two series–-both reporting a median survival of 20 months for the treatment of primary tumors. In the initial RTOG phase I trial of radiosurgery for recurrent tumors, there appeared to be a large difference in results favoring GK in the treatment of a mixed bag of primary brain tumors [51]. This is not borne out in the much larger and more homogeneous experience of the RTOG 93-05 randomized trial [52]. In this trial, in a subgroup analysis, patients treated with linac systems had a 14-month survival, not statistically different from the 12.1-month survival of GK patients (Fig. 17-2). If patients are to be treated with radiosurgery, the delivery system should not be an issue.

212

d. roberge and l. souhami 100 Linac

Survival Rate

80

Gamma Knife

60

40

20

0 0

6

12

18 24 30 36 Months FIGURE 17-2. Survival by SRS treatment delivery system in RTOG 93-05. (Reprinted from Souhami L, Seiferheld W, Brachman D, et al. Randomized comparison of stereotactic radiosurgery followed by conventional radiotherapy with carmustine to conventional radiotherapy with carmustine for patients with glioblastoma multiforme: report of Radiation Therapy Oncology Group 93-05 protocol. Int J Radiat Oncol Biol Phys 2004; 60:853–860. With permission from Elsevier.)

SRS in the Primary Management of HGGs The 1990s saw the publication of several small, retrospective and prospective, series of patients treated with SRS boost in the primary treatment of malignant glioma (Table 17-4) [53–58]. Ages ranged from 3 to 84 years and Karnofsky performance score (KPS) from 50 to 100. Overall, the median survival of patients treated with SRS was quite encouraging, although some series reported no beneﬁt in survival when comparing results with historical controls. However, in a disease where more than 95% of patients will ultimately die of their disease, the median survival is dictated more by patient-related than therapy-related factors [4]. By the very nature of SRS, patients are selected for small tumors, more complete resection (size determination being based on postoperative imaging), and a good response to initial therapy (when SRS is administered after EBRT ± chemotherapy, patients with progressive disease or decreasing performance status are likely to be excluded). In addition, trials commonly require a good performance status (KPS of 50 [49], 70 [50, 55], or 90 [56]). Other potential sources of bias include high patient motivation, more favorable tumor biology, and aggressive treatment of recurrences. Taking these selection factors into account, only one-tenth to one-quarter of GBM patients will be eligible for radiosurgical boost [59]. The results of radiosurgical series were viewed with skepticism. In an attempt to reduce bias, authors sought out retrospective control populations. One of the largest reviews [60] attempted to retrospectively stratify 115 patients from 3 institutions (Harvard, the University of Florida, and the University of Wisconsin) according to the prognostic classes of the RTOG recursive partitioning analysis of patients enrolled on RTOG 74-01, 79-18, and 82-02 [61–63]. This analysis concluded that there was a signiﬁcant improvement in both 2-year and median survival favoring SRS-treated patients (Table 17-5). Unfortunately, along with other ﬂaws, the RTOG classes are broad, do

not include all known prognostic factors (most notably tumor size), and convert important linear variables into binary ones (age, mental status, and KPS). Thus, this approach, along with all retrospective comparisons, is inherently ﬂawed. Irish et al. [59] published an elegant analysis of 101 consecutive patients seen at the London regional cancer center. Of these, 27% were deemed eligible for radiosurgery. The median survival of these patients was 23.4 months compared with 8.6 for the radiosurgery-ineligible patients. This study highlights the signiﬁcant effects of patient selection. It also demonstrates that a patient group can be selected that performs better than the most favorable group of grade IV tumors in the RTOG recursive partitioning analyses–-the RTOG class III patients (age >50, KPS ≥90) for whom the expected median survival is 18 months. The bottom line is that only randomization can control efﬁciently for all confounding factors. Having previously demonstrated the feasibility of a multiinstitutional radiosurgery trial [51] and encouraged by the favorable survival of patients treated on phase I/II protocols, the RTOG opened protocol 93-05 in 1994. This was a prospective randomized trial evaluating upfront SRS followed by EBRT with carmustine (BCNU) (arm 1) versus EBRT and BCNU (arm 2). Eligibility criteria required patients to be at least 18 years of age, have a KPS of at least 60%, and to have a histologically proved supratentorial, unifocal GBM. All lesions were to be 4 cm or less in maximal diameter. Patients presenting with tumors larger than 40 mm preoperatively were eligible only if the postoperative imaging studies showed residual tumors of ≤40 mm in maximal diameters. In the investigational arm, radiosurgery was to be given up front, within 5 weeks of surgery. External beam radiation was the same in both arms: a volume encompassing the tumor, surrounding edema (for the ﬁrst 46 Gy), and a margin was treated to a total of 60 Gy with daily fractions of 2 Gy. BCNU was administered at a dose of 80 mg/m2 on days 1, 2, and 3 of EBRT and then repeated every 8 weeks for a total of 6 cycles. The ﬁrst patient was treated in February 1994 and the study was closed in June 2000 after the 203rd patient was enrolled (thus meeting the accrual target of 200 patients). Results were published in 2004 [52]. Figure 17-3 illustrates the case of a patient treated on the experimental arm. At a median follow-up of 61 months (Fig. 17-4), the median survival for arm 1 was 13.5 months (95% CI: 11.0 to 14.9) and it was 13.6 months (95% CI: 11.3 to 15.2) for arm 2 (p = 0.53). The 2-year actuarial survival rates were 21% for arm 1 and 19% for arm 2. Disappointingly, patterns of failure were not inﬂuenced by the therapy with 90% of patients presenting with a component of local failure. There was also no difference in the mini-mental or Spitzer [64] indices (a validated quality of life index). Despite the exclusion from analysis of seven patients having progressed prior to SRS, no subgroup was identiﬁed as beneﬁting from SRS. An intention-to-treat analysis including all randomized patients produced nearly identical results. Both acute and late RTOG grade 3 toxicities were more frequent on arm 1, although not signiﬁcantly so. There were no incidences of radiation-related grade 4 toxicity in either arm. Five patients did die of chemotherapy-related toxicity on arm 1 compared with two patients on arm 2.

17.

213

high-grade gliomas

TABLE 17-4. SRS in the primary management of HGGs. Authors

Selch [53]

Masciopinto [54]

Gannett [45]

Buatti [56]

Kondziolka [49]

Shenouda [57]

Shrieve [50]

Nwokedi [58]

Institution

UCLA

University of Wisconsin

University of Arizona

University of Florida

University of Pittsburgh

McGill University

Harvard University

University of Maryland

Date of publication

1993

1995

1995

1995

1997

1997

1999

2002

Number of patients

18

31

30

11

107 (65 primary lesions)

14

78

31

Histology

12 GBM 6 AA

GBM

17 GBM 10 AA

6 GBM 5 AA

45 GBM 20 AA

GBM

GBM

GBM

KPS

100% >70

57% >70

97% >70

All >90%

Mean KPS 90 (50–100)†

79% >70

Median 90 (50–100)

61% >70

Median tumorvolume (cm3)

20 (8–46)

16 (2–60)

24 (2–115)

14 (6–23)

6.5 (1–31)†

<34

10

25

Median MPD (range)

30 (15–35)

12 (10–20)

10 (0.5–18)

13 (10–15)

16 (12–25)†

20

12 (6–24)

17

Sequence

Post-EBRT

Pre-EBRT 12 Post-EBRT17 No EBRT 2

Within 8 weeks (median 4)

12–109 days post-EBRT (typically 2–3 weeks)

GBM median 6.2 months postdiagnosis; AA median 3.9 months postdiagnosis

Pre-EBRT

Post-EBRT, median 14.2 weeks from diagnosis (range 1–42)

Within 4 week spostEBRT

GBM 13 AA 28

17

GBM 20 AA 56

10

19.9

25

Median survival (months)

9

9.5

1-year survival (%)

GBM 33 AA 100

37

GBM 43 AA 64.5

—

—

43

88.5

—

2-year survival (%)

GBM 33 AA 100

—

GBM 8 AA 53

—

GBM 41 AA 88

—

3