Extracranial Stereotactic Radiotherapy and Radiosurgery

Extracranial

Stereotactic Radiotherapy
and Radiosurgery

DK2966_FM.indd 1
Process CyanProcess
CyanProcess Magenta
MagentaProcess
Process Yellow
YellowProcess
Process Black

8/31/05 12:13:01 PM
Extracranial
and Radiosurgery
Edited by
Ben J. Slotman
Department of Radiation Oncology, VU University Medical Center
Amsterdam, The Netherlands
Timothy D. Solberg
Department of Radiation Oncology, University of Nebraska
Omaha, Nebraska, U.S.A.
Dirk Verellen
Department of Radiotherapy, AZ-VUB
Brussels, Belgium
DK2966_FM.indd 2
Process CyanProcess
CyanProcess Magenta
MagentaProcess
Process Yellow
YellowProcess
Process Black
8/31/05 12:13:01 PM
DK2966_Discl.fm Page 1 Monday, August 29, 2005 11:02 AM
Published in 2006 by
Taylor & Francis Group
270 Madison Avenue
New York, NY 10016
2006 by Taylor & Francis Group, LLC
No claim to original U.S. Government works
Printed in the United States of America on acid-free paper
10 9 8 7 6 5 4 3 2 1
International Standard Book Number-10: 0-8247-2697-9 (Hardcover)
International Standard Book Number-13: 978-0-8247-2697-3 (Hardcover)
This book contains information obtained from authentic and highly regarded sources. Reprinted material is
quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable efforts
have been made to publish reliable data and information, but the author and the publisher cannot assume
responsibility for the validity of all materials or for the consequences of their use.
No part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic,
mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and
recording, or in any information storage or retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.com
(http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC) 222 Rosewood Drive,
Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and registration
for a variety of users. For organizations that have been granted a photocopy license by the CCC, a separate
system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only
for identification and explanation without intent to infringe.
Library of Congress Cataloging-in-Publication Data

Catalog record is available from the Library of Congress
Visit the Taylor & Francis Web site at

http://www.taylorandfrancis.com
Taylor & Francis Group
is the Academic Division of T&F Informa plc.
Preface
A few years ago, out of common interest in the emerging eld of extracranial body radiosurgery, the editors considered it worthwhile to summarize
the on-going efforts by different research centers. An added benet was that
the editors themselves have different backgrounds illustrating the truly
multidisciplinary character of radiation oncology. Moreover, as the editors
live and work in either Europe or the United States the subtle differences in
approaching clinical issues will be represented in the work creating a bridge
between both continents. As always the subject of this work is not new, yet
technological evolution often creates new possibilities, and innovative centers have put large efforts in nding individualized solutions to these common challenges. Inevitably, a myriad of treatment strategies can be found
in the literature; the current work aims at being a comprehensive overview
of emerging developments in this sub-specialty in radiation oncology, as
well as illustrating possible clinical applications of these new and challenging technologies and approaches. As the basic concept was to generate a
general, objective, and comprehensive overview without overlooking any
possible strategy, great care has been taken to invite specialists in their
respective elds to act as contributing authors. In retrospect the editors
believe this objective has been reached and hope the reader will nd this
book to be a truly practical reference work on the topic.
Enjoy reading
Ben J. Slotman
Timothy Solberg
Dirk Verellen
iii
Contents
Preface . . . . iii
Contributors . . . . ix
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Ben J. Slotman, Timothy Solberg, Reinhard Wurm, and
Dirk Verellen
2. The ELEKTA Stereotactic Body Frame . . . . . . . . . . . . . . 5
Jorn Wulf
Introduction . . . . 5
System Description . . . . 6
Accuracy of Patient and Target Reproducibility
in the SBF . . . . 9
Treatment Planning and Delivery . . . . 13
Conclusion . . . . 17
References . . . . 17
3. Novalis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Paul Medin and Dirk Verellen
System Description . . . . 20
Patient Positioning . . . . 21
Verication and Clinical Validation . . . . 28
Treatment Planning and Treatment Delivery . . . . 36
Treatment Verication . . . . 40
v
19
vi
Contents
Treatment of Moving Targets . . . . 57

Capabilities and Limitations of the Technique . . . . 64
4. Whole-Body Radiosurgery with the CyberKnife . . . . . . . .
Achim Schweikard, Hiroya Shiomi, Minaro Uchida, and
John R. Adler
System Overview . . . . 75
Inverse Planning . . . . 75
Respiration Tracking . . . . 78
Clinical Trials . . . . 84
Conclusion . . . . 86
5. Serial Tomotherapeutic Approaches to Stereotactic Body
Radiation Therapy . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Bill Salter and Martin Fuss
System Description and Operation . . . . 89
Patient Immobilization and Alignment . . . . 91
Treatment Planning . . . . 94
Treatment Verication . . . . 100
Treatment of Moving Targets . . . . 104
Logistics of the Method . . . . 106
Future Directions . . . . 107
71
89
6. Anatomical and Biological Imaging . . . . . . . . . . . . . . . . 109

Frank J. Lagerwaard and Suresh Senan
Appropriate Selection of Patients . . . . 110
General Principles of Target Denition . . . . 113
Generating Representative Planning CT Scans . . . . 114
Repeated Treatment Planning in Between Fractions . . . . 119
Site Specic Comments . . . . 120
Follow-Up Imaging . . . . 122
Conclusions . . . . 122
Contents
7. Radiobiological Considerations of Stereotactic Body

Radiotherapy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Steve P. Lee, H. Rodney Withers, and Jack F. Fowler
Introduction: Clinical Radiobiology as a
Four-Dimensional Problem . . . . 131
Radiobiology Problem Along the Temporal Domain:
Time, Dose, and Fractionation . . . . 133
From Temporal Problem to Spatial Consideration: The
Double-Trouble Effect . . . . 146
Radiobiology Problem Along the Spatial Domain:
The Volume Effect . . . . 148
Radiobiology Problem in Four-Dimensions . . . . 162
Specic Issues Regarding SRS and SRT . . . . 164
Note Added in Proof . . . . 169
vii
131
8. Stereotactic Radiation Therapy of Liver Tumors . . . . . . . 177

Klaus K. Herfarth and Martin Fuss
Incidence . . . . 178
The Role of Stereotactic Body Radiation Therapy
in the Multi-Disciplinary Treatment Arsenal . . . . 179
Organ-Specic Difculties . . . . 181
Doses and Clinical Outcome . . . . 183
Partial Liver-Radiation Using
Stereotactic Setup . . . . 185
Ongoing Studies . . . . 189
Future Research . . . . 190
9. Stereotactic Radiotherapy of Lung Tumors . . . . . . . . . . .
Robert D. Timmerman and Jorn Wulf
Lung Metastases . . . . 202
Radiobiology . . . . 203
Radiation Tolerance of Pulmonary Tissues . . . . 207
Treatment Outcome . . . . 216
Local Control and Survival . . . . 220
Conclusion and Future Directions . . . . 226
197
viii
Contents
10. Prostate Tumors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231

Raymond Miralbell, Meritxell Molla`, Lluis Escude,
Dirk Verellen, Guy Soete, and Guy Storme
The TEKNON Experience . . . . 231
Introduction and Rationale . . . . 231
Hypofractionated Boost with Intensity Modulated X-Ray Beams
Under Stereotactic Conditions: A Pilot Study . . . . 233
Final Comments and Conclusions . . . . 244
The AZ-VUB Experience . . . . 246
Introduction and Rationale . . . . 246
Treatment Protocol and Irradiation Technique . . . . 246
Clinical Follow-Up . . . . 248
11. ECSRT for Spinal Tumors . . . . . . . . . . . . . . . . . . . . . .
Samuel Ryu and Fang-Fang Yin
Introduction and Clinical Indications . . . . 257
Techniques for Spinal Radiosurgery . . . . 260
Clinical Applications and Outcomes . . . . 275
257
12. Stereotactic Radiotherapy of Head and Neck Tumors . . .

Robert Smee and Reinhard Wurm
Background . . . . 286
Technique and Indications . . . . 289
Incidence . . . . 308
Organ Site-Specic Difculties . . . . 309
Fractionated Dose Selection . . . . 311
Treatment Delivery . . . . 313
Logistics . . . . 314
Future Use and Research . . . . 316
285
Index . . . . 321
Contributors
John R. Adler Stanford University Medical Center, Stanford,

California, U.S.A.
Lluis Escude Servei de Radio-oncologia, Instituto Oncologico Teknon,
Barcelona, Spain
Jack F. Fowler Department of Human Oncology, University Hospital,
Madison, Wisconsin, U.S.A.
Martin Fuss Department of Radiation Oncology, The University of Texas
Health Science Center, San Antonio, Texas, U.S.A.
Klaus K. Herfarth Department of Radiation Oncology, University of
Heidelberg, Germany
Frank J. Lagerwaard Department of Radiation Oncology, VU University
Medical Center, Amsterdam, The Netherlands
Steve P. Lee Department of Radiation Oncology, UCLA Medical Center,
Los Angeles, California, U.S.A.
Paul Medin Department of Radiation Oncology, UCLA Medical Center
Raymond Miralbell Servei de Radio-oncologia, Instituto Oncologico
Teknon, Barcelona, Spain and Service de Radio-oncologie, Hopitaux
Universitaires, Geneva, Switerland
ix
Contributors
Meritxell Molla` Servei de Radio-oncologia, Instituto Oncologico Teknon,

Barcelona, Spain
Samuel Ryu Department of Radiation Oncology and Neurosurgery,
Henry Ford Hospital, Detroit, Michigan, U.S.A.
Bill Salter Department of Radiation Oncology, The University of Texas
Health Science Center, and Cancer Therapy and Research Center, San
Antonio, Texas, U.S.A.
Achim Schweikard
Luebeck University, Luebeck, Germany
Suresh Senan Department of Radiation Oncology, VU University

Hiroya Shiomi
Osaka University Hospital, Osaka, Japan
Ben J. Slotman Department of Radiation Oncology, VU University

Robert Smee Department of Radiation Oncology, Prince of Wales
Hospital, Randwick, New South Wales, Australia
Guy Soete
Department of Radiotherapy, AZV-UB, Brussels, Belgium
Timothy Solberg Department of Radiation Oncology, University of

Nebraska, Omaha, Nebroska, U.S.A.
Guy Storme Department of Radiotherapy, AZV-UB, Brussels, Belgium
Robert D. Timmerman Department of Radiation Oncology, Indiana
University School of Medicine, Bloomington, Indiana, U.S.A.
Minaro Uchida Stanford University Medical Center, Stanford,
California, U.S.A.
Dirk Verellen
H. Rodney Withers Department of Radiation Oncology, UCLA

Medical Center, Los Angeles, California, U.S.A.
Jorn Wulf Department of Radiotherapy, University of Wurzburg,
Wurzburg, Germany
Contributors
Reinhard Wurm Abteilung Strahlentherapie, Universitat Klinikum,

Charite, Berlin, Germany
Fang-Fang Yin Department of Radiation Oncology, Duke Medical
Center, Detroit, Michigan, U.S.A.
xi
1
Introduction
Ben J. Slotman
Department of Radiation Oncology, VU University Medical Center,
Timothy Solberg
Department of Radiation Oncology, University of Nebraska, Omaha,
Nebraska, U.S.A.
Reinhard Wurm
Abteilung Strahlentherapie, Universitat Klinikum, Charite, Berlin, Germany
Dirk Verellen
Department of Radiotherapy, AZ-VUB Brussels, Brussels, Belgium
The concept of stereotactic radiosurgery was rst described by Lars Leksell

in 1951, as a single-fraction irradiation of intracranial targets, which, in
selected patients, would replace surgery. Stereotactic radiosurgery (SRS) is
characterized by the delivery of a high radiation dose to a small volume
in a short time with high accuracy and high conformality. This is performed
using a stereotactic technique, in which the location of a target is related to a
three dimensional Cartesian coordinate system. Based on this concept, any
intracranial localization can easily be identied in relation to the frame,
which is xed to the head.
Following the rst use of an orthovoltage dental X-ray tube in 1951
and the investigation of protons and early-generation linear accelerators
(linacs), Leksell and Larsson started their clinical work in 1967 with the
gamma knife. The latter in its current version consists of 201 cobalt-60
1
Slotman et al.
sources focused at one locus. The clinical usefulness of the GammaKnife

for the treatment of cranial base tumors and vascular malformations lead
to further technical developments and the GammaKnife is presently in use
in over 100 sites throughout the world. SRS using the GammaKnife carries,
depending on the size, shape, and number of lesions, the need to use multiple
isocenters with different beam diameters (collimator helmets), to produce
dose plans that conform to irregular lesions. A derivative of the GammaKnife, the RGS system with 30 cobalt-60 radiation sources contained in a
revolving hemispheric shell, was recently developed. The secondary collimator
is a coaxial hemispheric shell with six groups of ve collimator holes that are
arranged in the same pattern as the radiation sources. By selection of which
one of the groups of collimator holes is aligned with the radiation sources,
different beam diameters can be produced avoiding the use of different collimator helmets to change beam diameters during treatments.
Protons for irradiation of deep-seated targets were rst explored in the
early 1950s. It was shown that larger proton beams may be benecial for the
treatment of malignant tumors, while narrow beams could be utilized for
SRS of small circumscribed targets in the brain. However, the physical
advantages of protons including the Bragg ionization peak, to reduce doses
to the tissue beyond the target to a minimum, have not been substantiated
by clinical data or in comparative trials. The high cost and limited number
of facilities available are certainly factors that preclude the more widespread
use of protons in SRS. Only a small percentage of patients currently treated
with protons receive stereotactic proton-beam radiosurgery.
The widespread availability of linear accelerators (linacs) has lead to
investigations on their use for SRS and stereotactic radiotherapy (SRT).
The term SRS is generally reserved for single fraction stereotactic treatments, while the term SRT is reserved for fractionated stereotactic irradiation. Leksell rst explored linac-based radiosurgery. In the early 1980s,
following the description of the dosimetry of subcentimeter elds, BarciaSalorio, Betti, Colombo, Sturm, and others developed radiosurgery equipment to adapt linacs to produce highly collimated narrow beams for SRS.
Linac-based radiosurgical technologies were since then further advanced
by incorporating improved stereotactic positioning devices and methods
to measure the accuracy of various components. Over 500 such systems
are thought to be presently in operation throughout the world. Most of
these systems use circular collimators in combination with multiple isocenters techniques in which the patient couch is set at different angles and the
radiation source describes an arc with the isocenter at its center, to deliver
radiation that enters the cranium at many different points. More recently,
different techniques have been developed to enhance the conformity of dose
planning and delivery using linac-based systems. These include the use of
micro-multileafcollimators for beam shaping and intensity modulation.
Beam shaping involves modication of the traditionally circular contour
Introduction
of the radiosurgical collimators, so that the contour more accurately conforms to the beams eye view of the target volume. Conformal blocks and
micro-multileafcollimators can be used for static or dynamic beam shaping
and require only a single isocenter. Using beam intensity modulation, the
intensity can be varied across the beam and is weighted to be proportional
to the target thickness, as assessed by the beams eye view to obtain target
conformity. The CyberKnife (Accuray, Sunnyvale, California, U.S.A.) uses
a 6-MeV linac attached to a six-axis robotic manipulator that positions the
linac at different source positions, always aiming the center of the radiation
beam at the target. During treatment, an image-processing system acquires
X-ray images of the cranium of the patient and compares the actual images
with images in a database, to determine the direction and amount of any head
motion. This information is transferred to a robot, which corrects for this
motion.
In linac radiosurgery, an invasive frame, which is attached to the head of
the patient using pins, was used as reference frame for imaging, localization,
and treatment. After placing this frame, or headring, a localizer frame is
attached to it. This localizer frame has a number of ducials, which enable
the transformation of image coordinates to stereotactic coordinates. The rigid
system allows very accurate (re)positioning of the patient and targeting of the
radiation beams. The disadvantages of this headring system are that it cannot
easily be used for fractionated treatments over a longer period of time and that
it can only be used for the treatment of intracranial lesions. For fractionated
treatments of intracranial lesions, non-invasive systems have been developed
making use of dental and occipital impressions, or xation using ears and nose
bridge. These xation systems can be used as a frame of reference for imaging,
treatment planning, and stereotactic treatment delivery.
The current growth of SRS is certainly connected to the tremendous
advances in computer technologies and imaging in the past decade. Current
dose-planning programs provide on-screen integration of (multimodality)
images with the isodose curves, signicantly reducing treatment-planning
time. Recent versions facilitate the use of inverse planning, in which the target is three-dimensionally dened and the software, based on constraints
and penalty-functions, generates a treatment plan, that can then be further
adjusted and optimized. The advances in imaging techniques have improved
long-term results due to better target denition and localization. Integrated
use of magnetic resonance imaging (MRI) is now regarded standard because
of its high-resolution and excellent tissue contrast providing improved
anatomic detail. However, the reliability of MRI for stereotactic procedures
is related to the stereotactic frame used and/or fusion of CT and MRI data.
Techniques, such as positron emission tomography and magnetic resonance
spectroscopy, provide important information for tumor localization and
contouring and may enhance our understanding of the radiobiological
effects of radiosurgery on different tissues even further.
Slotman et al.
The great success of stereotactic radiosurgery and stereotactic radiotherapy for intracranial lesions has led to an interest in the use of these
techniques for the treatment of lesions outside the brain. The pioneering
work in this area was done by the groups of Lax and Blomgren at
Karolinska Hospital in Stockholm. Compared with intracranial SRS and
SRT, extracranial SRS and SRT are more difcult, due to motion of targets
and normal tissues. In the last decade, various approaches to deliver radiation to extracranial targets with stereotactic precision have been developed.
These include the use of elaborate immobilization techniques and introduction of so-called image-guidance technology aiming at reducing patient
set-up errors, and assessment of organ motion and organ changes during the
course of treatment. Internal organ and tumor movement during treatment
not only introduces an added risk of missing the target, but also introduces
errors in the dose delivery, which in itself may have become variable in time
for most intensity modulated techniques. With the introduction of intensity
modulation not only the possibility of geographic miss will be a matter of
concern, but also the possible interplay between target motion and the temporal dose delivery, and real-time knowledge of the target volume becomes
of utmost importance. Stereotactic body frames emphasize on immobilization where the device is used for patient xation, external reference system
for determination and localization of the stereotactic coordinates, and a
mechanical tool for reduction of breathing mobility. Image guidance in turn
emphasizes on real time, 3D knowledge of target localization during treatment and guiding the dose delivery accordingly. These techniques include
the use of optically-guided tracking devices, in-room CT (fan beam and cone
beam), ultrasound, stereoscopic X-ray devices, or combinations thereof.
These techniques, in order to be realistically applied require a thorough
understanding of anatomy and physiology and cannot be implemented without proper preparation (preferably based on multimodality and 4D imaging
techniques).
Extracranial stereotactic radiotherapy is a rapidly expanding treatment modality, being offered to an increasing number of patients for an
increasing number of indications. In this book, the various techniques,
including linac-based systems using the ELEKTA Body System , Novalis ,
CyberKnife , and tomotherapy, will be described in detail. The radiobiological aspects of stereotactic radiotherapy and the use of new imaging modalities are discussed and the clinical results of extracranial stereotactic
radiotherapy for various tumor sites, including liver, lung, prostate, spine,
and head and neck are presented.
2
The ELEKTA Stereotactic
Body Frame
Jorn Wulf
Department of Radiotherapy, University of Wurzburg, Wurzburg, Germany
INTRODUCTION
The stereotactic body frame (SBF) 23has been created and developed by
Ingmar Lax, Ph.D. and Henric Blomgren, M.D., Ph.D. at Karolinska Hospital, Stockholm, Sweden. It is the merit of these two authors having introduced the successful concept of stereotactic irradiation of cerebral lesions
into treatment of extracranial targets at the beginning of the 1990s (14).
While high precision of stereotactic irradiation of cerebral tumors is
achieved by sharp xation of the skull or tight mask systems, it is more
difcult in extracranial tumors, e.g., in the lung or liver. Sharp xation of
the patients body is impossible and the targets and organs at risk are potentially mobile due to breathing motions or changing organ llings. Additionally, marks on the patients body surface are less accurate than marks attached
to a mask system due to subcutaneous fat tissue, different llings (e.g., of
the abdomen) or breathing motions of the body. Nevertheless, stereotactic
irradiation according to the concept of Blomgren and Lax consists of precise
application of very high fraction doses, e.g., 2 l5 or 310 Gy prescribed to
the PTV-enclosing 65%-isodose and normalized to 100% at the isocenter (14).
Therefore, introduction of this concept into clinical practice requires a highprecision approach. For that purpose the SBF was developed and since the
mid-1990s, marketed by ELEKTA Instruments (ELEKTA AB, Stockholm,
Sweden). It has come into clinical use by groups all over the world (18).
ELEKTA Instruments, ELEKTA AB, Stockholm, Sweden.
Wulf
SYSTEM DESCRIPTION
The technical concept of the SBF addresses three basic requirements for
extracranial stereotactic radiotherapy, i.e., the need for:
1. patient xation,
2. an external reference system for determination, localization, and
alignment of the stereotactic coordinates, and
3. a mechanical tool for reduction of breathing mobility.
Patient Fixation
Patient xation is achieved by a vacuum mattress, which is molded to the
patients individual body contour. Two sizes of 25 or 40 L are available to
address different patient sizes. The mattress is xed to a plastic shell by two
screws. The shellmattress unit is inserted into the body frame by a system of
tongues and grooves. This allows a very easy and fast change of the shell
mattress unit for different patients without disconnecting the vacuum pillow.
Repositioning of the patient in the vacuum mattress for treatment is
supported by an SBF-attached laser system at the trunk and the legs. After
molding of the mattress, the patients position is marked by small marks tattoos at the trunk (preferably the sternum) and both tuberositae tibiae. The
position of the tattoos is indicated by a trunk laser attached to the stereotactic arc and a leg laser at a chosen position (Fig. 1, Nos. 6 and 7). For repositioning, the tattoos are aligned to the SBF-attached laser system at
previously determined positions.
The body frame itself is open ventrally and at the head and foot ends
(Fig. 1). It achieves rigidity by a honeycomb structured paper center embedded
in a berglass surface. The wooden edges are rounded to avoid artefacts, e.g.,
at CT scans. The low-density material of the SBF sidewalls is aimed to reduce
artefacts and to minimize absorption of irradiation. According to Lax et al.
(3), the geometrical specications of the SBF are within 0.5 mm. The outer
dimensions of the SBF are 111 cm in length, 50 cm in width, and 40 cm in
height. The complete system including rulers, indicators, and a bottom plate
for level control has a weight of about 9 kg.
The level control consists of a bottom plate loosely attached to the
SBF left ground side and a rubber bladder, which is pushed between the bottom plate and ground side of the SBF. Inating or deating air into the
bladder by a pump system (similar to a cuff for measuring blood pressure)
allows precise alignment of the SBF in the anteriorposterior direction to
the laser system of the CT or linac within a range of 5 mm.
The Stereotactic Reference System
The SBF is not only used for patient xation but also as external reference system for identication of the stereotactic isocenter. It consists of a
ELEKTA Stereotactic Body Frame
Figure 1 The ELEKTA stereotactic body frame (SBF). (1) Sidewall containing
oblique and horizontal copper wires for CT-based measurement of longitudinal
stereotactic coordinate. (2) Longitudinal stereotactic scale. (3) Stereotactic arc for
lateral and AP coordinates. (4) Arc and scaled screw for diphragm control. (5) Level
control. (6,7) SBF attached laser system (leg and trunk) for assistance at patient
repositioning. (8) Vacuum pillow.
longitudinal scale in millimeters, which is bilaterally xed to the outside of

the frame sidewalls. The lateral and anteriorposterior position in millimeters can be read from a stereotactic arc, which is attached to both sidewalls of the SBF at a chosen longitudinal position. The stereotactic arc is
also marked with a central notch (Fig. 2B), which allows precise alignment
of the SBF at a dened coordinate to the laser systems of uoroscopy, CT,
or linear accelerator.
This outer system for adjusting the SBF to the isocenter at a certain
stereotactic coordinate corresponds to an internal system, which allows
deriving the stereotactic coordinate from each CT slice. This internal
reference system consists of horizontal and oblique copper wires, which
are embedded in a plexiglas shell attached to both sidewalls of the SBF
(Fig. 2C). The same system is available with copper sulfate ll for use in
MRI. The copper wires appear as ducials in each CT slice (Fig. 2D). From
these ducials, the isocenter coordinates for each target can be calculated
(Figs. 2D and 3).
Breathing Control
Breathing mobility of targets in the lung, liver, or abdomen can be reduced
by a simple but effective mechanical tool: A pentagonal template is
Wulf
Figure 2 (A) Patient immobilized in the SBF with stereotactic arc and diaphragm
control. (B) Isocenter alignment of the SBF. The stereotactic arc is attached to the
frame at the planned longitudinal position, the AP and lateral coordinate can be read
at the stereotactic arc. (C) The longitudinal position in the SBF can be seen in each
CT slice due to a system of horizontal and oblique copper wires (D) e.g., ve
horizontal dots 500, distance to the oblique wire 70 mm: longitudinal position
570). (E) CT-slice of a patient immobilized in the SBF. ( F ) Two sizes of of templates and three sizes of screws are available. (G) Diaphragm control: A pentagonal
template is pushed into the patients epigastrium to increase abdominal pressure.
pushed into the patients abdomen by moving a scaled screw xed to an SBFattached arc. With this procedure the abdominal pressure is increased, leading to decreased motions of the diaphragm muscle. Instead of large breathing
motions up to 20 mm, the patient breathes with many smaller motions of
about 5 mm. For adjustment of this diaphragm control to different patient
sizes, a small and large template and three different lengths of screws are
available. Theoretically, the amount of pressure can be reproduced just by
pushing the screw to the same position as planned.
Dose Absorption of the SBF

For evaluation of dose absorption of the SBF, treatment plans including
and excluding the SBF in the calculation model were compared. Two
phantoms (Alderson phantom and thoracic phantom as described in the
ICRU Report 42) were used. By means of ionization chambers and TLD
measurements, the dose and dose distributions were compared with the
results of the 3D-treatment planning system Helax TMS. While the difference
between calculation and measurement was 5% at maximum if the SBF was
Figure 3 Schematic view of the SBF. The stereotactic coordinate can be derived and
calculated from a system of oblique and horizontal copper wires in the SBF sidewalls. This internal system corresponds to marks in millimeters on the outer sidewall
and the stereotactic arc, which allows precise alignment of the SBF to the isocenter of
a CT or linac according to the calculated coordinates.
included into the calculation model, the difference increased by additional

10% if the SBF was excluded (9). Therefore, the SBF should be included into
the calculation model. This approach additionally allows clinical estimation
of the skin dose for targets close to the patients surface, because the buildup effect is calculated more realistically.
ACCURACY OF PATIENT AND TARGET

REPRODUCIBILITY IN THE SBF
The accuracy of target reproducibility for stereotactic irradiation of extracranial targets not only depends on the xation device itself but also on
the users experience and the mobility of the targets chosen for treatment.
Principally, external xation by a vacuum pillow is more reliable in slim
patients than in obese individuals. Nevertheless, an accuracy of about 2 mm
can be achieved if the vacuum pillow is molded properly. Hence, a double-S-shape design of the vacuum pillow is desired (Fig. 4). It is achieved
laterally by molding the pillow very tightly to the patients waist, and longitudinally by molding the material to the patients lordosis of the lumbar
spine. The patient is asked to perform snake-like movements in the smooth
vacuum pillow. To achieve vacuum, the pillow material is tightly pressed
laterally to the patients surface. A comfortable position for both arms
has to be ensured while performing this.
10
Wulf
Figure 4 Accuracy of patient repositioning in the SBF is dependent on the quality

of molding of the vacuum pillow. For a maximum of reproducibility, an S-shaped
pillow is recommended. This is achieved by preparing a sufcient amount of pillow
material at the anticipated waist and lordosis of the patients back prior to positioning. By asking the patient to move like a snake with his/her back, the material is
moved to the correct site molded to the patients anatomy and the pillow can be
deated.
Accuracy of treatment in the SBF can be evaluated at three different

levels:
1. alignment of the SBF to the isocenter system of a CT or linac.
2. reproducibility of patient repositioning in the SBF especially in
fractionated treatment.
3. reproducibility of the target.
The alignment of the SBF to the isocenter system, e.g., of the
CT scanner, can be measured by comparing the aligned longitudinal coordinate to the measured coordinate in the corresponding CT slice. The results
not only depend on the SBF but also on the accuracy of the laser system.
Lax et al. (3) reported an accuracy of alignment of less than 1 mm derived
from 10-mm CT slices. Our own evaluation revealed a standard deviation of
1.4 mm derived from a 5 mm slice thickness. Nevertheless, maximum deviations of up to 3.9 mm were observed, indicating that misalignment occasionally might occur and that correct alignment should be proved prior to
proceeding with treatment planning or the treatment itself (10).
Accuracy of patient repositioning can be measured by comparing
immobile patient structures such as bones to the external reference system
of the SBF. Hence, for treatment planning the bony structures in the CT
are compared to the position in a CT-verication at a dened coordinate.
11
The differences can be measured using the external reference system of the
SBF. In our department, the rst 32 targets were evaluated (10). The mean
deviation of bony structures as derived from CT-verication was 2.9 mm
(SD 2 mm) in longitudinal, 2.2 mm (SD 1.8 mm) in anteriorposterior, and
2 mm (SD 1.9 mm) in lateral directions. The mean 3D-vector was 4.7 mm
(SD 2.6 mm). The reproducibility of bony structures can also be measured
by comparison of digitally reconstructed radiographs (DRR) from the planning CT to portal lms. In our analysis of 9397 verication lms for 32
targets, we found a mean deviation of 1.5 mm (SD 4.2 mm) in longitudinal,
0.1 mm (SD 2.3 mm) in anteriorposterior, and 0.1 mm (SD 2.5 mm) in
lateral directions.
While repositioning accuracy in the SBF is about 2 mm, it is the concept of extracranial stereotactic radiotherapy, as introduced by Blomgren
and Lax, to overcome isocenter verication relative to bony landmarks, a
common practice in conventional radiotherapy. The only relevant structure
to verify is the target relative to the stereotactic system and coordinates of
the SBF. Lax reported the reproducibility of 30 tumors evaluated with
48 verication CTs treated in the current version of the SBF. The targets
were located in the lung, liver, retroperitoneal space, and skeleton. The
mean deviation of the target was 3.4 mm in the transverse and 5.5 mm in
the longitudinal plane. In 98%, the transversal deviation was within 5 mm.
The longitudinal deviation was at 95% within 8 mm and 100% within
10 mm (3). Additionally, according to Lax, it was generally possible to keep
breathing mobility of mobile targets within 5 mm using the diaphragm control
device. Based on these results, most groups use security margins for PTVdenition to address target deviation of 5 mm in transversal and 510 mm
in longitudinal directions.
In our own analysis of 32 targets in the lung, liver, abdomen, pelvis,
and bones, the SD of all targets was 3.4 mm in anteriorposterior (mean
1.1 mm), 3.3 mm in lateral (mean 0.7 mm), and 4.4 mm in longitudinal directions (mean 1.5 mm), which corresponded well to the results of Lax. Nevertheless, we occasionally found maximum deviations of up to 12 mm leading
to a proportion of targets reproducible within 5 mm of 84% in anterior
posterior, 88% in lateral, and 91% in longitudinal directions. About 98%
of targets deviated within 10 mm in anteriorposterior and lateral directions
and 94% in the longitudinal direction. If a security margin for target variability of 5 mm in axial and 10 mm in longitudinal directions was used, these
results indicate that about 1216% of targets might be missed partially in
anteriorposterior and lateral directions and 9% of targets in the longitudinal
direction. Therefore, the conclusion was to recommend CT-verication prior
to irradiation to detect those targets with decreased reproducibility.
Isocenter verication relative to bony landmarks seemed to be inappropriate, at least for mobile targets, because this approach implies that deviation of bony structures is representative for target deviation. To prove this
12
Wulf
hypothesis, the CT-verication data were analyzed and it was postulated that
target deviation should be within 5 mm relative to bony landmarks. This was
true in only 62.5% of all 32 targets. Differentiated to different types of targets
as bony targets, soft tissue targets xed to bony structures, and mobile soft tissue targets breathing control device, major difference were observed. While
100% of bony targets and 80% of xed soft tissue targets deviated within 5 mm
relative to bony reference structures, this was the case in only 37.5% of mobile
soft tissue targets without breathing control and only 28.5% with breathing
control (10). These results again indicate the importance of CT verication
to eventually correct for major target deviation.
Nevertheless, the presented data are based on only one CT verication
prior to treatment. Therefore, it seemed necessary to evaluate target reproducibility over a complete course of treatment, e.g., three fractions as rst
described by Blomgren and Lax. A study including three CT verications
in each of 22 lung tumors and 21 liver tumors was performed (11). The main
goal of dening a planning target volume (PTV) is to ensure that the clinical
target volume (CTV) will be covered by the prescribed reference dose despite
target mobility. Therefore, the target reproducibility was evaluated by analyzing the CTV-dose by dosevolume histograms (DVH). For that purpose
the CTVs derived from repeated CT-verications were matched into the
planning study using the ducials of the SBF sidewalls as an independent,
external matching system. Major target deviation exceeding the PTV-related
reference isodose should result in decreased target dose to the CTV. Prior to
this evaluation, the conformity of the stereotactic dose distribution to the
PTV has been analyzed (12). For PTV denition, the commonly used security margins of 5 mm in axial and 510 mm in longitudinal directions were
added to the CTV. The study revealed a decrease of target coverage to
the CTV to less than 95% (5% of the CTV were not covered by the reference
isodose) in 3 of 60 simulations for lung targets (5%) and 7 of 58 verications
for liver targets (12%). Related to targets in 2 of 22 lung tumors (9%) and in
4 of 21 liver tumors (19%), a decreased target coverage <95% was observed
in at least one fraction. Two out of three major deviations in lung tumors
were observed in a single patient after pneumonectomy and in liver targets
six out of seven major deviations occurred in targets with a CTV >100 cm3.
These results again indicate the importance of CT verication to detect
occasionally occurring major target deviation beyond the reference isodose.
According to our study, patients with large liver tumors or mobile lung
tumors, in whom breathing mobility cannot be sufciently suppressed by
the breathing control device (e.g., after pneumonectomy), are at higher risk
for major target deviation (11). Theoretically an increase of security margins
would be able to compensate for increased target mobility. Nevertheless,
this seems not to be an appropriate approach, because the high-dose area
should be kept as small as possible especially in a patient group not amenable to surgery due to impaired medical condition.
13
TREATMENT PLANNING AND DELIVERY

Patient Positioning in the SBF and CT-Planning Study
Treatment planning can be performed within 30 min and starts with
premolding of the vacuum pillow. It can be performed at the uoroscopy or
CT unit. First, the patient is asked to use a hospital shirt ventrally open to
avoid too much clothing between the patient and the pillow. The hospital
shirt also allows easier and more accurate moving of the patient by pulling
the shirt as necessary. If application of i.v. contrast media is planned (e.g.,
for targets in the liver), the informed consent and an i.v. needle should be
placed prior to positioning the patient in the SBF. Depending on the
patients size, the 25 or 40 L mattress should be chosen. The pillow should
be prepared with a slight vacuum to avoid dislocation of the pillow material
under the patient during the positioning procedure. After premolding, the
pillow is attached to the plastic shell, which is inserted into the SBF. The
SBF is positioned on the end of CT-/uoroscopy-couch and the patient is
asked to rst sit down and then lay back onto the pillow. The SBF containing the patient is slipped to the scanning position, a standardized support
device is positioned under the knees, and the arms are comfortably moved
behind the head. In this position, the prevacuum is deated and the patient
is asked to perform some snake-like movements to achieve an optimal embedment in the vacuum pillow. Now the mattress can be deated maximally while
pressing the material laterally to the patient and supporting the arm to maximize comfort. At this point, the comfort of the patient should be queried
because a change of the mattress or patient position from this point is impossible without repeating the complete procedure. If there is doubt on the optimal embedment of the patient, an investigatory CT slice can be performed to
assess the quality of the embedment: The patient should be well enclosed by
the vacuum mattress with full contact of the patients back to the base of lateral and dorsal surface to the pillow. When the vacuum pillow is optimal, little
tattoos are pricked into the patients skin at the trunk and tibiae to ease later
repositioning for treatment. For this purpose the leg and trunk laser are
attached to the SBF at the chosen position, preferably at the sternum and
tuberositas tibiae. At the trunk, tattoos not only at the midline but also laterally are recommended to detect rotational errors at the time of repositioning.
If patient positioning is performed at the uoroscope, breathing
mobility of the target should be evaluated there andif necessarydiaphragm control should be used. If patient positioning is performed at the CT
unit, breathing mobility is evaluated later. In the CT situation, the SBF is
aligned to the isocenter of the CT by the use of the stereotactic arc. Correct
alignment is proved by comparing the aligned position to the measured position of the internal ducials in the SBF sidewalls. The planning study should
be performed, only if alignment is correct, otherwise irreproducible oblique
slices would be achieved.
14
Wulf
The CT study for treatment planning starts with obtaining a reference

slice, usually at the initially aligned position. From that position the target
must be found by moving the scanner directly to the estimated target position. This procedure can be performed without obtaining a scout scan. After
locating the target, breathing mobility must be evaluated if not performed
previously. While tumors in the lung are usually easy to locate, targets in
the liver might be assessed relative to vessels or the diaphragmatic dome.
CT evaluation of target mobility can be performed optimally by multiple
slice technique in modern CTs (13). Otherwise it can be evaluated by slow
CT scanning over some seconds (14) or by continuous dynamic scanning
at the same table position over a chosen period of time, e.g., 615 sec
(10,11). While axial mobility can easily be measured in each CT slice, cranio-caudal mobility must be assessed by comparison of the target shape in
several CT slices.
If breathing mobility is assessed to be more than 5 mm, the breathing
control device can be used. A pentagonal template (two sizes are available)
is positioned on the patients epigastrium. The exible arc is positioned at
the SBF and a scaled screw (three sizes are available) is attached. The screw
is moved on the template until a sufcient and patient-tolerable pressure is
achieved. Breathing mobility is evaluated again until an optimum is
achieved. Of course other modalities to decrease breathing mobility can
be used, such as oxygen-supplied shallow breathing or the use of an active
breathing control device (active breathing coordinatorTM, ELEKTA AB,
Stockholm, Sweden), and evaluated alternatively (1517). After breathing
control is achieved the patient is ready for the CT-planning study.
Depending on the size and location of the target, the CT for treatment
planning can be performed in 35 mm slice thickness. For lung tumors, i.v.
contrast is only necessary if the target has to be differentiated from normal
tissue structures; i.v. contrast is always required for liver tumors. To achieve
an optimum of contrast, the planning study should be performed under
almost diagnostic conditions. Therefore, an arterial and a portal-venous
phase should be achieved using a volume scan. Sometimes, an additional
venous phase is helpful to differentiate the target volume from normal liver
tissue. After the planning study is nished the patient is removed from the
SBF after being photographed to document the setup, e.g., the exact position of the arms.
Target Definition and 3D-Treatment Planning
3D-treatment planning requires denition of a patient model, CTV, PTV, and
organs at risk. As described previously (9), the SBF should be included into
the patient model. This not only increases reliability of the dose calculation
but also allows assessment of skin dose due to a more realistic calculation
of the build-up effect. No consistent data have been published concerning
15
target denition. While some authors report adding security margins for PTV
denition of 510 mm to the GTV, others add it to the CTV. Nevertheless, the
goal of stereotactic irradiation is to achieve local control. Therefore, local
tumorincluding microscopic diseaseshould receive the planned dose to
avoid local failure. The security margins for PTV-denition are exclusively
necessary to address potential target mobility and setup inaccuracy. Generally, the security margins should be added to the GTV or CTV using a digital
tool to ensure a 3D PTV-model.
For 3D-treatment planning of dose distribution, again different
doses, irradiation techniques, and normalization procedures are used.
Generally, a reference isodose is created, which should encompass the
PTV as conformal as possible. For that, 57 static beams individually modeled by multileaf collimators or rotational beams are created. Non-coplanar
beams and/or wedges can be added, but it should be considered that these
approaches might prolong treatment time. Especially in lung tumors, dose
calculation should be performed by a collapsed cone (point-kernel)
algorithm with low photon energy, because the widely used pencil beam
algorithm neglects the secondary charged particle disequilibrium at the
tumorlung interface and therefore might overestimate the target dose
considerably (1821).
Target Verification and Irradiation
A treatment session usually lasts for 3060 min depending on the target
verication procedure and the irradiated dose. While patient setup and repositioning in the SBF can be performed within ve minutes of target verication, the check of the correct isocenter coordinates relative to the target is
more time consuming. As described above, CT verication is preferred
compared to isocenter verication relative to bony landmarks.
Usually CT verication is performed at the CT scanner with subsequent transport of the patient in the SBF to the treatment room. This
approach requires additional isocenter verication at the linac to detect
patient dislocation due to transport and to document the correct isocenter
coordinates. Because patient dislocation can be detected by comparing bony
structures relative to the stereotactic coordinates, this procedure can be
performed by comparison of digital reconstructed radiography (DRR) from
the CT-verication study to portal images.
For CT verication the patient is repositioned in the SBF. The accuracy
of repositioning is eased by matching the laser of the trunk and leg to the tattoos on the patients skin at the determined positions. The diaphragm control
device is also positioned and adjusted as determined at the CT for treatment
planning. The SBF is aligned to the CT at the planned longitudinal stereotactic isocenter coordinate. Beginning from this position, one or multiple CT
slices are generated to nd the slice with the target shape that best matches
16
Wulf
the shape of the isocenter slice from treatment planning. For mobile tumors,
sufcient suppression of breathing mobility by the diaphragm control device
has to be controlled also. If breathing mobility is increased compared to the
planning study the diaphragm control device must be adjusted by increasing
pressure until a sufcient result is achieved. Under this condition it is usually
necessary to repeat the procedure to reproduce the target shape at the
planned isocenter slice. Finally, the stereotactic coordinates for irradiation
of the current target position can be measured relative to the internal SBF
reference system (Figs. 3 and 5). While the procedure determining the isocenter coordinates by CT verication is identical, performing this process
directly on the treatment couch has obvious advantages: The patient will
not dislocate due to transport, the immobilization time of the patient is shortened. Therefore potential changes, e.g., the abdominal pressure by the
diaphragm control are less probable. An example of CT verication at
the treatment couch using a mobile CT with gantry movement is shown in
Figure 5.
After CT- and/or isocenter verication are performed, the patient is
aligned to the isocenter of the linac at the current coordinates. Depending
on the practice of each department, the correct eld size, MLC positions,
and plan parameters should be checked prior to irradiation. Ideally, this
Figure 5 The CT verication of a small lung metastasis in the left lower lobe. The
verication is performed directly on the treatment couch using a mobile CT (Philips
Tomoscan M). The CT gantry moves over the patient positioned in the SBF as if for
treatment. A carbon ber couch allows scanning without artefacts in diagnostic quality. Compared to the planning study (right) according to CT-verication (left) the
target is dislocated 5 mm cranial (SBF longitudinal position 610 vs. 605), 4 mm
ventrally, and 4 mm medially of the isocenter position in the center of the tumor.
Again, the effectiveness of the breathing control device can be evaluated on the treatment couch by using the dynamic scan procedure.
17
can be performed parallel to evaluation of the CT verication to shorten the

overall immobilization time.
CONCLUSION
The SBF is an easy to handle and reliable stereotactic system. During 6 years
of use with more than 150 patients, no unforeseen events occurred, e.g., leakage of the vacuum pillow or breakage of important parts. The patient can be
immobilized sufciently upto an hour, which might be occasionally necessary
for patient repositioning, target verication, and irradiation. Nevertheless,
initial procedures such as molding of the vacuum pillow, tattooing, and setup
can be performed within 15 min due to the easy-to-handle system. Patient
setup at the linac as well is possible within 10 min.
With a properly molded vacuum pillow, repositioning accuracy of
about 2 mm can be achieved. Nevertheless, individual accuracy depends
on the patients condition and is more accurate in slim than in obese
patients. Target reproducibility as inuenced by breathing mobility can be
sufciently decreased to 58 mm using the mechanical breathing control
device and again is dependent on individual factors. Therefore, target mobility and reproducibility should be individually evaluated and controlled
using CT verication, if the commonly used security margins for PTV denition of 5 mm in axial and 510 mm in longitudinal direction are chosen.
REFERENCES
1. Blomgren H, Lax I, Goranson H, Krapelien T, Nilsson B, Naslund I,
Svanstrom R, Tilikidis A. Radiosurgery for tumors in the body: clinical experience using a new method. J Radiosurg 1998; 1(1):6374.
2. Blomgren H, Lax I, Naslund I, Svanstrom R. Stereotactic high dose fractionation radiation therapy of extracranial tumors using an accelerator. Acta Oncol
1995; 34:861870.
3. Lax I, Blomgren H, Larson D, Naslund D. Extracranial stereotactic radiosurgery of localized targets. J Radiosurg 1998; 1(2):135148.
4. Lax I, Blomgren H, Naslund I, Svanstrom R. Stereotactic radiotherapy of
malignancies in the abdomen. Methodological aspects. Acta Oncol 1994; 33:
677683.
5. Wulf J, Haedinger U, Oppitz U, Thiele W, Ness-Dourdoumas R, Flentje M.
Stereotactic radiotherapy of targets in the lung and liver. Strahlenther Onkol
2001; 177:645655.
6. Nagata Y, Negoro Y, Aoki T, Mizowaki T, Takayama K, Kokubo M,
Araki N, Mitsumori M, Sasai K, Shibamoto Y, Koga S, Yano S, Hiraoka M.
Clinical outcomes of 3D conformal hypofractionated single high dose radiotherapy for one or two lung tumors using a stereotactic body frame. Int J
Radiat Oncol Biol Phys 2002 Mar 15; 52(4):10411046.
7. Lee SW, Choi EK, Park HJ, Ahn SD, Kim JH, Kim KJ, Yoon SM, Kim YS,
Yi BY. Stereotactic body frame based fractionated radiosurgery on consecutive
18
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
Wulf
days for primary or metastatic tumors in the lung. Lung Cancer 2003 Jun;
40(3):309315.
Timmerman RD, Papiez L, McGarry R. Extracranial stereotactic radioablation: results of a phase I study in medically inoperable stage I non-small cell
lung cancer. Chest 2003; 125(5):19461955.
Richter J, Haedinger U, Bratengeier K, Flentje M, Wulf J. QA of stereotactic
treatment techniques in the body region. Radiother Oncol 1998; 48(suppl l):733.
Wulf J, Haedinger U, Oppitz U, Olshausen B, Flentje M. Stereotactic radiotherapy of extracranial targets: CT-simulation and accuracy of treatment in
the stereotactic body frame. Radiother Oncol 2000; 57:225236.
Wulf J, Haedinger U, Oppitz U, Thiele W, Flentje M. Impact of target reproducibility on tumor dose in stereotactic radiotherapy of targets in the lung and
liver. Radiother Oncol 2003; 66:141150.
Haedinger U, Thiele W, Wulf J. Extracranial stereotactic radiotherapy: evaluation of PTV coverage and dose conformity. Z Med Phys 2002; 12:221229.
Hof H, Herfarth KK, Muenter M. The use of the multislice CT for the determination of respiratory lung tumor movement in stereotactic single-dose
irradiation. Strahlenther Onkol 2003: 179(8):542547.
Lagerwaard FJ, Van Sornsen de Koste JR, Nijssen-Visser MR, SchuchhardSchipper RH, Oei SS, Munne A, Senan S. Multiple slow CT-scans for incorporating lung tumor mobility in radiotherapy planning. Int J Radiat Oncol Biol
Phys 2001 Nov 15; 51(4):932937.
Wong JW, Sharpe MB, Jaffray DA, Kini VR, Robertson JM, Stromberg JS,
Martinez AA. The use of active breathing control (ABC) to reduce margin
for breathing motion. Int J Radiat Oncol Biol Phys 1999; 44(4):911919.
Dawson LA, Brock KK, Kazanjian S, Fitch D, McGinn CJ, Lawrence TS, Ten
Haken RK, Balter J. The reproducibility of organ position using active breathing control (ABC) during liver radiotherapy. Int J Radiat Oncol Biol Phys 2001
Dec 1; 51(5):14101421.
Onishi H, Kuriyama K, Komiyama T, Tanaka S, Ueki J, Sano N, Araki T,
Ikenaga S, Tateda Y, Aikawa Y. CT evaluation of patient deep inspiration selfbreath-holding: how precisely can patients reproduce the tumor position in the
absence of respiratory monitoring devices? Med Phys 2003 Jun; 30(6):11831187.
Scholz C, Schulze C, Oelfke U, Bortfeld T. Development and clinical application of a fast superposition algorithm in radiation therapy. Radiother Oncol
2003; 69:7990.
De Jaeger K, Hoogeman MS, Engelsman M, Seppenwoolde Y, Damen EM,
Mijinheer BJ, Boersma LJ, Lebesque JV. Incorporating an improved dosecalculation algorithm in conformal radiotherapy of lung cancer: re-evaluation
of dose in normal lung tissue. Radiother Oncol 2003 Oct; 69(1):110.
Haedinger U, Krieger T, Flentje M, Wulf J. Inuence of the calculation model
on dose distribution in stereotactic radiotherapy of pulmonary targets. Int J
Radiat Oncol Biol Phys 2005 Jan 1; 61(1):239249.
Wulf J, Haedinger U, Oppitz U, Thiele W, Mueller G, Flentje M. Stereotactic
radiotherapy for primary lung cancer and pulmonary metastases: a noninvasive
treatment approach in medically inoperable patients. Int J Radiat Oncol Biol
Phys 2004 Sep 1; 60(1):186196.
3
Novalis
Paul Medin
Department of Radiation Oncology, UCLA Medical Center, Los Angeles,
California, U.S.A.
Dirk Verellen
Department of Radiotherapy, Oncology Center, Academic Hospital,
Vrije Universiteit Brussels (AZVUB), Brussels, Belgium
INTRODUCTION
The history of radiation therapy is one of continuous development of new
skills and new approaches. Often many of the desirable concepts were
understood years ago but it is only with recent developments in physics,
engineering, and computing that the techniques have become practicable.
The latest developments in radiotherapy have allowed surgically precise
delivery of radiation dose distributions to cure the patient without damaging
healthy tissue. Conformal radiation therapy (CRT) and intensity-modulated
radiation therapy (IMRT) have been the subject of many research projects
during the last decade and are becoming clinically available today. High
resolution IMRT treatments will probably result in improved outcomes
and denitely better quality of life for patients compared to treatments
based on conventional planning and dose delivery. The dynamic delivery
of intensity-modulated beams provides homogenous dose coverage of
the lesion as well as a much steeper dose fall-off at the lesions boundaries. Yet, the current positioning techniques do not match the accuracy
needed to perform CRT/IMRT adequately. In fact, difculties with accurate target localization have represented the most signicant obstacles to full
exploitation of the capabilities of CRT/IMRT treatments. The clinical
Novalis System, BrainLAB AG, Heimstetten, Germany.
19
20
Medin and Verellen
implementation of CRT is only feasible when two important requirements are

fullled simultaneously: the ability to generate conformal dose distributions
and to ensure accurate target localization (clinical knowledge of the target
region as well as ensuring positioning accuracy during treatment). The latter
requires an appropriate safety margin around the clinical target that increases
with increasing uncertainty of the actual position of the tumor during irradiation. Margins of typically 1 cm are needed to make sure that the tumor will be
irradiated sufciently. This is in contradiction to the millimeter precision that
can be achieved with modern radiotherapy techniques. Only when both
aspects are covered adequately can one truly use the term stereotactic body
radiation surgery or stereotactic body radiation therapy (SBRT).
SYSTEM DESCRIPTION
The Novalis system (BrainLAB AG, Heimstetten, Germany) provides the user
with an integrated system featuring treatment planning (BrainSCAN), automated patient positioning, and image-guidance (ExacTrac/Novalis Body)
and treatment delivery (Novalis) for non-invasive stereotactic radiosurgery
(SRS) and SBRT. All aspects of the treatment chain are seamlessly integrated
in the process, avoiding common problems related to communication between
different systems and vendors. Target delineation (based on multimodality imaging), image-guided target positioning (referencing the treatment isocenter to
external or internal localization markers), virtual treatment simulation, dose
calculation, and treatment verication are all supported by the treatment planning system. All treatment parameters (including the desired patient position
necessary for automated patient setup) can be transferred to the treatment
machine. The latter (Novalis) is a 6 MV single photon energy dedicated linac
with build-in mini-multileaf collimator capable of different treatment modalities, such as conformal beam, dynamic eld shaping arc, and IMRT. The eld
size ranges from 3.0 3.0 mm2 to 98.0 100.0 mm2 at isocenter distance with
inner leaves of 3.0 mm (the mini-MLC features two banks with 26 leaves each,
with varying leaf width from 3.0 mm for the 14 central pairs, to 4.5 mm for the
next three, and naly 5.5 mm for the outer three leaf pairs). Each leaf can over
travel by 50 mm, and a central straight edge allows the leaves to touch when
closed, with full interdigitation possible. In addition, a set of conical collimators
can be attached to the head of the machine for very small tumors and functional
treatment. Image-guided radiation therapy (IGRT) is provided by the ExacTrac/Novalis Body system that automatically aligns the target volume with
the treatment beam based on infrared tracking of external IR/CT bodymarkers
and/or automated registration of bony structures and implanted radio-opaque
markers (using stereoscopic X-ray imaging). The systems also offers dedicated
tools for quality assurance (QA), which will be explained in later chapters.
A detailed summary of the different features of the Novalis system is
provided.
Novalis
21
PATIENT POSITIONING
Description
Introduction
The SBRT or CRT generally is clinically not feasible without the appropriate
target localization and tools for patient setup that matches the accuracy
needed, especially if one considers that, to increase the therapeutic range,
smaller and smaller margins are used between the clinical target volume
(CTV) and planning target volume (PTV) (1,2). Often these margins are
reduced with the introduction of CRT and IMRT techniques, in spite of
the fact that the applied margin should reect the accuracy that can be realistically obtained in clinic. Moreover, with the introduction of highly
sophisticated and computer-guided treatment modalities, patient positioning procedures should evolve and develop equally, and should, ideally, be
integrated in the treatment planning process.
Geometric accuracy for the SBRT/CRT/IMRT procedure is basically
image-guided, and several solutions have been proposed in the last decade.
Electronic portal imaging devices (EPIDs) (35) have been embraced
with the expectation of achieving the required accuracy. A comprehensive
overview of existing EPID techniques was published recently by Herman
et al. (4), which acknowledges that the initial expectation has not led to widespread clinical application of EPIDs. Most studies present developments by
research centers (in collaboration with manufacturers) to cover their individual needs, and commercial systems are often (arguably) limited to replacements of portal lm and do not allow automated correction of setup errors.
The clinical application of EPIDs for patient setup verication can generally
be classied into two approaches: on-line (or intra-fractional) (619) and offline (inter-fractional) (2024). The latter, also coined adaptive radiation therapy (ART), monitors the position of the patient during a limited number of
fractions and adapts the safety margins and/or treatment plan accordingly.
This approach does not allow for decreasing the treatment margins sufciently for aggressive SBRT. The intra-fractional approach offers the possibility of reducing all treatment execution errors (both systematic and random),
yet is considered to be time consuming, requiring automated control of the
treatment couch and mostly limited to two-dimensional setup errors
(79,1518). While EPIDs suffer from the lack of soft tissue imaging, ultrasound (US) (25) and kilovoltage X-raybased (2629) image-guidance systems
have been proposed as a promising alternative. Some kV X-raybased solutions under investigation include:
1. computed tomography (CT), either in-line CT scanners installed
inside the treatment room (30,31) or kV cone-beam CT (26),
2. stereoscopic X-ray imaging systems (2729).
22
Medin and Verellen
The ExacTrac 3.0/Novalis Body system resides in the latter classication as it combines visualization of internal structures based on stereoscopic
X-ray imaging with real-time infrared (IR) tracking of the patients surface.
The system is designed to be a positioning tool ensuring accurate positioning
a priori fullling the following basic requirements: (a) integrated in the
treatment planning process, (b) perform as a fully automated positioning
tool (not verication tool) allowing highly accurate positioning of the target
volume based on treatment planning data, (c) does not increase the number
of actions required for patient setup compared to conventional methodologies (believed to be one of the major reasons for the limited clinical use of
EPIDs to date), and (d) perform this task within an acceptable time frame
(i.e., a typical treatment including positioning should not exceed 15 min)
(29,32,33). Some centers combine this technique with minimal patient xation (32,33), while others prefer more elaborate xation devices (see Chapter
11 on spinal tumors). This chapter will be restricted to a detailed description
of the image-guided technology with a limited summary of immobilization
tools that can be used in combination.
For the readers interest the positioning hardware of the Novalis
system can be classied as ExacTrac 1.0/Novalis Body (IR system only),
ExacTrac 2.0/Novalis Body (IR and stereoscopic X-ray imaging, with only
one amorphous silicon (AmSi) detector mounted to the treatment couch),
and ExacTrac 3.0/Novalis Body (IR and stereoscopic X-ray imaging
making use of two AmSi detectors mounted to the ceiling). Basically, the
positioning algorithms used in the latter two systems are based on similar
principles but adapted to the hardware differences. All versions allow for
automated computer-guided movement of the treatment couch.
Real-Time IR Tracking System
Hardware: The real-time IR tracking device is a system developed for
automated positioning of patients by detecting IR-reective/CT markers
placed on the patients surface, comparison of marker location with stored
reference information, and instructing the treatment machine to move the
patient to a preplanned position by moving the treatment couch (ExacTrac
1.0). The markers are visible by two IR cameras and one video camera that
are mounted to the ceiling of the treatment room (Fig. 1). The patients
movements can be monitored in 3D real time with the IR cameras in the
room, and consequently the patients position can be controlled on-line
either using a hand-pendant or computer-assisted commands
Software and settings: The IR-reective/CT markers are automatically
localized in the treatment planning system (Fig. 2) and the planned isocenter is
referenced to this marker conguration. The IR tracking system is able
to match a variable number of IR-reective markers, visible at the time of
Novalis
23
Figure 1 Room with Novalis system.

Note the IR-cameras and AmSi detectors
mounted to the ceiling and the X-ray tubes
embedded in the oor. (ExacTrac 3.0/
Novalis Body courtesy of BrainLAB, AG.)
positioning, onto a set of markers (not necessarily of equal number) that were
detected in the planning CT scan. As such the planned isocenter (based on the
marker conguration detected during planning) can be localized with respect to
the treatment isocenter of the linac (Fig. 3). The algorithm uses an unsorted
Figure 2 Left: Patient with IR reective marker. (Note that the camera system has
identied the markers indicated by the circles, and the coincidence with the planned
position indicated by the small crosses.) Right: CT-image showing IR marker (localized by software), contours of CT, PTV, and rectum, and position of the treatment
isocenter. (See color insert.)
24
Medin and Verellen
Figure 3 Illustration of the graphical interface of ExacTrac 3.0/Novalis Body with

the patient on the treatment couch prior to treatment setup. Note the detection of
transversal and rotational patient position at the right side and bottom of the image.
The circles indicate the actual position and the crosses indicate the planned position
where the patient will be moved. (See color insert.)
points match to solve the problem of matching the CT-localized marker with
the corresponding visible IR-reective markers. The markers are subjected to
possible skin shift and do not represent a rigid body and, therefore, a number
of assumptions and settings are needed. One of these settings is the maximum
distortion of the marker conguration that will be accepted for a successful
match. This setting reects the search area for the software to identify the
IR-reective markers and has no noticeable inuence on the positional accuracy of the isocenter. The latter is performed by least-square-ts, calculated
by using different subsets of the markers. From the isocenters that correspond
to the different subsets, the algorithm calculates a weighted average that
becomes the isocenter where the patient will be positioned. The latter allows
the system to recognize when one or more markers are shifted too much to
be taken into consideration. Another, more important setting with respect to
positional accuracy is the level of required accuracy. This setting denes the limits within which the isocenter position is considered acceptable, preventing innite trials in reaching the exact isocenter position by computer-controlled couch
movement. As mentioned earlier, the system allows for detection of patient
movement as well as being used for computer-controlled couch movement to
adjust the patients position in real time.
Novalis
25
Stereoscopic X-Ray Imaging Device

Hardware: The X-ray imaging system is fully integrated into the IR
tracking device described above and consists of a generator, two X-ray tubes
(MP 801 X-ray generator and comet X-ray tubes; K&S Rontgenwerk,
Bochum, Germany) embedded in the oor, and two amorphous silicon
(AmSi) detectors (PerkinElmer Optoelectronics GmbH, Wiesbaden,
Germany) mounted to the ceiling (Fig. 1). The angle between both X-ray
tubedetector pairs is approximately 90 , and approximately 42 tilted from
the horizontal. In addition, a key board-controlled interface (using a particular key combination) has been developed allowing remote computerassisted control of patient movement to predened positions (nal treatment
position) from outside the treatment room. The X-ray system produces
diagnostic photon beams ranging from 40 to 150 keV in exposure mode
and from 40 to 125 keV in uoroscopic mode, and projects a eld size of
approximately 20 20 cm2 on the AmSi-detector. The detectors have an
active area of 22 22 cm2. A calibration is needed to dene the spatial relationship between X-ray tubes and AmSi detectors on one hand, and the relationship with respect to the treatment machines isocenter on the other. The
spatial relationship with respect to the treatment isocenter is established by
dening a relationship between the X-ray system and the IR tracking system
with radio-opaque markers inside and IR-reective markers outside a specially designed calibration phantom. Patient and treatment couch movements are controlled by real-time tracking of the IR-reective markers.
This system is referred to as the ExacTrac 3.0/Novalis Body system. The
ExacTrac 2.0/Novalis Body refers to a previous setup with the X-ray tubes
mounted to the ceiling and only one AmSi detector mounted on the treatment couch.
Software and positioning procedure: Once the patient is positioned on
the treatment couch, two options are provided: (a) Using the predened IRreective marker conguration that denes the planned isocenter based on
CT data (see earlier). The advantage is that, once the patient is recognized
by the system, the initial position does not need to be close to the actual
treatment position. From that point on the entire positioning procedure is
computer driven and performed from outside the treatment room. (b) A
preliminary isocenter position is dened by the user that does not require
a predened IR-reective marker conguration on the patients skin and
can even consist of IR-reective markers rigidly xed to the treatment couch.
The latter has the advantage that breathing movements are eliminated and is
therefore more stable. The disadvantage is that the system is no longer able
to track patient movements during the positioning process and assumes that
the patient is perfectly motionless. This option requires the entire positioning
procedure to start from an a priori assumed correct treatment position
(which will be corrected afterwards), after which the remaining procedure
26
Figure 4 (Caption on facing page)
Medin and Verellen
Novalis
27
is again remote controlled and computer assisted. A video image can be provided during the automated setup procedure for patients safety.
Once both X-ray images have been acquired, again two options are
provided: (a) automated fusion of the actual X-ray images and DRRs representing the ideal patient position, and (b) matching implanted radio-opaque
markers. The former procedure is considered an improvement in patient
setup compared to conventional methods, although it is not able to cope
with internal organ movements and therefore still requires a substantial
internal target margin (ITV) (2). The implanted markers offer a more realistic assessment of the target volumes actual position and therefore enables
reduction of treatment margins suitable for SBRT. A clinical example of
appropriate treatment margins for treatment of the prostate follows, offering a detailed description of both procedures, with an illustration of the
procedure given in the owchart (Fig. 4).
Automated fusion of X-ray images and DRRs. In this setup, a 2D/3D
co-registration algorithm is applied to align a 3D CT patient data set with two
X-ray images. Assuming that all components of the system are properly
calibrated (i.e., the exact position of the X-ray tubes and detectors are known
with respect to the machines isocenter), it is possible to generate digitally
reconstructed radiographs (DRRs) from the planning CT (representing the
ideal patient position) and compare these with the acquired X-ray images.
For accurate positioning both the location and orientation of the patient need
to be assessed, taking into account all six degrees-of-freedom (6 DOF) for the
image co-registration (translations as well as rotations). An automated fusion
algorithm is used based on gradient correlation, which optimizes a similarity
measure for each image pair (34). The similarity measure relies primarily on
edges and gives a high response if strong edges are visible in the same place.
In a rst phase, the two pairs of corresponding X-rays and DRRs are fused
and the amount of 2D translations necessary to register the image pairs can
be used to compute a rst course 3D correction vector (this is possible since
the spatial relations and magnication factors between X-ray tubes and
patient are known). This was the only option in the ExacTrac 2.0/Novalis
Body system. This 2D/3D correction vector is then used as a starting value
for the second phase, being the 6 DOF co-registration. The latter is obtained
from an iterative optimization cycle to determine values for the rotation and
the translation of the 3D CT data set to maximize the similarity measure
Figure 4 (Facing page) Flowchart illustrating the different steps in the positioning
procedure using ExacTrac 3.0/Novalis Body. From top to bottom: (A) Patient
on the treatment couch with IR reective markers. (B) Acquisition of X-rays (only
one shown). (CD) Calculation of 3D correction vector based on either automated
fusion of X-ray images with DRRs representing the ideal position (left) or matching
of implanted radio-opaque markers (right). (E) Automated patient positioning. (See
color insert.)
28
Medin and Verellen
between the corresponding DRRs (each time re-calculated from the previous
values for rotations and translations) and the actual X-ray images. The latter
requires an efcient algorithm for rendering DRRs (since some hundred
DRRs will be used in the registration process), an efcient optimization,
and automated fusion algorithm. If the automated fusion should fail, a
backup procedure is offered to manually shift the DRR images until an
acceptable registration is obtained; the user can dene regions of interest in
the images (eliminating regions of high contrast that are not related to the
patients anatomy such as patient immobilization devices that may inuence
the automated fusion), or limit the search area (avoiding that the system
drifts-off to nd an unrealistic solution).
Matching implanted markers. Again, assuming a calibrated X-ray
system, implanted radio-opaque markers previously located in the planning
CT volume set will be projected on the X-ray images (Fig. 4). When the
initial patient setup is correct these projections will coincide with the images
of the markers on the X-ray image. In case of a setup error, each marker
projection can be clicked and dragged by mouse to coincide with the
corresponding image of the actual position. The combined marker translations/rotations in each X-ray projection allow for calculation of a full 6
DOF correction assuming a rigid conguration. If the marker conguration
deviates too much from the expected conguration (indicating possible marker migration), the system will fail to match the markers and the migrated
marker will have to be eliminated in the software.
VERIFICATION AND CLINICAL VALIDATION
Phantom Measurements
Verication tests have been performed on anthropomorphic phantoms
containing a humanoid skeleton to assess the precision and accuracy of the
positioning system. A summary will be given of the results that have been
published previously (29) on the systems performance for detection and
correction of known translational setup errors with and without rotational
errors in the pelvic region. In this study a segmented phantom (Alderson
Rando Phantom for radiotherapy: Radiology Support Devices, CA) has
been used consisting of 25-mm thick axial segments and allowing insertion
of hidden targets to evaluate the entire procedure from CT scanning to
treatment and verifying the alignment of treatment beam and target (assessment of the residual setup error). The phantom also allowed for insertion of
radio-opaque markers to test the systems performance with matching of
implanted markers. CT scans with 2-mm slice thickness and spacing between
consecutive slices were acquired of the phantom together with the IR-reective
markers. The image data sets were transferred to a dedicated treatment planning system (BrainScan V 5.1: BrainLAB AG, Heimstetten, Germany) to
dene an appropriate treatment isocenter, after which these data, in turn were
Novalis
29
transferred to the positioning system on the treatment machine. Due to the

segmented construction, inter-segment movements may have affected the
obtained results and a rigid anthropomorphic phantom (a human skeleton
embedded in resin with the shape of a human torso) had been introduced to
validate the results obtained with the segmented phantom for the automated fusion of bony structures, and to investigate the possible inuence of
slice thickness in the acquisition of CT data (2 vs. 5-mm slice thickness and
spacing).
Known setup errors have been applied to the phantoms and, after
automated positioning based on either fusion of bony structures or matching implanted radio-opaque markers, the residual error was obtained with
the hidden target test (HTT) known for quality assurance of stereotactic
radiosurgery procedures (35,36). A lead bead of 2-mm diameter had been
inserted in the segmented anthropomorphic phantom and dened as being
the treatment isocenter. The phantom followed the entire procedure from
CT scanning to treatment planning and automated positioning with the
Novalis Body system. The treatment beam, collimated with a 10-mm circular
collimator, had been used to generate portal lms (X-OMATIC cassette with
T-MAT L/RA lm: Kodak, Rochester, NY) from 0 to 90 gantry angles.
The center of the projected image of the lead bead had been measured with
respect to the center of the circular eld with a ruler (0.5 mm precision) on lm
with a magnication of 1.5 (focusisocenterdistance 100 cm; focuslm
distance 150 cm). This procedure not only yielded an experimental estimate
of the corrected setup error but also offered a comprehensive test of the entire
treatment procedure, including, target localization on CT images (accuracy
limited by voxel size) and mechanical uncertainties of treatment table and
linear accelerator.
The following tests have been performed to assess the setup accuracy
of the Novalis Body system using both the automated fusion of bony structures and matching of implanted radio-opaque markers:
1. automated positioning in the absence of shifts
2. automated positioning in the presence of shifts, each coordinate
has been evaluated separately
3. automated positioning in the presence of combined shifts in the
three principle directions
4. automated positioning in the absence of shifts with a rotation
around one of the principle axes
5. automated positioning in the presence of combined shifts with a
rotation around one of the principle axes
6. automated positioning in the presence of combined shifts and
combined rotational setup errors
As mentioned before the automated fusion of bony structures is only
applicable for SBRT of lesions that do not show much internal movement
30
Medin and Verellen
with respect to the bony structures used for positioning purposes. An

investigation of the rst version of the fusion algorithm (based on correcting
three degrees-of-freedom onlyomitting rotationswhere the amount of
2D translations necessary to register the image pairs was used to compute
a 3D correction vector for the patient) yielded an overall 3D accuracy (for
combinations of both translational and rotational setup errors: up to
30 mm and 4 about all three axes) of 0.7 mm (SD: 0.8 mm) and 0.4 mm
(SD: 0.9 mm) with the segmented and rigid phantom, respectively
(Table 1). Reducing rotational setup errors below 0.5 prior to the positioning process increases this accuracy to 0.4 mm (SD: 0.5 mm) and 0.5 mm (SD:
0.7 mm). Breaking apart the measurements (rotations about a single axis in
combination with translational setup errors) revealed an increased sensitivity
with respect to rotations around the table axis (i.e., a 3D setup deviation vector of 2.0 (SD: 0.3 mm) versus 0.9 mm (SD: 0.4 mm) for rotations around the
longitudinal and lateral axes). The largest residual error observed for combinations of induced translational and rotational setup errors was 2.7 mm, whereas
the use of a 5 mm CT data set resulted in a 3D setup deviation vector of 2.0 mm
(SD: 0.9 mm). Note that the DRRs were not re-sampled during the matching
process and only a shift has been calculated rather than a full 6 DOF motion.
The latest 6 DOF software version as described earlier has not yet been analyzed in detail but preliminary results look promising. The use of implanted
radio-opaque markers increases the accuracy even more to 0.3 mm (SD:
0.4 mm) and 0.5 mm (SD: 0.3 mm) with and without rotational setup errors.
The largest residual error observed for combinations of induced translational
and rotational setup errors was 1.7 mm. Separate analyses of individual translations in combination with a particular rotation did not show any specic
directional sensitivity. It must be recognized that although detected, the rotational setup errors have not been corrected for. The systems includes the
detected rotational error in the calculation of the required translations to position the planned isocenter with respect to the treatment machines isocenter.
The phantom setup procedures have not been timed explicitly, but typically
required less than 5 min and 3 min using the ExacTrac 2.0/Novalis Body or
ExacTrac 3.0/Novalis Body system, respectively.
Patient Studies
A study has been performed in three steps to investigate the clinical performance of the ExacTrac/Novalis Body system for treatment of prostate
cancer (see Chapter 10 on prostate and other pelvic tumors). The aim of the
study was to reduce rectal toxicity while maintaining equal outcome for the
patient. The latter can be realized only when the conventional PTV margins
can be reduced without compromising the dose coverage of the CTV. As the
PTV margin is introduced to cope with positional uncertainties (both internal
organ movementinternal margin or IMand uncertainties related to patient
Novalis
31
positioning and alignment of the treatment beamsetup margin or SM) (2) a

reduction can be warranted only by increased knowledge of the patients position and/or the actual position of the target volume. Once the positional uncertainty is quantied, an appropriate PTV margin can be realized by means of
any appropriate CRT technique. Both dynamic eld shaping arc and IMRT
(37) have been adopted by the AZ-VUB for treatment of the prostate, and
the reader is again referred to Chapter 10 on prostate and other pelvic tumors
for more details on the treatment protocols. The different phases in this study
were related to the introduction of more accurate methodologies of target positioning and dening the appropriate PTV margins for each technique:
Step 1: Patient positioning based on IR-skin markers using ExacTrac
1.0. This study included 19 patients (553 treatment fractions in
total). All patients were treated for prostate cancer by dynamic eld
shaping arc with the mini-MLC of Novalis. Patients were treated in
the supine position, stabilized by a vacuum cushion (12/19) or with
conventional head and knee supports only (7/19). A comparison
was made between conventional (using skin drawings and room
lasers) and IR positioning (32).
Step 2: Patient positioning based on automated fusion of X-rays and
DRRs using ExacTrac2.0/Novalis Body. Fifteen patients were
followed with a total of 261 treatment sessions. All patients were
treated in the supine position and stabilized with conventional
head-and-knee support. A comparison was made between conventional IR positioning and the DRR-fusion (33).
Step 3: Patient positioning based on matching radio-opaque implanted
markers using ExacTrac 3.0/Novalis Body. The study included 12
patients (122 treatment fractions analyzed), again treated in the
supine position with conventional head-and-knee support. A comparison was made between conventional IR positioning, DRRfusion, and implanted marker matching. An objective verication
(residual error after automated positioning) of the patient positioning was performed on orthogonal megavoltage lms taken at gantry
0 and 90 .
In the step 1 and 2 study the distances of the isocenter to the midline and to
lines tangential to the superior and ventral border of the os pubis were measured on coronal and sagittal reconstructions of the planning CT data set
(with an intrinsic voxel size of 1.0 1.0 3.0 mm3) (Fig. 5). These measurements were repeated on the megavoltage lm by a radiation oncologist using
a standard ruler (1 mm scaling). For the step 3 study, the coordinates of the
implanted markers with respect to the isocenter (both on the CT-data set
and orthogonal megavoltage lms) were used to dene the residual setup
error. In all cases the patient was assumed to remain in the initial treatment
position during acquisition of the additional megavoltage lms. Again, one
32
Medin and Verellen
Table 1 Average Residual Setup Error (mm) and Standard Deviation (in Parentheses) With and Without Rotational Setup Errors After Automated Positioning Based
on One of Both Calculation Algorithms (Fusion of DRR and X-Ray Image or
Matching Implanted Markers)
DRR fusion
Marker matching
No
rotation
Rotation
All
No
rotation
Rotation
0.32
(0.65)
0.39
(1.12)
0.05
(0.92)
0.15
(0.65)
0.33
(0.41)
0.20
(0.52)
Longitudinal
0.09
(0.56)
1.05
(1.38)
0.46
(0.06)
0.16
(0.33)
0.17
(0.40)
0.17
(0.38)
Lateral
0.36
(0.62)
1.68
(0.99)
0.48
(0.80)
0.38
(0.77)
0.14
(0.45)
0.21
(0.55)
Vertical
All
The residual error is obtained from hidden target data representing the remaining error
measured with portal lms at gantry angles 0 and 90 with a 10 mm circular beam of the
isocenter represented by a 2 mm bead, after automated positioning based on the corrected shift.
must note that the DRR fusion used in these studies was not the full 6 DOF
version.
In this chapter, for comparison purposes, the results from the three
studies have been pooled into one data base and re-analyzed yielding an overall
3D residual error equal to 1.1 mm (SD: 11.7 mm), 1.4 mm (SD: 7.1 mm),
Figure 5 Illustration of the distances taken to dene the position of the treatment
isocenter with respect to bony structures for verication with portal lm. The
distance of the isocenter to the midline and to the lines tangential to the superior
and ventral border of the os pubis are measured according to the dotted line. (See
color insert.)
Novalis
33
0.5 mm (SD: 4.6 mm), and 1.2 mm (SD: 3.8 mm) for positioning based on conventional methods, infrared, DRR fusion, or marker matching, respectively.
For the rst three results the comparison was based on bony references,
whereas the last gure results from the actual marker coordinates and as such
the only indication of the actual target positioning including organ movement.
These results show a striking reduction in the spread of data going from
conventional to marker matching method. To obtain an estimate of the distribution of systematic errors in setup for all patients, the standard deviation
(SD) of the mean deviation of individual patients was calculated. The random
component was determined by calculating the SD of the individual deviations
(pooled data) after subtractions of their corresponding mean. These results
are shown in Figure 6. Table 2 shows the percentage of moderately large
(5 mm) and large errors (10 mm) for the pooled patient data. Only the
third study allowed assessment of the difference between positioning based
on bony structures and implanted markers, which can be interpreted as an
indication of organ movement. Overall differences of 1.6 mm (latero-lateral),
2.8 mm (antero-posterior), and 2.3 mm (cranio-caudal) have been observed
between both positioning methods in this patient data set. Based on these
results the following rules for PTV margin have been proposed at the
Figure 6 An estimate of the distribution of setup errors for prostate treatments

resulting from positioning with (left to right) skin markers and room-laser alignment,
IR tracking, automated fusion between DRR and actual X-ray images, and matching of implanted markers. The systematic error is calculated as the SD of the
mean deviation of individual patients. The random error is dened as the SD of
the individual deviations of all patients after subtraction of the corresponding mean.
34
Medin and Verellen
Table 2 Percentage of Moderate (5 mm) and Large (10 mm) Errors for the
Pooled Data Base Along One of the Principle Axes
Conventional
IR
DRR fusion
Marker matching
Moderately large
errors (5 mm)
Large errors (10 mm)
41
21
8
3
12
2
1
0
AZ-VUB: 6.0 mm latero-lateral, and 10.0 mm antero-posterior and craniocaudal when DRR-fusion is used for positioning; 5.0 mm antero-posterior
and cranio-caudal, and 3.0 mm latero-lateral when implanted markers are
used for positioning. Clinical use of the ExacTrac 2.0/Novalis Body sytem
required a total linac time (patient entering treatment room to patient leaving
treatment room) of 140 5100 (SD: 40 1800 ). The X-ray-assisted patient positioning
required 70 5400 (SD: 30 4300 ) (33). No specic time measurements have been
performed since the introduction of the ExacTrac 3.0/Novalis Body system,
but the total linac time never exceded 110 , the X-ray-assisted setup was
below 40 .
Discussion
The approach of using diagnostic X-rays for verifying treatment setup is
not new (3840) and offers a twofold advantage: (a) image quality [a welldocumented problem in EPIDs (4,17,19)] is no longer an issue, especially in
combination with AmSi detectors (4,41); (b) patient dose becomes less
important compared to daily megavoltage images acquired with EPIDs.
Dose measurements have been performed with an appropriate ionization
chamber (Dosimax, Welhofer Dosimetrie, Schwarzenbruck, Germany)
covering a range of 50125 kV, 50160 mA, and 501250 mS, yielding values
between 22.9 mSv and 1.640 mSv per X-ray image. A typical clinical setting
of 100 kV, 100 mA, and 100 mS resulted in 0.513 mSv per image. Based on
the work of Motz and Danos (42) and Rogers (43), Herman et al. (4) have
shown a strong link between signal-to-noise ratio (SNR), spatial resolution,
and patient dose. For the same dose to the patient, the SNR is much lower
at megavoltage energies than that at diagnostic energies. In the example of
a 78 Gy prostate treatment, some simple mathematics show that (assuming
3 MUor 30 mSv at depth of maximum doseper EPI and requiring
minimal two images for 3D information) patient doses of 2340 versus
40 mSv are delivered with electronic portal imaging and kilovoltage imaging,
respectively, i.e., a ratio of 58. Moreover, the combination with real-time
Novalis
35
monitoring of patient positioning is not limited to patient observation, but

also offers the possibility of controlling the treatment beam based on that
information (28,29,44,45). The feasibility of implanted radio-opaque markers has been investigated in this and other studies (6,29,4650) and shows
promising results. It is evident, however, that a detailed clinical study must
be performed to investigate the inuence of migration and possible variations in target shape on setup accuracy.
The prostate studies shown in the previous discussion made use of
immobilization devices only in the rst study where a subset of patients were
treated supine with a vacuum cushion and another subset with the
conventional head-and-knee support. Although it was not the aim of that
study (32), there was no evidence that the immobilization devices altered
the positional accuracy, which is conrmed in other studies (51). The use of
implanted radio-opaque markers not only proved to be more accurate it also
allowed localization of the actual organ, and hence assessment of internal
organ movement. Observations from the AZ-VUB study showed average
deviations of 1.6 mm (latero-lateral), 2.8 mm (antero-posterior), and 2.3 mm
(cranio-caudal) between bony landmarks and implanted markers. These
results agree with observations made with consecutive CT measurements
(52,53) and assessments of intra-fractional movement based on MRI (54).
Ultrasound studies (55,56) show signicantly larger (i.e., a factor 2) organ
motions, which may be attributed to the technique more than actual internal
organ movement (57,58). All studies show that organ movement is largest in
the antero-posterior and cranio-caudal directions. So far no marker migration
has been observed in the AZ-VUB studies (CT prior, during, and after termination of treatment). Yin et al. performed a similar study on the usefulness of
the ExacTrac 2.0/Novalis Body system for radiosurgery for spinal tumors
(59). This group did use the system in combination with a vacuum body-xing
device (combining a vacuum bag and a piece of special plastic wrap from
Medical Intelligence) and an alpha cradle. A verication of positional accuracy obtained from comparison with orthogonal portal lms and DRRs from
the planning CT data set yielded a 3D isocenter deviation of less than 2 mm.
The deviation was attributed by the authors to various sources such as CT
slice thickness (2 mm), patient breathing, and positioning system accuracy.
The systems performance for other extracranial sites such as liver metastases
and lung tumors is currently under investigation. First results at the AZ-VUB
show promising results, yet one limitation of the Novalis Body system can
already be identied. Contrary to pelvic lesions where a sucient amount of
bony landmarks is present, abdominal or thoracic lesions that are located laterally in the patient do not offer sufcient bony landmarks and failures of the
automated fusion algorithm occur more often. Implanted radio-opaque markers become a necessity in these cases. Moreover, lesions where the vertebrae
are the major bony landmarks the automated fusion algorithm might be off
by one vertebra due to the cylindric symmetry.
36
Medin and Verellen
TREATMENT PLANNING AND TREATMENT DELIVERY

Dynamic Arc
The accelerator is equipped with an integrated miniMLC consisting of two
banks with 26 leaves each. The leaf width varies at isocenter distance from
0.30 cm for the 14 central pairs, 0.45 cm for the next three, and nally 0.55 cm
for the outer three leaf pairs. A maximum eld size of 9.80 10.00 cm2 can
be generated, again at isocenter distance. Each leaf can overtravel by 5.00 cm.
A central straight edge (the other two straight edges correspond to the eld
divergence at full extension and full retraction) allows the leaves to touch when
closed and full inter-digitation is possible. A detailed description of the
miniMLC is given by Cosgrove et al. (60) and Xia et al. (61).
The dynamic conformal arc technique is a merging of traditional arcing radiosurgery and conformal beam irradiation. Radiation is delivered
while the linac gantry rotates around the patient and the eld shape is modied
dynamically by the Novalis multileaf collimator to conform to the target
shape for each gantry angle (37). The dynamic conformal arc concept is illustrated in Figure 7, which shows eight consecutive beams eye views (BEVs)
from a 70 arc in 10 increments. The yellow line surrounding the target
(green) indicates the conformal eld shape created by the multileaf collimator.
The spinal cord (magenta) is shown running vertically through each frame.
Parameters such as arc length, dose, and the margin between the eld edge
and the tumor are specied by the user. Dose distributions may be customized
Figure 7 Beams eye views of a dynamic arc (10 gantry steps). The yellow line
surrounding the target (green) indicates the conformal eld shape created by the multileaf collimator. The spinal cord (magenta) is shown running vertically through each
frame. Parameters such as arc length, dose, and the margin between the eld edge
and the tumor are specied by the user. Dose distributions may be customized using
software tools that allow for preferential sparing of organs at risk (OARs) and for
graphical editing of eld shapes in any BEV. (See color insert.)
Novalis
37
Figure 8 Beams eye view of a dynamic arc with an organ at risk (OAR). The
eld shape (yellow) intentionally blocks part of the target (purple) in order to
minimize the dose to the OAR behind it. (See color insert.)
using software tools that allow for preferential sparing of organs at risk
(OARs) and for graphical editing of eld shapes in any BEV. An arc that
incorporates OAR sparing is shown in Figure 8. The eld shape (yellow) intentionally blocks part of the target (purple) in order to minimize the dose to the
OAR behind it. The dynamic conformal arc combines the most advantageous
aspects of traditional radiosurgery and conformal beam irradiation by minimizing the dose to surrounding structures by use of a large number of beam
angles and by effectively custom shielding healthy tissue in every beam. The
dynamic conformal arc is the most expedient delivery method because the
dose can be delivered to a large number of beam angles in one uid motion.
The 9.8 10.0 cm2 maximum eld size of the Novalis collimator obviates
the need for multiple isocenters in the case of larger lesions.
The TPS (BrainSCAN V 5.1: BrainLAB AG, Heimstetten, Germany)
comes with the shaped beam radiosurgery linac, which allows for both forward and inverse planning. The former is used to calculate the dose resulting
from dynamic conformal arc treatments. The dose calculation is based on
the pencil beam algorithm; two different calculation grids are used: (a) an
adaptive grid (i.e., the grid size is locally reduced in high dose gradient areas)
that can be rened with the zoom function for 2D display of isodoses, and
(b) a xed grid of 0.2 0.2 0.2 cm3 for calculation of the CDVH.
IMRS
The Novalis system is able to generate intensity-modulated treatment elds
by means of SMLC and DMLC treatment delivery (again, a sliding window).
The latter was created with the interpreter developed by Agazaryan et al.
38
Medin and Verellen
(62), which corrects for transmission through MLC leaves by an iterative

method. Leakage between opposing and neighboring leaves is also minimized and individual leaves are synchronized to reduce the tongue-andgroove (TaG) effect. The inter-digitation and 5.00 cm overtravel yield
maximum IMRT eld sizes of 9.80 10.00 cm2. The resolution of the beam
elements (pixel) in the direction of leaf travel can be specied arbitrarily and
has been set to 0.20 cm for this study. The system guarantees that only bixels
smaller or equal to the set value will be used. Perpendicular to the direction
of leaf travel the resolution of the bixels is dened by the leaf width (i.e.,
0.30, 0.45, and 0.55 cm).
The TPS (BrainSCAN V 5.1: BrainLAB AG, Heimstetten, Germany),
used for the dynamic conformal arc in the previous section, yet in the inverse
mode, was also used but with using the so-called treatment constraints. The optimization is based on the dynamically penalized likelihood method developed by
Llacer (63), which is a variation of the maximum likelihood estimator known
from inverse problems in positron emission tomography. A ltering term is
included in the optimization loop that also penalizes solutions that yield bixel
weights substantially different from their neighbors (64).
The Novalis approach to IMRS can be outlined as follows:
Planning targets and organ(s) at risk are identied.
An isocenter and beam angles are dened analogous to conformal
beam planning.
Various calculation parameters are prescribed.
Clinical dosevolume constraints are specied.
Multiple planning solutions are calculated with different priority
relationships between treatment volumes and OARs.
Planning solutions are compared and one is selected.
The treatment plan is veried and delivered.
Only objects with PTV, OAR, or Boost status are considered during
IMRS planning. All of these objects are contoured and dened on MR/
CT images with Novalis software tools. The IMRS treatment planning
process begins in the same way as conformal beam planning. An isocenter
is dened and beam angles are selected. A maximum of 24 beams may be
added to each isocenter but excellent dose distributions are usually achieved
using 6 to 8. Unlike standard conformal SRS/SRT planning, the distribution of beams around the isocenter is not necessarily driven by OAR constraints. Often the best plans result when beams are non-coplanar with
maximal geometric separation. One advantage of the Novalis system over
others is that a wide range of angles is available over which to spread beams
including the patients posterior. In most cases, a full 360 of gantry angle is
available for treatment in the axial plane and approximately 50 of anterior
gantry angle is available for nonaxial planes.
Novalis
39
Figure 9 Novalis inverse planning wizard.
Once the physical beam directions have been set, inverse planning
calculation parameters are specied using a planning wizard (Fig. 9). The ve
major parameter groups include: leaf sequencing, calculation grid, normal
tissue restriction, sharp edge smoothing, and hot beamlet restriction.
A detailed description of all inverse planning parameters is beyond the
scope of this chapter but the following example from UCLA is typical for
prostate and spine treatments:

step-and-shoot, approximate 15 segments

closed gap position behind jaws 5.0 mm
tongue-and-groove (TaG) optimization ON
tag MU result above MU w/o TaG 25%
sharp edge smoothing lter parameter 5%
hot beamlet restriction maximum 150200%.
Dosevolume constraints are prescribed graphically for the PTV and

each OAR by dragging adjustment points to shape the desired dosevolume
histogram (DVH). The dosevolume constraint window for a prostate plan
is shown in Figure 10. Dosevolume constraints for individual objects can
be saved and recalled so that planning for a group of patients can be done
under the same criteria. The relative importance of critical objects can be
weighted for optimization using the Guardian slider.
40
Medin and Verellen
Figure 10 Clinical dosevolume constraints for a prostate plan.
Once constraints have been specied, Novalis software generates four

treatment plans with optimization based on differing priorities for the
OARs. The four results are called PTV only, OAR low, OAR normal, and
OAR high. Dose distributions and DVHs from the four plans are displayed
side by side for evaluation (Fig. 11). Forward calculations are processed for
whichever plan is selected.
TREATMENT VERIFICATION
The AZ-VUB Approach
Introduction
Intensity-modulated radiation therapy is now accepted by the radiotherapy
society as a feasible treatment technique and is gaining momentum in the
clinical environment. Indeed, with the clinical implementation of IMRT
the attention is shifting from feasibility studies toward patient-based
studies and the investigation of treatment efciency. However, off-the-shelf
systems are still scarce and the clinical implementation of IMRT requires a
substantial effort from the individual centers. The technology has not yet
reached maturity and the step from phantom verication to patient treatment is, in many respects, a jump in the dark. The clinical implementation
Novalis
41
Figure 11 Plan optimization window. (See color insert.)
of IMRT requires the establishment of a complete chain of processes,

starting with inverse planning and going right through to verication.
Analysis of the QA needs to include careful delineation of the planning and
delivery process, documenting where important decisions are made, how
information is transferred, what kind of errors are likely or possible, and
the sensitivity of various parts of the process to errorsin short hazard
analysis. It is important to understand what has not been veried with
the applied QA procedure. The QA program can be divided into three
classes: machine-related QA, pretreatment QA, and treatment QA. Only
the latter two, in principle, involve patient-based issues. Unfortunately,
the conventional methods are no longer valid and the intuition from conventional radiotherapy is lost. In vivo dosimetry becomes difcult and hand
calculation is no longer feasible due to the complexity of the treatment.
Moreover, target localization and target volume motion become major
parameters in the delivered dose distribution, which is difcult to assess.
There is no ideal solution and some of the options are either to perform individualized extensive verication tests prior to each treatment, or generalized
verication of the so-called class solutions. The verication procedure in
turn can be designed to analyze all variables of the treatment process in
detail or to be comprehensive. The machine-related QA can be seen as a
pyramid-shaped approach in that the upper level is build on the quality of
42
Medin and Verellen
the level underneath. The base level comprises basic QA of the linac and
MLC; level 2 covers small eld dosimetry, small amounts of MUs, and
leaf-control properties; level 3, dosimetry of IM beams; and level 4, 3D
IMRT verication. In this chapter, treatment verication will be limited
to verication of dose delivery as target localization has been extensively
covered in elsewhere in this book. Moreover, special emphasis will be given
to verication of IMRT treatment delivery, which with its complexity
requires a more extensive evaluation. Verication of conventional CRT
and dynamic arc techniques can be considered special cases requiring some
of the tools described in this chapter.
The Novalis system (BrainLAB AG) offers a number of CRT
techniques such as IMRT by using Static MultiLeaf Collimation (SMLC)
or Dynamic MultiLeaf Collimation (DMLC) both based on the sliding
window technique. Studies have shown that DMLC is preferred over SMLC
for several MLCs (65,66). IMRT planning at the AZ-VUB has been limited
to DMLC only (again, the evaluation procedures shown here can equally be
applied to or easily adapted for SMLC techniques). The Novalis system
consists of a single energy (6 MV photons) linear accelerator (l) (Varian
Clinac 600) with the integrated mini-MultiLeaf Collimator (mMLC) m3TM
(BrainLAB AG), which is described in detail by Cosgrove et al. and Xia
et al. (60,61), the Treatment Planning System (TPS) BrainSCAN (BrainSCAN V 5.1, BrainLAB AG, Heimstetten, Germany), and the automated
positioning tool Novalis Body as described in the previous chapter. The interface between the linac and the TPS is realized with the VARIS record and verify system (Varian, Medical Systems, Milpitas, CA). The inverse planning
modality for IMRT has also been described earlier. After arranging the static
beam conguration in the conformal beam and preplanning and dening the
calculation parameters and constraints to be used, the inverse planning optimizes four plans with different priorities to OARs using a dynamically penalized likelihood algorithm (63). The calculated uence maps are used in the
forward calculation of the dose distribution using the pencil beam dose calculation algorithm (6769). The user can chose one of the four calculated plans
or change the calculation parameters and constraints to achieve the required
dose distribution.
The use of the mMLC enables a higher degree of conformity of the
dose distributions produced by the BrainSCAN IMRT planning system
(37,70). With the increasing degree of complexity and sophistication of a
treatment technique such as the IMRT technique, the clinical implementation requires a more comprehensive QA procedure than conventional radiation therapy. At the moment there is no general protocol established for the
QA procedure of IMRT, therefore each clinical site is forced to create an
individualized QA procedure to verify the correctness and accuracy of the
TPS calculations and treatment delivery. This chapter summarizes the procedure followed at the AZ-VUB for the QA of IMRT with the mMLC.
Novalis
43
The QA procedure for IMRT at the AZ-VUB is divided into the

following major parts:
Non-patient-related QA:

Evaluate the overall functionality of the system (similar to acceptance testing).

Analyze class solutions for different treatment sites (similar to
commissioning of the complete system).
Patient-related QA:

Pretreatment tests of the conrmed plan for a patient prior to the

rst fraction.
Prefraction tests designed to eliminate the risk of mistreatment.
To ensure that the system is operating within specications, general nonpatient-related tests are regularly performed as well as sample verications of
actual patient treatments (e.g., once a month). The patient sample verication
consists of a comprehensive test that includes mapping of the treatment plan
into a cubic or anthropomorphic phantom with absolute (TLD and ionization
chamber measurements) and relative (EDR-2, gamma evaluation of dose
distributions) dose verication. This procedure allows sufcient condence
in the systems performance and reducing the patient-specic QA. The latter
consists of verication of uence maps and independent calculation of monitor units (MU). The AZ-VUB policy is to perform a comprehensive QA
procedure of class solutions for the different treatment sites and to limit the
verication prior to each patient treatment.
Evaluation of Overall Functionality
In MLC-based IMRT, complex movement of the MLC leaves is used to
deliver the desired nonuniform dose distribution in the treatment eld. This
complex leaf movement is controlled by the operating system of the linac
(Varis, Varian Medical Systems, Milpitas, CA, U.S.A. for the Novalis
system). Due to these complex leaf movements, the acceptance of the
mMLC requires additional tests compared to conventional acceptance tests
for the verication of the linacs accuracy and reliability to ensure the accuracy of the mMLC when delivering intensity-modulated treatment elds.
The validation of the mMLC includes tests for: (a) leaf positioning accuracy,
(b) speed stability, (c) beam on/off stability, (d) gravity inuence, and (e)
MLC reliability.
a. The calibration of an MLC to be used for the delivery of intensity
proles needs special attention because the tolerances of the leaf
positioning accuracy used for conventional treatments (12 mm)
are no longer stringent enough for IMRT treatments. The leaf
positioning is especially critical for the step-and-shoot technique.
44
Medin and Verellen
b.
c.
d.
e.
If the shaping of every segment is not accurate, the dose of the

total intensity map can be changed. The leaf position is checked
by static light eld projection on graph paper ruled in millimeters
positioned at the couch top at isocenter distance, or with radiographic lm positioned at isocenter distance horizontally between
solid water phantom at the depth of maximum dose (dmax). The
positional accuracy of the leaves of the mMLC is smaller than
0.50 mm, the acceptable accuracy for IMRT treatments.
The leaf speed stability is critical for the DMLC technique since
this parameter denes the formation of the segments for the
sliding window during the irradiation. The leaf speed test is
conducted by moving the leaves over a dened distance during
a specied time, or while irradiating a certain amount of monitor
units (MU) at a certain dose rate. Use the test les provided by
Varian Medical Systems for the MLC dynamic treatment acceptance procedure to perform this test. With the dynamic le
viewer application available from Varian, an error histogram
table is generated for the performed tests that allows one to readily evaluate the tests.
To deliver intensity-modulated beams (IMBs) with either step-andshoot or dynamic multileaf collimation accurately, the beam holdoff (i.e., the period where there is no irradiation because the leaves
are in motion) must be controlled correctly. The MLC dynamic
treatment acceptance procedure from Varian also provides test les
to verify if the periods of beam hold-off appear at the planned times.
To ensure that the mMLC operates accurately under variable
inuences of gravitation (variable gantry angles), a gravity test
must be performed. This can be done by verifying the leaf speed
with a maximum and minimum inuence of gravitation. Therefore, the leaf speed test is conducted at three different gantry
angles of 0 , 90 , and 270 .
The mMLC reliability or repositioning accuracy is veried by running an autocycle of a number of mMLC elds where the last eld
setup is the same as the initial one. After about 50 cycles, the leaf
positions of the rst eld are compared with those of the last eld.
The reposition accuracy of the leaves of the mMLC is within
0.50 mm.
The accuracy must be veried on a regular (e.g., weekly) basis. Therefore it is desirable to have a test for routine QA that is simple and quick to
give an overall assessment of the accuracy of all leaf pairs of the MLC simultaneously by visual inspection of the irradiated radiographic lm.
During the installation of the BrainSCAN V5.1 software, an IMRT
phantom was copied to the TPS. On this phantom some uence distributions
Novalis
45
are predened to test the performance of the MLC in both step-and-shoot

IMRT and DMLC IMRT. The uence distributions simulate a chessboard
pattern and a garden fence pattern as shown in Figure 12. The procedure
to deliver these uence distributions is described in the Acceptance Checklist for BrainSCAN IMRS working instruction provided by BrainLAB.
When these uence distributions are delivered to radiographic lm,
the resulting dose pattern will show whether or not there is a positional error
for a particular leaf or leaf pair. For instance, in the garden fence test a
positional error will result in either the location or the thickness of the dark line
that will be different from the lines generated by the other leaf pairs. To be able
to compare the performance of the mMLC in step-and-shoot IMRT with
DMLC IMRT, the same dose must be delivered to the radiographic lm in
both delivery types to result in the same amount of blackening of the lm.
Another QA pattern is created at the AZ-VUB based on patterns
already in use by other medical departments. Figure 13 displayes the daily
QA pattern that is delivered with the collimator rotated to 90 so that the
leaves move from left to right over the pattern. All proles in the pattern
give an idea about the positional accuracy of the leaves. The wedge proles
help to evaluate the leaf speed stability by measuring the relative doses of
the different steps of the proles. Relative doses can also be veried with
the two black areas, the gray area, and the white circle that have a 100%,
50%, and 0% dose, respectively. The TaG underdosages will appear clearly
in the gray area and allow one to judge the inuence of the TaG effect.
When the leaf patterns are evaluated, it is more important to look at
the resulting geometry than at absolute doses. Absolute doses are veried
during the acceptance of the pencil beam calculation algorithm, of the TPS.
For the acceptance of the pencil beam calculation algorithm, the beam
parameters of the Novalis linac are measured as prescribed in the Beam Measurements for Pencil Beam operating instruction provided by BrainLAB and
Figure 12 Chessboard and garden fence intensity map.
46
Medin and Verellen
Figure 13 Daily QA pattern used at the AZ-VUB to check the mMLC

performance.
dened in the BrainSCAN planning system. The beam data measurements

require accurate small eld dosimetry with a micro-ionization chamber or diamond detector to ensure that the TPS will accurately calculate the dose delivered by small segments often appearing in IMBs. At the AZ-VUB, the beam
data measurements were performed with the NAC007 micro-ionization
chamber (Wellhofer Dosimetrie GmbH, Schwarzbruck, Germany), which is
a waterproof chamber with a high spatial resolution due to its small active
volume (0.007 cm3) that can be used for both relative and absolute dosimetry.
The small active volume makes the chamber particularly suitable for measurements of small elds and high-dose gradients.
The dened beam and the calculation algorithm are veried in
accordance with the AAPM TG23 report (71). 2D dose proles are measured with the NAC007 chamber in a cubic water phantom and compared
with the distributions planned on a cubic phantom. Absolute doses were
also veried at several positions with respect to depth and off-axis position.
The agreement between the calculated and the delivered dose was within 2%
or 2 mm in a water phantom, as recommended in the ICRU42 report (72).
In addition to verication in a water phantom, the algorithm was also veried for calculations for different locations on an anthropomorphic phantom (Alderson Rando Phantom for Radiotherapy, Radiology Support
Devices, California, U.S.A.). Again the agreement between the calculated
and actual delivered dose was within 2% or 2 mm.
Novalis
47
Analysis of Class Solutions

The verication of the quality of the resulting treatment plan can be performed
in two ways. A rst possibility is to work topdown with an absolute and
relative verication of the complete plan (the comprehensive approach).
verication is broken into separate parts of the calculation and delivery
process if a discrepancy in the result of the comprehensive test has been
observed. A second possibility is to work bottomup, where all parts of the plan
are veried separately to achieve complete plan verication. Analysis of the
verication is straightforward with the latter approach but the procedure is
time consuming. The rst approach requires less time, yet the discrepancy analysis is much more complicated; it has been adopted by the AZ-VUB.
Phantoms: Two types of phantoms are used at the AZ-VUB for the
verication of IMRT. The rst is a homogeneous polystyrene cubic phantom
with external dimensions of 16.50 16.50 17.80 cm3 containing 20 polystyrene spacers of 13.90 12.70 0.60 cm3 for inserting lm sheets. The spacers
can be oriented according the three orthogonal directions allowing transversal,
coronal, and sagittal measurements of the relative dose distribution. In some
of the spacers, holes with a 0.50 cm diameter and 0.08 cm depth are drilled,
where TLDs can be placed in for absolute dosimetry. An adaptation
of the phantom allows insertion of the NAC007 micro-ionization chamber.
The second phantom is a humanoid phantom (Alderson Rando Phantom
for Radiotherapy) and is used to verify clinically relevant treatment plans
where tissue inhomogeneities can have a large inuence on the resulting dose
distribution. Film sheets can be inserted between the slabs of this phantom
and, at specied locations in some of the slabs, holes are drilled to allow performance of absolute dosimetry with TLDs. Sequential CT images with
0.20 cm slice width and 0.20 cm slice spacing were generated with both phantoms. The obtained sets of images were imported in the TPS for the calculation
of the verication plans. Target localization and phantom positioning for all
verication measurements were performed by ExacTrac3.0/Novalis Body
(BramLAB AG, Heimstetten, Germany) for both the homogeneous cubic
phantom and the humanoid phantom.
Dosimeters and calibration: The dose distributions were examined relatively with radiographic lm (X-OMAT V and EDR-2 ready pack; Kodak,
Rochester, New York, U.S.A.) and absolute dosimetry performed with
ThermoLuminescent Detectors (TLD) [LiF: 700 pellets (Vinten Instruments,
Surrey, U.K.)] with 0.50 cm diameter and 0.08 cm thickness and the NAC007
micro-ionization chamber (Wellhofer Dosimetrie GmbH, Schwartzbruck,
Germany).
The prescribed dose of the verication plans was 0.50 or 2.00 Gy for the
irradiation of the X-OMAT V lm and the EDR-2 lm, respectively, resulting
in a net optical density within the approximate linear range of the response of
48
Medin and Verellen
the radiographic lm. The original radiographic lm was cut into smaller sheets
that t into the phantom. To create a sensitometric curve allowing the conversion of the optical densities into absorbed dose distributions, a calibration
procedure was developed and validated in our department yielding an accuracy
of 3% (1 SD). The lm sheets used to calculate the sensitometric curve were
calibrated with 5.00 5.00 cm2 elds against ionization chamber measurements. The calibration lm sheet was developed (Kodak X-Omat 3000RA
processor: Kodak, Rochester, New York, U.S.A.) simultaneously with the
measurement lm sheets. The lms were digitized using the WP102 lm scanner
with an aperture of 0.08 cm and the WP700V3.51 software (Wellhofer Dosimetrie, Schwarzenbruck, Germany), and compared to calculated dose distributions using an in-house developed version of the gamma-method (4% dose
difference/4 mm distance-to-agreement tolerances) (73), dose difference maps,
and cumulative dose histograms in a MATLAB environment (MATLAB V5
Student Edition: The MathWorks Inc., Littleeld, Texas, U.S.A.).
For the absolute dosimetry with TLD, the prescribed dose was set to
1.00 or 2.00 Gy, the dose to which the TLDs were calibrated. The detectors
were individually calibrated with the 6 MV photon beam of the Novalis
linac, yielding a reproducibility of 3% (1 SD). The TLDs were positioned
in the transversal plane. Alternatively, the plan could be mapped into
the cubic phantom allowing insertion of the NAC007 micro-ionization
chamber.
Verication plans: Verication plans have been evaluated for different treatment sites such as prostate, head and neck, and brain lesions, and
once a month a sample plan is run, selected from the patients under treatment. These cases have been simulated on the anthropomorphic phantom
and/or mapped to the cubic phantom. The latter is a useful evaluation
feature of the TPS where the complete set of treatment parameters is superimposed on another image set, e.g., an anthropomorphic or a geometrical
phantom. In this case the isocenter of the patients plan is placed into an
appropriate place in the phantom to ensure a relevant verication measurement of the treatment. The treatment plan (geometrical settings, beam
arrangement, leaf settings, MUs are used from the patients plan) is then
recalculated using the contours and densities of the phantom, allowing
relative and absolute dosimetry of the patients plan. Special care needs to
be given to the choice of the phantom; when important variations in tissue
density are present at the treatment site, the homogeneous phantom may not
be appropriate and offer a false sense of condence as seen in Figure 14. The
dose export tools of the BrainSCAN TPS allow the comparison of the
calculated with the measured doses and dose distributions. The dose distributions of the overall plan as well as the uence distributions of every treatment beam used in the plan can be exported from the TPS. The
correspondence between the planned and the measured doses and dose
Novalis
49
Figure 14 The importance of the choice of an appropriate phantom when mapping

treatment parameters from a patient treatment to a phantom for verication of dose
distribution is illustrated here. The left-hand pane shows the result from mapping
into an anthropomorphic phantom; the white erased pixels indicate regions where
the gamma tolerance (4% DD, 4 mm DTA) is not met between measured and calculated dose distributions. The right-hand pane shows the same treatment mapped into
a homogeneous cubic phantom; the bold pixels indicate regions where the gamma
tolerance is not met. The homogeneous mapping does not show possible errors due
to tissue heterogeneities and might give a false sense of condence. (See color insert.)
distributions has then been analyzed using the gamma-method, dose

difference maps, and cumulative dose histograms (Fig. 15).
Pretreatment Verification
Once an acceptable condence level with respect to the overall performance
of the entire system is achieved and maintained (test-pattern tests, class
solution verication, and regular sample tests), the individual patient QA
is reduced to verication of uence patterns and absolute dose calculation.
The absolute dose check is performed prior to every patient treatment
by mapping the patients plan to a homogeneous polystyrene cubic phantom
(identical to the one used for commissioning) that is modied so that the
NAC007 micro-ionization chamber ts in the phantom. The patients plan
is recalculated and delivered to the phantom and the calculated dose to
the chamber is compared with the actual delivered dose measured by the
chamber. Care should be taken while positioning the dose distribution,
allowing an absolute dose measurement in a low-dose gradient region of
the dose distribution.
A second patient-specic test is an independent MU verication. A
simple spreadsheet-based program has been developed that uses as few
as possible parameters from the plan to re-calculate the dose given in the
50
Medin and Verellen
Figure 15 Illustration of in-house developed tool for verication of dose distribution (measured and calculated) at the AZ-VUB showing percent difference, absolute
difference, and gamma map overlayed with both dose distributions and a cumulative
dose histogram. (See color insert.)
normalization point by the IMBs used in the treatment plan (65). The parameters imported in the program from the treatment plan are the MU per
IMB, shapes and relative weights for each segment (written in the beam
shape le that can be exported with the Export Beam Data tool from
the TPS), and the equivalent depth to the normalization point. The parameters used in the program that are not imported from the treatment plan
are the tissue maximum ratio (TMR) data, the output factors (OF), and the
off axis ratios (OR). These parameters are taken from the original measured
beam data.
First, the contribution of each segment of the IMB to the dose in the
normalization point is determined. Therefore the relative weight of that segment (wi) is dened by subtracting the indices from the successive segments
in the exported beam shape le. For each segment the contribution of the
segment (coni) to the normalization point dose is dened by determining
if it covers the normalization point fully or partially (e.g., if the normalization point is located in the penumbra of the segment). For the segments that
block out the normalization point, the leakage contribution is calculated. By
Novalis
51
multiplying the relative weight and the contribution with the total amount
of MU of the IMB, the amount of MU of the IMB is divided over the
contributing segments.
MUi MUIMB wi coni
where MUIMB is the total MU for IMB; MUi, MU for segment i; wi, relative
weight of segment i; and coni, contribution of segment i.
The dose that every segment delivers to the normalization point is
calculated with this formula:
Di
MUi TMRi OFi ORi

Dref CFSAD
where Di is the dose of segment i; MUi, MU for segment i; TMRi,

tissue maximum ratio for segment i; OFi, output factor for segment i;
ORi, off axis ratio for segment i; Dref, total dose in the normalization point;
and CFSAD, conversion factor for isocentric treatments.
CFSAD SSDnorm dnorm =SID2 :
The OF of every segment is dened according the equivalent square

eld that corresponds with the surface of every segment. The equivalent
square eld of the segment and the equivalent depth are used for looking
up the TMR. To nd the OR for every segment, the coordinates of the gravity point of the segment are calculated with respect to the coordinates of the
normalization point. The OR for the segment is found in function of the
equivalent depth and the distance to the normalization point.
To dene the dose delivered by that IMB in the normalization point,
the doses of all segments of the IMB are summed together.
This approach of recalculating the dose of an IMB in the normalization point eliminates the use of some parameters calculated by the TPS, such
as the leaf sequencing factor (LSF), which is a parameter that is hard to verify, and any inaccuracy introduced by the recalculation of TMR data to
PDD data done by the TPS.
The verication of the uence patterns of the IMBs used in a treatment
plan is the third part of the patient specic QA. With the use of the Varis
MLC Shaper application, every IMB can be delivered with the adjusted
amount of MU to result in a dose appropriate for the detector. The uence
patterns are veried with radiographic lm positioned horizontally between
solid water phantom at a depth of at least 5.0 cm. This depth is necessary
since the beam prole of the Novalis at depth of maximum dose (1.6 cm)
shows a dip in the center of the eld of about 5%, which can greatly
inuence the measured uence pattern. The measured pattern is compared
with the pattern exported from the TPS. This exported uence pattern
is calculated at an entered equivalent depth (the depth used for the
52
Medin and Verellen
measurement) with a resolution of 0.5 mm and saved in a le that is spreadsheet compatible.

Figure 16 shows an example of one of the uence maps of a prostate treament. Figure 16A displays the exported uence pattern; Figure 16B, the uence pattern obtained with lm measurement. Based on a visual comparison
of both patterns, one can determine that the correct pattern is deliverd by that
specic treatment beam.
Prefraction Verification
The prefraction verication is performed by the radiation technologists by
comparing, during the creation of the chart, the shape of the rst segment
of each treatment beam to the printout of the plan. The DMLC scheme of
that treatment beam is simulated with the Varis MLC shaper application
to ensure that there are no initial problems in the scheme. Another technologist reviews all chart components to discover possible errors. Prior to the
rst fraction, a dry-run of the complete treatment is carried out to ensure that
there is no equipment collision during the treatment. After the rst positioning of the patient (either with the Target Positioner Box for brain or with
ExacTrac 3.0/Novalis Body for extracranial treatments), the lasers are
dened on the patients mask or skin to create a visual reference for positioning of the patient during all following fractions of the patients treatment.
Figure 16 Comparison of uence maps for a prostate treatment. (A) Fluence

pattern exported from treatment planning system. (B) Measured uence pattern.
Novalis
53
The UCLA Approach

Rigorous QA measurements are performed to verify the accuracy of dose calculations and radiation delivery of every IMRS computer plan. The Novalis
IMRS QA protocol at UCLA has been discussed in detail by Agazaryan et
al. (62). A combination of lm and ionization chamber measurements is performed to evaluate the absolute dose delivered at isocenter, the composite dose
distribution, and the uence maps of individual elds. Absolute dose and
composite dose distributions are measured in one of two phantoms. The Benchmark IMRT PhantomTM (MED-TEC Inc., Orange City, Iowa, U.S.A.) is used
for targets that are surrounded by largely homogeneous tissue and the Thorax
IMRT Phantom (CIRS Inc., Norfolk, Virginia, U.S.A.) is used for targets that
require irradiation through lung. Both phantoms are constructed with multiple
axial planes for lm measurements and multiple inserts for ionization chambers. After a treatment plan has been completed, Novalis software is used to
map all planning parameters from the patient plan to a CT scan of the appropriate phantom. The treatment plan in Figure 17 is shown mapped to the
MED-TEC and CIRS phantoms in Figures 18 and 19, respectively. The dose
distributions in Figures 17 to 19 differ in appearance because identical treatments are delivered to each object; therefore, the resulting distributions are factors of the objects size, shape, and density. In-phantom dose distributions are
exported to a le for digital qualitative comparison with lm measurements.
The dose to phantom isocenter is a standard printed plan parameter. Composite dose distributions are measured in the axial plane at isocenter using EDR2
lm (Eastman Kodak Corp., Rochester, New York, U.S.A.). Exposed lms are
digitized and compared to the calculated dose distributions with the aid of software developed for this task. The measured and calculated dose distributions
Figure 17 Eight-eld IMRS dose distribution for T1l metastasis on fused CT/MR.
The 30%, 50%, 80%, 90%, and 105% isodose lines are displayed. The maximum dose
is 105%. (See color insert.)
54
Medin and Verellen
Figure 18 An eight-eld
IMRT plan mapped to a
MED-TEC
benchmark
phantom. The 30%, 50%,
80%, and 90% isodose lines
are displayed. (See color
insert.)
are superimposed and positioned graphically using shift, rotate, and mirror
tools, or by specifying isocenter coordinates and using ducial marks. Dose difference, distance-to-agreement, and the gamma index (73) (minimum scaled
multidimensional distance between a measurement and a calculation point
determined in combined dose and physical distance space) are calculated along
a specied isodose line. At UCLA, 3% dose difference and 3 mm distance is
used as a scaling acceptability criterion. The results of an IMRT dose distribution analysis using the gamma index are shown in Figure 20. The solid black
lines represent the calculated 20%, 50%, and 80% isodose lines while colorwash
indicates the corresponding isodose lines from lm. Dark green spots indicate
areas where the 3%/3 mm criterion was exceeded. Absolute dosimetry for each
Figure 19 An eight-eld
IMRT plan mapped to a
CIRS thorax phantom.
The 30%, 50%, 80%, 90%,
and 105% isodose lines
are displayed. (See color
insert.)
Novalis
55
Figure 20 Comparison of a
lm measurement to calculation
using gamma-index analysis for
an IMRT dose distribution.
Solid black lines are the calculated isodose lines; colorwash
is the corresponding isodose
lines from lm. Dark green are
the area where the dose criterian
was exceeded. (See color insert.)
composite plan is performed using a 0.015 cm3 ionization chamber inserted in

the appropriate phantom. Fluence maps of individual elds may be analyzed
either qualitatively or quantitatively.
Quantitative lm dosimetry analysis was performed for the rst 100
IMRT patients at UCLA using the same method described for composite
distributions but this was replaced by qualitative visual analysis. It was
decided that the information gained by quantitative analysis was not sufcient to justify the signicant additional expense in time and lm. Currently,
individual elds are delivered to a sheet of XV2 lm (Eastman Kodak Corp.,
Rochester, New York, U.S.A.) perpendicular to the beam for visual comparison with Novalis calculated uence maps as shown in Figure 21.
Dynamic arc or static conformal elds are preferable for more
regularly shaped tumors that do not wrap around the cord. Both treatment
options are more expedient to deliver, require fewer MU, and require a lesser degree of QA than IMRS. IMRS delivery times are essentially twice as
long as dynamic arc and conformal elds because the number of MU is
approximately doubled and the beam is switched off between eld segments.
An increase in delivery time also increases the risk that a patient may move
or become too uncomfortable to nish treatment. Additional monitor units
result in a proportionally higher whole-body leakage dose to the patient that
is considered undesirable. Patient-specic QA measurements are not generally performed for dynamic arc and static eld therapy. Commissioning data
along with daily, monthly, and annual QA (American Association of Physicists in Medicine, Task Group-40) (74) are considered adequate. The treatment planning principles followed when using dynamic arcs or conformal
56
Medin and Verellen
Figure 21 Calculated and measured uence maps for an IMRT eld. The unirradiated canal through the center of this uence map corresponds to the position
of the spinal cord.
elds are the same as in intracranial SRS. Increasing the number of noncoplanar arc degrees or xed elds spreads the dose over healthy tissue and
provides tighter dose conformation to the target. Dynamic arc plans with
three arcs and a 13-eld conformal beam plan are shown in Figures 22
and 23, respectively.
Once a treatment plan is nished and approved, it is transferred to
the ExacTrac computer in the linac control area. Practical information, such
as the height of the isocenter above the treatment table and the distance to
Figure 22 Dynamic arc dose distribution for L2 schwannoma on fused MR/CT.

The 10%, 30%, 50%, 80%, and 90% isodose lines are displayed. The maximum dose
Novalis
57
Figure 23 Conformal eld dose distribution for L5 metastasis on fused MR/CT.

The 30%, 50%, 85%, 90%, and 95%, isodose lines are displayed. The maximum dose
ExacTrac reectors, is recorded as an additional means of patient

positioning verication.
TREATMENT OF MOVING TARGETS
The characteristics of a novel, IR-based stereophotogrammetry system for
patient positioning (ExacTrac, BrainLAB AG, Heimstetten, Germany) have
been described previously (75). Subsequent investigations have been conducted at UCLA to evaluate the capabilities of the Novalis ExacTrac systems for tracking motion in real time (76). For these studies, patients
were placed in a supine position and four to seven IR-reecting markers
were attached at various locations on the anterior chest, abdomen, and
pelvis. Prior studies have shown that the ExacTrac system is capable of
reproducing the position of a point to within 0.2 mm when the markers
are attached to a rigid object (75). On a nonrigid object, system accuracy
is clearly degraded. However, Wang et al. demonstrated that using an overdetermined set of markers (greater than 3) could improve localization
accuracy (75). Subjects were monitored for approximately 20 min during
which time the three spatial coordinates of each marker were sampled at
a rate of 5 Hz. Figure 24 shows the displacement in the vertical direction
of three markers afxed to a single volunteer as a function of time. For
clarity only the rst 60 sec are shown, though the sinusoidal pattern
corresponding to respiratory motion, with a frequency of approximately
0.20.3 Hz, can be observed for the entire 20-minute study duration.
Additional studies were performed to investigate patient motion
characteristics in 3D. In this case, the individual coordinates of the IR
58
Medin and Verellen
Figure 24 Vertical isocenter displacement as a function of time for three markers

attached to the anterior surface of one patient. Though data are shown only for
the rst 60 sec, the pattern repeats over the entire 20 min. (See color insert.)
Figure 25 Marker displacement as a function of time for a single marker attached

to the anterior surface of one patient. The steresophotogrammetry system is capable
of reporting marker coordinates at a frequency of up to 10 Hz. (See color insert.)
Novalis
59
marker locations were recorded, again at a frequency of 5.0 Hz. Not surprisingly, the greatest motion was observed in the anteriorposterior (A/P) and
superiorinferior (S/I) directions; motion in the lateral direction was essentially non-existent. Figure 25 shows an example of the real time, 3D motion
characteristics of one patient. Subsequent observations in many patients
indicated signicant variation of the amplitudes of the A/P motion relative
to the S/I motion; that is, different patients breathe in different ways. Based
on this observation, the 3D motion function (Eq. 4) described by Baroni et
al. (77) was adopted for many subsequent interventions (including gating
studies).
(
F
N h
X
Zj x xj y yj z zj
1
2
i2
)12
4
To facilitate gated delivery using the Novalis, a software module designed

at UCLA was added to the ExacTrac control system (78). The module displays
Figure 26 The gating control software monitors respiration through the ExacTrac,
calculates and displays ths Baroni F-function, establishes gating windows or thresholds, and triggers the Novalis via the MHOLDOFF/status bit. (See color insert.)
60
Medin and Verellen
the 1D or 3D position of any marker or combinations thereof as well as the

Baroni F-function calculated from the marker positions. Gating is facilitated
through the selection of an amplitude-based gating window measured in absolute units of mm, as in percentage of respiratory amplitude, or as a percentage
of duty cycle (Fig. 26). The modied ExacTrac system was used to control the
gated operation through a commercial respiratory gating interface installed on
the Novalis. The gating interface consists of a 25-pin connector inserted between
the dynamic MLC (dMLC) cable and the console backplane. A 3-pin connector
provides access for a 12 V DC signal to the MHOLDOFF/status bit on the console backplane. A beam inhibit is triggered whenever this bit is low. Gating is
controlled through a switching device connected between pins 1 and 2 of the
3-pin connector. An open switch generates an initial position (IPSN) interlock
on the accelerator controller. For dynamic treatments, the absence of an IPSN
interlock ensures that neither the dose nor the collimator position varies more
than 0.2 MU and 0.2 cm, respectively. Thus this interlock halts both radiation
and leaf motion and is cleared immediately upon closure of the switch. System
operation has been described in some detail previously (76,79).
Gated operation and dosimetry was initially investigated through the
use of a simple signal generator. For open elds, FWHM, atness and
Figure 27 2D data were measured using the amorphous silicon device. Gating
frequencies of 0.2, 0.5, and 1.0 Hz were used in addition to non-gated conditions.
Proles are shown for an open 10 10 cm2 square eld (top left), a dynamic wedge
(top right), and arbitrary intensity map (bottom right). (See color insert.)
Novalis
61
symmetry as well as absolute dose were evaluated as a function of gating

frequency (76,79). Images were obtained using lm and using an amorphous-silicon (a-Si:H) imaging system (Scanditronix Medical/Wellhofer
Dosimetrie, Uppsala, Sweden). The a-Si:H system consists of a 256 256
at panel array coupled to a personal computer. The size of individual detector elements is 0.8 mm per pixel. Each image was corrected for pixel
sensitivity and the dark signal was removed appropriately based on the
number of acquired frames. Dosimetric properties of this device have been
described previously (80). Measurements were repeated for MU ranging
from 25 to 200 with no gating and at the gating frequencies between 0.2
and 1.0 Hz. At each MU and gating frequency level, non-gated images were
subtracted from gated to facilitate analysis. Frames were acquired every
80 msec and integrated to obtain an absolute dose map. Dynamic wedge
and IMRT delivery were evaluated in a similar manner. Proles for an open
10 10 cm2 eld, a dynamic wedge, and an arbitrary IMRT intensity proles are shown in Figure 27.
Figure 28 The gating stage was constructed to verify proper operation of the gating
system and to evaluate imaging and dosimetry characteristics. The stage accommodates
a variety of phantoms; note the infrared makers attached to the MedTec phantom. The
stage can be programmed to move in a periodic manner or can accommodate patientspecic motion acquired with the ExacTrac system.
Figure 29 Composite lm dosimsetry results for a 5-eld IMRT liver

plan. The dosimetry phantom is moving according to the liver function
described by Lujan et al. (81,82). The
Novalis is operated in either normal
(non-gated) or gated mode. Measurement in colorwash is overlaid on calculations (solid lines with the 80%, 50%,
and 20% levels displayed). Dark areas
indicate poor agreement between measurement and calculation as quantied
using the g index. Source: From
Ref. 78.
62
Medin and Verellen
Novalis
63
Using the modied ExacTrac system interfaced to the Novalis,

Hugo et al. have subsequently studied gated IMRT dosimetry in a series
of extracranial sites (78). For this work, a gating stage (Fig. 28) was constructed to provide respiratory motion to phantoms of various designs.
Investigations were performed using several motion patterns including a
1D liver function described by Lujan et al. (81,82) and a 2D lung function.
Several treatment locations were considered, including IMRT delivery for a
liver target. Figure 29 shows one example of the resulting dosimetric characteristics of gated delivery. This strongly supports the assertion that gated
delivery can signicantly reduce the dosimetric errors associated with extracranial radiosurgery and IMRT in anatomical sites that may be inuenced
by respiratory motion. However, as pointed out subsequently by Hugo et
al., equal consideration must be given to imaging studies (e.g., gated CT
acquisition) if treatment planning is to be performed properly (83).
Figure 30 Example of target localization for a lung lesion using co-registration

information from PET and CT imaging, with cumulative dosevolume histogram
resulting from a coplanar dynamic conformal arc treatment. (See color insert.)
64
Medin and Verellen
CAPABILITIES AND LIMITATIONS OF THE TECHNIQUE

By far the largest limitation of the Novalis system with respect to SBRT is
its limited eld size of 9.8 10.0 cm2. Although arguments can be found to
dismiss tumors exceeding these dimensions for aggressive stereotactic
radiotherapy or radiosurgery, a work-around has been presented by
Linthout et al. (84). This study showed that using partially overlapping
intensity-modulated beams assigned to different isocenters to enlarge the
treatment region was indeed feasible, and introduced smaller dose inhomogeneities in the resultant dose distribution than in the case of abutting treatment elds. The resultant dose distribution proved to be less sensitive to
positioning errors of the used treatment isocenters. Based on lm and
TLD measurements, the magnitude of the maximum dose inhomogeneities
varied by 8% per mm (for shifts ranging from 3 to 3 mm).
An interesting feature of the Novalis system is the dynamic conformal
arc technique, which is extremely efcient both in treatment planning (see the
section on Treatment Planning and Treatment Delivery). The latter has
previously been shown by Verellen et al. (37) for prostate cases without
concavities in the target volume, where comparable dose distributions can
be achieved as with IMRT techniques, yet with considerably less beamon time and overall linac time. The technique also proved to be very useful
Figure 31 Illustration of a four non-coplanar

dynamic conformal arc technique for a meningioma. The 30%, 50%, 90%, 98%, and 100%
isodose lines are shown. (See color insert.)
Novalis
65
for hypofractionated treatment of lung (Fig. 30) and liver metastasis in combination with the high positional accuracy that can be obtained with the
ExacTrac 3.0/Novalis Body system. Due to possible collisions with non-zero
table angles, the dynamic conformal arc technique for abdominal and pelvic
regions is limited to coplanar arcs only. Therefore, the technique is limited in
its performance for target volumes that present large concavities. However,
for head-and-neck and cranial treatments the dynamic conformal arc technique is capable of competing with most IMRT proposals. Figure 31 illustrates
a treatment of a meningioma using four non-coplanar beams, yielding high
conformality without compromising dose homogeneity.
ACKNOWLEDGMENTS
The authors wish to express their gratitude to the nurses (for daily setup of
the patients) and the members of the respective physics teams for support.
In particular Drs. Nadine Linthout, Koen Tournel, and Swana Van Acker
who largely collaborated in the work presented here. Dr. Guy Soete is greatly
acknowledged for setting up the prostate studies at the AZ-VUB, gathering
the raw data (input of some thousands of gures), and making the data available for the author for the pooled analysis. We also thank S. Froehlich and A.
Wackerle and their teams from the BrainLAB company for a very dynamic
and interesting collaboration on the Novalis Body project.
REFERENCES
1. International Commission on Radiation Units and Measurements: Prescribing,
Recording and Reporting Photon Beam Therapy. ICRU Report 50. Maryland:
Bethesda, 1993.
2. International Commission on Radiation Units and Measurements: Prescribing,
Recording and Reporting Photon Beam Therapy. ICRU Report 62. Maryland:
Bethesda, 1999.
3. Boyer AL, Antonuk L, Fenster A, et al. A review of electronic portal imaging
devices (EPIDs) Med Phys 1992; 19(1):116.
4. Herman MG, Balter JM, Jaffray DA, et al. Clinical use of electronic portal imaging: Report of AAPM radiation therapy committee task group 58. Med Phys
2001; 28(5):712737.
5. Munro P. Portal imaging technology: past, present, and future. Semin Radiat
Oncol 1995; 5(2):115133.
6. Alasti H, Petric MP, Catton CN, et al. Portal imaging for evaluation of daily
on-line setup errors and off-line organ motion during conformal irradiation of carcinoma of the prostate. Int J Radiat Oncol Biol Phys 2001; 49(3):
869884.
7. Bel A, Petrascu O, Van de Vondel I, et al. A computerized remote table control
for fast online patient repositioning: implementation and clinical feasibility.
Med Phys 2000; 27(2):354358.
66
Medin and Verellen
8. De Neve W, Van den Heuvel F, Coghe M, et al. Interactive use of on-line portal
imaging in pelvic radiation. Int J Radiat Oncol Biol Phys 1993; 25:517524.
9. De Neve W, Van den Heuvel F, De Beukeleer M, et al. Routine clinical on-line
portal imaging followed by immediate eld adjustment using a tele-controlled
couch. Radiother Oncol 1992; 24:4554.
10. Ezz A, Munro P, Porter AT, et al. Daily monitoring and correction of radiation
eld placement using a video-based portal imaging system: a pilot study. Int J
Radiat Oncol Biol Phys 1992; 22(1):159165.
11. Gildersleve J, Dearnaley DP, Evans PM, et al. A randomised trial of patient
repositioning during radiotherapy using a megavoltage imaging system. Radiother Oncol 1994; 31(2):161168.
12. Herman MG, Abrams RA, Mayer RR. Clinical use of on-line portal imaging
for daily patient treatment verication. Int J Radiat Oncol Biol Phys 1994;
28(4):10171023.
13. Petrascu O, Bel A, Linthout N, et al. Automatic on-line electronic portal
image analysis with a Wavelet-based edge detector. Med Phys 2000; 27(2):
321329.
14. Van de Steene J, Van den Heuvel F, Bel A, et al. Electronic portal imaging with
on-line Correction of setup error in thoracic irradiation: clinical evaluation. Int
J Radiat Oncol Biol Phys 1998; 40(4):967976.
15. Van de Vondel I, Coppens L, Verellen D, et al. Microprocessor controlled
limitation system for a stand-alone freely movable treatment couch. Med Phys
1998; 25(6):897899.
16. Van de Vondel I, Coppens L, Verellen D, et al. Remote control for a standalone freely movable treatment couch with limitation system. Med Phys 2001;
28(12):25182521.
17. Van den Heuvel F, De Neve W, Coghe M, et al. Relations of image quality
in on-line portal Images and patient individual parameters for pelvic eld
radiotherapy. Eur Radiol 1992; 2:433438.
18. Van den Heuvel F, De Neve W, Verellen D, et al. Clinical implementation
of an objective computer-aided protocol for intervention in intra-treatment
correction using electronic portal imaging. Radiother Oncol 1995; 35:232239.
19. Verellen D, De Neve W, Van den Heuvel F, et al. On-line portal imaging: image
quality dening parameters for pelvic elds - A clinical evaluation. Int J Radiat
Oncol Biol Phys 1993; 27:945952.
20. Bel A, van Herk M, Lebesque JV. Target margins for random geometrical treatment uncertainties in conformal radiotherapy. Med Phys 1996; 23:15371545.
21. Bel A, Vos PH, Rodrigus PT, et al. High-precision prostate cancer irradiation
by clinical application on an ofine patient setup verication procedure, using
portal imaging. Int J Radiat Biol Phys 1996; 35:321332.
22. van Herk M, Bruce A, Kroes AP, et al. Quantication of organ motion during
conformal radiotherapy of the prostate by three dimensional image registration.
Int J Radiat Oncol Biol Phys 1995; 33:13111320.
23. Yan D, Wong JW, Gustafson G, et al. A new model for accept or reject strategies in off-line and on-line megavoltage treatment evaluation. Int J Radiat
Novalis
67
24. Yan D, Wong J, Vicini F, et al. Adaptive modication of treatment planning to

minimize the deleterious effects of treatment setup errors. Int J Radiat Oncol
Biol Phys 1997; 38:197206.
25. Holupka EJ, Kaplan ID, Burdette EC, et al. Ultrasound image fusion for external beam radiotherapy for prostate cancer. Int J Radiat Oncol Biol Phys 1996;
35(5):975984.
26. Jaffray DA, Siewerdsen JH, Wong JW, et al. Flat-panel cone-beam computed
tomography for image-guided radiation therapy. Int J Radiat Oncol Biol Phys
2002; 53(5):13371349.
27. Shirato H, Shimizu S, Shimizu T, et al. Real-time tumor-tracking radiotherapy.
Lancet 1999; 353:13311332.
28. Shirato H, Shimizu S, Kitamura K, et al. Four-dimensional treatment planning
and uoroscopic real-time tumor tracking radiotherapy for moving tumor. Int J
29. Verellen D, Soete G, Linthout N, et al. Quality assurance of a system for
improved target localization and patient set-up that combines real-time infrared
tracking and stereoscopic X-ray imaging. Radiother Oncol 2003; 67:129141.
30. Aoki Y, Akanuma A, Karasawa K, et al. An integrated radiotherapy treatment
system and its clinical application. Radiat Med 1987; 5(4):131141.
31. Takeuchi H, Yoshida M, Kubota T, et al. Frameless stereotactic radiosurgery
with mobile CT, mask immobilization and micro-multileaf collimators. Minim
Invasive Neurosurg 2003; 46(2):8285.
32. Soete G, Van de Steene J, Verellen D, et al. Initial clinical experience with
infrared reecting skin markers in the positioning of patients treated by conformal radiotherapy for prostate cancer. Int J Radiat Oncol Biol Phys 2002; 52(3):
694698.
33. Soete G, Verellen D, Michielsen D, et al. Clinical use of stereoscopic X-ray
positioning of patients treated with conformal radiotherapy for prostate cancer.
Int J Radiat Oncol Biol Phys 2002; 54(3):948952.
34. Penney GP, Weese J, Little JA, et al. A comparison of similarity measures for
use in 2D3D medical image registration. IEEE Trans Med Img 1998; 17(4):
586595.
35. Schell MC, Bova FJ, Larson DA, et al. AAPM Report No. 54, Stereotactic
Radiosurgery: Report of AAPM Task Group 42, 1995.
36. 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
37. Verellen D, Linthout N, Soete G, et al. Considerations on treatment efciency
of different conformal radiation therapy techniques for prostate cancer. Radiother Oncol 2002; 63:2736.
38. Biggs PJ, Goitein M, Russell MD. A diagnostic X-ray eld verication
device for a 10 MV linear accelerator. Int J Radiat Oncol Biol Phys 1985; 11:
635643.
39. Schewe J, Lam K, Balter J, et al. Development of a room-based diagnostic
imaging system for use in radiotherapy. Med Phys 1995; 22:939940.
68
Medin and Verellen
40. Shiu AS, Hogstrom KR, Janjan NA. Technique for verifying treatment elds
using portal images with diagnostic quality. Int J Radiat Oncol Biol Phys
1987; 13:15891594.
41. Munro P, Bouius DC. X-ray quantum limited portal imaging using amorphous
silicon at-panel arrays. Med Phys 1998; 25(5):689702.
42. Motz JW, Danos M. Image information content and patient exposure. Med
Phys 1978; 5(1):822.
43. Rogers DW. Fluence to dose equivalent conversion factors calculated with
EGS3 forelectrons from 100 keV to 20 GeV and photons from 11 keV to 20
GeV. Health Phys 1984; 46(4):891914.
44. Adler JR, Chang SD, Murphy MJ, et al. The cyberknife: a frameless robotic
system for radiosurgery. Stereotact Funct Neurosurg 1997; 69:124128.
45. Murphy MJ. An automatic six-degree-of-freedom image registration algorithm
for image-guided frameless stereotaxic radiosurgery. Med Phys 1997; 24(6):
857866.
46. Balter JM, Lam KL, Sandler HM, et al. Automated localization of the prostate
at the time of treatment using implanted radiopaque markers: technical feasibility. Int J Radiat Oncol Biol Phys 1995; 33(5):12811286.
47. Balter JM, Sandler HM, Lam K, et al. Measurement of prostate movement
over the course of routine radiotherapy using implanted markers. Int J Radiat
Oncol Biol Phys 1995; 31(1):113118.
48. Gall KP, Verhey LJ. Computer-assisted positioning of radiotherapy patients
using implanted radiopaque ducials. Med Phys 1993; 20(4):11531159.
49. Lam KL, Ten Haken RK, McShan DL, et al. Automated determination of
patient setup errors in radiation therapy using spherical radio-opaque markers.
Med Phys 1993; 20(4):11451152.
50. Vigneault E, Pouliot J, Laverdiere J, et al. Electronic portal imaging device
detection of radiopaque markers for the evaluation of prostate position during
megavoltage irradiation: a clinical study. Int J Radiat Oncol Biol Phys 1997;
37(1):205212.
51. Song PJ, Washington M, Vaida F, et al. A comparison of four patient immobilization devices in the treatment of prostate cancer patients with three dimensional conformal radiotherapy. Int J Radiat Oncol Biol Phys 1996; 34:213219.
52. Stroom JC, Koper PC, Korevaar GA, et al. Internal organ motion in prostate
cancer patients treated in prone and supine treatment position. Radiother
Oncol 1999; 51(3):237248.
53. Zelefsky MJ, Crean D, Mageras GS, et al. Quantication and predictors of
prostate position variability in 50 patients evaluated with multiple CT scans
during conformal radiotherapy. Radiother Oncol 1999; 50(2):225234.
54. Mah D, Freedman G, Milestone B, et al. Measurement of intrafractional prostate motion using magnetic resonance imaging. Iht J Radiat Oncol Biol Phys
2002; 54(2):568575.
55. Chandra A, Dong L, Huang E, et al. Experience of ultrasound-based daily
prostate localization. Int J Radiat Oncol Biol Phys 2003; 56(2):436447.
56. Little DJ, Dong L, Levy LB, et al. Use of portal images and BAT ultrasonography to measure setup error and organ motion for prostate IMRT: implications
for treatment margins. Int J Radiat Oncol Biol Phys 2003; 56(5):12181224.
Novalis
69
57. Langen KM, Pouliot J, Anezinos C, et al. Evaluation of ultrasound-based prostate localization for image-guided radiotherapy. Int J Radiat Oncol Biol Phys
2003; 57(3):635644.
58. Van den Heuvel F, Powell T, Seppi E, et al. Independent verication of ultrasound based image-guided radiation treatment; using electronic portal imaging
and implanted gold markers. Med Phys 2003; 30(11):28782887.
59. Yin FF, Ryu S, Ajlouni M, et al. A technique of intensity-modulated radiosurgery (IMRS) for spinal tumors. Med Phys 2002; 29(2):28152822.
60. Cosgrove VP, Jahn U, Phaender M, Bauer S, Budach V, Wurm RE. Commissioning of a micro multi-leaf collimator and planning system for stereotactic
radiosurgery. Radiother Oncol 1999; 50:325336.
61. Xia P, Geis P, Xing L, Ma C, Findley D, Rorster K, Boyer A. Physical characteristics of a miniature multileaf collimator. Med Phys 1996; 26(1):6570.
62. Agazaryan N, Solberg TD, DeMarco JJ. Patient specic quality assurance for
the delivery of intensity modulated radiotherapy. J Appl Clin Med Phys 2003;
4(l):4050.
63. Llacer J. Inverse radiation treatment planning using the dynamically penalized
likelihood method. Med Phys 1997; 24(11):17511764.
64. Llacer J, Solberg T, Promberger C. Comparative behaviour of the dynamically
penalized likelihood algorithm in inverse radiation therapy planning. Med Phys
Biol 2001; 46:26372663.
65. Linthout N, Verellen D, Van Acker S, Van de Vondel I, Coppens L, Storme G.
Assessment of the acceptability of the Elekta multileaf collimator (MLC) within
the Corvus planning system for static and dynamic delivery of intensity modulated beams (IMBs). Radiother Oncol 2002; 63(1):121124.
66. Low DA, Sohn JW, Klein EE, Markman J, Mutic S, Dempsey JF. Characterization of a commercial multileaf collimator used for intensity modulated radiation therapy. Med Phys 2001; 28(5):752756.
67. Mohan R, Chui C, Lidofsky L. Differential pencil beam dose computation
model for photons. Med Phys 1986; 13(1):6473.
68. Mohan R, Chui C, Lidofsky L. Energy and angular distributions of photons
from medical linear accelerators. Med Phys 1985; 12:592597.
69. Mohan R, Chui C. Use of fast fourier transforms in calculating dose distributions for irregularly shaped elds for three-dimensional treament planning.
Med Phys 1987; 14:7077.
70. Bortfeld R, Oelfke U, Nill S. What is the optimum leaf width of a multileaf
collimator? Med Phys 2000; 27(11):24942502.
71. American Association of Physicists in Medicine (AAPM). Radiation treatment
planning dosimerty verication. AAPM Report 55 of Task Group 23 of the
Radiation Therapy Committee. Woodbury, NY: American Institute of Physics,
1995.
72. International Commission on Radiation Units and Measurements (ICRU). Use
of Computers in External Beam Radiotherapy Procedures with High-Energy
Photons and Electrons. ICRU Report 42. Baltimore, MD: ICRU, 1987.
73. Low DA, Harms WB, Mutic S, Purdy JA. A technique for the quantitative
evaluation of dose distributions. Med Phys 1998; 25(5):656661.
70
Medin and Verellen
74. Kutcher GJ, Coia L, Gillin M, Hanson WF, Leibel S, Morton RJ, et al. Comprehensive QA for radiation oncology: report of AAPM radiation therapy
committee task group 40. Med Phys 1994; 21(4):581618.
75. Wang LT, Solberg TD, Medin PM, Boone RA. Infrared patient positioning for
stereotactic radiosurgery of extracranial tumors. Comput Biol Med 2001; 31:
101111.
76. Solberg TD, Paul TJ, Boone RA, Agazaryan NN, Urmanita T, Arellano AR,
Llacer J, Fogg R, DeMarco JJ, Smathers JB. Feasibility of Gated IMRT.
Proceedings of the World Congress on Medical Physics and Biomedical
Engineering, Chicago, IL, July 2328, 2000.
77. Baroni G, Ferrigno G, Orecchia R, Pedotti A. Real-time three dimensional
motion analysis for patient positioning verication. Radiother Oncol 2000;
54:2127.
78. Hugo GD, Agazaryan N, Solberg TD. An evaluation of gating window size,
delivery method, and composite eld dosimetry of respiratory-gated MRT.
Med Phys 2002; 29:25172525.
79. Solberg TD, Paul TJ, Agazaryan NN, Urmanita T, Arellano AR, Llacer J,
Boone RA, Fogg R, DeMarco JJ, Chetty I, Smathers JB. Dosimetry of Gated
Intensity Modulated Radiotherapy. In: Schlegel W, Bortfeld T, eds. The Use of
Computers in Radiation Therapy. Berlin: Springer-Verlag, 2000:286288.
80. Paul TJ, Solberg TD, Leu MY, Rosemark PJ, Smathers JB. Independent real-time
verication of dynamically shaped intensity modulated radiotherapy (IMRT)
using an amorphous silicon (a-Si:H) detector array. Med Phys 1999; 26:1065.
81. Lujan AE, Baiter JM, Ten Haken RK. Determination of rotations in three
dimensions using two-dimensional portal image registration. Med Phys 1998;
25(5):703708.
82. Lujan AE, Larsen EW, Baiter JM, Ten Haken RK. A method for incorporating
organ motion due to breathing into 3D dose calculations. Med Phys 1999;
26:715720.
83. Hugo GD, Agazaryan N, Solberg TD. The effects of tumor motion on planning
and delivery of respiratory-gated MRT. Med Phys 2003; 30:10521066.
84. Linthout N, Verellen D, Van Acker S, et al. Dosimetric evaluation of partially
overlapping intensity modulated beams using dynamic mini-multileaf collimation. Med Phys 2003; 30(5):847855.
4
Whole-Body Radiosurgery
with the CyberKnife
Achim Schweikard
Luebeck University, Luebeck, Germany
Hiroya Shiomi
Osaka University Hospital, Osaka, Japan
Minaro Uchida and John R. Adler

Stanford University Medical Center, Stanford, California, U.S.A.
INTRODUCTION
In the 1950s, Professor Lars Leksell of the Karolinska Institute in Sweden
coined the term radiosurgery to dene a neurosurgical procedure that
combined precision targeting with a large number of cross-red beams of
ionizing radiation. By directing a very large dose of highly collimated radiation at a discrete location in the brain, the objective of Leksells operation
was to make a lesion (ablate brain tissue) without cutting, arguably
thereby achieving the ultimate in minimally invasive surgery. At the time
(1950s and 1960s) the clinical motivation for such a procedure was a class
of operations broadly referred to as functional neurosurgery, which includes
thalamotomy for the tremor of Parkinsons disease and anterior capsulotomy for treating obsessivecompulsive disorder.
Leksell investigated multiple irradiation technologies for fullling
his vision, including linear accelerator and heavy particle based concepts,
before focusing his development efforts on the radioactive cobalt-based
CyberKnife , Accuray, Sunnyvale, California, U.S.A.
71
72
Schweikard et al.
GammaKnife , (ELEKTA AB, Stockholm, Sweden). The eventual design

of this instrument was such that it could only be applied to the brain. The
rst GammaKnife was built in the late 1970s, and has since undergone a series of changes to accommodate a gradual evolution in usage. Leksell always
envisioned that his system would be used for non-invasive functional
neurosurgery. However, in the 1970s, computed tomography (CT) scanning
arrived and, with it, the opportunity to visualize mass lesions in the brain,
especially tumors. Quickly the role of the GammaKnife and the principles
of radiosurgery were redirected to facilitate the treatment of brain tumors
and vascular malformations. By incorporating the effectiveness and
minimally invasive nature of radiosurgery into the neurosurgical armamentarium, the past 20 years have witnessed a revolution in the management
of many neurosurgical procedures. Over this period new radiosurgical
technologies evolved, including more spatially accurate linear acceleratorbased methods (compared to Leksells earliest experience) and heavy
particle beams. Furthermore, the inuence of radiosurgery on neurosurgery continues to expand as indications and treatment parameters are
rened.
Technological Limitations
The near ubiquity of radiosurgery within modern neurosurgical practices,
there are inherent shortcomings to present-day technology that limit potential new applications. In particular, the GammaKnife and other related
linac-based technologies all require stereotactic frames for accurate beam
targeting. Such skeletal xation causes enough pain to preclude exible
treatment fractionation. Since the principle of administering a dose of radiation over more than one session is an essential element of all modern
radiation oncology, this limitation is not trivial; basic radiobiology and clinical experience have demonstrated the superiority of using fractionation to
treat most malignant tumors and/or lesions involving critical yet sensitive
anatomic structures, such as the optic apparatus. Furthermore, current
radiosurgical instruments are isocentric-based, constraining all beams to
converge on a common point. This design restricts some types of more exible and conformal treatment planning. While these limitations by themselves might not serve as an impetus for designing a new concept in
radiosurgery, an even greater rationale is provided by the opportunities to
perform radiosurgery outside the brain.
At its core, radiosurgery involves two elemental principles: (a) precision
targeting and (b) the application of ablative doses of radiation using limited
if any fractionation. From a biologic and therapeutic perspective such
principles could be of generic interest to all surgeons. Therefore, to address
the above limitations and meet the needs of potential new extracranial
therapies, a system for image-guided robotic radiosurgery was implemented.
CyberKnife
73
Why Image-Guided Robotic Radiosurgery?

In developing a system for extracranial targeting and treatment, it was critical
to begin by dening the specic clinical rationale for such a technology.
Not surprisingly it was decided that the specications for an extracranial
radiosurgical system should be consistent with Leksells original denition
for radiosurgery. In particular, such an instrument should enable the
highest possible targeting accuracy (essentially near millimeter RMS errors)
and enable ablative doses of radiation to be administered with a maximal
use of solid angle. It is important to emphasize that the intent was to administer ablative doses of radiation and thereby enable surgeons to replace open
surgical resection or invasive tissue destruction with a non-invasive procedure. This radiosurgical objective is to be distinguished from precision
radiation therapy and intensity-modulated therapy (IMRT), where accuracy
is less important, and the principles of conventional fractionation play a more
important role in clinical outcome. An instrument utilizing image-guided
robotics seemed uniquely able to achieve these objectives.
By extrapolating from the huge prior experience with brain radiosurgery,
various surgical specialists are nally in a position to develop new extracranial
radiosurgical applications for the CyberKnife (Accuray, Sunnyvale, California, U.S.A.). However, similar to other forms of surgery, the limits of human
operative capability are not dened by tools alone. The imagination and
diligence with which these instruments are directed towards relieving human
suffering remain critical aspects of improving patient outcome.
This chapter gives a brief overview of the CyberKnife technology, and
then presents two of the most innovative technical components of the
CyberKnife: the planning sub-system and the motion tracking system for
respiration compensation.
Related Work
Although radiosurgery is rapidly changing the scope of surgery, the vast
majority of therapeutic irradiation is administered as part of a regimen of
conventionally fractionated external beam radiotherapy. Medical linear
accelerators (linac systems), which use a gantry construction for moving
the linear accelerator, are the standard technology for delivering conventional radiotherapy. However, this mechanical construction was designed
more than 40 years ago to deliver radiation from a limited number of directions during a single treatment. Meanwhile, its ability to compensate for
target motion during treatment is inherently very limited. Several additional
factors limit the accuracy of gantry-based systems, most notably mechanical
ex and lack of fully computerized position/motion control.
Conventional linac-systems have six motion axes, which are sufcient to
target any point within a given workspace from any angle. However, four of
74
Schweikard et al.
the six axes of the linac-gantry are built into the patient table. Compensating
for respiratory motion is difcult with this construction. Motions of the patient
table (especially lateral table tilting) are likely to cause involuntary countermotion of the patient. Such involuntary motions lead to muscle contraction,
changes in breathing patterns, and may cause substantial additional inaccuracy.
Given the kinematic and size limitations of such systems, most researchers have
investigated methods for motion detection rather than active motion tracking.
Respiratory gating is a technique that attempts to address the problem
of breathing motion with conventional linac-based radiation therapy.
Gating techniques do not directly compensate for breathing motion; that
is, the therapeutic beam is not moved during activation. Instead the beam
is switched off whenever the target is outside a predened window. One
of the disadvantages of gating techniques is the increase in treatment time.
A second problem is the inherent inaccuracy of such an approach. One must
ensure that the beam activation cycles are long enough to obtain a stable
therapeutic beam.
Kubo and Hill (1) compared various external sensors (breath temperature sensor, strain gauge, and spirometer) with respect to their suitability for
respiratory gating. By measuring breath temperature, it is possible to determine whether the patient is inhaling or exhaling. Kubo and Hill veried that
frequent activation/deactivation of the linear accelerator does not substantially affect the resulting dose distribution. However, the application of such
a technique still requires a substantial safety margin for the following reason:
the sensor method only yields relative displacements during treatment, but
does not report and update the exact absolute position of the target during
treatment.
Tada et al. (2) report on the use an external laser range sensor in connection with a linac-based system for respiratory gating. This device is
used to switch the beam off whenever the sensor reports that the respiratory
cycle is close to maximal inhalation or maximal exhalation. However, typical variations in the respiratory motion patterns of 12 cm for the same
patient (in pediatrics), and in the duration of a single respiratory cycle of
25 sec are reported in Ref. 3.
As noted above, respiratory motion is difcult to track with conventional linac-based systems, and the accuracy of such an approach would
inherently be very limited. In contrast, modem robotic manufacturing relies
on highly accurate motion control and high unit numbers. In the CyberKnife
system, the radiation source (6 MV linear accelerator) is mounted on a
robotic arm that can move with six degrees-of-freedom.
Recent research has extended CyberKnife brain and spine radiosurgery in such a way that respiratory motion can be compensated for by active
motion of the robot. The basic advantage of this approach is that it is now
possible for the robotic arm to track the motion of a lesion, and is therefore
not necessary to gate the treatment beam.
CyberKnife
75
To enable dynamic respiratory tracking, the stereo X-ray imaging

system of the CyberKnife is combined with infrared tracking. X-ray imaging
senses internal anatomy, while infrared tracking provides information on the
motion of the patient surface. While X-ray imaging gives very accurate information about the internal location of a target, it is very difcult to obtain realtime motion information from X-ray imaging alone. In contrast, the motion
of the patient surface can be tracked in real-time with commercial infrared
position sensors. The central idea of our approach is to use a series of images
from both sensors (infrared and X-ray), synchronizing one with the other.
Thus, both X-ray image pairs and infrared data have time stamps. From a
series of sensor readings and corresponding time-stamps, we can determine
a motion pattern. This pattern correlates external to internal motion, and is
both patient and site specic. A series of preclinical and clinical trials has since
veried that we can accurately infer the placement of an internal target from a
precalculated (and continuously updated) motion pattern.
SYSTEM OVERVIEW
Figure 1 shows the components of the CyberKnife. A robot arm moves a
linear accelerator generating a 6 MV radiation beam. An X-ray camera
system (two X-ray cameras with nearly orthogonal visual axes) records
the position of natural or articial markers (i.e., bony landmarks or
implanted gold ducials). An infrared tracking system records the external
motion of the patients abdomen.
For treatments of brain or spinal lesions, no respiration compensation is
needed. In this case, infrared tracking is not active and the position of the target
is computed from the stereo X-ray images alone. Two nearly orthogonal X-ray
images show the position of the patients skull within a carefully calibrated imaging space. Meanwhile, the position of the target is known with respect to the
skull from the tomographic images. To compute the exact intratreatment position of the target, a registration step is necessary. This step spatially correlates
the intratreatment (live) X-ray images to the pretreatment three-dimensional
(3D) tomographic images. To this end, synthetic X-ray images are computed
from the CT scans, for a series of positions and angles. Live X-ray images
are then compared to synthetic images, to nd the best matching pair of synthetic images. Because each synthetic image pair has one and only one position
and angle associated with it, we can infer angular or positional displacement of
the patients skull, and hence perform the registration.
INVERSE PLANNING
During treatment the linear accelerator is moved in as step-and-shoot fashion to a series of xed points in space. At each such point (also called node),
the beam is activated for a certain duration. This duration determines the
weight of each beam. Notice that we assume each node to represent a xed
76
Schweikard et al.
Figure 1 CyberKnife system overview. (The arrows are pointing at the infrared
tracking system camera bar and the left X-ray camera detector). Infrared tracking
is used to record external motion of the patients abdominal and chest surface. Stereo
X-ray imaging is used to record the 3D position of internal markers (gold ducials)
at xed time intervals during treatment. A robotic arm moves the beam source to
actively compensate for respiratory motion.
position in space. However, this position may be adjusted during the

respiration tracking process. Thus, at each node, the robotic arm performs
very small corrective motions to compensate for respiratory motion when
this compensatory system is activated.
To make full use of the kinematic exibility of the six degree-of-freedom
robotic arm, an inverse planning system has been designed. The planning
process consists of two steps:
1. select nodes (beam congurations, positions, and orientations of
the beam source),
2. adjust beam weights (i.e., activation durations of the beam).
CyberKnife
77
Figure 2 Collimator selection menu.
Phase 1: Node Selection

In phase 1, it is assumed that the tumor has been delineated previously in the
tomographic image series. This results in a series of polygonal contours,
where one contour corresponds to each tomographic slice. Furthermore,
a collimator size is selected. The beam is cylindrical with collimator sizes
ranging from 5 to 60 mm (Fig. 2). After delineation of the tumor, the process
of beam selection is fully automatic. Up to 1200 beams are selected; greater
numbers of beams are used for more complex, non-convex tumor shapes,
or tumors near critical healthy structures. The rationale for the method
underlying beam selection for targets of arbitrary shape is a mathematical
extension of the (ideal) process that would be used if the tumor were strictly
spherical. After beam selection, the user can interactively modify selected
beam directions. For the principles underlying the automatic beam
selection, the reader is referred to Ref. 4.
Phase 2: Beam Weighting

In phase 2, it is assumed the beam selection process has been completed. As
noted above, this process is fully automatic once the tumor has been delineated in the tomographic images. To compute the optimal weights of the
individual beams, it is necessary to delineate critical structures in the vicinity
of the tumor. Furthermore, it is necessary to specify dose constraints. To
this end, upper and lower threshold values for the target region are entered.
Furthermore, threshold values are also entered for the critical regions, and
for individual beams. After specifying the constraints, the process of computing weights is fully automatic and based on linear programming. Linear
programming is a technique that has been proposed by several authors (5) in
the context of radiation therapy planning due to the completeness
properties of this method. Completeness in this context means that the
78
Schweikard et al.
Figure 3AC Inverse planning for robotic radiosurgery. (A) (See color insert.) Manual delineation of the target. Beam directions for optimized treatment of specic
tumor shape (up to 1200, lower left corner) are computed automatically by the
inverse planning system. (Continued)
algorithm is guaranteed to nd appropriate weight distributions exactly fullling the given constraints, if such a distribution exists. Extensions of this
method are described in Refs. 4 and 6.
An example for the planning process is given in Figure 3AC.
RESPIRATION TRACKING
Tumors in the chest and the abdomen move during respiration. The ability of
conventional radiation therapy systems to compensate for respiratory motion
by moving the radiation source is inherently limited. Because safety margins currently used in radiation therapy increase the radiation dose by a very large
amount, an accurate tracking method for following the motion of the tumor is
of utmost clinical relevance. To track respiratory motion, the following basic
method is used (7). Prior to treatment, small gold markers visible in X-ray images
CyberKnife
79
Figure 3 (B) Computing beam weights with linear programming. Dose threshold
selection menu. (Continued)
are anchored in proximity to the target organ. Stereo X-ray imaging is used
during treatment to determine the precise spatial location of the implanted gold
markers via automated image analysis. Using stereo X-ray imaging, precise
marker positions can be established once every 10 sec. This time interval is too
long to accurately follow respiratory motion.
In contrast, external markers (placed on the patients skin) can be
tracked automatically with optical methods at very high speed. Updated
80
Schweikard et al.
Figure 3 (C) (See color insert.) Both beam directions and beam weights are
computed automatically, once target and critical regions have been delineated, and
upper/lower dose thresholds have been entered.
positions can be reported to the control computer more than 60 times per
second. As noted above, external markers alone cannot adequately reect
internal displacements caused by breathing motion. Large external motion
may occur together with very small internal motion, and vice versa. In addition the direction of the visible external motion may deviate substantially
from the direction of the target motion (Fig. 4).
Because neither internal nor external markers alone are sufcient for
accurate tracking of lesions near the diaphragm, X-ray imaging is synchronized with optical tracking of external markers. The external markers are small
active infrared emitters (IGT Flashpoint 5000, Boulder, Colorado, U.S.A.)
attached to a vest. Notice that the individual markers are allowed to change
their relative placement. The rst step during treatment is to compute the exact
relationship between internal and external motion, using a series of timestamped snapshots showing external and internal markers simultaneously.
Although infrared tracking is combined with X-ray imaging, it is not necessary to detect the position of the infrared emitters in an X-ray image. Time
CyberKnife
81
Figure 4 Tracking of external markers/motion curves corresponding to inhalation

and exhalation. While external motion is in an anterior/posterior direction, the internal motion of the target may be in a left/right direction.
stamps permit the positions of both marker types to be established simultaneously, and can therefore be used to determine the pattern of respiratory
motion. Such patterns are patient specic and can be updated during treatment.
During the initialization stage of the procedure, a deformation model
is computed, which describes the correlation between internal and external
motion. To obtain the deformation model, we proceed as follows. A series
of (stereo) X-ray image pairs of the target region are taken while the patient
is breathing. When activating the X-ray sources, we record the point in time
Figure 5 Infrared emitters attached to the patients chest.
82
Schweikard et al.
Figure 6 Dependency between number of model points in a correlation model and

tracking accuracy. Test data were taken with an infrared tracking system alone,
where a correlation model correlating external chest motion to external abdomen
motion was used. The experiment suggests that very few model points are necessary
to obtain an accurate model. The gure shows data for three different infrared emitters placed on a test persons chest, while inferring chest position from abdominal
position alone via correlation.
at which the sources were activated. We also record the current position of
the external sensors at the time of X-ray image acquisition.
Assume ve X-ray image pairs have been taken in this way. These
images have a sequence order, namely the sequence in which they were
taken. In each image, we compute the absolute position of the gold internal
markers. Note that we obtain their exact spatial position, as the X-ray imaging reports stereo images. Consider the rst of the gold markers. By linear
interpolation, the positions of this marker determine a curve in space. We
compute such an interpolation curve for each of the gold markers.
There are six external markers and we compute the center of mass for
these markers. During a motion, the series of center points thus obtained
give rise to a single curve describing the motion of the external markers.
As the treatment proceeds the curves for internal and external markers
CyberKnife
83
are used in the following way. A new pair of X-ray images is acquired every
10 sec. Let there be a time point at which no X-ray image is taken. We must
then determine a predicted target placement based on the given deformation
model. At the given time point, we read the external sensor positions. After
computing the center of mass for the external markers, we can locate the
closest point on the curve for the external motion in our deformation model.
By linear interpolation this point determines corresponding points on the
curves for the internal markers.
Linear interpolation extends directly to the case, where one or both
curves are not line segments. In this case, points are connected by line
segments, and we determine the lengths of the line segments in each curve.
The parameter interval for each curve is then partitioned according to these
lengths and the interpolation proceeds as above. This interpolation is sufciently fast to update the given deformation model as the treatment
proceeds. Specically, as new X-ray images and matching sensor readings
become available, the curves for the internal markers are updated. The
updating scheme will be described in more detail in subsequent sections.
A difculty arising in this context is to distinguish between voluntary
patient movement and breathing motion. Clearly, the two types of motion
must be processed in different ways. Patient movement can occur, e.g., as
a result of muscle relaxation, sneezing, or voluntary movement. A patient
shift must cause a shift of the deformation model, i.e., each curve must be
shifted. In contrast, normal breathing should not shift these curves.
At rst glance, distinguishing the two types of motion may seem difcult. However, the internal markers represent a ground truth. Thus, our
deformation model gives a predicted position for the internal markers,
based on the position of the external markers. The location of internal
markers can be predicted not only for time points in between X-ray imaging,
but also for the times at which images are taken. Any deviation (exceeding a
xed threshold value d) between predicted and actual placement is thus
regarded as displacement caused by patient motion. Thus, the above linear
interpolation scheme yields a predicted placement for each internal marker
This predicted placement is a point in space. We compute the distance from
the actual placement of this point. If this distance is larger than d, the beam
is switched off. A new deformation model is computed after respiration has
stabilized.
Small values for d give better accuracy but enforce frequent re-computations of the deformation model. Given differences in patterns of breathing
between patients, it is reasonable to determine an appropriate value for during
initialization, when the patient is asked to breathe regularly. The deviations
observed during this interval of regular breathing (here for the external sensors
alone) are used to determine d.
A practical improvement of this technique for detecting patient
motion allows for updating the deformation model without interrupting
84
Schweikard et al.
the treatment. Assume the deformation model consists of ve X-ray image

pairs together with matching infrared position data. Because we take a new
pair of X-ray images at xed time intervals during treatment (i.e., every
10 sec), the deformation model can be updated continuously in the following
way. A table of position data with ve entries is maintained. Each entry has
a time stamp. In this table, we remove the earliest entry whenever a new set
of position data (image pair with matching infrared positions) becomes
available. The computation of the deformation model via linear interpolation can be carried out in real time after each update of the table data. This
method can compensate both for systematic drift of the patient position and
for small changes in the pattern of respiration during treatment.
CLINICAL TRIALS
Figure 7 shows representative results for respiration tracking in a clinical
case. The gure shows the total correlation error. Thus, based on the
Figure 7 Total target excursion (top curve), and correlation error (bottom curve) in
millimeters for a clinical case. The x axis gives the treatment beam direction number
(X-ray live shot number); the y axis gives the error in millimeters.
CyberKnife
85
Figure 8 Treatment of a hepatocellular carcinoma (HCC) with ducial-based

respiration compensation as described in the text. (A) Before treatment; (B) 3 months
after treatment. Total treatment time was 40 min per fraction for three fractions with
80 beam directions and 39 Gy.
correlation model, we compute the current position of the target based on

the external infrared sensor signal alone. At this same time point, we also
acquire a pair of X-ray images. We then plot the distance in millimeters
from the placement inferred by the correlation model and the actual placement determined from the implanted ducial markers in the image. The top
curve in this gure shows the corresponding target excursion.
Figure 9 Treatment of an adenocarcinoma with ducial-based respiration tracking

(Osaka University Hospital), three fractions at 39 Gy.
86
Schweikard et al.
CONCLUSION
Experience reported in the literature (8) suggests that robotic radiosurgery
with cylinder collimators can achieve distributions with higher conformality
than conventional multileaf collimators or micro multileaf collimators.
There are two practical reasons for this: (a) Cylinder collimators have excellent penumbra characteristics. In contrast, penumbra remains problematic
for multileaf collimators. (b) For spherical targets (a large percentage of
all tumors), cylinder collimators can achieve near optimal distributions.
The system selects beam congurations and beam weights automatically.
However, beam congurations and input thresholds may be adjusted in a
variety of ways. Selecting adequate parameters requires skill and thorough
understanding of the principles underlying the inverse planning process.
The clinical experience collected at several leading institutions worldwide conrms our hypothesis that respiratory motion of internal organs
can be correlated to visible external motion. It is necessary that the correlation
model be updated automatically during treatment. To update the correlation
model we use intratreatment stereo X-ray images. The clinical experience
further suggests that any dose margin placed around a tumor to compensate
for respiratory motion can be reduced by a very substantial amount. Because
dose is directly proportional to volume, this could allow for higher doses in
the tumor, and much lower doses in surrounding healthy tissue. For a variety
of cancers with grim prognoses, this robotic technique could thus lead to
far-reaching improvements in clinical outcome.
REFERENCES
1. Kubo HD, Hill BC. Respiration gated radiotherapy treatment: a technical
study. Phys Med Biol 1996; 41:9391.
2. Tada T, Minakuchi K, et al. Lung cancer: intermittent irradiation synchronized
with respiratory motionresults of a pilot study. Radiology 1998; 207(3):779783.
3. Sontag MR, Lai ZW, et al. Characterization of respiratory motion for pedriatic
conformal 3D therapy. Med Phys 1996; 23:1082.
4. Schweikard A, Bodduluri M, Adler JR. Planning for camera-guided robotic
radiosurgery. IEEE Trans Robotics Automation 1998; 14(6):951962.
5. Rosen II, Lane RG, Morrill SM, Belli JA. Treatment plan optimization using
linear programming. Med Phys 1991; 18(2):141152.
6. Hilbig M, Hanne R, Schweikard A. IMRT-Inverse planning based on linear
programming. Z Medizinische Physik 2002; 12:8996.
7. Schweikard A, Glosser G, Bodduluri M, Adler JR. Robotic motion compensation for respiratory motion during radiosurgery. J Comput-Aided Surg 2000;
5(4):263277.
8. Webb S. Conformal intensity-modulated radiotherapy (IMRT) delivered by
robotic linac-conformality versus efciency of dose delivery. Phys Med Biol
2000; 45:17151730.
CyberKnife
87
9. Bzostek A, Inoesco G, Carrat L, Barbe C, Chavanon O, Troccaz J. Isolating

moving anatomy in ultrasound without anatomical knowledge: application to
computer-assisted pericardical punctures. Medical image-computation and
computer-assisted interventionMICCAI 98, Lecture Notes in Computer
Science 1998; 1496:10411048.
10. Carol MP, Targovnik H 3D planning and delivery system for optimized conformal therapy. Int J Radiat Oncol Biol Phys 1992; 24:885887.
11. Hofstetter R, Slomoczykowski M, Sati M, Nolte LP. Flouroscopy as an imaging means for computer-assisted surgical navigation. Comput-Aided Surg
1999; 4:6576.
12. Lujan AE, Larsen EW, Balter JM, Haken RKT. A method for incorporating
organ motion due to breathing into 3D dose calculations. Med Phys 1999;
26(5):715720.
13. Morrill SM, Langer M, Lane RG. Real-time couch compensation for intratreatment organ motion: theoretical advantages. Med Phys 1996; 23:1083.
14. Winston KR, Lutz W. Linear accelerator as a neurosurgical tool for stereotactic
radiosurgery. Neurosurgery 1988; 22(3):454464.
5
Serial Tomotherapeutic Approaches to
Stereotactic Body Radiation
Therapy
Bill Salter
Department of Radiation Oncology, The University of Texas Health Science Center,
and Cancer Therapy and Research Center, San Antonio, Texas, U.S.A.
Martin Fuss
Department of Radiation Oncology, The University of Texas Health Science Center,
San Antonio, Texas, U.S.A.
SYSTEM DESCRIPTION AND OPERATION

Serial tomotherapy represents an unusual delivery technique relative to the
more traditional approaches typically employed in radiation therapy. As a
result, the specics of this approach are less familiar to many in our eld.
Most notably, the system is unusual in that it uses a combination of a moving
gantry and intensity modulation. While many static-gantry systems are
capable of delivering intensity modulated treatments, tomotherapy takes
advantage of both the semi-innite number of angles of approach afforded
by an arcing method of delivery, and the ability to create convex-shaped distributions of dose, which comes from modulation of intensity. The Peacock
system delivers its intensity modulated pencil beams through a binary
collimator called the MIMiC (NOMOS Corp., Crannberry Township,
Pennsylvania, U.S.A.) (Fig. 1A). The collimator is binary, i.e., each vane
of the collimator can assume one of only two states, either opened or closed.
Unlike so-called sliding leaf collimators, which can also be used to deliver
89
90
Salter and Fuss
Figure 1 (A) Image depicting

MIMiC (A in gure) and controller computer (B in gure).
(B) Image looking into mouth
or opening of MIMiC. Vanes
are arranged in a checkerboard
(opened and closed) pattern.
intensity modulated treatments, the MIMiCs leaves are pneumatically driven in very rapid fashion between these open and closed states, with typical
open/close times of less than 100 ms. Figure 1B depicts a view of the
collimator looking into the mouth or opening from which radiation emerges.
The collimator consists of two rows of 20 tungsten vanes (40 total vanes)
which are 8 cm in thickness. The leaves are stair-stepped in design, creating
an interdigitation of adjacent vanes that allows for very low interleaf
leakage values. Leaf/pencil-beam widths are constant at 10 mm and the pencil beam length can be varied among three user-selected dimensions (4, 8.5,
and 17 mm) by varying the physical or effective length of retraction of the
vane (1). During the treatment planning phase, pencil beam dimensions
are chosen appropriate to the treated lesions size and irregularity of shape.
Modulation of intensity is accomplished by varying the amount of time that
each pencil beam is open from a given gantry angle, and the uence map for
all pencil beams may be updated as frequently as every 5 of gantry rotation.
Each of the 40 pencil beams are controlled independently from the others,
Serial Tomotherapy
91
thus allowing each arc of the gantry to deliver two completely different and
separate uence maps, corresponding to each of the two rows of vanes, each
of which can be tailored to the unique shape of the target on that particular
treatment slice. For the 340 arcs (net 320 ) of radiation delivery typically
used at our center, this allows for 128 different pencil beams for each delivered gantry arc, or 64 per treated slice of the target. This can be contrasted
with typical values for static gantry approaches of 57, equivalent to the
number of static gantry ports utilized. Because serial tomotherapy employs
an arcing rotation of the slit-like MIMiC collimator, each rotation of the
gantry treats a slice of the patient, much like axial computed tomography
(CT) images a slice of the patient. Because of this similarity, serial tomotherapy has often been likened to CT image acquisition. A key difference
between the two approaches is that for CT acquisition the input beam uence is uniform, for the purpose of measuring the exiting non-uniformity
as an indication of the tissue density encountered by the beam, whereas
for intensity modulated radiation therapy (IMRT) the input beam uence
is intentionally non-uniform, for the alternate purpose of producing a uniform and conformal buildup of dose within the target. Because each rotation of the gantry typically treats a subsection of the target of thickness
equal to 2 4 mm 8 mm, 2 8.5 mm 17 mm, or 2 17 mm
34 mm, depending on the selected pencil beam size, the treatment
table/patient must be sequentially, or serially, incremented following each
delivered arc. Such increments must be performed very precisely to ensure
an accurate abutment of each adjacent treatment slice to avoid the creation of hot or cold strips between the delivered slices. Gantry rotations
are performed and the table is subsequently incremented until the entire
target has been treated. Because current linear-accelerator-inherent couch
positioning systems are not typically capable of positioning the patient
to the precision required for serial tomotherapy (required accuracy
0.1 mm), the vendor supplies a system referred to as either the Crane
or AutoCrane (NOMOS Corp., Crannberry Township, Pennsylvania,
U.S.A.) (2) for performing this function (Fig. 2). The device performs
the precise indexing of the patient necessary for serial tomotherapy, and
is available in an automated version that can be controlled at the treatment console, obviating the need to return to the treatment vault for
between-arc incrementing of the patient. The serial tomotherapeutic delivery approach has been shown to be capable of producing very conformal
dose distributions (14).
PATIENT IMMOBILIZATION AND ALIGNMENT
A prerequisite to the utilization of conformal distributions of dose is the use
of accurate patient xation and alignment methods. The challenges of
patient xation and alignment that are unique to extracranial applications
92
Salter and Fuss
Figure 2 AutoCrane mounted on linear accelerator table. The AutoCrane

performs remote incrementing of the treatment couch and the patient.
have been discussed in previous chapters. The methods employed at our

particular institution for serial tomotherapeutic delivery of extracranial
stereotactic treatments will now be described.
Patient alignment and immobilization is facilitated through the use of
a stereotactic, whole-body, double-vacuum-assisted immobilization system
(BodyFIX, Medical intelligence, Schwabmunchen, Germany). Comprehensive characterizations of the repositioning accuracy of the system have been
previously published (4,5).
In general, the system consists of a composite material base plate
(variably sized), sealed vacuum cushions, a clear plastic foil covering the
patients torso and lower extremities, and a vacuum pump (Fig. 3). Depending on the need for stereotactic localization and targeting, an arch-like
attachment can be afxed to the base plate, providing CT, magnetic resonance (MR), and positron emission tomography (PET) visible ducials, as
well as an integrated targeting system for stereotactic alignments in the
treatment vault.
For immobilization device implementation, in preparation for treatment simulation, the patient is placed supine onto a vacuum cushion which
is registered to the base plate by machined, clear-plastic registration bars that
Serial Tomotherapy
93
Figure 3 BodyFIX immobilization system with vacuum pump.
insert onto protruding pins. The vacuum cushion, which is lled with small
styrofoam BBs and has a valve attachment, allows for evacuation of the
enclosed air space through a vacuum pump. A clear plastic cover sheet is
attached by a sticky rubber strip to the left, right, and bottom/inferior sides
of the vacuum cushion, covering the patients lower body like a blanket up to
the abdomen or thorax. The air between the clear plastic sheet, the patient,
and the base vacuum cushion is evacuated, while the base cushion retains
its enclosed air, causing the soft cushion to mold to the patients posterior
surface, providing a negative mold. Once the vacuum cushion is molded to
the patients back and sides, the enclosed air is evacuated from the cushion,
creating a rigid, semi-permanent body cast. The vacuum cushions are available in various sizes, typically ranging in dimension from 180 65 to
220 80 cm, with a constant Styrofoam-BB to air ratio maintained for all
cushion sizes. Cushions are selected according to the patients dimensions,
thus enabling immobilization of both slender and relatively large patients in
the same system. Once the full-body cast of the patient has been created, evacuation of the air between the clear plastic sheet, the patient, and the full-body
cast creates a signicant and valuable immobilization of the patient in the
94
Salter and Fuss
mold. Vacuum pressure can be adjusted between 0 and 120 Mbar to accommodate individual patient tolerance, and values most commonly used for
our patients range between 60 and 100 Mbar. In addition to standard isocenter cross hairs and BBs that might be placed on the mold system for facilitation of treatment alignment, several other small BBs are typically placed onto
the mold system, on each side of the patient, at a cranio-caudal location near
the level of the tumor to facilitate CT/CT fusion for verication of treatment
position, as will be discussed in a later section on Treatment Verication.
At our institution, the patients BodyFIX mold system is often formed
in the PET imaging suite prior to acquisition of a PET image study which
can be utilized for fusion with the treatment-planning CT. Acquisition of
the PET image set with the patient in the precise position and orientation,
as will be used for the treatment planning CT, has proven to greatly facilitate the accurate registration of the two data sets, thus facilitating an accurate delineation of the metabolically active target. Initial creation of the
mold system in the PET suite is necessitated by the smaller bore size of
the Siemens ECAT-HR PET scanner (Siemens Corp., Berlin, Germany)
(64 cm) versus the 70-cm aperture of the Philips PQ5000 CT scanner (Philips
Medical Systems, Andover, Maryland, U.S.A.) utilized at our facility. If the
mold is initially created in the CT suite, there is no guarantee that the device
and patient will t through the smaller PET aperture later. Ironically, while
the PET aperture is smaller than the 70 cm CT aperture, the CT FOV of the
PQ5000 is only 48 cm, causing the CT-imaged FOV to be smaller than that
of the PET data set. This necessitates that special attention be paid to creating locations on the BodyFIX system for the placement of alignment and
registration markers that will be visible on the 48 cm FOV CT image set.
TREATMENT PLANNING
The widely recognized ability of intensity-modulating delivery approaches to
achieve conformality in even convex-shaped targets is achieved through an
exploitation of the increased degrees of freedom afforded by such delivery
schemes. Inherent to the increase in degrees of freedom is an enormous increase
in solution space and, thus, complexity. Fortunately, the computational power
of modern computing platforms has evolved to be capable of sufciently addressing such complex solutions. Most, if not all, vendors of intensity-modulationcapable planning systems now employ so-called inverse planning approaches.
The NOMOS Peacock serial tomotherapy approach used at our institution is
supported by the Corvus Inverse Treatment Planning System (ITPS) Version
5.0 (NOMOS Corp., Crannberry Township, Pennsylvania, U.S.A.). The planning system employs a dose volume histogram (DVH)-based objective function
and offers a choice between fast simulated annealing (FSA) and gradient descent
optimization approaches. Because the more computationally intense FSA
approach still progresses very quickly on modern hardware (as quickly as
Serial Tomotherapy
95
56 min) and because FSA affords an optimization approach that can avoid
becoming trapped in the local minima normally associated with DVH-based
objective functions, we use the FSA approach for all clinical treatment plans.
Typical to all ITPSs, the Corvus objective function requires the entry of
optimization parameters that characterize the nature of the problem to be
solved by optimization. The Corvus system requires that the following parameters be entered for the target: (a) goal dose (the desired prescription dose
to the target), (b) % volume below (the percentage volume of target tissue
that can be below the goal value), (c) minimum dose (the minimum dose
to the target that can be tolerated), and (d) maximum dose (the maximum
target dose that can be tolerated). In addition to the previously listed parameters, the system also requires that a target type be selected. Inherent
to the DVH-based objective function employed by the software are weighting
arguments that assign importance to the previously listed user-supplied parameters. The importance of the maximum target dose, for instance, would be
greater for a fractionated case (for which large hot spots are not acceptable),
than for a radiosurgical application (for which hot spots inside the target
might even be seen as advantageous). The selection of target type allows
for a customized weighting of the importance arguments to the particular
clinical application at hand. Interestingly, the implementation of such customized target types into the Corvus software evolved from our recognition in
1998 that in order to achieve clinically acceptable intracranial IMRS plans,
such a customized approach would need to be implemented.
In addition to the target optimization parameters listed previously, the
Corvus ITPS also requires that parameters be entered for all critical structures. As a minimum, the system always requires parameters to be entered
for healthy tissue, which represents, as a critical structure, all non-specic
healthy tissue types. If other critical structures, which need to be individually
controlled below specic dose levels, are present, parameters must be entered
for them as well. These parameters include: (a) limit (the dose limit that the
structure should be held below), (b) % volume above (the percentage of the
structure that can be allowed to rise above the structure dose limit), (c)
minimum (the structure dose below which there is limited value to further
decrease of dose), and (d) maximum (the maximum tolerable structure dose).
As for the target, each critical structure must be assigned a structure type,
and multiple structure types are available, with each dening customized
importance arguments unique to various clinical applications.
For the serial tomotherapeutic treatment of extracranial lesions at our
institution, the following values for the previously described optimization
parameters are typically used:
Target goal 3660 Gy in three fractions
Target % volume below 3%
Target minimum dose 9597% of target goal
96
Salter and Fuss
Target maximum 200% of target goal and target type ) IMRS

Tissue limit Target goal
Tissue volume above 0%
Tissue minimum 0%
Tissue maximum Target goal and tissue type ) IMRS
These values generally correspond to the vendors recommendation
for the IMRS target and tissue types, and may be set up one time in the software as a user-named template that can be easily recalled by a click of the
mouse. Because conformality and homogeneity are known to be competing
factors, our utilization of the IMRS target and tissue types allows for the
creation of extremely conformal distributions of dose, with an associated
acceptance of increased hot spots inside the target. Interestingly, the large
number of angles of approach afforded by the tomotherapy approach leads
to what would still be considered by many to be relatively low-valued hot
spots for such conformal plans. Hot spots, dened as the highest dose to
any single dose voxel, are almost always less than 30% and are, most often,
on the order of 1015% of the prescribed dose value.
Other relevant TPS parameters typically used include:
Heterogeneity correction ON during optimization and dose
calculation
Number of couch angles 1 (up to 3 for upper lung lobe tumors)
Couch angle 180 (Varian)
Gantry angles 35010 (Varian)
Pencil beam size 1 cm
Leaf transmission 0100% in 10% steps
Gantry step (for uence map update) 5
Cost function DVH
Optimizer Continuous annealer
Iterations More
Efciency 0
Benchmarks (i.e., relocation of isocenter) ) As necessary to achieve
target coverage from all gantry angles.
PTV margins are patient specic and evaluated based on full inhale
and exhale CT evaluation, as discussed in the section on Treatment of
Moving Targets. Typical values are 5 mm anterior, posterior, left, and right,
with 1015 mm inferior and superior.
For purposes of illustration an example case will now be presented.
Case Report
A 66-year-old female with diagnosis of stage 1 (T1N0M0) non-small-cell lung
cancer (NSCLS), histopathologically conrmed in ne-needle aspiration. The
Serial Tomotherapy
97
patient was referred for SBRT due to medical inoperability second to severe
chronic obstructive pulmonary disease (COPD).
The patient was immobilized using the BodyFIX double-vacuum
whole body immobilization system for both CT simulation and FDG-PET
imaging. For treatment planning, target volume delineation, anatomical
tumor information derived from CT data, and metabolic tumor information
derived from PET imaging were co-registered using a mutual information
image-fusion software inherent to the AcQSim (Philips Medical Systems,
Andover, Maryland, U.S.A.) virtual simulation software. Following target
delineation, the CT data and the associated target volume were exported
to the Corvus inverse planning platform by DICOM RT data transfer.
Inverse treatment planning parameters utilized were as follows:
Target
Target
Target
Target
goal dose 60 Gy (3 20 Gy)

volume below 3%
minimum dose 96% of target goal (or 57.5 Gy)
maximum 200% of target goal
The target volume type was specied as IMRS target, emphasizing

dose conformality over in-target dose homogeneity. The target volume was
grown into a PTV by adding margins of 5 mm transversally and 10 craniocaudally. Parameters for organs-at-risk (OAR) were set as follows:
Tissue
Tissue
Tissue
Tissue
Tissue
limit Target goal (60 Gy)

volume above 0%
minimum 0%
maximum Target goal (60 Gy)
type ) IMRS
Figure 4 depicts the Corvus prescription page with relevant critical

structure parameters.
The pencil beam dimension chosen was 8.5 10 mm (1-cm mode) and
the range of gantry rotation was 35010 , with pencil beams delivered from
340 to 20 . While the presented treatment was delivered over a single,
straight couch angle (Varian 180 ), upper lobe lesions are sometimes treated
using three couch angles (180 , 150 , and 210 , respectively).
The treatment plan was optimized utilizing typical ITPS parameters
described previously, with a benchmark move equal to 40 mm (i.e., patient
lowered from isocenter mark by 4 cm to allow laterally delivered pencil
beams to have access to target). Figure 5 depicts the resulting dose distribution, with 100%, 90%, 70%, and 50% isodose lines displayed, and Figures 6
and 7 present the achieved/delivered structure statistics and DVHs. The dose
distribution was normalized so that 95% of the PTV received 100% of the
prescription dose. Thus, the GTV received no less than 61.4 Gy, with mean
and maximum doses of 68.8 and 76.9 Gy, respectively.
98
Salter and Fuss
Figure 4 Corvus prescription page for the case report example containing ITPS
input parameters.
OARs, namely the esophagus and the spinal cord, were predicted by
the TPS to receive average doses of 6.9 and 4.2 Gy, with maximum doses
of 13.9 and 4.2 Gy, respectively.
Following repeat control CT scanning in the immobilization system
immediately prior to treatment to conrm accuracy of patient and target
setup (as described in the following section of Treatment Verication), the
treatment was delivered in three fractions separated by 48 hr using a 6 MV
linear accelerator with 600 MU/min delivery capability. The number of couch
indices, or treatment slices, was four. Figure 8 depicts the pencil beams and
associated intensity levels utilized from a subset of ve of the 64 actual arcing
segments used for treatment. Note that for table indices 1, 2, and 3 (numbered
Serial Tomotherapy
99
Figure 5 Isodose distribution for the case report example. Shown are the CTV
(bright red), PTV (darker red), esophagus protection region (green), spinal cord protection region (blue), and isodose lines: 100% (dark blue) 90% (red), 70% (yellow),
and 50% (green). (See color insert.)
at the left of Fig. 8 in white), both banks of pencil beams are delivering radiation and are modulated independently. At table index 4, only the cranially
oriented bank of pencil beams sees the target, and accordingly only these
pencil beams are utilized by the inverse planning system. As mentioned, the
ve arc-segments depicted are a subset of a total of 64 segments delivered
for each gantry arc (each segment delivered over 5 gantry rotation). The
central portion of Figure 8 also displays all utilized pencil beams over the
entire gantry rotation, with each white dot on the patient surface representing
an actual pencil beam delivered with non-zero intensity.
100
Salter and Fuss
Figure 6 Statistics page for the case report example.
Total monitor units delivered were 43,037 for a net radiation delivery
time of approximately 72 min. Including patient setup, control CT/image analysis, and treatment delivery, each fraction was completed in less than 100 min.
TREATMENT VERIFICATION
As mentioned previously, the effective and safe delivery of radioablative
doses to extracranial targets requires a precise placement of the delivered
high-dose region. While exhaustive efforts are undertaken at our institution
to assure an adequate alignment and immobilization of the patient, as
described previously, extensive experience gained from the greater than
200 SBRT treatments delivered at our facility (as of March 2004) have
Serial Tomotherapy
101
Figure 7 Cumulative dose volume histogram for the case report example. Shown
(from back to front) are the CTV, PTV, esophagus, and cord. (See color insert.)
convinced us that such efforts must include a detailed verication of patient

treatment position immediately prior to administration of treatment. While
our previously published evaluation of the accuracy and reproducibility
of the BodyFIX immobilization system used here for such treatments
demonstrated mean target translations of 2.9, 2.3, and 3.2 mm in the x, y,
and z directions, respectively, the occasional observance of larger values,
which are capable of compromising adequate coverage by the previously
stated PTV margins, is evidence of the need for pretreatment verication (4).
At our institution we have adopted a pretreatment verication
approach which utilizes a three-dimensional (3D) data set provided by a
repeat CT of the patient. Immediately prior to each days treatment, the
Figure 8 Depicting ve of the 64 different intensity patterns used for the

case report example (gantry angles
340 , 260 , 180 , 100 , and 20 ). Note
that the patient was treated with a
straight couch (i.e., 180 Varian),
and four couch increments (or treatment slices). The shaded squares represent pencil beam intensities utilized,
with brighter shades indicating higher
intensities. The central region of the gure graphically depicts pencil beams
used, as dots on the surface of the
patient. Note that the nal arc of the
treatment uses only the cranially
located row of MIMiC vanes because
inferior row of pencil beams does not
see the target volume. (See color
insert.)
102
Salter and Fuss
Serial Tomotherapy
103
patient is returned to the CT simulation suite, placed into the BodyFIX

whole-body immobilization system, and re-imaged in the treatment position. The resulting CT data set, referred to as the Control CT, is transferred
to the Corvus ITPS for fusion with the original treatment planning CT data
set. The ducial point matching fusion algorithm inherent to the Corvus
ITPS is utilized to produce a registration of the two data sets relative to the
BodyFIX whole-body cast. For such an approach to be valid, we must
assume the BodyFIX cast to represent a rigid-body, which does not change
shape during the course of a typical three-fraction treatment course, and the
experience gained from 200 such fusions has supported the validity of this
assumption within reasonable limits (4). The registration of images is performed using the small BBs that were attached to the mold system during
the treatment planning simulation, as described in a previous section.
Fusion of the two CT data sets, based on the BodyFIX cast system, results
in a near-perfect registration of the BodyFIX cast system between the two
image sets, with any variation of patient position between the two imaging
sessions apparent as movement of the patient anatomy observed when
toggling between data sets. The Corvus ITPS allows for an overlay of the
planned isodose distribution to be displayed on the Control CT dataset for
that particular days imaging session. Assessment of the pretreatment alignment quality is a straightforward matter of evaluating the coverage of the
relevant isodose lines relative to the target and any other critical structures
of interest. Based on such analysis, x, y, and z direction shifts required to
correctly re-center the target in the high-dose region may be determined
using the measurement tool inherent to the Corvus software. In our previously
published report on the accuracy of the BodyFIX system, we observed that
PTV margins of 5 mm transversally and 10 mm cranio-caudally would have
provided adequate (i.e., target covered by at least the 90% isodose line) uncorrected target coverage in 87.1% of the 109 control CT data sets evaluated (4).
Following determination of the 3D shifts necessary to return the target to
the correct location within the delivered isodose distribution, all patients are
shifted accordingly in the treatment room.
The utilization of a pretreatment control CT has the advantage of
providing a high-resolution, high-contrast, 3D image set for precise evaluation of target location relative to treatment isocenter. The method allows for
visualization of treatment isodose lines overlaid on the patient/target position for that particular days CT control, thus providing reassurance of a
proper 3D target position prior to treatment. The method has as a disadvantage the assumption of the validity of the CT control as a surrogate for the
patient position on the treatment table. This assumption is valid only if one
believes that the patient does not move within the BodyFIX cast between
the time of imaging and treatment delivery. Because we currently do not
have the ability to acquire a 3D image set in the treatment room, the CT
control must be performed in the virtual simulation CT suite, which is
104
Salter and Fuss
located down a long hall from the treatment vault. Following CT control
image acquisition, the patient is transferred to a transport-gurney by sliding
the BodyFIX treatment board assembly from the CT table onto the gurney.
The patient can then be wheeled down the hallway to the treatment vault
where a similar procedure can be used to transfer the patient from the gurney to the linear accelerator treatment table. Ideally, the schedules of the CT
suite and treatment vault are coordinated such that the patient may be
transferred directly to the treatment vault. In reality this is often, but not
always, the case. In such instances the patient may need to wait in a patient
holding area, adjacent to a nurses station, for as much as 1015 min, thus
further testing the assumption of the CT control position as an acceptable
representation of patient position at treatment time. As a coarse evaluation
of this assumption, we also perform an assessment of patient position on the
treatment table by comparison of pretreatment port lms and CT-simulation-generated digitally reconstructed radiographs (DRRs). Any signicant
patient movement within the BodyFIX mold between the time of CT control
and placement of the patient onto the treatment table is believed to be visible on such images. Ideally, of course, we would acquire the 3D CT data set
in the treatment room, on the treatment table, and linear accelerator-based
cone beam imaging and CT-on-rails approaches may hold promise for
such applications, as will be discussed in a later section entitled Future
Directions.
TREATMENT OF MOVING TARGETS

As discussed in previous chapters, the treatment of thoracic and upper
abdominal targets must necessarily address the problem of respiration
related motion, and numerous reports have characterized the nature and
magnitude of target motion due to respiration (58). Current approaches
aimed at addressing the treatment delivery aspects of such motion are seen
to center around three general philosophies. The rst is that of attempting to
stop, or control, the motion to some degree by the application of breathing
control approaches. Such approaches may be as aggressive as physically limiting when, and to what degree, the patient may breath; as moderate as reducing respiratory amplitude through the use of abdominal pressure applied
via belts, plates or cushions; or as passive as coaching the patient to minimize the depth and frequency of respiration. In any case, investigators have
reported an ability to effectively reduce the degree of target motion to acceptable limits. Such motion controlling approaches have the advantage of
ensuring that the target remains in an acceptable treatment location, thus
maximizing the amount of time spent delivering dose to the target, and minimizing dose delivered to healthy surrounding tissue. These approaches
suffer from the disadvantage that the majority of patients undergoing
Serial Tomotherapy
105
treatment for thoracic lesions are compromised, to some degree, in their

breathing efciency. Many such patients are incapable of tolerating such
limitations to their breathing cycle.
A second approach to addressing respiration related target motion is
that of pulsing, or gating, the treatment beam. Such approaches may monitor the respiratory cycle of the patient through the use of various respiration
sensors (e.g., belts, spirometers, and chest markers) and then quickly turn
the treatment beam off when the target has moved beyond tolerable limits.
Target position can also be ascertained through real-time tracking of
implanted markers. Such approaches have the advantage of gating the treatment such that dose is not delivered when the target is out of position, but
suffer from the disadvantage of increased treatment time due to time spent
waiting for the target to return to an acceptable treatment position.
Additionally, methods which rely on the rise and fall of the chest wall as
an indication of target position must assume that the chest wall is a valid
indicator of target position, and this assumption may be challenged by such
factors as phase and amplitude differences between target and chest wall
motion. The third general approach to dealing with target motion is, arguably, the least sophisticated. This approach addresses the problem of target
motion by simply increasing the PTV margin to sufciently accommodate
the motion of the target. Numerous methods exist for dening the magnitude and nature of PTV increase, from sophisticated time, motion, and
probability-related denitions, to methods which dene the PTV as the
physical intersection of the full inhale and full exhale PTV volumes, to even
simpler methods that measure the maximum extent of motion in the three
principle directions on the full inhale and exhale CTs. A so-called slow
CT scan is often performed to generate a treatment-planning CT data set
which represents the target as a larger blurred volume due to the slow
scan-time, relative to target motion frequency. Because we currently lack
technology necessary to support more sophisticated approaches, we are
employing a combination of the latter of the previously described PTVgrowth methodologies with abdominal pressure applied via the BodyFIX
immobilization system. PTV margins are designed, on a patient specic
basis, by the evaluation of full normal inhale and exhale maximum extent
of target motion while immobilized in the BodyFIX system, and treatment
planning CT data sets are acquired using slow CT scanning techniques.
Moderate abdominal pressure, applied by the vacuum cushions inherent
to the BodyFIX system, has been observed to provide a measure of respiratory amplitude reduction. While such an approach has proven clinically
effective, it is, admittedly, less than ideal, and we are currently budgeted
for and exploring the options for implementation of gating-related approaches. Such approaches appear to be capable of allowing for reductions in
PTV margin and, therefore, subsequent reductions in volume of healthy
tissue treated to prescription dose levels.
106
Salter and Fuss
LOGISTICS OF THE METHOD

The method described here for stereotactic treatment of extracranial targets
by use of a serial tomotherapeutic approach has been successfully employed
at our institution since August of 2001 for the treatment of 82 patients by
260 delivered fractions. The tomotherapy approach utilized is, in all relevant
regards, identical to the treatment approach utilized for the 50 conventionally fractionated IMRT patients we treat each day. As such, it represents a
proven delivery method which has been shown to present no signicant
technological challenges. The use of an arc-based delivery approach does
require that consideration be given to linear accelerator/treatment-couch
clearance issues, and we currently utilize a Varian ETR-Exact couch combination that has not posed any signicant challenges to such clearance for
treatment of lesions of the torso. The use of an intensity-modulating
approach allows for the creation of extremely conformal distributions of
dose, with the anticipated loss of treatment efciency associated with such
approaches. Monitor units (MU) required for delivery of any intensity
modulated treatment may be increased by a factor of four to six times over
standard, unmodulated delivery schemes, due to dose wasted to modulation.
Serial tomotherapys slicewise delivery approach further increases inefciency
by an additional factor approximately equal to the number of separate treatment arcs, due to its separate treatment of each slice of target. As a result,
delivered monitor units for a 12-Gy fraction typically range from 15,000 to
25,000 MU, and beam-on times from 25 to 45 min on the 600 MU/min Varian 600CD accelerator used for such treatments at our facility. The average
time from procedure start, dened as patient placed in the BodyFIX mold
for CT control, to procedure nished, dened as all dose delivered, is on
the order of 1 hr and 45 min. As such, the beam on time represents roughly
2443% of the entire procedure time. Any reduction of the treatment procedure time is, of course, valuable in that it can allow for a reduction in the
probability of patient movement, and for greater patient comfort. We are
mindful, however, that a 25% reduction in beam-on time will still represent
only a 611 min, or a 610%, reduction in procedure time. We are currently
exploring methods for reducing beam-on time, as will be discussed in the
Future Directions section, but such efforts must be weighed against improvements in conformality which are achieved through the use of such sophisticated, yet inefcient, delivery methods. Perhaps a more effective method
for a signicant reduction of procedure time would entail elimination of the
need to perform two patient setups, with associated transport and wait time,
subsequent to performing the control CT in a separate suite. Acquisition of
the 3D positional data set on the treatment table could, theoretically, reduce
the procedure time by 2030 min, and such an approach is of signicant
interest to us, as will be discussed in the following section.
Serial Tomotherapy
107
FUTURE DIRECTIONS
The Patient Immobilization and Alignment section alluded to logistical
challenges associated with limited aperture and FOV issues that necessitated
the creation of the BodyFIX mold in the smaller bore of the PET suite.
In the near future it is envisioned that our acquisition of a large bore and
extended FOV CT-PET combined unit will render such simulation arrangements unnecessary, allowing for a single imaging session and patient position for both the PET and CT planning data sets. For patients who do
not require a fused PET data set, acquisition of a large-bore CT unit with
80 cm aperture and 65 cm FOV is envisioned to eliminate current FOV
concerns, again, in the near future.
The section on Treatment Planning discussed the trade off between
improvements in conformality associated with intensity-modulating
approaches, such as Peacock approved, and the subsequent increases in treatment and procedure times due to reduced efciency and increased required MU
of such treatments. The Corvus ITPS possesses inherent capabilities which
allow for user selection of the complexity desired in the resulting intensitymodulated treatment plan. The software accepts user input through the positioning of a slider bar, which ultimately affects the modulation frequency of
the developed plan. Greater complexity equates to greater variation in the uences of the delivered pencil beams used for treatment and, therefore, to larger
numbers of wasted monitor units. We are currently systematically exploring
the trade-offs between plan quality and increased efciency. As mentioned in
the previous section, improvements in efciency can equate to reduced treatment time, which can subsequently result in reduced potential for patient
movement and discomfort. However, such improvements must not come at
the expense of clinically signicant reductions in conformity.
The need for ascertaining the correct and accurate position of the patient
was discussed in the section on Treatment Verication. In this section we
presented our method for verifying pre-treatment patient position through
the use of a Control CT. The reliance of this approach on an assumed validity
of the CT Control as an acceptable surrogate for the position of the patient on
the treatment table, at treatment time, was described. In the near future we
anticipate the elimination of the need for such assumptions through the installation of a linear accelerator with in-room CT and/or on-board cone beam
capabilities. Such in-room, three-dimensional imaging capability holds the
promise of allowing us to explore the potential of these exciting new image guidance approaches, and to provide nearly real-time verication of patient treatment position, with associated reduction of total procedure time. Lastly, the
section on Treatment of Moving Targets characterized our currently lessthan-sophisticated approaches to addressing respiratory related target
motion. We are currently budgeted for, and anticipate the installation of,
respiratory-gated technology in the very near future. Such technology is
108
Salter and Fuss
believed to be capable of allowing for a clinically signicant reduction in PTV

margin and, therefore, dose to healthy tissue. The obstacles to implementation
of gated technology in a tomotherapeutic environment are anticipated, and
alternative scenarios are currently being explored.
REFERENCES
1. Salter BJ. NOMOS Peacock IMRT utilizing the Beak post collimation device.
Med Dosim 2002; 26(1):3745.
2. Salter BJ, Hevezi JM, Sadeghi A, Fuss M, Herman TS. An oblique arc capable
patient positioning system for sequential tomotherapy. Med Phys 2001; 28(12):
24752488.
3. Fuss M, Salter BJ, Sadeghi A, Bogaev CA, Hevezi JM, Herman TS. Fractionated stereotactic Intensity-modulated radiotherapy (FS-IMRT) for small intracranial target volumes at the example of acoustic neuromas. Med Dosim 2002;
27(2):147154.
4. Fuss M, Salter BJ, Rassiah P, Cheek D, Cavanaugh SX, Herman TS. Repositioning accuracy of a commercially available double vacuum whole-body immobilization system. Technol Cancer Res Treat 2004; 3(2):161170.
5. Nevinny-Stickel M, Sweeney RA, Bale RJ, Posch A, Auberger T, Lukas P.
Reproducibility of patient positioning for fractionated extracranial stereotactic
radiotherapy using a double-vacuum technique. Strahlenther Onkol 2004;
180(2):117122.
6. Balter JM, Lam KL, McGinn CJ, et al. Improvement of CT-based treatmentplanning models of abdominal targets using static exhale imaging. Int J Radiat
7. Balter JM, Ten Haken RK, Lawrence TS, et al. Uncertainties in CT-based
radiation therapy treatment planning associated with patient breathing. Int
J Radiat Oncol Biol Phys 1996; 36:167174.
8. Davies SC, Hill AL, Holmes RB, et al. Ultrasound quantitation of respiratory
organ motion in the upper abdomen. Br J Radial 1994; 67:10961102.
9. Kubo HD, Hill BC. Respiration gated radiotherapy treatment: a technical study.
Phys Med Biol 1996; 41:8391.
6
Anatomical and Biological Imaging
Frank J. Lagerwaard and Suresh Senan
Department of Radiation Oncology, VU University Medical Center,
INTRODUCTION
The growing clinical interest in extracranial stereotactic radiotherapy
(ECSRT) is a logical consequence of the high control rates observed with
hypofractionated intracranial stereotactic radiotherapy (SRT). In principle,
similar SRT fractionation schedules can also be applied to extracranial sites,
provided that treatment volumes are small and critical normal structures are
avoided. Knowledge of patient-setup errors and (residual) mobility of both
tumor and normal structures is essential in order to determine the smallest
possible treatment planning margins without losing adequate tumor coverage. However, organ motion remains a major problem for extracranial sites,
and methods to address this problem require considerable resources and are
the subject of ongoing research.
It will be evident that the consequences of errors in the denition of
target volumes and adjacent critical normal tissues will be far greater for
hypofractionated stereotactic treatments than for conventionally fractionated treatments, and the characteristic steep dose-gradients obtained with
stereotactic irradiation will increase even further the clinical impact of such
geographical errors. In order to minimize the probability of such errors, two
important issues that are specic to implementing ECSRT have to be solved:
(1) the accuracy and reproducibility of patient positioning and (2) the
internal mobility of tumors and normal tissues.
109
110
Lagerwaard and Senan
Optimal imaging techniques that precisely localize both the target

volume and adjacent critical structures are a prerequisite for ECSRT.
Although some form of image-based information has always been used to
guide the radiotherapy treatment process, the availability and improved
quality of digital data and the computerized 3D reconstruction of the
patients anatomy have permitted current image-guided radiotherapy.
Image guidance can be performed not only for diagnosis and treatment
planning, but also for visualization during treatment delivery using either
external reference systems that are attached to the patient (frame-based
ECSRT) or using images that are registered relative to the patients anatomy
(frameless ECSRT).
The implementation of image-guided approaches has expanded from
using anatomical imaging techniques, such as computed tomography
(CT), magnetic resonance imaging (MRI), and ultrasound, to include several biological imaging techniques, e.g., positron-emission tomography
(PET) or magnetic resonance spectroscopy (MRS). However, the exact role
of biological imaging in ECSRT planning is currently investigational, but
co-registration of images from different anatomical and/or biological techniques can provide additional information in the determination of target
volumes and critical structures.
This review will address current and ongoing developments in the role
of image guidance in the phases of pretreatment selection, treatment planning, treatment delivery, and follow-up in ECSRT. As the topic of imaging
using all available techniques is extensive, and many tumor sites can be treated with stereotactic radiotherapy, this chapter will only focus on three key
target areas, i.e., lung, liver, and vertebral lesions, in order to highlight the
main principles.
APPROPRIATE SELECTION OF PATIENTS
ECSRT is a labor-intensive process and appropriate patient selection is
required in order to identify patients who are most likely to benet from this
approach. Patients with small primary tumors or limited metastatic disease
are the key candidates for ECSRT, and an important aspect of imaging
studies for patient selection is the exclusion of patients who have more extensive or rapidly progressing disease. In the absence of data from prospective
randomized clinical trials of ECSRT for metastatic disease, the surgical literature may offer useful pointers for patient selection criteria (16). In general,
patients who have controlled primary tumors with a limited number of metastases (3 or less) that are restricted to a single organ system, and have a good
performance score, can be considered candidates for ECSRT.
Staging accuracy for patients with cancer has been improved substantially by the availability of PET. PET has the ability to detect cancer on the
basis of biochemical and molecular processes within tumor tissues, and can
111
provide quantitative information, not only on metabolic activity but also on

perfusion (7,8), oxygenation (911), and proliferation (1214). A literature
review suggests that 18FDG-PET is between 10% and 20% more accurate
than conventional imaging for diagnosing, staging, and restaging of many
tumor types (15), including tumors of the lung (1623), colorectal cancer
(2426), head and neck cancer (27), breast cancer (28,29), and melanoma
(30,31).
In the following sections, the aspects of imaging in the patient selection
will be highlighted separately for three tumor sites, namely, lung cancer,
hepatic malignancies, and vertebral metastases.
The Role of Imaging in the Selection of Patients with Lung Cancer
Local control rates of only 3055% have been reported for conventionally
fractionated radiotherapy for stage I non-small cell lung cancer (NSCLC)
(32,33). In contrast, local control rates as high as 8090% have been
reported for hypofractionated ECSRT (3437). However, accurate staging
is essential as ECSRT would be an inappropriate treatment if occult hilar
or mediastinal nodal metastases were present. After a clinical diagnosis of
stage I NSCLC using conventional staging procedures, only around 6571%
of patients will have the same pathological stage, with upstaging to stage II
in 15%, stage III in 16%, and stage IV in 4% of patients (17,38).
18
FDG-PET has been shown to be superior to conventional CT scans
in the staging of patients with NSCLC (16,1823,39,40). However, as falsepositive ndings may occur in granulomatous disease such as tuberculosis or
sarcoidosis (41,42), 18FDG-PET uptake alone is not a sufcient basis to
select patients with solitary pulmonary lesions. Similarly, false-negative ndings have been reported for broncho-alveolar carcinoma, carcinoid, and in
NSCLC measuring less than 1 cm (43), and histological or cytological
conrmation is advisable prior to ECSRT regardless of the ndings on
18
FDG-PET scans. As ECSRT is usually restricted to either medically
inoperable patients with early stage lung cancer or patients refusing surgery,
mediastinoscopy is rarely performed and staging is conned to non-invasive
methods. 18FDG-PET scans are more reliable than CT scans for the detection of mediastinal or hilar lymph node metastases, with a mean sensitivity
and specicity of 79% 3% and 91% 2% for PET and 60% 2% and
77% 2%, respectively, for CT (39). The accuracy of nodal staging can
be increased even further using co-registration of CT and PET scans, or
an integrated PET-CT (4446).
In patients with early-stage NSCLC, endobronchial ultrasound
(EBUS) can be useful in excluding metastases to mediastinal and hilar nodes
(47). In addition, EBUS may also have a role in the assessment of patients
with peripheral nodules in whom no denite histological diagnosis is available (48). In such patients, EBUS permits the visualization of the internal
112
structure of peripheral pulmonary lesions, information that appears to

correlate with histology of the lesion.
The Role of Imaging in the Selection of Patients
with Hepatic Malignancies
The suitability of local treatments for liver metastases has traditionally been
based upon their number, distribution, and size using anatomical imaging
studies such as contrast-enhanced CT scans, MRI, and ultrasonography.
The sensitivity of transabdominal ultrasonography is relatively poor in comparison to intraoperative ultrasonography, in particular due to the failure to
detect smaller metastases (4951). In a series of 157 patients with liver
metastases, 247 out of a total of 290 lesions found with intraoperative ultrasound, intraoperative palpation, and histopathologic examination had been
correctly identied by helical CT, yielding an overall detection rate of 85%
(52). The high positive predictive value of 96% indicates that histological or
cytological conrmation prior to ECSRT is required only for equivocal
lesions on contrast-enhanced CT scans. 18FDG-PET scanning improves
the selection of candidates for local treatment of liver metastases from colorectal cancer by identifying extrahepatic disease sites, (53) and in detecting
unresectable disease (5457). However, anatomical imaging studies remain
superior for identifying small liver metastases, as the detection rate for
lesions of 1 cm or smaller was only 25% with FDG-PET scans, compared
to 85% for lesions larger than 1 cm (55,58). The low sensitivity for detection
of smaller liver metastases is partly a result of the high hepatic background
activity produced by the high metabolic activity of normal hepatocytes.
The Role of Imaging in the Selection of Patients
with Vertebral Metastases
The appropriateness of ECSRT for vertebral metastasis is determined by the
multiplicity of bone metastases. While determination of the number of
involved sites is usually based on conventional technetium-99m methylene
diophosphate (Tc-99m MDP) whole body bone scans, several authors have
reported a substantially higher sensitivity and a better accuracy of 18FDGPET scans in the detection of bone metastases (5962). In addition,
18
FDG-PET revealed more metastatic lesions than bone scanning, independent of the type of cancer or location of bone involvement (63). Thin-section
CT scans can be used to determine the local extension of vertebral metastases, and coronal and sagittal multiplanar reconstructions can assist in
determining the local extent of bony destruction and provide detailed information about inltration and potential spinal instability. The visualization
of the local extension of vertebral metastases, particularly with respect to
the epidural and paravertebral regions and the relation to the spinal cord,
is superior using gadolinium-enhanced multiplanar MRI scans (64).
113
GENERAL PRINCIPLES OF TARGET DEFINITION

The steep dose-gradients, characteristic of ECSRT, imply that the accurate
denition of target volumes is of utmost importance in order to avoid marginal
misses. To ensure that representative target volumes are generated for ECSRT
treatment planning, imaging techniques will have to be used that reect the
method of treatment delivery. For instance, if treatment delivery is performed
during abdominal compression or respiratory gating, planning CT scans
should also be performed under identical conditions. In addition, an imaging
procedure of a lung tumor lasting a few minutes may not provide
representative target volumes if the lesion is treated with respiration-gated
single fraction SRT which may take up to 2 hours (65,66).
The small margins used for ECSRT dictate that careful attention
must be paid to limiting variations and inconsistencies in contouring GTVs.
Interobserver variability in contouring target volumes appears to be a major
factor contributing to geometric inaccuracy (67). In addition, with respect to
the critical dose distributions in ECSRT, variations in contouring adjacent
normal tissues, which can also be substantial (68,69), should be taken into
account. Although some residual interobserver variability remains inevitable, measures to diminish such variations, including standard contouring
protocols, the use of standardized optimal window-level settings for contouring, thin slice planning CT scans, and multimodality image registration
techniques are of paramount importance (70,71). The development of reliable segmentation software tools may in the future also decrease the impact
of human error.
Although only minimal acute toxicity has been reported after ECSRT,
late toxicity may become important in patients who are treated with curative
intent. A signicant risk of late radiation-induced Grade 35 complications
has been reported after high-dose irradiation or chemoradiotherapy of the
thorax (72). While these ndings were reported after fractionated conventional radiotherapy, it indicates that doses to normal tissues should be minimized in ECSRT by using optimal techniques, and these doses should be
carefully documented. In view of the large fraction sizes used in ECSRT,
establishing the actual dose delivered to critical normal tissues is important
in order to derive normal tissue tolerance doses. The ICRU Report 62
recommends drawing margins around organs at risk to produce planning
organ at risk volumes (PRVs) to account for geometric uncertainty in the
radiotherapy treatment process (73). The use of such PRVs requires that
data on organ mobility and systematic uncertainties in patient setup are
available. However, this is generally not the case at present.
Occasionally, ECSRT is performed for residual disease after induction
chemotherapy, e.g., for metastatic disease or T2N0M0 primary NSCLC.
There are presently no recommendations available as to whether to irradiate
the pre- or postchemotherapy target volumes. Only a careful study of
114
disease recurrence patterns will show if treatment of postchemotherapy

volumes is acceptable. Whenever the choice is made to irradiate the prechemotherapy volume, the use of multimodality fusion can aid in an accurate
reconstruction of this volume (70).
GENERATING REPRESENTATIVE PLANNING CT SCANS
The challenge of generating representative target volumes for mobile organs
has most extensively been addressed for lung lesions. The following section
aims to describe potential solutions for this tumor site, although not all
solutions are suitable for other mobile tumor sites.
Individualized assessment of mobility, instead of using standard
margins, is required for ECSRT, as an analysis of the mobility of peripheral lung tumors has revealed that no clear correlation exists between
the anatomical location of lung tumors and the extent of mobility (74).
It is commonly assumed that the correct, or at least the mean, position
of a normal organ is captured by a single planning CT scan, and that this
position is reproduced throughout the treatment. However, planning CT
scans may show considerable artifacts due to internal mobility during the
generation of the scans (Fig. 1). In addition, fast spiral CT scans will generate a random snapshot position of both target volumes and normal tissues, and dose distributions calculated in such a static model are unlikely
to represent the actual dose distribution (75). A study in which an abdominal CT scan was performed in a patient who was breathing quietly and
freely, and was followed by a CT scan with breath-holding at normal inhalation and normal exhalation, showed that the absolute volume of the liver
varied by as much as 12% between studies, due to sampling error. The
Figure 1 Movement of a solitary lung nodule during CT scanning (1 second/slice)

results in discontinuous imaging of the lesion, as seen on a digitally reconstructed
radiograph (left panel) and consecutive CT slices (right panel).
115
normal tissue complication probabilities (NTCP) calculated from static

exhale and inhale studies in six patients varied randomly from 3% to 44%
for doses that resulted in a 15% NTCP on the free-breathing studies (76).
As standard imaging techniques are lacking in information on the time
factor, several approaches have been used to ensure adequate target
coverage, including:
1. a full characterization and incorporation of internal mobility,
2. restriction of respiration-induced mobility during both pretreatment imaging and treatment delivery,
3. respiration-gated planning CT scan and gated treatment delivery,
4. real-time tumor tracking.
Full Characterization and Incorporation of Internal Mobility
Characterization and incorporation of all intrafractional tumor mobility,
although time consuming, represents the simplest of measures to deal with
the problem of internal mobility. An obvious disadvantage of this approach
is that an unnecessarily larger volume will be irradiated, particularly if
extremes of mobility are incorporated. For an adequate dosimetric analysis,
the full incorporation of internal mobility should not only include the target
volume, but also adjacent normal structures in order to integrate all timeweighted changes in anatomy.
Fluoroscopy is not an appropriate method for characterizing mobility
in high precision radiotherapy, as only limited mobility data can be generated (if at all) on superiorinferior and medio-lateral movements (77,78).
More importantly, the observed mobility cannot be accurately linked to
the geometry of planning CT scans.
The fusion of target volumes generated on two-phase CT scans,
obtained at either deep or quiet inspiration and expiration, has also been used
to characterize internal mobility (7881). However, summation of target
volumes generated at deep inspiration and deep expiration will not only lead
to an overestimation of the actual target volume (82), but the reproducibility
of such target volumes is questionable (66). However, it has been argued that
two-phase planning, by removing the need for large symmetric, population-based margins, allows for a reduction in the amount of irradiated normal
tissue, and could thereby improve the reliability of patient data for DVH modeling (81). An alternative method of obtaining an individualized internal target
volume (ITV) is the summation of target volumes generated on multiple rapid
planning CT scans (Fig. 2). The exact number of CT scans needed for generating
an optimal ITV is unknown, and as the fusion of multiple snapshot positions
is unlikely to cover the full range of mobility, a small additional internal and
setup margin for generating planning target volumes (PTVs) will be required.
Another method that is restricted for use in lung tumors is the generation of slow CT scans that are performed with a CT revolution time of
116
Figure 2 Generating target volumes for a stage I lung tumor. The ITV (orange
contour on left panel) encompasses all GTVs contoured on six consecutive multi-slice
CT scans (light yellow contours on left and right panel). The ITV was expanded with
a 3 mm margin to derive the PTV (red contour on right panel). (See color insert.)
4 sec/slice (83,84). The GTVs generated from single slow CT scans are generally located in a central position relative to an optimal CTV generated using
six separate CT scans during quiet respiration. Being centrally located, GTVs
generated using slow CT scans can enclose the 6-scan volume by applying a
symmetrical 3D margin of 5 mm (85). A complete breathing cycle in patients
with lung cancer has been reported to range between 1.53.5 sec and
3.6 0.85 sec (86,87), and the need for an additional margin (of 5 mm) may
reect factors such as minor variations in mobility between respiratory cycles
and contouring variations. This 5 mm margin reects the internal margin
only; additional margins to account for patient-setup are required as well. As
symptomatic radiation pneumonitis is very uncommon in patients irradiated
for stage I NSCLC (33), the modest increase in PTV for a plan based upon a
slow CT scan with an additional margin is not clinically relevant.
Minimizing Respiratory Motion
A simple and frequently used method for restricting respiration-induced
tumor mobility is the application of abdominal pressure, both during
117
pretreatment imaging and treatment delivery, for decreasing the mobility of

both lung and liver lesions (8891).
SRT during patient breath-holding, usually at end-inspiration, has
been used as a method for minimizing the mobility of targets in the lung
and liver (92,93). Although theoretically attractive, this method has a number of practical disadvantages. Cooperative patients have to be coached
thoroughly in order to generate reproducible results, and an individualized
assessment is required for each patient. In addition, a considerable number
of medically inoperable patients with compromised respiratory function
cannot tolerate breath-holding (65,92,94). When performing breath-holding,
some residual mobility persists due to variations in breath-holding and
cardiac action, and this movement must also be incorporated. In addition,
tumor drifts have been reported during treatment, which is particularly
relevant for SRT in view of the long fractional treatment time (95).
Recently, Onishi et al. (96) described a self-breath-holding system,
which is based on the control of the radiation beam by patients themselves.
Planning CT scans during patients self-breath-holding were repeated three
times, and the tumor positions on these scans showed that the technique was
reproducible within a 2 mm distance. In addition to patient-controlled
breath-holding, several authors have reported minimizing mobility using
active breathing control (ABC) (97,98), or even using general anesthesia
and high-frequency ventilation (34).
Respiratory Gating
Respiratory gating constitutes an attractive alternative to the breath-holding
techniques, although advanced gating equipment at the CT scan and linear
accelerator is mandatory. The approach of prospective gating, i.e., performing respiration-triggered CT scans for radiotherapy planning, has
major limitations (99). Firstly, CT sessions will have to be long, as only
one slice is acquired at each table position per breathing cycle, and multiple
scans of similar duration would be required if many respiratory phases are
to be imaged. The long image acquisition time for each CT set increases the
likelihood of patient movement during the scanning process. Another drawback is the fact that the respiratory phase at which the scan is to be
performed has to be predetermined before image acquisition.
A major recent advance in CT scanning technology has been the use of
multi-slice CT scans to characterize the 4D position, which includes the time
factor, of tumors or normal organs from co-registered respiratory signals.
This requires multiple CT slices to be performed at each relevant table position (over-sampling) for at least the duration of one full respiratory cycle,
while simultaneously recording signals which are generated using a respiratory motion monitoring system (100). Each image from such a scan is sorted
into an image bin that corresponds with the phase of the respiratory cycle in
118
Figure 3 A 4D CT dataset is derived by performing multiple CT slices at each table

position (over-sampling), while simultaneously recording respiratory signals. Each
image is then sorted into a bin that corresponds with the phase of the respiratory
cycle at which the image was acquired.
which the image was acquired (retrospective gating). The complete set of
such image bins accumulated over a respiratory cycle constitutes a 4D dataset (Fig. 3). Motion artifacts are signicantly reduced in the 4D dataset compared to 3D images, and reconstruction volumes match those expected on
basis of stationary-phantom scans to within 5% in all cases (101). The initial
reports concerning the use of 4D CT scanning for radiotherapy planning
described the use of a single-slice spiral CT and mainly phantom studies
(100,101), but clinical experience in a single patient with lung cancer was
also reported (112).
Tumor Tracking Radiotherapy
One of the most advanced methods for coping with the problem of mobile
target volumes is uoroscopic real-time tumor-tracking radiation therapy
(RTRT). Prior to performing a planning CT scan, radio-opaque gold markers are inserted in or near the tumor. This is performed using bronchoscopic insertion for peripheral lung lesions, transcutaneous insertion for
liver lesions, or cystoscopic or percutaneous insertion for the prostate. The
robustness of the treatment plan is dependent upon a constant relationship
119
between the ducials and tumor isocentre. Using uoroscopic tracking, the
system triggers the linear accelerator to commence and stop irradiation only
when the marker(s) are located within a predetermined coordinate-range.
The method has a number of limitations, however. The accurate insertion of markers in tumors can be difcult or may be contraindicated.
Bronchoscopic implantation of markers in centrally located lung tumors
is often unsuccessful due to problems with early displacement (102,103),
and the insertion of markers is restricted into small peripheral bronchi in
or adjacent to the tumor (103). The difculty in inserting any marker into
lung tumors was highlighted by a report in which it was only possible to insert
markers in the proximity of tumors in ve (of seven) patients with T1 lung
tumors (87). Transthoracal insertion of markers for lung lesions is associated
with a substantial risk for pneumothorax (37), a complication that may be
life-threatening in patients with compromised pulmonary function.
Ideally, four ducial markers are required in order to accurately detect
tumor rotation and volumetric changes during treatment (104). In patients
with prostate cancers, a reduction in the distance between markers that was
consistent with tumor regression has been observed during treatment
(103,105). Preliminary data suggests that marker migration does not appear
to be a major problem in lesions of the liver and prostate (106).
REPEATED TREATMENT PLANNING IN BETWEEN FRACTIONS
Interfractional variation in location of target volumes may play an important role in hypofractionated SRT, in particular when the overall-treatment
time is longer than a few days, but may also be relevant for single fraction
treatment when the time between imaging and actual treatment delivery is
substantial. For hypofractionated SRT, the geometrical accuracy can be
improved by irradiating a target of the day, which relies on image guidance that identies the target volume position before each fraction. This
can be performed in instances when a change in size or position of the tumor
or normal tissues is anticipated during treatment, using ultrasound imaging,
radiographic imaging of implanted radio-opaque markers, or CT imaging.
The general principle of this approach is to adapt the treatment elds to
the daily position of the target as detected by each imaging procedure. While
attractive, the use of markers only accounts for variations in the spatial
position of the target volume and cannot correct for (radiation-induced)
changes in volume and shape of the target volume. Data on potential volumetric changes during the course of radiotherapy, including both radiationinduced enlargement and tumor shrinkage, are required to assess the
importance of such volumetric changes on dosimetry. Volumetric measurements and beam adjustments may become feasible with the implementation
of cone-beam CT scanners, which are linked to the linear accelerator.
In instances where volumetric changes can be expected to be present, or
120
Figure 4 Changes in ITVs seen on weekly CT scans in two patients with peripheral
lung tumors when ve fractions of stereotactic radiotherapy were delivered in
5 weeks (12 Gy/fraction). (See color insert.)
in situations where markers cannot be used, repeated imaging and treatment

planning in order to re-establish both the PTV and PRV before subsequent
fractions may be warranted.
The reproducibility of target volumes has been studied using repeated
CT scans prior to each of three fractions of SRT in 60 CT-simulations for
lung targets, and 58 CT-simulations for hepatic targets (91). This analysis
showed that had all treatments been based upon a plan derived from the
initial scan, an adequate (95% or greater coverage of the target volume by
the reference isodose) was only attained in 91% of lung tumors, and 81% of
liver tumors. The authors suggested that pulmonary targets with increased
breathing mobility and liver tumors exceeding 100 cm3 should undergo
repeated evaluation of sufcient margins
Preliminary data from our series in general showed stable ITVs during
the 35 weeks course of ECSRT; however, an occasional transient but
signicant increase in ITV after the rst fraction of ECSRT was observed
(Fig. 4). Additional investigations into the magnitude and impact of timetrends during the course of hypofractionated SRT are awaited.
SITE SPECIFIC COMMENTS
Table: Site-Specific Aspects of Imaging for ECSRT
Spiral CT-scanning (Fig. 5) should be performed during either the
late arterial or portal venous phase (scanning delay 4580 sec) for
optimal visualization of both hypovascular and hypervascular liver
metastases (107,108).
121
Figure 5 Four-phase hepatic CT scans showing a liver metastasis in the right lobe.
The hypodense region seen on the scan without contrast (upper left panel) shows distinct contrast enhancement in the late arterial phase (upper right panel) and to a lesser
degree in the portal venous phase (lower left panel). The contrast enhancement has
disappeared in the late venous phase (lower right panel).
A trans-abdominal ultrasound-based stereotactic guidance system

that accurately localizes the prostate and hepatic lesions is a rapid
and direct method to correct for patient position and organmotion error (109,110).
Magnetic resonance spectroscopy (MRS) has been used to identify
tumor-bearing regions on the basis of an elevated choline/citrate
ratio (111,112). Co-registration of this information may permit a
biological target volume to be generated. However, the spatial
resolution of MRS is modest.
122
FOLLOW-UP IMAGING
Sequential posttreatment imaging is required in order to characterize the
sites of disease recurrence and to optimize the planning margins and dosefractionation schemes used for ECSRT. No elective irradiation is delivered
using ECSRT, and some centers omit using separate margins for potential
microscopic tumor extension in stereotactic radiotherapy, which could
increase the rate of marginal and regional failures.
Volumetric measurements using images after stereotactic radiotherapy
can be misleading, as early changes may arise from treatment-induced
edema. Transient increases in volume and/or changes in patterns of contrast
enhancement have been reported after SRT for brain metastases (113,114),
pituitary adenomas (115), vestibular schwannomas (116,117), gliomas (118),
and craniopharyngiomas (119). The radiological ndings reported following
extracranial SRT range from local radiation pneumonitis to marked brosis
or atelectasis (36,120,121). Difculties in differentiating between residual
tumor and radiation-induced changes mean that the absence of progression
on serial studies may be a better denominator of local control than the
disappearance of abnormalities on imaging (122).
There is currently limited data available on the indications and timing
of PET scans in assessing local control after ECSRT. However, 18FDG-PET
has been shown to be superior to using anatomic imaging during follow-up
after radiofrequency ablation of liver tumors (123,124).
CONCLUSIONS
ECSRT is a relatively new high-precision treatment modality, and optimal
imaging studies are essential for optimizing the technique and results.
Optimal imaging plays an important role in patient selection, treatment
planning, and verication and follow-up.
REFERENCES
1. Davidson RS, Nwogu CE, Brentjens MJ, et al. The surgical management of
pulmonary metastasis: current concepts. Surg Oncol 2001; 10:3542.
2. Friedel G, Pastorino U, Ginsberg RJ, et al. Results of lung metastasectomy
from breast cancer: prognostic criteria on the basis of 467 cases of the International Registry of Lung Metastases. Eur J Cardiothorac Surg 2002;
22:335344.
3. Headrick JR, Miller DL, Nagorney DM, et al. Surgical treatment of hepatic
and pulmonary metastases from colon cancer. Ann Thorac Surg 2001; 71:
975979.
4. Porte H, Siat J, Guibert B, et al. Resection of adrenal metastases from nonsmall cell lung cancer: a multicenter study. Ann Thorac Surg 2001; 71:981985.
5. van Halteren HK, van Geel AN, Hart AA, et al. Pulmonary resection for
metastases of colorectal origin. Chest 1995; 107:15261531.
123
6. Wedman J, Balm AJ, Hart AA, et al. Value of resection of pulmonary metastases in head and neck cancer patients. Head Neck 1996; 18:311316.
7. Hoekstra CJ, Stroobants SG, Hoekstra OS, et al. Measurement of perfusion in
stage IIIA-N2 non-small cell lung cancer using H(2)(15)O and positron emission tomography. Clin Cancer Res 2002; 8:21092115.
8. Lodge MA, Carson RE, Carrasquillo JA, et al. Parametric Images of Blood
Flow in Oncology PET Studies Using [150] Water. J Nucl Med 2000;
41:17841792.
9. Bentzen L, Keiding S, Nordsmark M, et al. Tumour oxygenation assessed by
18F-uoromisonidazole PET and polarographic needle electrodes in human
soft tissue tumours. Radiother Oncol 2003; 67:339344.
10. Dehdashti F, Mintun MA, Lewis JS, et al. In vivo assessment of tumor
hypoxia in lung cancer with 60Cu-ATSM. Eur J Nucl Med Mol Imaging
2003; 30:844850.
11. Hoebers FJ, Janssen HL, Olmos AV, et al. Phase 1 study to identify tumour
hypoxia in patients with head and neck cancer using technetium-99m BRU
5921. Eur J Nucl Med Mol Imaging 2002; 29:12061211.
12. Barthel H, Cleij MC, Collingridge DR, et al. 30 -deoxy-30 -[18F]uorothymidine
as a new marker for monitoring tumor response to antiproliferative therapy in
vivo with positron emission tomography. Cancer Res 2003; 63:37913798.
13. Dittmann H, Dohmen BM, Paulsen F, et al. [(18)F]FLT-PET for diagnosis
and staging of thoracic tumours. Eur J Nucl Med Mol Imaging 2003; only
internet version yet.
14. Wells P, Gunn RN, Alison M, et al. Assessment of proliferation in vivo
using 2-[(11)C]thymidine positron emission tomography in advanced intraabdominal malignancies. Cancer Res 2002; 62:56985702.
15. Gambhir SS, Czernin J, Schwimmer J, et al. A tabulated summary of the FDG
PET literature. J Nucl Med 2001; 42(suppl 5):1S93S.
16. Bury T, Dowlati A, Paulus P, et al. Whole-body 18FDG positron emission
tomography in the staging of non-small cell lung cancer. Eur Respir J 1997;
10:25292534.
17. DCunha J, Herndon JL, Herzan DL, et al. Poor correlation between clinical
and pathological staging in stage I non-small cell lung cancer: results from
CALGB 9761, a prospective trial. Lung Cancer 2005; 48:241246.
18. Fritscher-Ravens A, Bohuslavizki KH, Brandt L, et al. Mediastinal lymph
node involvement in potentially resectable lung cancer: comparison of CT,
positron emission tomography, and endoscopic ultrasonography with and
without ne-needle aspiration. Chest 2003; 123:442451.
19. Pieterman RM, van Putten JW, Meuzelaar JJ, et al. Preoperative staging
of non-small-cell lung cancer with positron-emission tomography. N Engl
J Med 2000; 343:254261.
20. Schmid RA, Hautmann H, Poellinger B, et al. Staging of recurrent and
advanced lung cancer with 18F-FDG PET in a coincidence technique (hybrid
PET). Nucl Med Commun 2003; 24:3745.
21. Scott WJ, Gobar LS, Terry JD, et al. Mediastinal lymph node staging of nonsmall-cell lung cancer: a prospective comparison of computed tomography and
positron emission tomography. J Thorac Cardiovasc Surg 1996; 111:642648.
124
22. Steinert HC, Hauser M, Allemann F, et al. Non-small cell lung cancer: nodal
staging with FDG PET versus CT with correlative lymph node mapping and
sampling. Radiology 1997; 202:441446.
23. von Haag DW, Follette DM, Roberts PF, et al. Advantages of positron
emission tomography over computed tomography in mediastinal staging of
non-small cell lung cancer. J Surg Res 2002; 103:160164.
24. Kalff V, Hicks RJ, Ware RE, et al. The clinical impact of (18)F-FDG PET in
patients with suspected or conrmed recurrence of colorectal cancer: a prospective study. J Nucl Med 2002; 43:492499.
25. Lonneux M, Reffad AM, Detry R, et al. FDG PET improves the staging and
selection of patients with recurrent colorectal cancer. Eur J Nud Med Mol
Imaging 2002; 29:915921.
26. Valk PE, Abella-Columna E, Haseman MK, et al. Whole-body PET imaging
with [18F]uorodeoxyglucose in management of recurrent colorectal cancer.
Arch Surg 1999; 134:503511.
27. Kresnik E, Mikosch P, Gallowitsch HJ, et al. Evaluation of head and neck
cancer with 18F-FDG PET: a comparison with conventional methods. Eur J
Nucl Med 2001; 28:816821.
28. Siggelkow W, Zimny M, Faridi A, et al. The value of positron emission tomography in the follow-up for breast cancer. Anticancer Res 2003; 23:18591867.
29. Vranjesevic D, Filmont JE, Meta J, et al. Whole-body (l8)F-FDG PET and
conventional imaging for predicting outcome in previously treated breast cancer patients. J Nucl Med 2002; 43:325329.
30. Gambhir SS, Czernin J, Schwimmer J, et al. A review of the literature for
whole-body FDG PET in the management of patients with melanoma. Q J
Nucl Med 2000; 44:153167.
31. Swetter SM, Carroll LA, Johnson DL, et al. Positron emission tomography is
superior to computed tomography for metastatic detection in melanoma
patients. Ann Surg Oncol 2002; 9:646653.
32. Lagerwaard FJ, Senan S, van Meerbeeck JP, et al. Has 3D conformal radiotherapy (3D CRT) improved the local tumour control for stage I non-small
cell lung cancer? Radiother Oncol 2002; 63:151157.
33. Qiao X, Tullgren O, Lax I, et al. The role of radiotherapy in treatment of stage
I non-small cell lung cancer. Lung Cancer 2003; 41:111.
34. Hof H, Herfarth KK, Munter M, et al. Stereotactic single-dose radiotherapy
of stage I non-small-cell lung cancer (NSCLC). Int J Radiat Oncol Biol Phys
2003; 56:335341.
35. Lee S, Choi EK, Park HJ, et al. Stereotactic body frame based fractionated
radiosurgery on consecutive days for primary or metastatic tumors in the lung.
Lung Cancer 2003; 40:309315.
36. Uematsu M, Shioda A, Suda A, et al. Computed tomography-guided frameless stereotactic radiotherapy for stage I non-small cell lung cancer: a 5-year
experience. Int J Radiat Oncol Biol Phys 2001; 51:666670.
37. Whyte RI, Crownover R, Murphy MJ, et al. Stereotactic radiosurgery for
lung tumors: preliminary report of a phase I trial. Ann Thorac Surg 2003;
75:10971101.
125
38. Hoffmann H. Invasive staging of lung cancer by mediastinoscopy and videoassisted thoracoscopy. Lung Cancer 2001; 34(suppl 3):S3S5.
39. Dwamena BA, Sonnad SS, Angobaldo JO, et al. Metastases from non-small
cell lung cancer: mediastinal staging in the 1990smeta-analytic comparison
of PET and CT. Radiology 1999; 213:530536.
40. Vansteenkiste JF, Stroobants SG, De Leyn PR, et al. Lymph node staging in
non-small-cell lung cancer with FDG PET scan; a prospective study on 690
lymph node stations from 68 patients. J Clin Oncol 1998; 16:21422149.
41. Lee J, Aronchick JM, Alavi A. Accuracy of F-18 uorodeoxyglucose positron
emission tomography for the evaluation of malignancy in patients presenting with new lung abnormalities; a retrospective review. Chest 2001;
120:17911797.
42. Pitman AG, Hicks RJ, Binns DS, et al. Performance of sodium iodide based
(18)F-uorodeoxyglucose positron emission tomography in the characterization of indeterminate pulmonary nodules or masses. Br J Radiol 2002;
75:114121.
43. Marom EM, Sarvis S, Herndon JE 2nd, et al. Tl lung cancers: sensitivity of
diagnosis with uorodeoxyglucose PET. Radiology 2002; 223:453459.
44. Aquino SL, Asmuth JC, Alpert NM, et al. Improved radiologic staging of lung
cancer with 2-(18F)-uoro-2-deoxy-D-glucose-positron emission tomography
and computed tomography registration. J Comput Assist Tomogr 2003;
27:479484.
45. Lardinois D, Weder W, Hany TF, et al. Staging of non-small-cell lung cancer
with integrated positron-emission tomography and computed tomography.
N Engl J Med 2003; 348:25002507.
46. Poncelet AJ, Lonneux M, Coche E, et al. PET-FDG scan enhances but does
not replace preoperative surgical staging in non-small cell lung carcinoma.
Eur J Cardiothorac Surg 2001; 20:468474.
47. Okamoto H, Watanabe K, Nagatomo A, et al. Endobronchial ultrasonography for mediastinal and hilar lymph node metastases of lung cancer. Chest
2002; 121:14981506.
48. Kurimoto N, Murayama M, Yoshioka S, et al. Analysis of the internal structure of peripheral pulmonary lesions using endobronchial ultrasonography.
Chest 2002; 122:18871894.
49. el Mouaaouy A, Naruhn M, Becker HD. Diagnosis of liver metastases from
malignant gastrointestinal neoplasms; results of pre- and intraoperative ultrasound examinations. Surg Endosc 1991; 5:209213.
50. Machi J, Isomoto H, Kurohiji T, et al. Accuracy of intraoperativee ultrasonography in diagnosing liver metastasis from colorectal cancer: evaluation with
postoperative follow-up results. World J Surg 1991; 15:551556.
51. Rafaelsen SR, Kronborg O, Larsen C, et al. Intraoperative ultrasonography in
detection of hepatic metastases from colorectal cancer. Dis Colon Rectum
1995; 38:355360.
52. Valls C, Andia E, Sanchez A, et al. Hepatic metastases from colorectal
cancer: preoperative detection and assessment of resectability with helical
CT. Radiology 2001; 218:5560.
126
53. Strasberg SM, Dehdashti F, Siegel BA, et al. Survival of patients evaluated by
FDG PET before hepatic resection for metastatic colorectal carcinoma: a prospective database study. Ann Surg 2001; 233:293299.
54. Desai DC, Zervos EE, Arnold MW, et al. Positron emission tomography
affects surgical management in recurrent colorectal cancer patients. Ann Surg
Oncol 2003; 10:5964.
55. Fong Y, Saldinger PF, Akhurst T, et al. Utility of 18F-FDG positron emission
tomography scanning on selection of patients for resection of hepatic colorectal metastases. Am J Surg 1999; 178:282287.
56. Rydzewski B, Dehdashti F, Gordon BA, et al. Usefulness of intraoperative
sonography for revealing hepatic metastases from colorectal cancer in patients
selected for surgery after undergoing FDG-PET. Am J Roentgenol 2002;
178:353358.
57. Topal B, Flamen P, Aerts R, et al. Clinical value of whole-body emission
tomography in potentially curable colorectal liver metastases. Eur J Surg
Oncol 2001; 27:175179.
58. Rohren EM, Paulson EK, Hagge R, et al. The role of F-18 FDG positron
emission tomography in preoperative assessment of the liver in patients being
considered for curative resection of hepatic metastases from colorectal cancer.
Clin Nucl Med 2002; 27:550555.
59. Hsia TC, Shen YY, Yen RF, et al. Comparing whole body 18F-2deoxyglucose positron emission tomography and technetium-99m methylene
diophosphate bone scan to detect bone metastases in patients with non-small
cell lung cancer. Neoplasma 2002; 49:267271.
60. Ohta M, Tokuda Y, Suzuki Y, et al. Whole body PET for the evaluation of
bony metastases in patients with breast cancer: comparison with 99TcmMDP bone scintigraphy. Nucl Med Commun 2001; 22:875879.
61. Wu HC, Yen RF, Shen YY, et al. Comparing whole body 18F-2-deoxyglucose
positron emission tomography and technetium-99m methylene diphosphate
bone scan to detect bone metastases in patients with renal cell carcinomasa
preliminary report. J Cancer Res Clin Oncol 2002; 128:503506.
62. Yang SN, Liang JA, Lin FJ, et al. Comparing whole body (18)F-2deoxyglucose positron emission tomography and technetium-99m methylene
diphosphonate bone scan to detect bone metastases in patients with breast
cancer. J Cancer Res Clin Oncol 2002; 128:325328.
63. Nakamoto Y, Osman M, Wahl RL. Prevalence and patterns of bone metastases detected with positron emission tomography using F-18 FDG. Clin Nucl
Med 2003; 28:302307.
64. Godersky JC, Smoker WR, Knutzon R. Use of magnetic resonance imaging in
the evaluation of metastatic spinal disease. Neurosurgery 1987; 21:676680.
65. Hara R, Itami J, Kondo T, et al. Stereotactic single high dose irradiation of
lung tumors under respiratory gating. Radiother Oncol 2002; 63:l59163.
66. Ozhasoglu C, Murphy MJ. Issues in respiratory motion compensation during
external-beam radiotherapy. Int J Radiat Oncol Biol Phys 2002; 52:13891399.
67. Weiss E, Hess CF. The impact of gross tumor volume (GTV) and clinical target volume (CTV) denition on the total accuracy in radiotherapy theoretical
aspects and practical experiences. Strahlenther Onkol 2003; 179:2130.
127
68. Collier DC, Burnett SS, Amin M, et al. Assessment of consistency in contouring of normal-tissue anatomic structures. J Appl Clin Med Phys 2003; 4:1724.
69. Fiorino C, Vavassori V, Sanguineti G, et al. Rectum contouring variability in
patients treated for prostate cancer: impact on rectum dosevolume histograms and normal tissue complication probability. Radiother Oncol 2002;
63:249255.
70. Lagerwaard FJ, van de Vaart PJ, Voet PW, et al. Can errors in reconstructing
pre-chemotherapy target volumes contribute to the inferiority of sequential
chemoradiation in stage III non-small cell lung cancer (NSCLC)? Lung
Cancer 2002; 38:297301.
71. Senan S, van Sornsen de Koste J, Samson M, et al. Evaluation of a target contouring protocol for 3D conformal radiotherapy in non-small cell lung cancer.
Radiother Oncol 1999; 53:247255.
72. Anscher MS, Marks LB, Shafman TD, et al. Risk of long-term complications
after TFG-betal-guided very-high-dose thoracic radiotherapy. Int J Radiat
73. International Commission on Radiation Units and Measurements. ICRU
Report 62: Prescribing, Recording and Reporting Photon Beam Therapy
(suppl to ICRU Report 50), Bethesda MD, 1999.
74. van Sornsen de Koste JR, Lagerwaard FJ, Nijssen-Visser MR, et al. Tumor
location cannot predict the mobility of lung tumors: a 3D analysis of data generated from multiple CT scans. Int J Radiat Oncol Biol Phys 2003; 56:348354.
75. Booth JT, Zavgorodni SF. Modelling the dosimetric consequences of organ
motion at CT imaging on radiotherapy treatment planning. Phys Med Biol
2001; 46:13691377.
76. Balter JM, Ten Haken RK, Lawrence TS et al. Uncertainties in CT-based
radiation therapy treatment planning associated with patient breathing. Int
77. Halperin R, Pobinson D, Murray B. Fluoroscopy for assessment of physiologic movement of lung tumors, a pitfall of clinical practice? [abstr]. Radiother
Oncol 2002; 65:Sl.
78. Stevens CW, Munden RF, Forster KM, et al. Respiratory-driven lung tumor
motion is independent of tumor size, tumor location, and pulmonary function.
79. Aruga T, Itami J, Aruga M, et al. Target volume denition for upper abdominal irradiation using CT scans obtained during inhale and exhale phases. Int
80. Onimaru R, Shirato H, Shimizu S, et al. Tolerance of organs at risk in smallvolume, hypofractionated, image-guided radiotherapy for primary and metastatic lung cancers. Int J Radiat Oncol Biol Phys 2003; 56:126135.
81. Yamada K, Soejima T, Yoden E, et al. Improvement of three-dimensional
treatment planning models of small lung targets using high-speed multi-slice
computed tomographic imaging. Int J Radiat Oncol Biol Phys 2002;
54:12101216.
82. Senan S, Lagerwaard FJ, Nijssen-Visser MR. Incorporating lung tumor
mobility in radiotherapy planning. Int J Radiat Oncol Biol Phys 2002; 52:
11421143.
128
83. Lagerwaard FJ, Van Sornsen de Koste JR, Nijssen-Visser MR, et al. Multiple
slow CT scans for incorporating lung tumor mobility in radiotherapy planning. Int J Radiat Oncol Biol Phys 2001; 51:932937.
84. van Sornsen de Koste JR, Lagerwaard FJ, Schuchhard-Schipper RH, et al.
Dosimetric consequences of tumor mobility in radiotherapy of stage I nonsmall cell lung canceran analysis of data generated using slow CT scans.
85. van Sornsen de Koste JR, Lagerwaard FJ, de Boer HC, et al. Are multiple CT
scans required for planning curative radiotherapy in lung tumors of the lower
lobe? Int J Radiat Oncol Biol Phys 2003; 55:13941399.
86. Chen QS, Weinhous MS, Deibel FC, et al. Fluoroscopic study of tumor
motion due to breathing: facilitating precise radiation therapy for lung cancer
patients. Med Phys 2001; 28:18501856.
87. Seppenwoolde Y, Shirato H, Kitamura K, et al. Precise and real-time measurement of 3D tumor motion in lung due to breathing and heartbeat, measured
during radiotherapy. Int J Radiat Oncol Biol Phys 2002; 53:822834.
88. Herfarth KK, Debus J, Lohr F, et al. Extracranial stereotactic radiation
therapy: set-up accuracy of patients treated for liver metastases. Int J Radiat
89. Lax I, Blomgren H, Naslund I, et al. Stereotactic radiotherapy of malignancies
in the abdomen. Methodological aspects. Acta Oncol 1994; 33:677683.
90. Negoro Y, Nagata Y, Aoki T, et al. The effectiveness of an immobilization
device in conformal radiotherapy for lung tumor: reduction of respiratory
tumor movement and evaluation of the daily setup accuracy. Int J Radiat
91. Wulf J, Hadinger U, Oppitz U, et al. Stereotactic radiotherapy of extracranial
targets: CT-simulation and accuracy of treatment in the stereotactic body
frame. Radiother Oncol 2000; 57:225236.
92. Murphy MJ, Martin D, Whyte R, et al. The effectiveness of breath-holding to
stabilize lung and pancreas tumors during radiosurgery. Int J Radiat Oncol
Biol Phys 2002; 53:475482.
93. ODell WG, Schell MC, Reynolds D, et al. Dose broadening due to target
position variability during fractionated breath-held radiation therapy. Med
Phys 2002; 29:14301437.
94. Barnes EA, Murray BR, Robinson DM, et al. Dosimetric evaluation of lung
tumor immobilization using breath hold at deep inspiration. Int J Radiat
Oncol Biol Phys 2001; 50:10911098.
95. Murphy MJ, Chang SD, Gibbs IC, et al. Patterns of patient movement during
frameless image-guided radiosurgery. Int J Radiat Oncol Biol Phys 2003;
55:14001408.
96. Onishi H, Kuriyama K, Komiyama T, et al. A new irradiation system for lung
cancer combining linear accelerator, computed tomography, patient selfbreath-holding, and patient-directed beam-control without respiratory monitoring devices. Int J Radiat Oncol Biol Phys 2003; 56:1420.
97. Balter JM, Brock KK, Litzenberg DW, et al. Daily targeting of intrahepatic
tumors for radiotherapy. Int J Radiat Oncol Biol Phys 2002; 52:266271.
129
98. Wong JW, Sharpe MB, Jaffray DA, et al. The use of active breathing control
(ABC) to reduce margin for breathing motion. Int J Radiat Oncol Biol Phys
1999; 44:911919.
99. Ford EC, Mageras GS, Yorke E, et al. Evaluation of respiratory movement
during gated radiotherapy using lm and electronic portal imaging. Int J
Radiat Oncol Biol Phys 2002; 52:522531.
100. Vedam SS, Keall RJ, Kini VR, et al. Acquiring a four-dimensional computed
tomography dataset using an external respiratory signal. Phys Med Biol 2003;
48:4562.
101. Ford EC, Mageras GS, Yorke E, et al. Respiration-correlated spiral CT:
a method of measuring respiratory-induced anatomic motion for radiation
treatment planning. Med Phys 2003; 30:8897.
102. Harada T, Shirato H, Ogura S, et al. Real-time tumor-tracking radiation
therapy for lung carcinoma by the aid of insertion of a gold marker using
bronchoberscopy. Cancer 2002; 95:17201727.
103. Shirato H, Harada T, Harabayashi T, et al. Feasibility of insertion/implantation
of 2.0-mm-diameter gold internal ducial markers for precise setup and real-time
tumor tracking in radiotherapy. Int J Radiat Oncol Biol Phys 2003; 56:240247.
104. Murphy MJ. Fiducial-based targeting accuracy for external-beam radiotherapy. Med Phys 2002; 29:334344.
105. Pouliot J, Aubin M, Langen KM, et al. (Non)-migration of radiopaque
markers used for on-line localization of the prostate with an electronic portal
imaging device. Int J Radiat Oncol Biol Phys 2003; 56:862866.
106. Kitamura K, Shirato H, Seppenwoolde Y, et al. Three-dimensional intrafractional movement of prostate measured during real-time tumor-tracking radiotherapy in supine and prone treatment positions. Int J Radiat Oncol Biol Phys
2002; 53:11171123.
107. Francis IR, Cohan RH, McNulty NJ, et al. Multidetector CT of the liver and
hepatic neoplasms: effect of multiphasic imaging on tumor conspicuity and
vascular enhancement. AJR 2003; 180:12171224.
108. Gualdi GF, Casciani E, DAgostino A, et al. Triphasic spiral computerized
tomography of the liver: vascular models of non-cystic focal lesions. Radiol
Med 1998; 96:344352.
109. Lattanzi J, McNeeley S, Donnelly S, et al. Ultrasound-based stereotactic
guidance in prostate cancerquantication of organ motion and set-up errors
in external beam radiation therapy. Comput Aided Surg 2000; 5:289295.
110. Trichter F, Ennis RD. Prostate localization using transabdominal ultrasound
imaging. Int J Radiat Oncol Biol Phys 2003; 56:12251233.
111. Mizowaki T, Cohen GN, Fung AY, et al. Towards integrating functional imaging in the treatment of prostate cancer with radiation: the registration of the
MR spectroscopy imaging to ultrasound/CT images and its implementation in
treatment planning. Int J Radiat Oncol Biol Phys 2002; 54:15581564.
112. Zaider M, Zelefsky MJ, Lee EK, et al. Treatment planning for prostate
implants using magnetic-resonance spectroscopy imaging. Int J Radiat Oncol
Biol Phys 2000; 47:10851096.
130
113. Huber PE, Hawighorst H, Fuss M, et al. Transient enlargement of contrast

uptake on MRI after linear accelerator (linac) stereotactic radiosurgery for
brain metastases. Int J Radiat Oncoi Biol Phys 2001; 49:13391349.
114. Peterson AM, Meltzer CC, Evanson EJ, et al. MR imaging response of brain
metastases after gamma knife stereotactic radiosurgery. Radiology 1999;
211:807814.
115. Tung GA, Noren G, Rogg JM, et al. MR imaging of pituitary adenomas
after gamma knife stereotactic radiosurgery. AJR Am J Roentgenol 2001; 177:
919924.
116. Nakamura H, Jokura H, Takahashi K, et al. Serial follow-up MR imaging
after gamma knife radiosurgery for vestibular schwannoma. Am J Neuro
radiol 2000; 21:15401546.
117. Szeifert GT, Massager N, DeVriendt D, et al. Observations of intracranial
neoplasms treated with gamma knife radiosurgery. J Neurosurg 2002;
97(suppl 5):623626.
118. Bakardjiev AI, Barnes PD, Goumnerova LC, et al. Magnetic resonance imaging changes after stereotactic radiation therapy for childhood low grade
astrocytoma. Cancer 1996; 78:864873.
119. Chung WY, Pan DH, Shiau CY, et al. Gamma knife radiosurgery for craniopharyngiomas. J Neurosurg 2000; 93(suppl 3):4756.
120. Choi EK, Lee S-W, Ahn SD, et al. Stereotactic body frame based fractionated
radiosurgery on consecutive days for intrathoracic tumors: preliminary results
of a phase I/II study. Lung Cancer 2003; 41(suppl 2):S10.
121. Fuss M, Salter BJ, Selva M, et al. Intensity-modulated hypofractionated
radioablation for lung lesions. Lung Cancer 2003; 41(suppl 2):S10S11.
122. Green MR, Ginsberg R, Ardizzoni A, et al. Induction therapy for stage III
NSCLC: a consensus report. Lung Cancer 1994; 11(suppl 3):910.
123. Anderson GS, Brinkmann F, Soulen MC, et al. FDG positron emission
tomography in the surveillance of hepatic tumors treated with radiofrequency
ablation. Clin NucI Med 2003; 28:192197.
124. Langenhoff BS, Oyen WJ, Jager GJ, et al. Efcacy of uorine-18-deoxyglucose
positron emission tomography in detecting tumor recurrence after local ablative therapy for liver metastases: a prospective study. J Clin Oncol 2002;
20:44534458.
7
Radiobiological Considerations of
Stereotactic Body Radiotherapy
Steve P. Lee and H. Rodney Withers
Department of Radiation Oncology, UCLA Medical Center,
Jack F. Fowler
Department of Human Oncology, University Hospital,
Madison, Wisconsin, U.S.A.
INTRODUCTION: CLINICAL RADIOBIOLOGY AS

A FOUR-DIMENSIONAL PROBLEM
Cancer radiotherapy works via biological mechanisms rather than by
physically eradicating tumors as practiced by surgeons. That is, radiation
therapy is bona de biological therapymore akin to the craft practiced
by medical oncologists using systemic chemotherapy. Furthermore, radiation therapy is done with precise quantication of dosage deposited at the
patients anatomic site of interest. This is feasible because ionizing radiation
particles are several orders of magnitude smaller than their subcellular
biological targets, so that the probability of such interaction is rare enough
to permit a straightforward, albeit simplied, biophysical interpretation.
Equipped with a consistent way of measuring the dose, one can analyze
the effect of radiation upon living cells after quantifying the resulting biological changes, of which the end-point is often dened operationally as clonogenic cell survival. Various quantitative models have been proposed to
explain the observed survival probability as a function of dose. Among
131
132
Lee et al.
these models, the linear-quadratic (LQ) theory has gained wide popularity
over the past few decades (1). In addition, a versatile theoretical framework
based on the LQ theory has been developed to correlate biological effects of
radiation treatments using a variable range of dose rates or fractionation
schemes (2,3). Perhaps it sufces to say that the problem of analyzing radiobiology along the time domain, as specied by any unconventional fractionation protocol, has been solved in its most general form. What is left is
the problem in the space domain, otherwise known as the volume effect,
elicited due to heterogeneous dose distribution in three-dimensional space.
In real life, the complexity arising from combined temporalspatial variation of radiation dose could signicantly dictate the practical application
of radiation biology in the clinical setting.
The organization of this chapter is structured to follow a thread of
synthesis of central ideas rather than a verbatim literature survey. We will
rst review the fundamental principles in clinical radiation biology along
the temporal domain. Specically, utilization of the LQ theory to quantify
biological effects will be discussed. This will lead to the topic of the so-called
double-trouble effect for which clinical application would call for biological correction despite precise physical dosimetry in three-dimensional
space, representing an inevitable problem encountered in the spatial
domain. Then, when ultra precision oriented therapy such as stereotactic
irradiation (whether single-dose or fractionated) or intensity-modulated
radiation therapy (IMRT) is introduced, heterogeneous dose distribution
often surfaces to exacerbate the problem of the volume effect and must be
considered. This is currently still at the forefront of clinical radiobiological
research, but some relevant practical issues will be discussed to hopefully
help readers gain insight into what they might face in clinical practice using
high-tech driven treatment technologies. Interested readers outside the
radiation oncology specialty are urged to consult teaching texts such as Hall
(4) or Withers and Peters (5) for a detailed review of basic radiobiology principles. For on-going research topics in quantitative radiation biology, there
are excellent collections of articles in Seminars in Radiation Oncology (issues
9:1, 11:3, 12:3, and 14:1) and other major radiation oncology or medical
physics journals.
The fact that clinically relevant radiobiology issues can be visualized as
a spacetime problem should be distinguished from what medical physicists
currently describe as four-dimensional radiation therapy. The former will
be dealt with in detail in this chapter, while the latter pertains to dosimetric
problems arising from motion uncertainties due to intra-fractional as well
as inter-fractional movements of radiotherapy targets and normal tissues.
For extracranial stereotactic radiation treatments, all these issues are of
relevance and should be considered in order to devise rational treatment
strategy for each patient.
Radiobiological Considerations
133
RADIOBIOLOGY PROBLEM ALONG THE TEMPORAL DOMAIN:

TIME, DOSE, AND FRACTIONATION
A clinical phenomenon early radiotherapists observed was that the best
therapeutic result can be realized by fractionation rather than single-dose
treatment (6). However, the biological effect of radiation does not appear to
be linearly additive when a course of treatment is split into multiple fractions
of smaller doses, each given separately in time. That is, some biological processes seem to occur during the time interval in-between the fractions, such
that the nal biological effect of these fractions is not equivalent to that of a
treatment using the total sum of the doses in one single fraction. We might view
this phenomenonthat biological effect of radiation depends on dose fractionation in timeas the problem of radiobiology along the temporal domain.
The 4 Rs of Fractionation Radiobiology
Over the past several decades, we have gained more insights into the biological processes occurring in-between the treatment fractions. They have been
summarized by Withers (7) conveniently as the 4 Rs of fractionation radiobiology: reoxygenation, repopulation, repair, and redistribution.
Reoxygenation
The damage of tissues by radiation depends largely on the formation of
hydroxyl radicals, which in turn depends on the availability of oxygen molecules in close proximity. Fractionation allows oxygen to diffuse into the
usually hypoxic center of an expanding tumor during the interval between
fractions, and thus enables more tumor cell killing during the subsequent
treatment. This mechanism, rst described by Thomlinson and Gray (8),
essentially equates fractionation as an effective hypoxic tumor radiation sensitizer (like some chemicals designed for the similar purpose).
Repopulation
All living cells have the potential to repopulate in number by mitotic growth
of clonogens. During the time of a radiation treatment course, both normal
tissue progenitor cells and malignant cells may repopnlate, and the outcome
of such competing processes may inuence the therapeutic efcacy signicantly. Moreover, Withers et al. have described a phenomenon of accelerated repopulation for irradiated cells, which is stimulated by such
therapeutic intervention (9). This identies the overall treatment time as
an important clinical variable affecting the chances of tumor control and
acute normal tissue reaction. As malignant cells quickly repopulate, protraction of overall treatment time is therapeutically disadvantageous when
134
Lee et al.
the dose-limiting normal tissue is not rapidly repopulating. In order to

combat this potential bottleneck for tumor control, clinicians sometimes
employ the strategy of accelerated fractionation, i.e., delivering a conventional level of total dose but shortening the overall treatment time with more
intensely fractionated patterns (10).
Repair
Cells can be equipped with repair machinery to reverse partial damage
caused by a small fraction of radiation dose. They would die if such damage
fails to be repaired sufciently and is exacerbated by further radiation insults.
Elkind and Sutton in 1959 described one possible contributing mechanism
called sublethal damage (SLD) repair (11). By splitting a single dose into
two fractions, and observing the amount of dose required for the second fraction to result in the same level of nal cell survival as a function of the interfractional time interval, they demonstrated the capability of cells to repair
such radiation injury. Furthermore, this repair process proceeds roughly
exponentially in time with a typical half-time of about 12 hr for clinically
relevant tissues. A corollary is that if the dose per fraction is decreased and
the interfractional time interval is long enough to allow for complete SLD
repair, the total dose required to achieve a xed level of cell death would
be higher. Fractionation thus spares cells from radiation damage in comparison with single-dose irradiation. If normal cells repair better than tumor
cells, there would be therapeutic gain by fractionation, and vice versa. Other
types of radiation damage repair mechanisms have been proposed (12).
Redistribution
Cells exhibit differential sensitivities towards radiation at different phases of
the cell cycle. Most mammalian cells are more sensitive at the junction
between G2 and M phases. After an initial fraction of dose, the cells at a
more resistant phase (e.g., late S) may survive but then proceed in time eventually to the sensitive phases, thus subjecting themselves to more efcient
killing during the next fraction. One can thus appreciate why fast cycling
cells like skin or mucosal cells are more prone to radiation killing than slow
or dormant ones such as muscle or skeletal cells, a phenomenon realized by
investigators in the rst decade of radiation oncology history and enshrined
as the Law of Bergonie and Tribondeau (13).
Empirical PowerLaws
Once early investigators realized the signicance of radiobiological effects
due to fractionation, some sort of quantitative guideline was clearly desirable.
Initially, however, clinicians have for a long while relied mainly on empirical
approaches. The rst was published in 1944 by Strandqvist (14). In attempting to theorize a quantitative relationship between the total radiation dose
135
needed to achieve a certain clinical effect (e.g., skin reaction) and the extent of
fractionation (in the early days often paraphrased as protraction in time
resulting from the often ultra-low intensity of the radiation sources), Strandqvist borrowed rather empirically the popular Schwarzschilds Law of Photochemistry (15): Given that the intensity of light (i.e., dose per time), I,
should be inversely proportional to the time of exposure, T, for a certain
degree of photochemical exposure (i.e., isoeffect), one might express the following simple isoeffect relationship:
I Tp C
where C is a constant characterizing the desired effect, and the exponent p is a

parameter characterizing such a powerlaw relationship (0 p 1).
Because the dose, D, is equivalent to I T, then:
D I T C T 1p
2
One can readily see that such an equation predicts a linear curve on a
log D versus log T displaya for the same effect C, with the slope of the line
dictated by the parameter p.
Although Strandqvists approach has been criticized mainly due to the
lack of reliable clinical data needed to construct his loglog graph, the methodology was nevertheless a useful endeavor many found easy to emulate. The
empirical nature of such an approach necessitated constant revisions, depending on the particular clinical scenario or biological tissue from which useful
data were extracted to t the curves. Various quantitative schemes were
invented over several decades to interpret complex clinical observations, yet
the approaches largely followed the original framework of Strandqvist, i.e.,
utilizing phenomenologically oriented entities with empirical assumption of
the powerlaws. First, after laboratory and clinical investigations showed
the importance of the number of fractions, Ellis added a fractionation factor
to the powerlaw and created the nominal standard dose (NSD) (16). This
concept soon encountered a problem: the entity, NSD, was not linearly
additive. That is, there was no easy way to sum the effects of several partial
treatments to predict a net effect. To account for the effect at each subtolerance dose level, Kirk et al. extended the concept of NSD and devised the term
cumulative radiation effect (CRE) (17). Orton and Ellis then developed, after
some cumbersome algebra, a linearly additive quantity called the time dose
fractionation (TDF) factor (18). Despite the mathematical obscurity, clinicians have generally welcomed the use of TDF in treatment planning, since
it and its powerlaw predecessors represented the only quantitative guidelines
available then, and because look-up tables or graphs were readily available.
a
Unless specied otherwise, all logarithmic expressions used in this work (including formulas
and gures) pertain to the natural logarithm, for which both abbreviations, log and ln,
are used interchangeably.
136
Lee et al.
For example, tissue-specic entities termed Neuret (19) and Optic-neuret

(20) were proposed later for the treatment of central nervous system tumors.
As more is learned about the underlying mechanisms of radiation
effect on living cells and tissues, the phenomenological guidelines based
on powerlaws have slowly been replaced by more mechanistically-sound
biological principles. More importantly, as wider variability is incorporated
into the worldwide practice of fractionation radiotherapy, it has become
quite difcult to gather universally applicable empirical guidelines based
solely on ones preferred treatment regimens.
The powerlaws have been given yet another chance for resurrection
when contemporary researchers use the similar phenomenologic approach
to model the biology of heterogeneous dose distribution in space, i.e., radiobiology problem in the spatial domain, to be described later in the section on
Equivalent Uniform Dose (EUD).
Mechanistic Models and Cell Survival Curves
As implied earlier, a salient feature of radiobiology, in contrast to the pharmacologic principle behind chemotherapy, is the relative ease of quantifying
biological effect as a function of radiation dose. This results primarily from
the feasibility of measuring the input (radiation dose) and output (cell survival fraction) of a biological process fairly accurately, with the intervening
mechanism interpreted readily using biophysical models. The basic assumption is that there exist within each cell critical targets, which when hit by
ionizing radiation particles in a random fashion may lead to consequential
loss of cellular reproductive integrity. This is the essence of the so-called
target-cell hypothesis or hit theory (21).
Desauer was perhaps the rst to conceive a mechanistic interpretation
of radiation action, suggesting in 1923 a point-heat theory that describes
radiation effects as resulting from electron impact on protein molecules with
microscopic energy absorption and local temperature elevation (22). The
discovery of the nature of the generic material three decades later then led
to the identication of the most probable targets as chromosomes or DNA
molecules. Meanwhile, many experimentally obtained in vitro and in vivo cell
survival curves were published, the rst ever mammalian survival curve being
reported by Puck and Marcus (23). When plotted as a semi-log graph of log
SF (survival fraction) versus radiation dose D, most revealed a very similar
shape with a curvy shoulder at the low-dose region in contiguity with a
relatively linear tail towards the high-dose region (Fig. 1A). Evidently, at least
two biophysical mechanisms seemed to be operating simultaneously to produce such a result. Various biophysical models have been proposed since,
and a clear champion is the LQ model (1). Before we embark on its detailed
description, it is perhaps pedagogically sound to introduce another less popular model: the single-hit, multitarget killing model.
137
Figure 1 (A) Typical radiation cell survival curve as logarithm of survival fraction (SF) versus dose in linear scale.
It appears to have a curvy shoulder
and a linear tail, thus reecting probably
two independent biophysical processes.
(B) Single-hit, multitarget model. The
extrapolation number, n, corresponds
to the intercept on the vertical axis of a
straight line back-extrapolated from the
linear tail portion of the survival curve.
(C) Linear-quadratic model. In contrast
to (B), the tail portion is not linear, but
continuously bending as governed by
the quadratic term.
The SingleHit, Multitarget Killing Model

This model adequately describes the shape of a typical radiation survival curve
with the assumption of two independent components of cell killing (5,21). The
rst, or single-hit, component comes from a straightforward theorization
based on a Poisson processb of radiation action on a critical target within each
cell, which after being hit by the ionizing particle alone will cause irreversible
cell death. A conceivable example might be double-strand break of a DNA
molecule. Interested readers may refer to the raindrops in the bucket
b
This is a ubiquitous stochastic process used by physical scientists to model many natural or
articial phenomena, each with a probability of occurrence that is fundamentally rare.
138
Lee et al.
analogy (5), and appreciate the expression for survival fraction due to this
process (SF1) as
SF1 eD=1 D0
where D is the radiation dose, and 1D0 is the dose at which the survival fraction
becomes e1 (37%) of its original value (i.e., a parameter dening the intrinsic
radiation sensitivity) which translates on the average to one lethal hit per target.
On a survival curve presented as a semi-log plot of log SF versus D, the negative
reciprocal of 1 D0 yields the slope of the expected straight line (Fig. 1B).
The second independent mode of cell killing by radiation is thought to
result from cell death due to cumulative (though not necessarily synchronous) injuries elicited at all of several, say, n, multiple intracellular targets.
A good example might be repairable single-strand breaks of DNA which
would require repeated similar damages to consolidate into a permanent
chromosome aberration; in such case, n 2. By repeated interpretations
using Poisson process for each of the n target-inactivations, one can derive
the following expression for the survival fraction (SFn):

n
D=n D0
SFn 1 1 e
where n D0 is another biological parameter characterizing the effectiveness of
the radiation particle to inactivate these targets. Thus, the combined effect
of radiation in this two-component model appears to be rather complicated
mathematically:

n
SF eD=1 D0 1 1 eD=n D0
3
The semi-log plot based on the above equation shows a survival curve
with an initially convex shoulder at low dose range, but eventually tends
to nearly a straight line as D increases (Fig. 1B).
The Linear-Quadratic (LQ) Model
Based on this model, the survival fraction after a single treatment of radiation dose D can be characterized by the following equation (1,2,24):
SF eaDbD
where a and b are tissue-specic parameters governing intrinsic radiation

sensitivity. The survival curve takes on a continuously bending downward
trend rather than becoming nearly a straight line at high dose range (Fig. 1C).
The ratio, a/b, which has a dimension of dose (Gy), can be found algebraically
to correspond to the dose at which the linear (a-) and the quadratic (b-) components contribute equally to cell killing. The linear component might be seen
139
as result from the single-hit mechanism described above. While the mechanistic
origin of the quadratic component has been somewhat controversial (25,26)
some may argue that the quadratic term is the consequence of xation between
two otherwise repairable DNA single strand breaks or damaged chromatids into
permanent chromosome aberrationthe LQ model is nevertheless useful
because of its simplicity and the fact that it does describe the shape of the survival
curves adequately (at least from a/b values of about 110 Gy).
One of the most attractive roles of the LQ model stems from its ability
to explain the differential sensitivities of the so-called acute- versus lateresponding tissues to fractionated radiotherapy (Fig. 2A), thus establishing
the theoretical rationale of fractionation when treating acutely-responding
malignant tumors embedded within late-responding normal tissues (27).
Figure 2 (A) Distinct shapes of survival

curves between acutely-responding and
late-responding tissues. The former is
characterized by larger a/b ratio and
dominated by the linear component, the
latter by a smaller a/b and dominated
by the quadratic component. (B) Fractionation effect and survival curve. Here the
effects of three fractionation schemes are
shown, with respective dose per fraction
of d0, 2d0, and 3d0. Upon fractionation,
the initial portion of the single-dose curve
is repeated successively, provided that
near-completion of sublethal damage
repair occurs in-between the fractions. Clearly, the smaller the dose per fraction, the
less cell lethality, thus more sparing of the tissue. (C) Differential sparing effects of
fractionation upon acute- vs. late-responding tissues. Fractionation spares the lateresponding tissue more because of the curvier shape of its survival curve (Ls cf.
LF), while the acutely-responding tissue is spared relatively less (As cf. AF).
140
Lee et al.
Indeed, for a fractionated radiotherapy course with n fractions of dose per

fraction, d, the overall effect is given by
2
SF enadbd :
On the survival curve plot, this amounts to repeating by n times the

initial fractional amount of cell survival resulting from dose d, each picking
up successively where it ends from the previous treatment after complete
repair of SLD during the interfractional interval (Fig. 2B). The net result
is the sparing of the irradiated tissues, with an effective slope of an essentially linear overall survival curve becoming less steep the more fractionated
the treatment is. Furthermore, it is seen that by fractionation the lateresponding tissue is spared relatively more than the acutely-responding
tissue due to its curvier shape (signifying higher capacity for SLD repair)
of its single-dose survival curve in the small-dose range of clinical interest
(Fig. 2C). The late-responding tissues are hence much more sensitive to
the variation in fractionation size than the acutely-responding tissues (24,27).
Equation (5) results from the assumption that SLD events are repaired
sufciently in-between fractions. With a typical half-life of such repair measured
in hours, the condition is usually satised for a daily (24-hour) inter-fractional
interval. A useful corollary can be derived from the consideration of incompletely
repaired events when the interfractional interval is shortened to a signicant
degree or during continuous low-dose rate (LDR) irradiation via brachytherapy.
The mathematical construct behind it will be discussed below, after we introduce
a useful concept of biologically effective dose (BED). We can see from Figure 2B
that, as the dose per fraction (or dose rate) approaches zero, the progressively
more spared survival curve becomes less steep and ultimately has a limiting
effective slope characterized by the single-hit component only (a or 1 D0 ), since
this is the component of damage that is not repairable in time (5).
Note that Eq. (5) only takes into account one R of the 4 Rs: i.e., repair.
To account for repopulation of cells which is especially relevant when the
tissue under consideration is an acutely-responding (i.e., rapidly proliferating) type, a treatment time factor is introduced (1,28):
SF e
ln2TTk
Tp
nadbd 2
where T is the overall treatment time, Tk is the kick-off time after the
treatment starts and before accelerated repopulation (9) to begin, and Tp
is the effective doubling time of the clonogenic cells.
Finally, since both redistribution and reoxygenation essentially result
in sensitizing cells for radiation killing upon fractionation, it is not surprising
that these two Rs have been grouped together and named resensitization
for the purpose of conceptual simplication when using mathematical
models (29).
141
Biologically Effective Dose (BED)

The negative of the exponent in the expression of Eq. (6), denoted E, where
SF eE , can be visualized as a quantitative measure of the effectiveness
of radiation cell-killing, since it directly reects how steep the effective slope
is on a semi-logarithmic survival curve. Thus,
E nad bd 2
ln 2T Tk
Tp
Based on the LQ model, Barendsen (30) and Fowler (1) have suggested
a quantity termed biologically effective dose (BED) (for Barendsen, it was
called extrapolated response dose, ERD) which proved to be very convenient in quantifying radiobiological effects and even enabled sensible comparisons among various clinical trials using different fractionation schemes (31).
With a dimension of dose (Gy), it is dened (1,30) as:

E
d
ln 2T Tk
BED nd 1
7

a
a=b
a Tp
For late responding tissues only, the treatment time factor (i.e., repopulation) can be neglected, and

E
d
BED nd 1
8
a
a=b
This abstract quantity can perhaps be conceptualized best by its representation on the multi-fractionation survival curves (Fig. 3A). One can see
that the numerical value of BED for any fractionation scheme is equivalent
to the total physical dose needed to cause the same degree of biological
effect (cell survival) using an ultrafractionated regimen in which d
approaches zero and n approaches innity (d ! 0, n ! 1 ) such that the
product, nd, equals the given total dose, D (30,32).
Using the single-hit, multitarget killing model, Withers and Peters (5)
have analyzed in detail the change in the effective slope of the multifraction
survival curve, D0(eff), as the size of dose per fraction changes. Such a cell
survival plotalthough more complicated mathematically [Eq. (3)]shows
a limiting maximal (least steep) slope characterized by the single-hit mechanism only 1 D0 as the dose per fraction (or dose rate) approaches zero. It is
thus analogous to the ultrafractionation scheme of which the total dose is
equivalent to the BED for a given biological effect, using the LQ model
(Fig. 3A). This supports the notion that BED is indeed a mechanistically
sound quantity measuring the isoeffect dose for any fractionation scheme
with respect to a particular process of radiation killing (i.e., single-hit or
the linear component in the LQ theory), which represents the non-repairable
damage to the chromosome that results directly in cell lethality (25,33). Thus,
given any fractionation regimen delivering a total dose D, its corresponding
142
Lee et al.
Figure 3 (A) Concept of biologically effective dose (BED) as

depicted on survival curves. Only
late effects are considered. Each
line from the origin represents
the effective survival curve for a
multifraction regimen using dose
per fraction, d. For a given total
physical dose, D, a BED as a function of d, BED(d), can be visualized as the isoeffective total
dose of a regimen in which d
approaches zero asymptotically.
(B) Concept of normalized isoeffective dose (NID) as depicted on
cell survival curves. As in (A),
except here NID(d) refers to the
isoeffective total dose of an arbitrarily chosen reference scheme in
which d is 2 Gy. Source: Redrawn
from Ref. 34.
BED is a unique entity quantifying the equivalent biological effect and free
from any arbitrarily chosen reference fractionation scheme.
To use BED for clinical application, we may intuitively correlate BED
at a given anatomical location with total physical dose deposited at that point,
because BED also has a unit of dose (Gy). Furthermore, since BED represents
the negative exponent of an exponential function governing the cell-survival
curve on a semi-log plot, the quantity is linearly additive for combination
of multiple independent treatment schemes. Because of the mathematical simplicity, one might construct a computer algorithm enabling BED isodose
(or iso-BED) display on commercial treatment planning system (34). However, one should remember that BED varies with the radiosensitivity parameters (i.e., a and b values) specic for the structure in question. Fowler (1)
advocated using a subscript to make such a distinction, e.g., Gy3 for BED
based on a/b of 3 Gy, and Gy10 for BED with a/b equals 10 Gy.
We can derive readily the isoeffect conversion relation for two fractionation regimens with respective total doses, D1, D2; doses per fraction, d1,
d2; and overall treatment times, T1 T2. For isoeffect, BED1 BED2 ; then,
from Eq. (7),
D1 D2
d2 a=b ln 2T2 T1

d1 a=b bTp d1 a=b
143
Note that the second term involving the treatment time factor is no
longer dependent in Tk and can be ignored entirely when T1 equals T2 or
when considering late-responding tissues only (35):
D1 D2
d2 a=b
d1 a=b
10
To allow oncologists to relate observed radiation treatment effects

with their own clinical experiences, they may translate the BED value implicitly to an equivalent quantity based on an arbitrarily chosen fractionation
scheme. The resulting biologically oriented entity has been given many
names, such as Linear-Quadratic Equivalent Dose based on 2-Gy fraction
(LQED2) (36) or Normalized Isoeffective Dose (NID) (34). Figure 3B illustrates the concept of NID as depicted on cell survival curves, similar to
Figure 3A except that the reference scheme is now set arbitrarily at d equals
to 2 Gy (thus numerically also equivalent to LQED2). We should emphasize
that these are derived merely through a process of normalization to the chosen reference scheme, using the mathematical formula based on the general
isoeffect conversion relations [Eq. (9) or (10)]. Nevertheless, there seems to
be a general consensus among many radiotherapists that a conventional
scheme is about 1.82 Gy per fraction, ve fractions per week. The clinical
wisdom regarding tumor-cure dosage or normal-tissue tolerance levels has
commonly been based on experiences using such a regimen. On the other
hand, many altered fractionation schemes have appeared worldwide
(1,10,31), and new treatment techniques may also necessitate manipulation
of fractionation patterns such as stereotactic radiotherapy vs. radiosurgery
(which differ from each other depending on whether one fractionates the
treatment or not). Hence there is seldom a fractionation regimen now a days
which can qualify as the standard, and BED may prove to be a more
versatile clinical tool to be used as the lingua franca in the quantication
of biological effects for a given tissue in cancer therapy.
The Incomplete Repair Model

Some radiotherapy techniques may deviate from the conventional daily
fractionation scheme, such that the induced SLD may not have sufciently
long (24-hr) inter-fractional interval for adequate repair. In addition, during
continuous LDR brachytherapy, there is concurrent competition between
repair and continuing radiation damage at a protracted pace. To account
for these complex effects in time, the incomplete repair (IR) model may
be utilized. It basically assumes that the efciency of repair follows rstorder kinetics, i.e., exponential in time (37,38). The concept of BED (in
conjunction with the LQ theory) can then be applied for much generalized
144
Lee et al.
use. For example, for brachytherapy:

gt D
BED D 1
a=b

2
1
mt
1 1 e
gt
mt
mt
11
12
where m is the rate constant for SLD repair, and t is the time of continuous
irradiation (33,38,39). Repair kinetics for many tissues of clinical relevance
have been studied, e.g., by split-dose experiment using various range of
inter-fractional intervals, with the values m or g(t) inferred and tabulated
(40). Comparing with Eq. (8), the additional time-dependent factor, g(t),
is incorporated in Eq. (11) to modify the value of BED during continuous
irradiation, or as the inter-fractional interval is decreased such that the
repair becomes less complete. One can see that g(t) ! 1 for t ! 0 (i.e.,
almost instantaneous exposure of radiation such as treatment via daily fractionation) and Eq. (11) reverts to Eq. (8).
With further extension of the IR model, a generalized theoretical framework has been developed to determine equivalent biological effects of LDR or
high dose rate (HDR) brachytherapy (2,3). Indeed, with the property of
linear additivity, BED can be used to quantify the overall effects of fractionated teletherapy, brachytherapy, and indeed any variable range of dose
rates or fractionation schemes (Fig. 4). In this sense, one might say that the
problem of analyzing radiobiology in the temporal domain (represented by heterogeneous distribution of dose rates along the time axis, with its area under
the curve yielding the radiation dose) has been solved in its most general form.
Figure 4 Heterogeneous spread of radiation dose rate along the time (t) axis, with each
ith episode of irradiation centering upon bi, dose rate of ri, and treatment interval of ci (thus
the dose is rici). Mathematical theory has been established to quantify this most general
form of radiation treatment scheme, whether by irregularly fractionated external beam
teletherapy or by brachytherapy of low or high dose rate. Source: Adapted from Ref. 2.
145
TCP and NTCP

It is important to note that BED, despite its versatile theoretical use, does
not directly reect clinical endpoints which can be measured readily in cancer medicine. Clinically, the terms tumor control probability (TCP) and normal tissue complication probability (NTCP) come closer in giving patients a
quantitative sense of radiobiological consequence. They are dened, according to Poisson statistics, as:
TCP eMSFM
13
NTCP eNSFN
14
where M is the number of clonogenic cells in the tumor (or tumorlets), N is

the number of functional subunits (FSUs, at times called tissue rescue units,
TRUs) for normal tissue, and SFM and SFN are the respective surviving
fractions as functions of radiation dose and intrinsic radiation sensitivity
[per Eq. (5) or (6)] (5). When plotted against dose (thus termed dose
response curves), both TCP and NTCP present as sigmoid curves. Furthermore, one can show that the number M or N would dictate primarily the
location of the curve along the abscissa (dose axis): the higher the number,
the more towards the right the steep portion of the curve is located. The
slope of either curve at its steepest portion can become attened when
heterogeneous distribution of the radiation sensitivity parameters (a, b,
etc.) is introduced within a patient population.
It is perhaps worthwhile to discuss here the mathematical expression
of the steepest slope for these doseresponse curves. For the sake of simplicity, we assume that a predominates such that b can be ignored (especially
for acutely responding tissues like tumors, or in general by considering the
effective slope of a multi-fractionation radiation survival curve). Without
considering the heterogeneity of a, one can show readily that the steepest
slope is given as ae1 and is located at the inection point of the curve
where the corresponding probability is e1 or 37%, and the dose, D37, is
equal to ln M/a (for the TCP curve). A more popular entity to associate
with the slope of the response curve is the gamma factor, dened as the
product between dose and the slope (hence a dimensionless quantity)
(41,42). g-50 is thus related to the slope where the probability is 50%, with
the corresponding dose, D50. It is trivial to show that this slope is equal
to a ln2/2, and
g50 D50 a ln2=2
where
D50
ln ln 2=M
a
15
146
Lee et al.
Clinicians often aim for maximum tumor cell killing while minimizing
damage to the adjacent normal tissues in order to achieve a better clinical
outcome by optimization of the therapeutic ratio, dened rather guratively
as the benetrisk ratio of TCP over NTCP. In fact, almost all innovative
ways to improve the outcome of radiation therapy to date has amounted
to widening the gap between TCP and NTCP doseresponse curves. A
useful concept is the so-called uncomplicated TCP (UTCP), dened as (43):
UTCP TCP 1 NTCP
It can be seen that UTCP has a shape of a bell curve, with its peak located
at an ideal dose where TCP is nearly maximized and NTCP minimized.
Clinically, the fact that the TCP doseresponse curve is sigmoid signies that dose escalation beyond the steepest portion of the curve rarely
pays off when the corresponding NTCP is signicant. On the other hand,
for a malignancy with relatively high cure rate at a well-established conventional level of dosage, any small amount of underdosage within the tumor
may compromise the TCP tremendously. These issues will be explored
further when the volume effect is also considered (see section on Heterogenous Dose Distribution and TCP).
FROM TEMPORAL PROBLEM TO SPATIAL CONSIDERATION:
THE DOUBLE-TROUBLE EFFECT
With the development of modern treatment planning systems, radiation
dose distributions within patients bodies are readily available and usually
displayed as two- or three-dimensional contour plots. These isodose plots
represent great assets to clinicians who often need to evaluate the actual
dosage deposited at a critical site. The value of dose at such point may be
drastically different from the dose prescribed originally at any particular site
(e.g., the isocenter or somewhere within the tumor target) or any isodose
level deemed appropriate by the clinician.
Nevertheless, taking care of the differences among physical dose
received at the point of prescription and those at various sites of interest
by isodose contour plots may not be sufcient for the clinicians to assess
the true biological effects at these points. As described above, the biological
effects will depend greatly on how the physical dosage is delivered in time,
namely, the fractionation scheme. Figure 5 illustrates a hypothetical case
of head and neck cancer from which one may appreciate the degree of
changes of biological effects on tumor target or normal tissues, depending
merely on the way clinicians choose to prescribe the treatment dose (34).
When physical doses are compared with what is deposited at the prescription
point, one should note that by fractionation, what gets hot, gets hotter biologically; and what gets cold, gets colder biologically. (Fig. 5, 34). Above all,
the magnitude of the deviation between physical and biological doses is more
147
Figure 5 The double-trouble effect is illustrated here with a display for treatment
using an orthogonal pair of wedged photon beams for a hypothetical tumor in the
maxillary sinus. When the physical dose at the isocenter (tumor) is set at 100%,
the mandible is seen to receive 110%, and the spinal cord only 20%. The results of
biological correction using BED are tabulated, with the dose prescribed at the tumor
(top) and the minimal peripheral dose of 90% (bottom), respectively.
pronounced for late-responding tissues than for acutely-responding ones due

to the different curvatures of their single-dose cell survival curves (Fig. 2C).
Radiation oncologists thus must consider the biological impact of
fractionation on any particular structure of interest, especially the lateresponding type. This is neglected on the conventional treatment plans
based on physical dosimetry alone. The added complexity due to radiobiology consideration has been described as the double-trouble effect by
Withers (36): the rst trouble comes from the difference between physical
dose prescribed and actual dose received at any point (spatial heterogeneity
of total dose), and the second trouble results from the variation of biological
effects with different dose per fraction (spatial heterogeneity of dose per
fraction). The distinction between these two troubles is a consequence of
fractionation (thus a temporal problem). The notion of the double-trouble
effect represents thus a prelude to treating clinical radiobiology issues as
problems existing within a spacetime frame of reference, i.e., a four-dimensional domain.
In dealing with the double-trouble effect, conscientious clinicians
armed with only physical dose dosimetry plans will often estimate biological
effects qualitatively and choose a particular treatment plan accordingly.
148
Lee et al.
Biological dosimetry, such as the above-mentioned BED-based treatment

planning (34), represents a rudimentary attempt to deal with this commonly
encountered clinical issue. Nowadays, with inverse planning for IMRT and
other complex high-precision oriented treatment technologies, a more
sophisticated quantitative guideline based on biological optimization so
far has remained elusive at best, but continues to be a hotly pursued
research endeavor in radiation oncology.
RADIOBIOLOGY PROBLEM ALONG THE SPATIAL DOMAIN:

THE VOLUME EFFECT
The biological effect of radiation upon a particular normal organ at risk
(OAR) can vary signicantly with the amount of volume as well as which
portion or region of the organ is treated. For example, by partial volume
irradiation of murine lungs, Travis et al. have shown that the degree of
lethal pneumonitis depends on both the percent of lung volume irradiated
as well as whether the irradiated region is the apex versus the base (44). This
is commonly termed the volume effect (45,46) and until recently has
largely been ignored during treatment planning because the complexity
involved would have been rather difcult to deal with. Instead, planning
physicists would strive to achieve uniform dose distribution as much as
possible. The rst step would be to cover the entire target volume with
homogeneous dose, then an attempt is made to minimize the dose to the
OAR to be at least below the perceived tolerance level for the whole organ.
The cumulative clinical experiences throughout past decades have thus pertained mostly to uniform dose distribution over any structure of interest.
However, with the advancement in treatment technologies like IMRT,
one can now manipulate dose distribution relatively at ease, especially via
inverse planning algorithm. But such action often results in heterogeneous
dose distributions, whether within the tumor or the OARs. Even during
forward treatment planning for stereotactic or conformal radiotherapy,
while it is feasible to conform tightly the dose coverage for the target
tumor, one might wish to push the tolerance dose of the surrounding
OAR to the extreme by partial organ irradiation. Conversely, selectively
escalating dose within the tumor (i.e., dose painting), especially when directed by functional imaging showing metabolically hyperactive spots, might
be perceived to be benecial. In all these cases, the volume effect may no
longer be negligible.
From Eqs. (13) and (14), we note that both TCP and NTCP depend on
size: i.e., the number of tumor cells or FSUs, respectively. First, assume
radiation dose is distributed heterogeneously within the tumor target or normal organ but the tumor cells or FSUs are organized isotropically uniformly
in space, what then might be the effect of partial volume irradiation? One
149
could perhaps appreciate that the total sum of biological effects due to irradiation of a collection of partial volumes with heterogeneous doses may not
be equal to the effect of delivering the averaged dose to the whole volume
homogeneously. That is, the biological effect of partial volume irradiation
is not linearly additive. Second, the condition of isotropic organization for
tumor cells is perhaps easier to accept since a tumor is usually conceptualized
as containing millions to billions of closely packed tumor cells, like identical
marbles packed closely in a jar. But for normal tissues this is not a trivial matter since obviously FSUs can be distributed anisotropically for some OARs,
thus representing a complex issue to consider for the analysis of the volume
effect, namely, tissue organization (see section on Tissue Organization).
Partial Organ Irradiation and Heterogeneous Dose Distribution
Dose Volume Histogram (DVH)
To deal with the effect of partial normal organ irradiation quantitatively, a
seemingly straight-forward solution would be to use an empirical approach
like the power laws again. A typical example is stereotactic radiosurgery
(SRS) for brain tumors. Clinicians performing this technique have used
the so-called Kjellbergs Diagram, which also follows the empirical concept of powerlaws (47). Here the amount of volume (related to proton
beam diameter) irradiated is assumed to be inversely proportional to the
radiation dose for a constant biological effect (e.g., 1% or 99% chance of
brain necrosis). It does not take into account dose distribution outside the
SRS treatment volume. Flickinger has suggested a more rened approach
using the integrated logistic model (48) to incorporate the effect of partial
brain irradiation (see next section), but it remains to be a variation of the
same theme as the powerlaw model, i.e., empirical approach with statistical
tting of clinical data and applicable specically for brain irradiation.
The FSUs in OARs like the brain can be considered more or less isotropically organized (although not cranial nerves), and during SRS the
volume receiving high dose is typically small relative to the whole brain,
with rapid drop-off of dose outside the treatment volume. Thus, using the
simple powerlaw relationship to direct clinical practice is probably acceptable. Having said that, we might point out that dose homogeneity is still
a debatable issue between specialists doing single dose SRS like Gamma
Knife (less homogeneity in general) and those preferring fractionated stereotactic radiotherapy (SRT) (more homogeneity usually), especially when the
treatment volume contains ne structures like cranial nerves (49).
However, when dose distribution is grossly heterogeneous across a
signicant portion of an OAR (relevant when doing extracranial SRS or
SRT), the degree of such heterogeneity might need to be addressed. A useful
quantitative tool is the dosevolume histogram (DVH). When shown in a
cumulative (integral) rather than a differential form, it reveals the cumulative
150
Lee et al.
volume of the structure receiving at least the corresponding dose on the

abscissa. It assumes a monotonically decreasing shape on a plot of fractional
volume versus dose. Clearly, for tumors the ultimate ideal situation would
be a DVH curve shaped like a rectangle, with 100% of the volume receiving
a certain prescribed dose or more. For OARs the ideal situation would be a
spike at the origin (or L-shape with a tail of zero thickness extending
along the abscissa), i.e., 100% of the volume receiving no dose at all. The
real situation is of course somewhere in-between, since optimizing radiation
therapy is by nature a quid pro quo process: whatever one does to make the
tumor DVH as rectangular as possible, the corresponding DVH for OARs
would as a result deviate from the L-shape more, and vice versa. Keeping
the ideal shapes for both curves in mind can nevertheless help clinicians
evaluate treatment plans using DVH displays as criteria of selection.
Since the biological effect of partial volume irradiation is not linearly
additive, and clinicians are more familiar with uniformly irradiated wholeorgan tolerance dose, a natural question to ask is: given a structure of interest with heterogeneous dose distribution, can one nd an effectively uniform
dose, i.e., a hypothetical dose uniformly distributed over the entire organ,
which would result in equivalent biological effect? One possible way, at least
mathematically, based solely upon dosimetric measures, is to perform
DVH reduction. For example, Lyman and Wolbarst (50,51) proposed
the following empirical recipe: by dividing the dose axis into tiny segments,
one can successively reduce the corresponding volume algebraically from
the extreme right end (maximum dose) of the DVH curve step by step until
the maximum (100%) volume is reached at a particular dose level, which
would essentially be the effective dose (Deff). Once the Deff obtained,
one can simply look up an available NTCP versus D curve for whole-organ
irradiation by assuming the dose to be at Deff. Several other mathematical
algorithms have been proposed for DVH reduction (52), depending on the
parameters (volume, dose, response, probability of complication, etc.) for
which the theorists chose to manipulate the algebra involved.
Even though DVH is a powerful tool to help clinicians handle the problem
of heterogeneous dose distribution in space, one must realize that it often cannot
be relied upon to reect the clinical consequences completely. To use a very simple example: assume that the only dose heterogeneity in an otherwise uniformly
irradiated structure, say, rectum, arises from 5% of its volume receiving excessive dose. Such a small hot spot may result in very distinct clinical sequalae:
inconsequential, if the 5% hot spot scatters rather widely within the rectal wall;
ulcer, if the hot spot is concentrated at a point; or stricture, if it is distributed
circumferentially around a segment of the rectum (53). Needless to mention, the way FSUs are organized in each OAR may be critical.
Another potential problem with the use of DVH is the denition of the
baseline volume of the structure of interest, since volume as denoted in a
DVH is often a fractional (percent) rather than an absolute volume. Thus,
151
different investigators reporting their clinical ndings of 5% hot spot in the

rectum may mean 5% of very different structures: rectal volume as dened
by multiplying the thickness of the rectal wall by the surface area (whole circumferential versus only the arc portion within the treated volume) cylindrical rectal volume with or without the bowel content, or rectum with a
longitudinal length dened differently, etc.
Heterogeneous Dose Distribution and NTCP
Investigators have long realized the importance of quantifying the dependence of NTCP on the amount of partial volume irradiated. Knowing that
a NTCP versus D plot, like TCP, should approximate a rising sigmoid curve,
several have tackled the problem using the phenomenologic approach.
Schultheiss et al. (54) have proposed the logistic model (which gives rise to
a sigmoid curve) empirically to describe whole organ irradiation:
NTCPn 1; D
1
1 D50 =Dk
16
where k characterizes the slope of the sigmoid curve. For radiating only a
partial volume n:
NTCPn; D 1 1 NTCP1; Dn
This logistic model was also the basis upon which Flickinger (48)
constructed his integrated logistic formula to model the normal tissue
effect of brain radiosurgery.
A sigmoid curve can also be visualized as an integral of a bell curve
(i.e., normal or Gaussian function). Thus, Lyman (55) proposed the integrated normal model:
Z t
1
2
NTCP p
ex =2 dx
17
2p 1
Where
t
D D50 n
mD50 n
D50 n
D50 n 1
nn
D50 is the dose giving rise to 50% NTCP, and D50 n is the D50 for irradiating partial volume n. One can see that the parameter m would characterize
the slope of the NTCP curve, and n is an index (0 n 1) for an assumed
powerlaw relationship between dose and irradiated volume.
152
Lee et al.
For heterogeneous dose distribution, one may use the effective volume
method of Kutcher and Burman (56) to perform the DVH reduction rst.
It differs from the Lyman and Wolbarst scheme of nding the Deff by
reducing the differential DVH sequentially to arrive at a single effective
volume, n eff, which receives the maximum dose of D1 uniformly, according
to the following summing formula:
n eff
X Di 1=n
ni
D1
i1
18
where n is again the index for the powerlaw as dened above. The value of
veff can be used to substitute for v in the Lyman equation [Eq. (17)]. To
assess the corresponding NTCP, one would need to refer to a NTCP versus
D curve for the particular amount of volume, n eff.
By differentiating Eqs. (16) and (17) to nd the slope at D50, one can
show that the logistic model and the integrated normal model are equivalent
when
4
k p
19
2p m
Furthermore, using the denition of the gamma factor [Eq. (15)], one
can nd:

@NTCP
1
k
20
g50 D50
p
@D
2p m 4
DD50
The interplay among the three variables NTCP, dose, and volume can
best be represented as a three-dimensional plot (Fig. 6). Using Lymans
empirical model, Burman et al. (57) t the data collected by clinicians (58)
and provided useful reference plots for assessing the effect of partial volume
irradiation for many critical organs. In particular, the empirical parameters,
D50, m, and n can be inferred from these data. One can also see that, among
several different normal tissues, the change in the NTCP curves as partial
volume changes can display distinct behavioral patterns. For example, as
the irradiated volume decreases, some organs show the NTCP curves to
become atter, while those for others tend to shift to the right with relatively the same slope. The reason for such a phenomenon may be attributed
to the different way various normal tissues are organized (to be discussed in
Section on Parallel vs. Series structures).
Heterogeneous Dose Distribution and TCP
In contrast to the empirical formulation of NTCP as a function of irradiated
volume and dose, TCP can be expressed in a more mechanistically oriented
fashion because Eq. (13) involves the term SF which can be formulated based
on the LQ theory, and tumors are inherently organized more or less
Figure 6 A display of the three-dimensional relationships among normal tissue complication probability (NTCP), dose, and partial
volume irradiated. Source: From Ref. 57.
153
154
Lee et al.
isotropically such that the volume effect becomes easier to handle intuitively.
Thus, one may write:
TCP erV SF
21
where r is the density of clonogenic cells (typically on the order of

108109 per cm3), and V is the tumor volume. For the sake of simplicity,
tumors can be assumed to be spherical with diameter d; hence, V = pd3/6.
For the expression of SF, Brenner (59) used the LQ model with
an SLD repair function, g, as in Eq. (12) to handle possible variation in
fractionation scheme:
SF eaDgbD
Four different malignancies were studied and the data were tted
with the above equation to yield three-dimensional (TCP vs. biological dose
versus tumor diameter) surface plots.
Other investigators (6062) examined in further detail the effects of
several realistic clinical factors: namely, heterogeneous distribution of
clonogenic cells within the volume, heterogeneous distribution of radiation
sensitivity among patient populations, as well as non-uniform or partial
volume irradiation for the tumor. This is feasible since the doseresponse
relationship per Eq. (13) is a relatively simple yet mechanistically sound
formulation, especially if one uses only the a component for cell killing
(neglecting b for acutely responding tissue like tumors, or using an effective
slope for fractionated treatment). To analyze non-uniform irradiation, it is
proposed to divide a tumor into tiny voxels or tumorlets and assign
each ith element with its own doseresponse function as:
Y
TCP
eNi SFi
22
i1
where
SFi eaDi
Ni ri Vi
The heterogeneities introduced in biological parameters like r and a
can be distinguished further as intra-tumor versus inter-tumor heterogeneity.
While the deterministic version of Eq. (13) can result in a sharply rising
sigmoid doseresponse curve, it can be shown that in general heterogeneity
in the biological parameters would atten its slope (i.e., decrease g50), especially for inter-tumor heterogeneity in the radiation sensitivity, a. The size
of the tumor as well as intra-tumor heterogeneity and non-uniform irradiation can affect the position of the TCP curve on the dose axis (i.e., D50)
(61,62).
155
To illustrate the effects of heterogeneous dose distribution on TCP at

least qualitatively, Withers discussed in detail the dependency of TCP on the
magnitude of tumor underdosage or overdosage, as well as the fractional
volume of the tumor affected (63,64). After comparing with the theoretical prediction of the TCP for an otherwise uniformly irradiated tumor, Withers made
the following points regarding tumor underdosage (i.e., cold spots) (Fig. 7):
1. The magnitude of the underdosage is the most powerful determinant of TCP.
2. The fastest rate of decline in TCP occurs when the volume underdosed is still small.
Figure 7 The effect of tumor underdosage on TCP as discussed by Withers (63) and
Withers and Lee (64). The key points based on this graph are listed in the text.
Source: From Ref. 63.
156
Lee et al.
3. Signicant inhomogeneity, especially to small volumes, is likely to

occur when high precision radiation treatment (e.g., IMRT) is
applied to mobile tumors.
4. Single dose treatment (e.g., SRS) can have a very tight margin of
error.
5. The rate of decline in TCP is greater in the midrange of the TCP
(i.e., the steep portion of the sigmoid curve, between 15 and 80%)
achieved with homogeneous irradiation.
6. The planning tumor volume (PTV) may contain lower densities of
clonogens, and while it may frequently be underdosed, there may
not be a rapid early decline in TCP with small volumes of
underdosage.
For tumor overdosage (i.e., hot spots), Withers concluded the
following (Fig. 8):
1. Dose escalation is not very useful when TCP is already high.
2. Dose escalation to small volumes is essentially worthless.
3. TCP is mostly determined by the percentage of tumor in which the
dose is not escalated.
4. The gain in TCP increases steeply with dose escalation beyond
50% of tumor volume, especially when the baseline TCP is low.
Therefore, regarding dose painting, Withers concluded the following:
1. It is primarily aimed for areas of increased clonogen density or
decreased radiosensitivity within the tumor, which may correspond
to enhanced or hypodense spots on functional imaging studies.
2. It is subject to geographical uncertainty.
3. It has minimal gain if the dose painting is aimed only at a small
volume, even with huge dose escalation.
4. It is more important to cover the entire tumor volume.
5. The higher the TCP with the standard treatment, the lower the
gain from dose painting.
6. It needs expensive functional imaging techniques, but for minimal
clinical benet.
It is important to note that the above analyses are mainly based on the
assumption that the biological parameters like radiation sensitivity (e.g., a)
or clonogenic density (r) are uniform. The introduction of heterogeneities
into these parameters is expected to alter the conclusions signicantly. As
far as inter-tumor heterogeneity is concerned, it is conceivable that a parameter like a be subjected to a normally (Gaussian) distributed heterogeneity. Accordingly, what Withers has described represents essentially the mean, or
deterministic, behavior and thus remains clinically relevant. However, the
157
Figure 8 The effect of tumor overdosage on TCP as discussed by Withers (63) and
Withers and Lee (64). The key points based on this graph are listed in the text.
possible existence of intra-tumor heterogeneity for parameters like r makes it

difcult for one to conjure up any a priori form of spatial distribution.c
When advocating dose painting guided by functional imaging, for example,
it might be prudent to know whether the metabolically or pharmacokinectically active spots within the target volume actually correspond to radiosensitive/resistant or highly clonogenic spots. In such a way, selective dose
escalation at these sites might be more meaningful beyond what Withers
has implied.
c
A potentially even more complicated source of heterogeneity would be in time: that is, the
biological parameters such as a or r might uctuate randomly in time during the course of
radiation treatment.
158
Lee et al.
Tome and Fowler (65,66) and Fowler (67) have presented a modeling
scheme to deal with issues related to heterogeneous dose distribution. For
hot spots, they found (65,67):
1. Boost doses up to 20% or 30% above the prescribed dose can
increase TCP signicantly, but beyond that level (30%50%)
the effect saturates.
2. Peak or boost doses inside tumors are unlikely to be harmful, from
experience in brachytherapy and SRT, unless it falls on critical
normal structures, e.g., urethra in prostate.
3. Dose escalation can be helpful if there exists within tumors radioresistant (e.g., hypoxic or GO phase cells) subvolumes, but the
amount by which the dose should be escalated may vary.
For cold spots, especially tumor edge misses, Tome and Fowler
described (66,67):
1. A 10% dose decit in 10% of target volume reduces TCP from 50%
to about 43%.
2. If the cold spot has a volume of 1%, a dose decit of 20% also
would reduce TCP to 43%. Any larger dose decits would reduce
TCP precipitously.
3. A 50% dose decit in only 1% volume reduces TCP to zero. A
25% dose decit in 2% of the volume reduces TCP to less
than 30%.
4. The gain in TCP increases steeply with dose escalation beyond
50% of tumor volume, especially when the baseline TCP is low.
Tissue Organization
Flexible vs. Hierarchical Structures
Factors relating to the variations in tissue organization play a signicant
role in repair and repopulation, and may thus contribute to the observed
radiation effect, especially for normal tissues. Michalowski and Wheldon
rst proposed the distinction between the so-called type-H (hierarchical)
and type-F (exible) tissues (68,69). Type-H tissues (e.g., bone marrow, skin,
and gastrointestinal tract) contain stem cells which are destined to mature in
a stepwise fashion into functional cells. As they lose clonogenicity in the
process, these cells become radioresistant because only the rapidly proliferating stem cells are likely to be sensitive to radiation killing. In contrast,
type-F tissues (e.g., lung, liver, and kidney) contain cells which can simultaneously maintain their proliferation capacity (thus remaining radiosensitive)
as well as serve their normal physiological function. The mathematical
models used to analyze the behaviors of these two tissue types are based
on physiological and cellular kinetics reasoning, and aided by using
159
time-dependent differential equations. The results predict that type-F tissues

can exhibit a dose-dependent kinetics-of-damage expressionthe higher the
dose, the earlier the time of expressionwhile that of type-H tissues is relatively independent of the dose. These predictions are in general agreement
with clinical observations. Even though this theory is not easily integrated
into clinical practice due to the mathematical complexity, it is somewhat
more mechanistically sound. Unfortunately, while the physiological and
radiobiological behaviors in time of critical cells under radiation exposure
are considered, no spatial variable is introduced to analyze the possible
effect of different tissue organization in three-dimensional space, i.e., the
volume effect. What may need to happen, if radiation oncologists wish to
stick with the avor of following mechanistic principles, is to advocate quantitative research into the realm of radiation pathophysiology and integrate
such knowledge into clinical radiation treatment planning. Using the rectum
(a predominantly type-H structure) as an example again, one might wish to
know whether its FSUs (stem cells) are organized roughly radially or circumferentially, or what a typical microscopic migratory distance and the associated kinetics for these FSUs happen to be in order to repopulate in a
radiation denuded area.
Parallel vs. Series Structures
To take care of the spatial orientation of normal tissue organization,
Withers et al. (45) rst suggested that a separation be made between parallel
and series structures. The former are typied by kidney, liver, and lung
(as well as tumors), while the later include nerves, spinal cord, and peritoneal sheath. From modeling viewpoint, parallel tissues have been modeled
along the so-called critical volume argument (70), namely that the total
amount of volume irradiated has direct impact on the chance of complication. The smaller the degree of partial organ irradiated (or the more volume
spared within the organ as the functional reserve), the more the dose
response curve is shifted to the right, i.e., the higher the dose needed to result
in the same complication rate. However, the slopes of these sigmoid curves
may remain relatively unchanged. When NTCP is plotted against partial
volume irradiated at a constant dose, it is seen that there is a threshold
volume above which the NTCP rises sharply (67). On the DVH, one sees
that irradiating a signicant volume of such tissue, even if with moderate
doses, would be more detrimental than giving an extremely high dose but
to only a small volume of the organ (Fig. 9). Some would thus loosely state
that parallel structures demonstrate more signicant volume effect, meaning the bulk of the volume irradiated does matter signicantly.
On the other hand, tissues in series have critical elements arranged in
chains upon which irradiating even a small volume of the structure to a sufciently high dose might incur a complication (71). The prime example would
160
Lee et al.
Figure 9 Dose volume histogram (DVH) showing qualitative difference of risk

assessment based on whether the structure of the normal organ is parallel or in series.
be spinal cord, which needs only a hot spot at a given segment to manifest
transverse myelitis. On the DVH, a long tail extending to the high dose region
might be relatively more serious than when most of the volume receives a
moderate dose, in contrast with parallel structures (Fig. 9). One can appreciate that the incidence of a complication increases in proportion to the
volume of serially arranged FSUs irradiated. The doseresponse curve
might shift to the left as more volume (the number of FSUs) is irradiated,
with the slope of the NTCP curve becoming steeper. On the NTCP versus
volume curve, however, no threshold volume is observed, while the rapidity
(slope) of the increase in NTCP as volume increases depends largely on
the initial level of response, i.e., intrinsic radiation sensitivity per volume
element (67).
Most normal tissues have mixed characteristics of both parallel and
series structures. Thus, a concept of relative seriality has been proposed
based on the perceived organization of FSUs (42). Figure 10 illustrates such
a concept and the corresponding values for various structures of interest
(72). The doseresponse curves for several organs, each with different degree
of partial organ irradiation, are shown as well. It can be seen that tumors are
ideal parallel structures, and liver, kidney, and lung are organized in an analogous fashion with relative seriality close to zero. The gastrointestinal tract
and the nervous tissues are organized more in series and thus have higher
values of relative seriality.
Using the rectum as an example, it might be seen that when a precisionoriented radiotherapy such as IMRT is used for prostate cancer and only
treats the anterior wall of the rectum instead of treating the whole circumference, it essentially converts a series type of injury to a parallel type (67).
161
Figure 10 NTCP curves for various structures, showing effects of partial organ
irradiation. The concept of relative seriality is shown, with the value tabulated for
tumor and various normal tissues. Source: From Ref. 72.
The NTCP models based on the consideration of tissue organization

are more mechanistically sound than empirical formulations like the Lyman
model. It can be shown, however, that the critical element scheme proposed
by Niemierko and Goitein (71) gives rise to the same prediction as the
Lyman and Wolbarst (LW; 52,53) model at low levels of NTCP (73), which
is applicable in usual clinical settings. It can predict Deff in the LW scheme
and veff in the Kutcher and Burman (KB; 56, 73) model.
162
Lee et al.
RADIOBIOLOGY PROBLEM IN FOUR-DIMENSIONS

The concept of effective dose, Deff, was introduced earlier [see section
Dose Volume Histogram (DVH)] and ways to determine its value by
DVH reduction algorithms were discussed. This is an empirical entity created mostly to satisfy the clinicians desire to nd an effectively uniform
dose deposited within a normal organ, by mathematical manipulation of
the existing heterogeneous distribution of the dose. No underlying biological
mechanism is implied. Without sufcient knowledge of radiation pathophysiology, a more mechanistic approach is understandably difcult to pursue,
even if one has some rough idea about the structural organization (type-H
versus F or parallel vs. serial). Another important issue is the incorporation
of fractionation effect, i.e. radiobiological consideration along the time axis.
Inevitably, in order to devise rational therapeutic strategies, one needs to
consider biological effects within the combined spatial and temporal
domainsa four-dimensional framework.
Equivalent Uniform Dose (EUD)
Niemierko (74) rst proposed injecting the consideration of fractionation
radiobiology into the volume effect by formulating the concept of equivalent
uniform dose (EUD), thus effectively linking the temporal problem with the
spatial issue of heterogeneous dose distribution. In his words: For any dose
distribution, the corresponding EUD is the dose (in Gy), which, when distributed uniformly across the target volume, causes the survival of the same
number of clonogens. Thus, if the surviving fraction, SF, is assumed to be
governed only by a single-hit component (or effective slope of a multifractionated scheme): SF eaD , then it is straightforward to show that,
in order to satisfy the denition of EUD as given above,
!
N
X
1
aDi
nie
EUD ln
23
a
i1
or
N
1
1X
eaDi
EUD ln
a
N i1
!
24
where N is the number of volume elements (voxels) and vi is the volume of

the ith voxel with corresponding dose Di.d
d
Note that Niemierko (74) used the term SF2, the surviving fraction after 2 Gy, rather than the
parameter a as in Eqs. (23) and (24), to denote the factor pertaining to intrinsic radiosensitivity.
The expressions he wrote are algebraically equivalent to, but a bit more cumbersome than, what
are shown above. Furthermore, they are tied to a rather arbitrarily chosen reference fractionation regimen (by referring to a 2-Gy per fraction scheme).
163
Niemierko proceeded to include the effects of variable tumor volumes,

non-uniform spatial distribution of clonogens (i.e., intra-tumor heterogeneity), unconventional fractionation schemes [by incorporating Eq. (10)], cell
repopulation [by incorporating treatment time factor similar to Eq. (9)],
and heterogeneous biological parameters in a patient population (i.e.,
inter-tumor heterogeneity).
Perhaps because of its mathematical complexity, EUD as depicted in
Eq. (23) or (24) has not gained enough popularity among clinicians. Niemierko
has thus advocated an alternative approach, though reverting back to the
paradigm of using power laws (75). In this version, EUD is dened as:
!1=a
N
1X
a
EUD
D
25
N i1 i
The parameter, a, is clearly an empirical parameter to be inferred from retrospective analysis of clinical data. Its values for various structures of interest,
including cancers, have been collected and tabulated (75). One might notice
that it has negative values for malignant tumors, smaller positive values
(0.53) for commonly considered parallel organs like liver and lung,
and higher values (1020) for serial structures like spinal cord and gastrointestinal tract. Thus, despite its phenomenological origin, EUD as formulated in Eq. (25) may be linked with more mechanistically oriented
quantities as in Eqs. (23) and (24), perhaps by considering factors stemming
from structural organization or radiation pathophysiology.
It remains to be seen whether the phenomenological approach of
Eq. (25) would gain the attention of clinical practitioners readily. As with
all power laws seen in the history of the specialty, such an approach may stir
up future controversy, since it is not clear whether some of the major pitfalls
associated with empirical laws might surface eventually. Nevertheless, as
Niemierko and Mohan pointed out (75), the difculty of applying a mechanistically oriented model is quite real. Since the development in high-precision
oriented treatment technique is here to stay and in fact is ourishing commercially, there is a sense of urgency in trying to implement biological optimization. In this regard, empirical guidelines may represent a quick and easy
solution, just like what Strandqvists plot did for dose fractionation more than
half a century ago.
Equivalent Uniform Biologically Equivalent Dose (EUBED)
Based on Niemierkos concept of EUD and Barendsens BED, Jones and
Hoban (76) introduced the equivalent uniform biologically effective dose
(EUBED), mainly for the need to incorporate a wide variety of fractionation schemes (much like what BED was designed for but EUBED adds
the consideration of the volume effect). This represents thus a rather bold
164
Lee et al.
step of combining the radiobiological problems in the four-dimensional

framework of time and space. Analogous to Eq. (23), and using the denition of BED as in Eq. (8), one can derive the following:
!
N
X
1
aBEDi
EUBED ln
nie
26
a
i1
where BEDi denotes the value of BED at the ith volume element.
With fractionation, using the term eud to denote the equivalent uniform
dose per fraction, one can write

eud
EUBED n eud 1
27
a=b
which is seen to be entirely analogous to Eq. (8). Furthermore, EUBED is
related to EUD by the following:

eud
EUBED EUD 1
28
a=b
The clinical applicability of the EUBED concept has been discussed in
detail by Jones and Hoban (76), especially with a comparison of the IMRT
optimization process between a physical dose-based and an EUBED-based
dosimetry (77). They also utilized the concept of integral biologically effective dose (IBED) advocated by Clark et al. (78). The latter group of investigators had used differential DVHs and the IBED concept to analyze the
toxicity of hypofractionated stereotactic radiotherapy to brain stem. Specically, IBED is dened as:
X
DBEDi
IBED
i
X
i
ndi
di
1
a=b
Dn i
V
29
where the dose axis on the differential DVH is divided into i dose-bands, each
with width of 1% of the maximum dose, and di is the dose delivered to the ith
band. The volume within each dose-band is denoted as Dn i and the total
volume is V. The fact that the total biological effect, IBED, can be determined
by summing up (integrating) individual BED elements (DBEDi) highlights
once again the versatility of BED as a linearly additive quantity.
SPECIFIC ISSUES REGARDING SRS AND SRT
Clinicians performing stereotactic radiation typically distinguish between
two types of approaches: single-dose (SRS) and fractionated (SRT)
165
treatments. Equipped with the knowledge of the radiobiological principles

discussed so far, we might begin to address several issues specic to SRS
and SRT.
General Guidelines in Choosing SRS vs. SRT
Based on what we have discussed about the biology of fractionation, i.e.,
that fractionation spares better late-responding tissues than the acutelyresponding type (see section on Mechanistic Models and Cell Survival
Curves), certain general guidelines might be appropriate regarding selection
of SRS versus SRT. First, whenever an aggressive tumor (thus, acutelyresponding) is found located in close proximity to or enclosing a critical
late-responding normal tissue, SRT would probably be more benecial than
SRS since the biological advantage of fractionation can be exploited. In this
scenario, the stereotactic treatment is truly a form of radiation therapy, i.e.,
biological therapy. On the other hand, if there is not much difference
between the tumor (e.g., a benign or low-grade lesion) and the surrounding
normal tissue as far as the shapes of their radiation survival curves (that is,
a/b) are concerned, it may be legitimate to offer patients SRS treatment,
especially when radical excision of the tumor is perceived to be useful but
contraindicated due to an unacceptably high operative risk. Stereotactic
irradiation used in this wayfor a purely tumor-ablation purposeserves
just like a surgical tool, hence the appropriate name of radio-surgery. We
should emphasize that, since fractionation always spares late-responding
tissues better than acute ones, and that tumors generally belong to a
more acutely-responding typewith the possible exception of prostate
cancer (79)SRT will in general have a theoretical advantage over SRS.
SRS is often favored for logistic reasons rather than biological considerations per se.
Tumor Debulking with SRS vs. Radical Field Coverage
Even though SRS can be used to substitute for real surgical resection, a fundamental tenet in surgical oncology still needs to be observed: that is, partial
tumor resection (tumor debulking, equivalent to partial SRS eld coverage)
is rarely helpful. Accepting the common oncologic notion that a 1 cm
diameter of solid tumor nodule harbors about 109 cells, then a 90%
debulking still leaves 108 cells behind, a 99% debulking leaves 107 cells,
etc. Thus, it is important to ensure that the SRS eld coverage of the lesion
be as radical (complete, with adequate margins) as possible. Radiographic
imaging to help clinicians delineate precisely the extent of the tumor is thus
crucial. In fact, functional imaging studies like positron emission tomography
(PET) or magnetic resonance spectroscopy (MRS) could probably augment
the efcacy of high-precision radiation therapy better if they can help detect
166
Lee et al.
previously unseen tumor edges rather than, or in addition to, picking out
metabolically active spots within a tumor.
Treatment Field Margins
A radiation treatment plan requires judicious choice for margins of eld
coverage, especially for SRS or SRT, where dose fall-off outside the treatment
volume is steep. This is based on the concern for undetectable tumor edge as
well as motion uncertainty. In modem treatment planning terminology, the
concepts of clinical target volume (CTV) and planning target volume (PTV)
are dened precisely for this purpose (80,81). For stereotactic treatment,
one of its most advertised virtues is its high precision and the extremely narrow margin achievable such that the surrounding normal tissue can be
spared to the fullest extent possible. However, one must be aware that the
tight eld margin also means that it is much less forgiving if one makes a
mistake when delineating the gross tumor volume (GTV) and CTV for treatment planning. A 5% shortage in diameter (e.g., by outlining a 9.5-mm
rather than a truly 10-mm diameter of a spherical tumor) at the periphery
would translate into about 14% of the volume at the outer shell being outside
the volume to which the tumor dose is prescribed (Fig. 11). Because SRS acts
like surgeons knife, the 1.4 108 cells in the outer portion have a very good
chance to be underdosed and survive (in particular if this peripheral region is
hypoxic. Furthermore, if the dosimetry is normalized to a prescribed dosage
at the edge of the CTV, then, based on the principle of fractionation biology
Figure 11 The volume of the outer shell of a missed spherical tumor is plotted
against the percent fraction of the radius missed at the periphery, when the radiation
is aimed precisely at the inner sphere. For a 5% miss of the radius, the outer shell
occupies 14% volume of the original sphere. For a 10% miss of the radius, more than
a quarter of the spherical volume would be underdosed.
167
described above (see the section entitled From Temporal Problem to Spatial
Consideration: The Double-Trouble Effect), the biological margin is even
tighter than what the physical margin would imply, because of the different
dose per fraction inside and outside the CTV. This augmentation of change
in biological dose in an inhomogeneously irradiated volume is not surprising
since inside the CTV the dose per fraction is higher than the prescribed level,
hence even higher after biological correction, and vice versa (much lower) for
dose outside the CTV (Fig. 12).
Unconventional Fractionation for SRT
The technology of SRT, often with removable body-xation frames, has been
developed to overcome the biological disadvantages of single-dose treatment
as used in SRS. However, perhaps due to the wide acceptance of SRS, or
because SRT is simply a more laborious procedure, clinicians might have a lingering desire to minimize the number of fractions for patient treatment. Thus,
rather than following the long-held practice of conventional fractionation by
using 1.82 Gy per fraction, unconventional fractionation schemes using a
signicantly higher fractional dose, i.e., hypofractionation, have been tried
for SRT (82,83). Brenner and Hall have supported the practice of accelerated
fractionation (with decreased overall treatment time) when using SRT, specically for intracranial lesions (84). They suggested that the spatial sparing of
the normal tissues by stereotactic technique alone may sufce to overcome the radiobiological disadvantage of hypofractionation. Certainly, when
Figure 12 The dose prole across and outside a spherical tumor, with the prescribed
dose specied as 100% at the edge of the mass. The rapid drop-off of physical dose, so
characteristic for high-precision oriented radiation treatment, in fact translates to an even
sharper drop-off after biological correction using, say, biological effective dose (BED).
168
Lee et al.
applying SRT to extracranial disease sites, care must be taken to minimize the
toxicity to various acutely-responding normal tissues like skin and mucosa if
accelerated treatment is contemplated. For late-responding tissues, both the
dose per fraction and the total dose are key factors to consider. The guidelines
as outlined in the section on Biologically Effective Dose (BED), with the concept of BED or the isoeffect conversion relations of Eqs. (9) and (10), can be
used to compare various fractionation schemes quantitatively (10,31) and provide a starting point for designing treatment protocols.
Coverage, Homogeneity, and Conformality

The effect of geographic miss (or, more correctly, underdosage) at the
peripheral edge of a spherical tumor is illustrated in Figure 11, showing
inadequate eld coverage. There may also be problems of conformality
or conformity between the intended target volume and the actual irradiated volume. Furthermore, homogeneity pertains to the issue of heterogeneous dose distribution, with the associated volume effect discussed
above (see section on Partial Organ Irradiation and Heterogeneous Dose
Distribution). These entities are well specied in the quality assurance
guidelines set forth by the Radiation Therapy Oncology Group (RTOG)
for SRS (85): Coverage is described as the percent of prescription isodose
completely encompassing the target. The conformity index is dened as the
ratio of the prescription isodose volume to the target volume, referred as
the PITV ratio. The homogeneity index is dened as the ratio of the
maximum dose to the prescription dose, referred to as the MDPD ratio.
Utilization of these concepts for dosimetric quality assurance of SRS is
discussed by authors elsewhere in this textbook, but it sufces to say that
Figure 13 Difference in homogeneity between SRT and SRS for a hypothetical

tumor such as acoustic neuroma, which may wrap around the eight cranial nerve.
The relatively higher degree of homogeneity achieved in SRT could possibly explain
the clinical observation of better hearing preservation. Source: From Ref. 49.
169
the incorporation of radiobiological considerations (especially for SRT)

would certainly introduce a higher level of complexity. Hopefully, the theoretical considerations as discussed in this chapter would assist readers to
gain some qualitative appreciation of the issues involved. As an example,
Figure 13 depicts a clinical scenario regarding the use of SRS versus SRT
for an acoustic neuroma, which is hypothesized to contain within its CTV
the eighth cranial nerve as a late-responding normal tissue. Because of the
higher homogeneity generally achievable by SRT rather than by SRS (thus
less differential between the prescribed dose to the tumor and the dose to
the cranial nerve, in particular after biological correction using BED, for
example), it is anticipated that the patients sense of hearing might be preserved better by SRT. The preliminary report for a subsetof patients with
this disease seems to support such an assertion (49).
NOTE ADDED IN PROOF

Simultaneous Integrated Boost (SIB)
With the advent of IMRT and inverse planning, heterogeneous dose distribution
might be planned on purpose in an approach called simultaneous integrated boost
(SIB) (75). The idea of SIB stems from ones desire to simultaneously escalate the
dose to the primary tumor while maintaining approximately the same conventional
level of dose to the regional structures, e.g., lymphatics, which might harbor subclinical metastases. This is in contrast to the traditional shrinking-eld techniquee in
which the primary tumor boost is done sequentially in time after a bigger treatment volume containing the tumor and the regional lymphatics has been given a
dose considered sufcient for the control of subclinical disease. The latter structure
might be described as CTV2, to be distinguished from CTV1 which is equivalent
to the earlier-dened CTV (see section on Treatment Field Margins) that represents the perceived direct microscopic extension of the malignant tumor beyond
the GTV. For example, the shrinking-eld technique for most head and neck cancers usually involves a boost to the gross primary tumor and its edges
(GTVCTV1) to about 70 Gy total, after such volume and the regional lymphatic
(CTV2) have been given approximately 50 Gy. With SIB, the treatment volume
would include both the primary and the cervical lymph nodes, each prescribed
with different dose level but for the entire time duration of the treatment course.
Clearly, this treatment volume can be quite extensive, a scenario not encountered
often by specialists using SRS or SRT for intracranial tumors. From the consideration of both the double-trouble effect and the treatment time factor as discussed earlier, the application of biological dosimetry for SIB technique may
prove to be more complicated. Some empirical guidelines have been extracted
from a few clinical reportsf (75). Recommendation from RTOG seems to dwell
upon delivering the conventional 2 Gy per fraction to the CTV1, higher (2.2 Gy)
e
Fletcher GH. Clinical radiation therapy. In: Fletcher GH, ed. Textbook of Radiotherapy.
3rd ed. Philadelphia: Lea and Febiger, 1980:228.
f
Blanco AI, Chao C. Principles & Practice of Radiation Oncology Updates, 3(3). Philadelphia:
Lippincott, 2002.
170
Lee et al.
for the GTV (hence equivalent to a form of accelerated fractionation strategy), and
lower (1.8 Gy) for CTV2 (75).
Dose Response for Subclinical Disease

Both CTV1 and CTV2, by denition, are perceived to harbor undetectedor subclinicaltumor cells. However, they have quite distinctive bases for their dose
response relationship. For CTV1, the importance to cover tumor edges with adequate
margins has been emphasized previously. Its dose-response depends directly on the
amount of malignant cells present within the irradiated eld (Eq. 13), while the ultimate chance of cure for the patient can also be hindered by those cells missed geographically during a planned precision-oriented treatment. On the contrary, the
subclinical disease conned within the CTV2 arises from the probabilistic dissemination and establishment of metastases. The dose-response curve for these subclinical
metastases has been reported to resemble a linear rather than a sharply rising sigmoid
shape as for bulky primary tumors.g A biophysical model based on the kinetics of
primary tumor growth and the probability of subsequent metastatic colony formation and growth has been proposedh in order to provide a more mechanistic way
of interpreting the clinical observations.
The Role of Ultra-Precision Oriented Radiation Therapy in

Cancer Medicine
With the advent of computer technology, ultra-precision oriented radiotherapy techniques like SRS, SRT, or IMRT have certainly fullled the goal long-held by oncologists to deliver adequate dose for tumor control while minimize toxicity to the
normal tissues. Is this however equivalent to nding the holy grail in cancer medicineas some enthusiastic technophiles so optimistically perceive? The answer is,
unfortunately, not quite. One needs to simply look at Table 1 to appreciate that
the bottleneck in cancer therapy lies largely in the control of metastatic disease rather
than primary tumor control. Clearly, when a malignancy is diagnosed relatively early
(i.e., localized), the cure rate (measured roughly as 5-year survival) has been rather
satisfactory for most common cancers even before the days of precision oriented radiation therapy. In contrast, when a patient presents with metastatic disease, the survival
outcome is often dismal. This underlies the fundamental tenet of cancer prevention and screening, that early diagnosis plus early treatment can translate into long
survival.
Nevertheless, from Table 1 one can also see that treatment results for some
malignancies remain poor despite the fact that the tumor is diagnosed at a localized
stage. While this may simply mean that the diagnostic resolution for micrometastases
is something to be improved, inadequate local control by curative surgery or radiation therapy does represent a signicant cause. This is highly conceivable especially
when one realizes that these malignancies are often gastrointestinal or pulmonary in
origin, for which radiation therapy is often limited in dose due to low tolerance of
the surrounding normal structures (small intestine or lungs, respectively) and
g
Withers HR, Peters LJ, Taylor JMG. Dose-response relationship for radiation therapy of
subclinical disease. Int J Radiat Oncol Biol Phys 1995; 31:353359.
h
Lee S. Ph.D. dissertation, University of California, Los Angeles, 2001.
171
Table 1 Comparison for Five-Year Survival Rates Between Local and Systemic
Stages for Some Common Cancers
Cancer Type
Local (%)
Systemic (%)
All Stage (%)
Prostate
Breast
Uterus
Bladder
Cervix
Colorectum
Stomach
Lung
Esophagus
Liver
Pancreas
99
97
95
93
91
91
61
48
22
13
13
30
20
26
6
9
7
2
2
2
2
2
87
84
84
81
69
61
21
14
11
6
4
Source: U. S. NCI Surveillance, Epidemiology, and End Results Program, 1996
its accuracy hindered by motion uncertainty introduced by breathing movement. The

corollary is, if one can introduce high dose to these tumors while minimizing normal
tissue damage, then enhancement of local control may translate into higher survival.
It may thus be highly productive to invest utilization of precision radiotherapy for
these particular cancers. The recent development in respiratory gating technique,
or broadly speaking image-guided radiation therapy (IGRT)to handle what physicists describe as 4-dimensional treatment, may be justied further in this regard.
Even for prostate cancer, for example, which has enjoyed very high survival
rate when early-stage patients are treated with old-standard radiation therapy, dose
escalation using conformal or IMRT technique may translates into higher distant
metastasis control, at least in theory. This phenomenon, that successful local control
can impact on distant disease outcome and thus survival, has been reported in a
modeling study using Goldie-Codman mechanism to simulate metastatogenesis.i
In summary, dose escalation via ultra-precision oriented radiation therapy
may not necessarily be the key to solving the cancer problem, in view of the fact that
distant metastasis remains to be the bottleneck. We can kill what we can see, it is
often what we cannot see that kills the patient. In this regard, it is not difcult to
see that when the zenith is attained, precision radiotherapy is at most equivalent
to bloodless cancer surgery, along with all the oncologic constraints (uncertain
tumor margins, uncertain metastasis, etc.) faced by surgeons. However, while we
must continue to advocate the importance of early diagnosis and intervention for
most cancers, we may also receive great dividend to simultaneously invest efforts
in implementing precision radiotherapy techniques such as IGRT for those malignancies which historically have not enjoyed adequate rates of local primary tumor
control.
i
Yorke ED, Fuks Z, Norton L, Whitmore W, Ling CC. Modeling the development of
metastases from primary and locally recurrent tumors: comparison with a clinical data base
for prostatic cancer. Cancer Res 1993; 53:29872993.
172
Lee et al.
REFERENCES
1. Fowler JF. The linear-quadratic formula and progress in fractionated radiotherapy. Br J Radiol 1989; 62:679694.
2. Brenner DJ, Hall EJ. Conditions for the equivalence of continuous to pulsed
low dose rate brachytherapy. Int J Radiat Oncol Biol Phys 1991; 20:181190.
3. Brenner DJ, Huang Y, Hall EJ. Fractionated high dose-rate vei-sus low doserate regimens for intracavitary brachytherapy of the cervix: equivalent regimens
for combined brachytherapy and external irradiation. Int J Radiat Oncol Biol
Phys 1991; 21:14151423.
4. Hall EJ. Radiobiology for the Radiologist. 5th ed. Philadelphia: Lippincott,
2000.
5. Withers HR, Peters LJ. Biological aspects of radiation therapy. In: Fletcher GH, ed.
Textbook of Radiotherapy, 3d ed. Philadelphia: Lea and Febiger, 1980:103180.
6. Coutard H. Roentgentherapy of epitheliomas of the tonsillar region, bypopharynx, and larynx om 1920 to 1926. Am J Roentgenol 1932; 28:313331 and
343348.
7. Withers HR: The 4Rs of radiotherapy. Lett JT, Alder H, eds. Advances in
Radiation Biology. Vol. 5. New York: Academic, 1975:241.
8. Thomlinson RH, Gray LH. The histological structure of some human lung
cancers and the possible implications for radiotherapy. Br J Cancer 1955; 9:
539549.
9. Withers HR, Taylor JMG, Maciejewski B. The hazard of accelerated tumor clonogen repopulation during radiotherapy. Acta Oncol 1988; 27:131146.
10. Ang KK, Thames HD, Peters LJ. Altered fractionation schedules. In: Perez
CA, Brady LW, eds. Principles and Practice of Radiation Oncology. 3d ed.
Philadelphia: Lippincott, 1998:119142.
11. Elkind MM, Sutton H. Radiation response of mammalian cells grown, in culture I. Repair of X-ray damage in surviving Chinese hamster cells. Radiat Res
1960; 13:556593.
12. Curtis SB. Lethal and potentially lethal lesions induced by radiationa unied
repair model. Radiat Res 1986; 106:252270.
13. Bergonie J, Tribondeau L. Comptes Rendus de lAlcademie des Sciences 1906;
143:983; Translation of original article: Interpretation of some results of radiotherapy and an attempt at determining a logical technique of treatment. Radiat
Res 1959; 11:587594.
14. Strandqvist M. Studieren uber die kumulative Wirkung der Rontgen-strahlen
bei Fraktionierung. Acta Radiol 1944; 55(suppl):1300.
15. Schwarzschild K. On the law of reciprocity for bromide of silver gelatin. Astrophys J 1900; 11:89.
16. Ellis F. Fractionation in radiotherapy. In: Deeley TJ, Woods CAP, eds. Modern Trends in Radiotherapy. Vol. 1. London: Butterworths, 1967:3451.
17. Kirk J, Gray WM, Watson ER. Cumulative radiation effect. Part I. Fractionated treatment regimens. Clin Radiol 1971; 22:145155.
18. Orton CG, Ellis F. A simplication in the use of the NSD concept in practical
radiotherapy. Br J Radiol 1973; 46:529537.
19. Sheline GE, Wara WM, Smith V. Therapeutic irradiation and brain, injury. Int
173
20. Goldsmith BJ, Rosenthal SA, Wara WM, Larson DA. Optic neuropathy after
irradiation of meningioma. Radiology 1992; 185:7176.
21. Alpen EL. Theories and models for cell survival. In: Alpen EL, ed. Radiation
Biophysics 2d ed. SanDiego: Academic Press, 1998:132287.
22. Desauer F. The cause of the action of X-rays and X-rays of radium upon living
cells. J Radiol 1923; 4:411415.
23. Puck TT, Marcus PI. Action of X-rays on mammalian cells. J Exp Med 1956;
103:653666.
24. Thames HD, Withers HR, Peters LJ, Fletcher GH. Changes in early and late
radiation responses with, altered dose fractionation: implications for dose
survival relationships. Int J Radiat Oncol Biol Phys 1982; 8:219226.
25. Brenner DJ, Hall EL. The origins basis of the linear-quadratic model. (Letter)
26. Yaes RJ, Patel P, Maruyama Y. Response to Brenner and Hall. (Letter) Int
27. Withers HR, Thames HD, Peters LJ. Differences in the fractionation
response of acute and late responding tissues. In: Karcher KH, Kogelnik HD,
Reinartz G, eds. Progress in Radio-Oncology II. New York: Raven Press,
1982:257296.
28. Travis EL, Tucker SL. Isoeffect models aid fractionated radiation therapy. Int
29. Brenner DJ, Hlatky LR, Hahnfeldt PJ, Hall EJ, Sachs RK. A convenient extension of the linear-quadratic model to include redistribution and reoxygenation.
30. Barendsen GW. Dose fractionation, dose rate and isoeffect relationships for
normal tissue response. Int J Radiat Oncol Biol Phys 1982; 8:19811997.
31. Fowler JF. Intercomparisons of new and old schedules in fractionated radiotherapy. Sem Radiat Oncol 1992; 2:6772.
32. Yaes RJ, Patel P, Maruyama Y. On using the linear-quadratic model in daily
clinical practice. Int J Radiat Oncol Biol Phys 1991; 20:13531362.
33. Lea DE, Catcheside DG. The mechanism of the induction by radiation of chromosome aberrations in Tradescantia. J Genet 1942; 44:216245.
34. Lee SP, Leu MY, Smathers JB, McBride WH, Parker RG, Withers HR. Biologically effective dose distribution based on the linear quadratic model and its
clinical relevance. Int J Radiat Oncol Biol Phys 1995; 33:375389.
35. Withers HR, Thames HD, Peters LJ. A new isoeffect curve for change in dose
per fraction. Radiother Oncol 1983; 1:187191.
36. Withers HR. Biologic basis of radiation therapy. In: Perez CA, Brady LW, eds.
Principles and Practice of Radiation Oncology 2d ed. Philadelphia: Lippincott,
1992:6496.
37. Lea DE. A theory of the action of radiations on biological materials capable of
recovery I. The timeintensity factor. Br J Radiol 1938; 11:489497.
38. Thames HD. An incomplete-repair model for survival after fractionated and
continuous irradiations. Int J Radiat Biol 1985; 47:319339.
39. Oliver R. A comparison of the effects of acute and protracted gamma-radiation
on the growth of seedlings of Vicia faha II. Theoretical calculations. Int J Radiat
Biol 1964; 8:475488.
174
Lee et al.
40. Thames HD, Hendry JH. Normal tissue tolerance: time, dose, and fractionation. In: Thames HD, Hendry JH, eds. Fractionation in Radiotherapy. London:
Taylor & Francis, 1987:218237.
41. Bentzen SM, Tucker SJ. Quantifying the position and steepness of radiation
doseresponse curves. Int J Radiat Biol 1997; 71:531542.
gren A, Brahmes A. Tumor and normal tissue responses to frac42. Kallman P, A
tionated non-uniform dose delivery. Int J Radiat Biol 1992; 62:249262.
43. Wambersie A, Hanks G, Van Dam J. Quality assurance and accuracy required
in radiation therapy: biological and medical considerations. In: Madhvanath U,
Parthasarathy KS, Venkateswaran TV, eds. Selected Topics in Physics of
Radiotherapy and Imaging. New Delhi: McGraw-Hill, 1988:124.
44. Travis EL, Liao Z-X, Tucker SL. Spatial heterogeneity of the volume effect for
radiation pneumonitis in mouse lung. Int J Radiat Oncol Biol Phys 1997;
38:10451054.
45. Withers HR, Taylor JMG, Maciejewski B. Treatment volume and tissue tolerance. Int J Radiat Oncol Biol Phys 1988; 14:751759.
46. Withers HR, Thames HD. Dose fractionation and volume effects in normal tissues and tumors. Am J Clin Oncol 1988; 11:313329.
47. Kjellberg RN, Hanamura T, Davis KR, Lyons SL, Adams RD. Bragg-peak
proton-beam therapy for arteriovenous malformations of the brain. N Engl
J Med 1983; 309:269274.
48. Flickinger JC. An integrated logistic formula for prediction of complications
from radiosurgery. Int J Radiat Oncol Biol Phys 1989; 17:879885.
49. Selch MT, Pedroso A, Lee SP, Solberg TD, Agazaryan N, Cabatan-Awang C,
DeSalles AAF. Stereotactic radiotherapy for the treatment of acoustic neuromas. J Neurosurg 2004. Accepted, for publication.
50. Lyman JT, Wolbarst AB. Optimization of radiation therapy 3. A method of
assessing complication probabilities from dosevolume histograms. Int J Radiat
51. Lyman JT, Wolbarst AB. Optimization of radiation therapy 4. A dosevolume
histogram reduction algorithm. Int J Radiat Oncol Biol Phys 1989; 17:433436.
52. Hamilton CS, Chan LY, McElwain DLS, Denham JW. A practical evaluation
of ve dosevolume histogram reduction algorithms. Radiother Oncol 1992;
24:251260.
53. Mark LB. Challenges dening structures of interest in radiation therapy treatment planning. In: Paliwal BR, Fowler JF, Herbert DE, Mehta MP, eds.
Volume & Kinetics in Tumor Control & Normal Tissue Complications. Madison: Medical Phys, 1998:1831.
54. Schultheiss TE, Orton CG, Peck RA. Models in radiotherapy: volume effects.
Med Phys 1983; 10:410415.
55. Lyman JT. Complication probability as assessed from dose volume histograms.
Radiat Res 1985; 104:S13S19.
56. Kutcher GJ, Burman C. Calculation of complication probability factors for
non-uniform normal tissue irradiation: the effective volume method. Int
57. Burman C, Kutcher GJ, Emami B, Goitein M. Fitting of normal tissue tolerance
data to an analytic function. Int J Radiat Oncol Biol Phys 1991; 21:123135.
175
58. Emami B, Lyman J, Brown A, Coia L, Goitein M, Munzenrider JE, Shank B,

Solin LJ, Wesson M. Tolerance of normal tissue to therapeutic irradiation. Int
59. Brenner DJ. Dose, volume, and tumour-control predictions in radiotherapy. Int
60. Deasy J. Inter-patient and intra-tumor radiosensitivity heterogeneity. In:
Paliwal BR, Fowler JF, Herbert DE, Mehta MP, eds. Volume & Kinetics in
Tumor Control & Normal Tissue Complications. Madison: Medical Physics
Publishing, 1998:363381.
61. Webb S, Nahum AE. The biological effect of inhornogeneous tumour irradiation
with inhomogeneous clonogenic cell density. Phys Med Biol 1993; 38:653666.
62. Suit H, Skates S, Taghian A, Okunieff P, Erd JT. Clinical implications of
heterogeneity of tumour response to radiation therapy. Radiother Oncol
1992; 25:251260.
63. Withers HR. Biological aspects of conformal therapy. Acta Oncol 2000;
39:569577.
64. Withers HR, Lee SL. The biology of IMRT. In: Paliwal BR, Herbert DE,
Fowler IF, Mehta MP, eds. Biological & Physical Basis of IMRT & Tomotherapy. Madison: Medical Physics Publishing, 2002:303311.
65. Tome WA, Fowler JF. Selective boosting of tumor subvolumes. Int J Radiat
66. Tome WA, Fowler JF. On cold spots in tumor subvolumes. Med Phys 2002;
29:15901598.
67. Fowler JF. Radiobiological issues in IMRT. In: Paliwal BR, Herbert DE,
Fowler IF, Mehta MP, eds. Biological & Physical Basis of IMRT & Tomotherapy. Madison: Medical Physics Publishing, 2002:822.
68. Michalowski A. Effects of radiation on normal tissues: hypothetical mechanisms of limitations of in situ assays of clonogenicity. Radiat Environ Biophys
1981; 19:157172.
69. Wheldon TE, Michalowski A, Kirk J. The effect of irradiation on function in
self-renewing normal tissues with differing proliferative organisation. Br
J Radiol 1982; 55:759766.
70. Niemierko A, Goitein M. Modeling of normal tissue response to radiation: the
critical volume model. Int J Radiat Oncol Biol Phys 1993; 25:135145.
71. Niemierko A, Goitein M. Calculation of normal tissue complication probability
and dose-volume histogram reduction schemes for tissues with a critical element
architecture. Radiother Oncol 1991; 20:155176.
72. Brahme A. Optimized radiation therapy based on radiobiological objectives.
Sem Radiat Oncol 1999; 9:3547.
73. Webb S. Conformal radiotherapy treatment planning, In: Webb S. In: The
Physics of Three-dimensional Radiation Therapy: Conformal Radiotherapy,
Radiosurgery and Treatment Planning. Bristol and Philadelphia: Institute of
Physics Publishing, 1993:138.
74. Niemierko A. Reporting and analyzing dose distributions: a concept of equivalent uniform dose. Med Phys 1997; 24:103110.
75. Niemierko A, Mohan R. Intenstiy-modulated radiation therapy. Part II. Radiobiological aspects. Refresher Course 303. 44th Annual Meeting of the American
176
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
Lee et al.
Society of Therapeutic Radiology and Oncology, New Orleans, LA, Oct
610,2002.
Jones LC, Hoban PW. Treatment plan comparison, using equivalent uniform
biologically effective dose (EUBED). Phys Med Biol 2000; 45:159170.
Jones L, Hoban P. A comparison of physically and radiobiologically based
optimization for IMRT. Med Phys 2002; 29:14471455.
Clark BG, Souhami L, Pla C, Al-Amro AS, Bahary J-P, Villemure J-G, Caron
J-L, Olivier A, Podgorsak B. The integral biologically effective dose to predict
brain stem toxicity of hypofractionated stereotactic radiotherapy. Int J Radiat
Brenner DJ, Hall EJ. Fractionation and protraction for radiotherapy of prostate carcinoma. Int J Radiat Oncol Biol Phys 1999; 43:10951101.
Prescribing, Recording, and Reporting Photon Beam Therapy. Bethesda: International Commission on Radiation Units and Measurements, Report 50,1993.
Prescribing, Recording, and Reporting Photon Beam Therapy (Supplement to
ICRU Report 50). Bethesda: International Commission on Radiation Units
and Measurements, Report 62, 1999.
Brada M, Laing RW, Graham J, Warrington AP, Hines F. Fractionated stereotactic radiotherapy in the treatment of recurrent high grade gliomadose escalation study. Acta Neurochir 1993; 122:151.
Pozza F, Colombo F, Chierego G, Avanzo RC, Marchetti C, Benedetti A,
Casentini L, Danieli D. Low grade astrocytomas: treatment with unconventional
fractionated stereotactic radiation therapy. Radiology 1989; 171:565569.
Brenner DJ, Hall EJ. Stereotactic radiotherapy of intracranial tumorsan ideal
candidate for accelerated treatment. Int J Radiat Oncol Biol Phys 1994;
28:10391041.
Shaw E, Kline R, Gillin M, Souhami L, Hirschfeld A, Dinapoli R, Martin L.
Radiation Therapy Oncology Group: radiosurgery quality assurance guidelines.
8
Stereotactic Radiation Therapy
of Liver Tumors
Klaus K. Herfarth
Department of Radiation Oncology, University of Heidelberg, Germany
Martin Fuss
Department of Radiation Oncology, University of Texas Health Science Center,
San Antonio, Texas, U.S.A.
INTRODUCTION
The rst reports of successful radiation treatment of liver tumors were
published 1954. Philipps et al. treated 36 patients with symptomatic liver
metastases with doses of 1936 Gy. The symptoms were reduced in more
than 50% of the cases (1). Since that time, the benets of palliative whole
liver radiation have been conrmed in multiple studies (24). However, dose
escalation of whole liver radiation from 27 to 33 Gy resulted in a signicant
increase of liver toxicity (5). Treatment-associated radiation induced liver
damage (RILD) as a dose limiting toxicity appears 48 weeks after radiation
therapy. Clinical symptoms include weight gain, increased abdominal girth,
ascites, and a substantial rise in alkaline phosphatase. Ingold et al. (6) rst
described the clinical picture in 1965. The mortality rate of RILD is approximately 1020%. The pathophysiological counterpart is veno-occlusive disease (VOD). Reed and Cox were the rst to characterize the histological
changes of RILD as marked congestion, which involves mainly the central
portion of each lobule with atrophy of the inner liver plates (7). The wall of
the small veins reveals a large number of ne reticulin bers that crisscross
the lumen of the vein and the adjacent afferent sinusoids (8). Chronic radiation
177
178
Herfarth and Fuss
liver damage is characterized by a distortion of the liver architecture: variable distances between the central veins and portal areas, brosis of the
central veins, and concentric brosis of portal areas (9).
The occurrence of RILD is not only dependent on the dose, it is also
dependent on the irradiated volume (10,11). Emami et al. estimated a TD 5/
5 (tolerance dose with 5% complications in 5 years) of 30 Gy for whole liver
radiation. If 1/3 or 2/3 of the liver could be spared, the TD 5/5 was estimated
to increase to 35 and 50 Gy, respectively (12). Based on more recent data by the
University of Michigan using three-dimensional treatment planning and conformal therapy, Dawson et al. estimated an even more pronounced volume
effect with a TD 5/5 of 31, 47, and 90 Gy for whole liver, 2/3 liver, and 1/3 liver
radiation, respectively (13). The basis for this calculation was a hyperfractionated (1.5 Gy bid) conformal treatment in conjunction with an intra-arterial
FdUrd chemotherapy in 43 patients. The total doses reached 90 Gy (14).
A stereotactic treatment approach for liver malignancies should
achieve better normal tissue sparing than conventional or conformal planning and delivery techniques. Therefore, a further dose escalation or hypofractionated dose delivery should be possible. First steps of stereotactic
radiation therapy of liver malignancies were performed at the Karolinska
Institute in Stockholm, Sweden (15).
Potential targets for stereotactic radiation in the liver are primary and
secondary liver tumors. The incidence of these tumors, indications for
stereotactic radiotherapy, liver specic difculties for the treatment, applied
doses, and current and future trials are described and discussed in the
following sections.
INCIDENCE
There are two major groups of liver tumors: primary liver tumors and
secondary liver tumors. Hepatocellular carcinoma (HCC) accounts for
about 8090% of the primary liver tumors. The incidence of the disease is
currently approximately 13/100,000 in the United States and in Europe,
with a recently observed growth of the incidence. The incidence increases
with age and is associated with chronic liver damage (cirrhosis), often
related to hepatitis C virus infection. Regions with high levels of hepatitis
infections (countries in the Far East) have a 100150 times higher incidence
of HCC. Other risk factors are aatoxins produced by the fungi Aspergillus
avus, which is a major cause of HCC in underdeveloped countries. HCCs
are often multinodular and multifocal tumors. About 30% of the patients with
HCC show distant metastases (mostly in the lungs, peritoneum, adrenal
glands, and bones) at the time of diagnosis.
Cholangiocellular carcinoma (CCC) has an incidence of 12/100,000.
This chapter focuses on the intrahepatic CCC only. CCC affecting the conuence of the right and left hepatic duct is also called a Klatskin tumor. Klatskin
Liver Tumors
179
have a high tendency of perineural and subendothelial spread. This characteristic makes it difcult to evaluate the whole tumor involvement using imaging
studies. About 3050% of the patients show lymph node involvement at
the time of diagnosis. One-third show distant metastasis (liver, lung, and
peritoneum) at this time.
Liver metastases, or secondary liver tumors, show up in about 35% of all
patients with solid tumors during the course of disease. Even higher numbers
are published in organ specic section statistics: the highest incidence for 1008
patients was seen in patients with pancreatic cancer (86%), followed by breast
cancer (60%), colorectal cancer (42%), lung cancer (39%), and stomach cancer
(34%) (16). Due to the portal vein drainage, the liver is the rst site of hematological spread of colorectal cancers. Therefore, liver metastases can be seen
as a kind of advanced loco regional disease in patients with colorectal cancer. The median survival for patients with untreated liver metastases varies
between 5 and 19 months (17,18). Survival is dependent on the primary
tumor and the extent of extrahepatic tumor involvement.
THE ROLE OF STEREOTACTIC BODY RADIATION THERAPY
IN THE MULTI-DISCIPLINARY TREATMENT ARSENAL
The current therapeutic gold standard for primary and secondary hepatic
tumors is surgical resection. The type of resection varies between an atypical
resection, segment resection, hemihepatectomy, extended hemihepatectomy,
and liver transplantation.
The resectability rate of HCC depends on the site of the tumor and the
existence of accompanying liver cirrhosis. Resection is not recommended for
patients with advanced cirrhosis (Child classication B or C). Five-year survival rates of up to 76% have been reported for resected primary HCC in
selected studies (19). For large and/or not resectable HCC, locally ablative
therapies (e.g., percutaneous ethanol injection, radiofrequency ablation) and
also radiation therapy have been described (2023). Radiation dose depends
on the volume of spared normal liver volume and is more limited for patients
with liver cirrhosis (see later this chapter). The sparing of normal liver tissue
favors a stereotactic approach of radiation therapy in these patients. Blomgren et al. treated nine patients with primary HCC using a stereotactic hypofractionated radiation approach. They reported no local failure with a
median follow-up time of 12 months. However, two patients with cirrhosis
developed RILD with non-tractable ascitic uid and died 1.5 and 2.5 months
after therapy. In one of these patients, a very large volume of nearly 300 cm3
was treated (24). Sato et al. stereotactically treated 18 patients with primary
HCC. They combined the radiation treatment with local chemotherapy. No
local failure was observed with a median follow-up of 10 months after the
frameless hypofractionated stereotactic radiation therapy. Only one patient
showed denite deterioration of serum liver function tests (25). The available
180
Herfarth and Fuss
data on stereotactic radiation therapy of HCC is still very limited and

follow-up times are short. However, these limited data show promising
results in inoperable patients if enough functional liver tissue can be spared.
For intrahepatic cholangiocarcinomas, 5-year survival rates of up to
44% have been published after surgical resection (26). However, due to
the subepithelial spread, only about 50% of the intrahepatic cholangiocarcinomas are suitable for resection (26). Experiences with radiation therapy
have not been very promising in non-resectable cholangiocarcinomas (27).
Due to the patterns of local tumor spread, only large margins around the
gross tumor volume may ensure sufcient local control. Therefore, sufcient
sparing of functional normal liver tissue is difcult to achieve, even when a
stereotactic approach is used. The number of patients with CCC treated
with stereotactic irradiation is very low in published series and follow-up
times are typically limited (24,28,29). Therefore, any indication for stereotactic radiation therapy of an intrahepatic primary cholangiocarcinoma
should be set on an individual basis.
Primary therapy for metastatic liver disease usually is systemic
chemotherapy. However, a local therapeutic approach might also have a curative intention in cases of metastatic colorectal cancer. Surgical resection is the
standard therapy in case of solitary secondary liver lesions. Wilson and Adson
retrospectively analyzed patients with limited liver metastases of colorectal
cancers. About 25% of the resected patients survived 5 years while none of
the comparable patients with unresected liver metastases were alive after
5 years (30). Similar observations were made by Adson et al. (18), with 25%
5-year survival in the resected group and 2.5% 5-year survival in the unresected
group. Patients with resections of metachronous liver metastases of colorectal
cancers have similar 5-year survival rates of 20% after each resection (3133).
Several minimal-invasive thermo-ablative approaches have been developed for inoperable metastases including radiofrequency ablation (34),
laser-induced thermotherapy (35), or cryotherapy (36). However, solitary
inoperable liver metastases are also the major indication for the only noninvasive cancer treatment, stereotactic body radiation therapy. Local tumor
control rates of 80100% have been published after stereotactic radiation
therapy of liver metastases with low treatment-associated morbidity
(24,28,29). Potential advantages compared with thermo-ablative procedures
include non-invasiveness, reduced risk of damaging of blood vessels (especially near the liver hilum), and lack of blood ow-mediated temperature
transport to distant liver regions. As of today, there are no denite indications
for stereotactic radiation therapy of liver metastases since the indications vary
between the published series. The major indication was inoperability due to
surgical or medical reasons. There is no established contraindication for centrally located tumors, mostly due to the fact that data for high-dose radiation
damage to centrally located liver structures is unavailable. One contraindication, however, is close proximity to other organs of the gastro-intestinal tract,
Liver Tumors
181
i.e., tumors located close to the surface of the liver. Blomgren et al. reported of
hemorrhagic gastritis or duodenal ulcers if the stomach or the duodenum were
irradiated with more than three times 5 Gy (24). The maximal tumor size
suitable for stereotactic body radiation therapy is controversial. While most
studies limit eligibility to metastases smaller than 56 cm in diameter, other
study groups have also successfully treated larger tumors (24,28). The indication for stereotactic body radiation therapy of larger liver tumors depends
mostly on the volume of the liver and the chances of sparing enough
functional liver tissue from the high dose area.
ORGAN-SPECIFIC DIFFICULTIES
Organ Motion
Liver radiation therapy and, even more so, conformal and stereotactic radiation therapy to targets in the liver have to account for several organ-specic
challenges. Organ motion secondary to diaphragm motion with the breathing
cycle is obviously the most problematic challenge. Traditionally, planning target volume safety margins are assigned to account for both inter-fraction and
intra-fraction liver motion (3739). While individually the range of liver
motion varies, appropriate safety margins range from about 10 mm craniocaudally to as much as 3 or 4 cm. The addition of such signicant safety margins
leads to the inclusion of relevant and dose limiting amounts of normal liver and
other tissues-at-risk volumes to provide for a high probability of target volume
dose coverage during radiation fraction delivery.
Several more or less technologically advanced strategies may be applied
to reduce the respiration-dependent liver motion predominantly occurring
during fraction delivery. Instructed deep breath-holding, or shallow breathing with support of oxygen via a mask may reduce diaphragm movement in
compliant patients (4042). While the reported clinical experience of
instructed breathing relates predominantly to radiotherapy of thoracic
tumors (4047), the resulting tumor immobilization can be directly translated
to abdominal targets since liver motion occurs as a function of diaphragmatic motion (13,48). Similar positive experiences have been reported by
use of a so-called Active Breathing Coordinator or institution-specic
breath-hold valve devices where the patient is coached to inhale to a predetermined depth and to hold this breathing volume for up to 20 sec, during
which time frame radiation delivery is enabled (13,49). Similar to shallow
breathing instructions, the use of such devices depends on patient compliance, and careful patient monitoring during treatment delivery is required.
More mechanistic approaches to reduce liver motion with inhalation/
exhalation depend on abdominal pressure devices attached to the stereotactic
body frame (5052). A plate, often triangular or trapezoid in shape, is pressed
onto the upper abdomen in an angle to constrain liver motion. While at least
182
Herfarth and Fuss
borderline-uncomfortable for the patient, these devices can effectively reduce

the respiration dependent liver motion and allow for reduction of PTV
margins to at or below 10 mm in the cranio-caudal direction (50).
Most recently, respiratory gating using software-controlled assessment
of the anterior abdominal or chest wall breathing-related movement has
been implemented at selected centers (40). Typically, camera-based systems
track the breathing cycle-related anterior/posterior motion of a small
indicator positioned onto the chest or anterior abdominal wall (5357).
The derived breathing amplitude can be used to determine the phases of
the breathing cycle, during which the least surface motion occurs, and subsequently this subset of the breathing cycle is dened as the time window for
beam delivery. Such approaches may signicantly reduce the impact of stillexisting organ motion but the associated cost is measured in the increase in
time to deliver the prescribed radiation dose. In clinical reality, up to 70% of
the breathing cycle has to be disabled for beam delivery, increasing the net
treatment delivery time by a factor of two or more. Since typical treatment
delivery times for stereotactic body radiation therapy already vary between
30 min and up to 2 hr, such prolongation may become relevant, as treatment
delivery time and patient compliance (in terms of patient motion on the
table or in the immobilization device) are closely related.
Imaging for Treatment Planning and Setup Assessment
Prior to Treatment Delivery
Image data for SBRT planning of liver malignancies are typically based on
CT imaging with and/or without intravenous contrast. In order to use any
strategy to reduce safety margins for breathing-related motion, any device
used must be in place during simulation imaging. Occasionally, this requirement may cause problems with the limited opening diameter of the used CT
scanner. Breathing motion during image data acquisition may cause a
variety of well-characterized imaging artifacts, which in the best of scenarios
may render the target volume larger than its true anatomical size (5860).
However, randomly, a liver lesion may be rendered smaller than its anatomical size in the resulting treatment planning image dataset, with the inherent risk for underestimation of the volume that needs to be treated. The
acquisition of a fast (spiral) breath hold scan in addition to slow helical
or sequential slice acquisition during free breathing may be helpful in
estimating the true target extent.
The use of intravenous contrast media should be mandatory in SBRT
treatment planning for liver lesions, although it needs to be appreciated that
especially liver metastases show a ll-in phenomenon that may suggest smaller lesion size in post-contrast CT slices than in the corresponding native CT
slice. Thus, a combination of a native scan with at least one contrast phase
scan is recommended. The addition of a multi-phase liver contrast scan,
Liver Tumors
183
such as a three-phase liver CT used typically for diagnostic purposes, may

further aid in the determination of lesion extent. If the patient setup is controlled immediately before treatment delivery, with the patient repositioned
in the body immobilization device, additional application of intravenous
contrast may not be feasible due to the associated secondary risk with loading excess contrast media and subsequent renal excretion. Thus, identication of the liver lesions in such control CT data may be compromised,
especially when primary liver tumors are the target.
Few experiences have been made with implementing MRI and PET data
into the SBRT planning process for liver tumors. The increasing availability of
fast abdominal imaging sequences for MRI may prove this imaging modality
to be of great aid in delineating the target volumes in the future. The superior
soft tissue contrast inherent to MRI over CT may prove especially helpful in the
delineation of HCC and CCC targets. The metabolic properties of most
primary and secondary liver tumors differ from the surrounding healthy tissue
of the liver and image data co-registration may not only aid in the process of
target delineation but also in the ultimate tumor response assessment in the
foreseeable future. New PET tracers based on amino-acid metabolism rather
than glucose consumption may prove to be more specic for proliferating
tumor tissues, enabling effective and appropriate biological tumor targeting.
DOSES AND CLINICAL OUTCOME
The development of dose escalation in conformal radiotherapy should be
highlighted rst since denite conclusions for a hypofractionated stereotactic approach can be drawn from these data, and the analysis of data
on partial liver irradiation gives more insight on the effects of dose escalation as it is intended in stereotactic radiotherapy.
Dose Escalation in the Normal Liver
The most extensive experience of partial liver radiation together with or
without whole liver radiation and intra-arterial chemotherapy has been
achieved at the University of Michigan (14,6062). A total of 203 inoperable
patients with normal liver function had been radiated for HCC (n 58),
CCC (n 47), and liver metastases (n 98) from 1987 to 1999. Forty-one
patients were treated with whole liver radiation (2436 Gy), 20 patients were
treated with whole liver radiation followed by a boost to a partial liver
volume (to 4566 Gy) and 142 patients were treated with partial liver radiation alone (4890 Gy) in doses of 1.51.65 Gy delivered twice daily (bid).
The median dose was 52.5 Gy (range 2490 Gy). Simultaneously, intraarterial chemotherapy with 5-FU (n 169) or bromodesoxyuridine (BUdR,
n 34) was administered. Treatment plans and total dose were adjusted
to an expected level of normal liver toxicity of 10% using a modied
184
Herfarth and Fuss
Lyman-NTCP-model (61,62). As predicted, in 19 patients (9%) RILD of

RTOG-grade 3 (treatment required) was observed. Six patients had
received whole liver radiation, six whole liver radiation plus local radiation
boost, and seven were treated by partial liver radiation alone. The strongest
parameter predicting liver toxicity was the mean liver dose. In patients with
hepatic toxicity, the mean liver dose was 37 Gy (NTCP 0.17) compared to
31 Gy in patients without RILD (NTCP 0.04), which is in accordance to
Emami et al.s (12) previously published volume-related doses for the whole
liver of the TD5/5 at 30 Gy and the TD50/5 at 40 Gy. Using the adjusted
LymanKutcherBurman NTCP model, the risk for RILD increased by
4% for a Gy increase of mean liver dose exceeding 30 Gy. The best t to clinical data was achieved using a TD50whole liver of 43.3 Gy, n (volume effect) of
1.1, and m (steepness of the doseresponse curve at TD50whole liver) of 0.18.
The volume effect (expressed by the term n) seems to be most important for
toxicity. According to the NTCP model n is close to 0 with low- and close to 1
with high-volume dependence. The authors have adjusted n from 0.32 to
0.69 (61) to now 1.1 as evaluations of a growing number of patients were
updated. Analyzing non-dosimetric prognostic factors by logistic regression,
Dawson et al. (62) found a signicantly increased risk for RILD for hepatobiliary carcinoma (compared to metastases), correlated with the use of
BUdR and male gender. In the group receiving 5-FU, male patients with
hepatobiliary cancer had the highest risk for RILD. In these subgroups, significantly different parameters for TD50whole liver, n and m could be derived.
Additionally, the results were found to be consistent with the threshold
hypothesis of Jackson et al. (63), who assumed that the risk for RILD could
be kept near 0 if the partial liver irradiation volume could be kept below a
threshold volume, regardless of the dose. Dawson et al. discuss that doses
as high as 100 Gy might be safely administered for small volumes of normal
liver tissue (approximately 1/3 of whole liver). This dose escalation might be
benecial, because the clinical results of the Michigan patients revealed an
improved local tumor control with increased dose (14).
Care must be taken if cirrhotic liver is irradiated. All University of
Michigan data were collected on patients with a normal liver function. However, when the liver function is impaired, the risk of developing RILD
increases, and more liver tissue has to be spared than in healthy liver patients.
Seong et al. (64) combined focal liver irradiation in 50 patients with HCC
(Child A n 38, Child B n 12) with transarterial-chemo-embolization
(TACE). The total dose (3060 Gy) was determined by the fraction of the
non-tumor liver volume receiving more than 50% of the prescribed dose
given in 1.8 Gy daily fractions. Six patients were observed with RILD, but
unfortunately were not analyzed concerning dosevolume relations. The
same group published an analysis of doseresponse relation in local radiotherapy for HCC in 158 patients (65). About 90% of patients had liver cirrhosis (Child A 74%, Child B 26%); patients with advanced liver cirrhosis Child C
Liver Tumors
185
were excluded. The tumor size ranges were <5 cm (11%), 510 cm (54%),
and >10 cm (35%). The average 3D-conformal planned dose was 48.2
7.9 Gy (25.259.4 Gy) in daily fractions of 1.8 Gy. While statistic evaluation
revealed that the total radiation dose was the only signicant factor determining
tumor response, hepatic toxicity was also increased with dose. Eleven patients
showed RILD: 4.2% (n l) of all patients in the category of <40 Gy, 5.9%
(n 3) from 4050 Gy and, 8.4% with doses >50 Gy (n 7). Liver cirrhosis of
Child B seemed to be a risk factor in development of RILD, but the number of
cases was small: 0/16 patients <40 Gy, 2/13 patients 4050 Gy (15.4%), and 2/
20 patients >50 Gy (10%). Nevertheless, the evaluation demonstrates that partial liver irradiation can be performed in considerable large volumes even in
patients with impaired liver function. However, the liver function should be
evaluated rst.
PARTIAL LIVER-RADIATION USING STEREOTACTIC SETUP

Hypofractionation
Blomgren and Lax were the rst who published data about stereotactic radiation of liver tumors. Their initial report from 1995 was followed by an update
in 1998 (24,50). After having had negative experiences with single-dose therapy, which is discussed later, they mainly used hypofractionated radiotherapy. The fractionation and the overall time of treatment varied greatly. The
dose ranged from 28 to 315 Gy or 410 Gy. The dose was prescribed
to the PTV encompassing 65% isodose, which resulted in maximal total
doses of 2082 Gy. The treatment time varied between 3 and 44 days (24).
The Swedish group treated 20 primary intrahepatic cancers in 11
patients. The median clinical target volume was 22 cm3 with a range of
3622 cm3. With a mean follow-up of 12 months, no local failures were
observed. However, two fatal cases of RILD in patients with liver cirrhosis
were reported. The rst patient presented with a 57 cm3 HCC nodule associated with hepatitis C and liver cirrhosis. The tumor was treated with
315 Gy to the periphery of the PTV. The patient developed ascites 20 days
after completion of the treatment and died the next month. The other patient
had a 293 cm3 large HCC treated with 310 Gy to the periphery of the PTV.
Also, this patient developed nontractable ascites in the rst 6 weeks after
treatment and died shortly after that. Unfortunately, there is no detailed
information about the size of the liver, the degree of pretherapeutic liver
impairment, or the mean liver dose. Therefore, no denite conclusions about
the risk assessment can be drawn from this published data. Apart from these
fatal side effects, patients experienced nausea, fever, or chills for a few hours
after radiosurgery.
About 10 patients with 20 metastases were also treated at the
Karolinska Institute using the hypofractionated stereotactic regimen. The
186
Herfarth and Fuss
median CTV was 24 cm3 with a range of 2263 cm3. Tumor response was
evaluated after a mean follow-up time of 9.6 months. All tumors showed
response to the therapy. One local recurrence was observed 6 months after
therapy. Again, patients experienced nausea, fever, and chill a few hours
after the procedure. These symptoms were assuaged with a prophylactic
treatment with acetaminophen (synonymous with paracetamol in Europe)
and anti-emetics later on. One patient suffered from a hemorrhagic gastritis
a few weeks after treatment. One-third of the stomach wall had been exposed
to 7 Gy for two treatment sessions. Parts of the duodenum were exposed to
45 Gy in another patient. This patient developed a duodenal ulcer, which
was treated conservatively. These early Stockholm data indicated the feasibility and the possible success rate of a hypofractionated stereotactic treatment for liver tumors. Unfortunately, no dosevolume constrains can be
drawn from these data due to the wide range of the applied dose and different
fractionation schemes. The Stockholm group has continued to treat patients
with hepatic cancer with the stereotactic approach. However, new data have
not been published. Wulf et al., from the University of Wurzburg in
Germany, adopted components of the Stockholm treatment approach (28).
They treated 24 patients with liver tumors (one CCC and 23 metastases).
The median clinical target volume was 50 cm3 with a minimum of 9 cm3
and a maximum of 512 cm3. All but one patient were treated with 310 Gy
to the 65% isodose at the periphery of the PTV. One patient was treated
by 47 Gy, also normalized to the periphery of the PTV. The reason for this
altered fractionation schedule was close proximity of the target to the esophagus. The crude local control was 83% at a mean follow-up of 9 months.
The actuarial local control after 12 months was reported to be 76%, with a
median survival of 20 months. Recurrences occurred 3, 8, 9, and 17 months
after treatment. All recurrences were initially treated with 310 Gy. Failure
of three of these targets occurred marginally. Treatment related morbidity
was low: 7/24 patients reported side effects of grade 1 or 2 according to
the WHO classication. Side effects were mostly observed following one
of three fractions and included fever, chills, and pain, with a typical onset
a few hours after irradiation. Additionally, nausea and/or vomiting might
occur at the same time. The symptoms ceased spontaneously or could
successfully be treated with acetaminophen or prednisolone. Only one
patient showed longer lasting fatigue, weakness, and loss of appetite.
Radiosurgery
The term radiosurgery implies a focused single-dose radiation therapy. Most
of the stereotactic treatments in the brain were successfully performed using
a radiosurgical approach (66,67). Blomgren and Lax also started with a
single dose therapy for liver tumors (50). Six tumors in ve patients were treated
radiosurgically. The median prescribed dose to the periphery of the PTV was
Liver Tumors
187
15.5 Gy, ranging from 7.7 to 30 Gy. No recurrences were observed during a
median follow-up of 5 months. However, one patient died 2 days after treatment. This patient had a 229 cm3 large HCC in a cirrhotic liver. The tumor
was treated with 30 Gy applied to the periphery of the PTV, with a corresponding
isocenter dose of 48 Gy. The patient already was icteric and showed signs of
ascites at the time of treatment. The other four patients showed marginal recurrences during follow-up as it is mentioned in a later paper of the Stockholm
group (24). These two circumstances forced Blomgren and Lax to abandon
the radiosurgical approach for large liver tumors.
In 1997, a phase I/II trial was initiated at the German Cancer
Research Center in Heidelberg (Germany) proving the feasibility and the
clinical outcome of a single-dose radiation therapy of liver tumors (29).
The inclusion criteria for the study were non-resectable tumors in the liver.
The number of liver lesions should not exceed three tumors (four, if two
tumors with less than 3 cm are close together). The size of a single lesion
should not exceed 6 cm, and none of the tumors should be immediately
adjacent to parts of the gastro-intestinal tract (distance >6 mm). The exclusion criterion was insufcient liver function. Thirty-seven patients were
included. A total of 60 tumors were radiosurgically treated at 40 occasions.
The targets included four primary hepatic tumors and 56 metastases (mainly
colorectal cancer or breast cancer). The median target size was 10 cm3
(1132 cm3). The dose was prescribed to the isocenter with the 80% isodose
encompassing the PTV. The dose was escalated from 14 to 26 Gy based on
the liver dose in the dosevolume histogram. After initial dose escalation, an
actuarial local tumor control of 81% at 18 months could be achieved with a
mean follow-up of 9.5 months. All patients received a prophylactic dexamethasone medication before and after radiation therapy. The actuarial
2 years survival was 59%. Patients with curative treatment intention showed
a signicant longer survival (actuarial 87% at 2 years) than patients with
additional extrahepatic tumor manifestations at the time of treatment
(median survival 12 months) (29). An update of these study patients with
a mean follow-up of 17 months was published in 2003 (68). Two patients
developed late local recurrences 4 years after therapy. The actuarial local
control remained unchanged with 81% after 18 months.
As described later in this chapter, a follow-up trial was initiated after
these promising initial results. More patients had been radiosurgically
treated according to the initial phase II protocol until recruitment of the
follow-up trial could be started. A combined total of 78 patients were
treated until spring 2003. The mean follow-up was 12 months and the
actuarial local tumor control dropped to 72% at 12 months. Analysis of
the increased failure rate revealed that patients with metastases of a colorectal cancer showed a signicant worse local tumor control than patients
with other histologies (68). Of special note, all 11 patients who already
had received chemotherapy using CPT-11 or oxaliplatine had shown local
188
Herfarth and Fuss
recurrences during the rst 15 months after therapy. These recurrences were
ineld and marginal recurrences. Therefore, higher doses and/or larger
safety margins should be used especially if colorectal cancer metastases
are treated.
Side effects of the treatment were minimal (29). They included mild
nausea or loss of appetite for 12 weeks in about one-third of the patients.
A singultus was observed in two patients and one patient developed fever.
There were signs of radiation induced liver disease. All patients who were
followed using multiphasic CT scanning showed a sharply demarcated focal
radiation reaction. Tumor and radiation reaction could be well differentiated in the portal-venous contrast-enhanced CT scans. Liver vessels ran
through the liver reaction and were not displaced, as is seen in case of an
expanding tumor. A detailed evaluation and characterization of this focal
radiation reaction in 36 of the Heidelberg patients was published in 2003
(69). The area of radiation reaction was hypodense in the majority of the
non-enhanced CT scans. Three different types of appearance of the reaction
could be dened based on the liver density in the portal-venous and the late
phase after contrast agent administration:
Type 1 reaction: Hypodensity in portal-venous contrast phase,
isodensity in the late contrast phase.
Type 2 reaction: Hypodensity in portal-venous contrast phase,
hyperdensity in the late contrast phase.
Type 3 reaction: Isodensity/hyperdensity in portal-venous contrast
phase, hyperdensity in the late contrast phase.
The onset of the reaction was after a median of 1.8 months. While type
1 or 2 reactions were usually observed earlier, type 3 reactions appeared
later than the other types. It was also seen that there was a shift of the
appearance during follow-up toward type 3 appearances. In addition, the
volume of the radiation reaction decreased with follow-up time. The most
dramatic shrinkage was observed during the rst months after appearance.
This lead to the speculation that the whole reaction goes through different
radiological stages (type 1, 2, and 3 appearances). The histological basis
of these stages was not determined since no biopsies were taken. However,
others had reported a type 2 appearance after single-dose radiation therapy
and it was histologically conrmed VOD (70).
Based on reconstruction of the dosevolume histograms, the mean
threshold dose was 13.7 Gy with a wide range between 8.9 and 19.2 Gy given
in a single fraction. One reason for this large variance might be the fact that
the volume decreased much between the initial detection and the further
follow-up examinations. The examination might have not detected larger
reaction volumes and, therefore, the calculated threshold doses might have
been overestimated. This was sustained by the signicant correlation between
the threshold dose and the time of detection (correlation coefcient r 0.709).
Liver Tumors
189
Apart from the time factor, other factors that could inuence the individual
radiation sensitivity (e.g., additional toxic liver agents like alcohol) might
have been another reason of the variance. More data are needed to strengthen
these threshold doses.
ONGOING STUDIES
As described earlier, there are two different strategies for stereotactic radiation of liver tumors: on one side, a hypofractionated approach with a more
or less inhomogeneous dose distribution within the PTV with maximum
dose of up to 150% (corresponding to a prescription to the 65% encompassing isodose). On the other side is the radiosurgical approach with a more
homogenous dose distribution within the PTV (80% isodose encompassing
PTV). The comparison of these two strategies has been the goal of a new
phase III trial that was initiated by the two major German groups engaged
in stereotactic body radiation therapy of liver targets. The StRaL-trial
(Stereotactic Radiation Therapy of Liver Metastases) is a prospective randomized multicenter trial, which has started patient recruitment in March
2003 with a planned enrollment of 276 patients over 5 years. Inclusion criteria are a maximum of three liver metastases, which are surgically inoperable. The maximal size of the tumors is dependent on the number of targets:
5 cm for one target, 4 cm for two targets, and 3 cm for three targets. The primary study goal is the comparison of the local tumor control. Secondary
goals are survival, morbidity, and quality of life. The study is designed to
prove the equivalence of both treatment arms. Patients in arm A receive a
single-dose radiation therapy of 28 Gy normalized to the isocenter with
the 80% isodose (22.4 Gy) encompassing the PTV. Patients in arm B receive
a hypofractionated therapy with 312.5 Gy normalized to the 65% isodose
(encompassing the PTV) (Table 1). This increase in dose over published
experiences is based on the recent internal updates of the initial phase II
data.
In the United States, two active phase I/II multicenter studies are being
conducted to determine the optimal dose and the maximally tolerated dose
(MTD) for hypofractionated treatment of HCC and liver metastases (71).
Both protocols (initiated by investigators from the Universities of Colorado
and Indiana) have a similar study design but investigate the two tumor entities separately, secondary to the perceived increased risk of treatment-related
toxicity in patients with primary liver malignancies. The initial dose level was
12 Gy delivered three times for a total minimal target dose of 36 Gy in 510
days. Dose escalation will be performed in steps of 2 Gy per fraction (6 Gy
total dose) up to a total dose of 60 Gy, or upon determination of an MTD.
The primary goal of both studies is the determination of the MTD by assessing the dose limiting toxicity (DLT). Secondary endpoints are: 6-month ineld tumor response, failure rate, disease free survival, and overall survival.
190
Herfarth and Fuss
Table 1 Target Doses and Normal Tissue Constraints for the German Prospective
Randomized Multicenter Trial StRaL
Arm A
128 Gy/isocenter
Arm B
312.5 Gy/65% isodose
Relative
dose/fx (%)
Absolute
dose/fx
Relative
dose/fx (%)
Absolute
dose/fx
100
80
43
25
43
43
43
43
43
28 Gy
22.4 Gy
12 Gy
7 Gy
12 Gy
12 Gy
12 Gy
12 Gy
12 Gy
100
65
36
26
36
36
36
36
36
19.2 Gy
12.5 Gy
7 Gy
5 Gy
7 Gy
7 Gy
7 Gy
7 Gy
7Gy
Isocenter
Minimum PTV
Liver (30% vol.)
Liver (50% vol.)
Esophagus (max.)
Stomach (max.)
Duodenom (max.)
Colon (max.)
Myelon (max.)
A maximum of 15 patients will be enrolled in the phase I portion of each trial

(a minimum of three at each dose level), and an additional 1335 patients will
be enrolled in the phase II portion of the studies.
FUTURE RESEARCH
While the methodology and procedural conduct of SBRT for primary and
secondary liver malignancies have been well established, the clinical role
of or the distinct indications for SBRT in this disease context has not yet
been well established. Some of this current shortcoming is certainly related
to the fact that SBRT represents a competing treatment modality for local
treatment concepts such as RFA, LITT, and cryoablation, and potentially
for the present gold standard, surgical resection.
Pending the results of the currently ongoing clinical trials, which are
designed to assess the equivalency of single dose and hypofractionated dose
scheduling as well as the MTD for three fraction SBRT planning and delivery, the impact of SBRT on disease-specic and overall survival in combination with other established treatment modalities or in comparison with one
or more modalities has to be tested. However, in the opinion of the authors
it seems especially intriguing to evaluate if combining SBRT with either
curative attempt surgery or RFA/LITT can improve upon the local failure
rates associated with the use of these modalities and if better local tumor
control can result in improved survival rates. A model of combining SBRT
with one modality could be to employ SBRT in the waiting phase before liver
transplant in small HCC. Thus, SBRT can serve as a desirable bridge-to-transplant, and, depending on the effective time window between SBRT and
Liver Tumors
191
harvesting of the diseased liver, a histopathological examination can reveal

local treatment efcacy in addition to hard outcome endpoints such as disease
specic and overall survival rates.
In summary, SBRT has been shown to provide for a feasible and completely non-invasive treatment modality complementing the present armamentarium in the ght against potentially curable localized primary and
secondary liver tumors. The acceptance and the ultimate success of this
new treatment modality will to a large extent depend upon our willingness
to prove its capabilities in the framework of multimodality treatment
approaches.
REFERENCES
1. Phillips R, Kamofsky DA, Hamilton LD, Nickson JJ. Roentgen therapy
of hepatic metastases. Am J Roentgenol Radiat Ther Nucl Med 1954; 71:
826834.
2. Borgelt BB, Gelber R, Brady LW, Grifn T, Hendrickson FR. The palliation
of hepatic metastases: results of the radiation therapy oncology pilot study.
Int J Radiat Oncol Biol Phys l981; 7:587591.
3. Leibel SA, Pajak TF, Massullo V, Order SE; Komaki RU, Chang CH, et al.
A comparison of misonidazole sensitized radiation therapy to radiation therapy
alone for the palliation of hepatic metastases: results of a radiation oncology group
randomized prospective trial. Int J Radiat Oncol Biol Phys 1987; 13:10571064.
4. Sherman DM, Weichselbaum R, Order SE, Cloud L, Trey C, Piro AJ.
Palliation of hepatic metastasis. Cancer 1978; 41(5):20132017.
5. Russell AH, Clyde C, Wasserman TH, Turner SS, Rotman M. Accelerated
hyperfractionated hepatic irradiation in the management of patients with liver
metastases: results of the RTOG dose escalating protocol. Int J Radiat Oncol
Biol Phys 1993; 27:117123.
6. Ingold JA, Reed GB, Kaplan HS, Bagshaw MA. Radiation hepatitis. Am
J Roentgenol 1965; 93:200208.
7. Reed GB, Cox AJ Jr. The human liver after radiation injury. A form of venoocclusive disease. Am J Pathol 1966; 48:597611.
8. Fajardo L, Colby T. Pathogenesis of veno-occlusive disease after radiation.
Arch Pathol Lab Med 1980; 104:584588.
9. Lewin K, Millis R. Human radiation hepatitis. A morphologic study with
emphasis on the late changes. Arch Pathol 1973; 96:2126.
10. Haddad E, Le Bourgeois JP, Kuentz M, Lobo P. Liver complications in
lymphomas treated with a combination of chemotherapy and radiotherapy:
preliminary results. Int J Radiat Oncol Biol Phys 1983; 9(9):13131319.
11. Poussin-Rosillo H, Nisce LZ, DAngio GJ. Hepatic radiation tolerance in
Hodgkins disease patients. Radiology 1976; 121(2):461464.
12. Emami B, Lyman J, Brown ALC, Goitein M, Munzenrider JE, et al. Tolerance
of normal tissue to therapeutic irradiation. Int J Radiat Oncol Biol Phys 1991;
21:109122.
192
Herfarth and Fuss
13. Dawson LA, Brock KK, Kazanjian S, Fitch D, McGinn CJ, Lawrence TS, et al.
The reproducibility of organ position using active breathing control (ABC) during
liver radiotherapy. Int J Radiat Oncol Biol Phys 2001; 51(5):14101421.
14. Dawson LA, McGinn CJ, Normolle D, Ten Haken RK, Walker S, Ensminger W,
et al. Escalated focal liver radiation and concurrent hepatic artery uorodeoxyuridine for unresectable intrahepatic malignancies. J Clin Oncol 2000; 18(11):
22102218.
malignancies in the abdomen. Acta Oncol 1994; 33(6):677683.
16. Blaker H, Hofmann WJ, Theuer D, Otto HF. Pathohistologische Befunde bei
Lebermetastasen. Radiologe 2001; 41:l7.
17. Jaffe BM, Donegan WL, Watson F, Spratt JS, Jr. Factors inuencing survival in
patients with untreated hepatic metastases. Surg Gynecol Obstet 1968; 127(1):111.
18. Adson MA, van Heerden JA, Adson MH, Wagner JS, IIstrup DM. Resection
of hepatic metastases from colorectal cancer. Arch Surg 1984; 119(6):647651.
19. Takenaka K, Shimada M, Higashi H, Adachi E, Nishizaki T, Yanaga K, et al.
Liver resection for hepatocellular carcinoma in the elderly. Arch Surg 1994;
129(8):846850.
20. Matsuura M, Nakajima N, Arai K, Ito K. The usefulness of radiation therapy
for hepatocellular carcinoma. Hepatogastroenterology 1998; 45(21):791796.
21. Matsuzaki Y, Osuga T, Saito Y, Chuganji Y, Tanaka N, Shoda J, et al. A new,
effective, and safe therapeutic option using proton irradiation for hepatocellular carcinoma. Gastroenterology l994; 106(4):l0321041.
22. Robertson JM, Lawrence TS, Dworzanin LM, Andrews JC, Walker S, Kessler ML,
et al. Treatment of primary hepatobiliary cancers with conformal radiation therapy
and regional chemotherapy. J Clin Oncol 1993; 11:12861293.
23. Seong J, Park HC, Han KH, Lee DY, Lee JT, Chon CY, et al. Local radiotherapy for unresectable hepatocellular carcinoma patients who failed with
transcatheter arterial chemoembolization. Int J Radiat Oncol Biol Phys 2000;
47(5):13311335.
24. Blomgren H, Lax I, Goranson H, Krpelien T, Nilsson B, Naslund I, et al.
Radiosurgery for tumors in the body: clinical experience using a new method.
J Radiosurg 1998; l(l):6374.
25. Sato M, Uematsu M, Yamamoto F, Shioda A, Tahara K, Fukui T, et al.
Feasibility of frameless stereotactic high-dose radiation therapy for primary
or metastatic liver cancer. J Radiosurg 1998; l(3):233238.
26. Nakeeb A, Pitt HA, Sohn TA, Coleman J, Abrams RA, Piantadosi S, et al.
Cholangiocarcinoma. A spectrum of intrahepatic, perihilar, and distal tumors.
Ann Surg 1996; 224(4):463473.
27. Bowling TE, Galbraith SM, Hateld AR, Solano J, Spittle MF. A retrospective
comparison of endoscopic stenting alone with stenting and radiotherapy in
non-resectable cholangiocarcinoma. Gut 1996; 39(96):852855.
28. Wulf J, Hadinger U, Oppitz U, Thiele W, Ness-Dourdoumas R, Flentje M.
2001; 177(12):645655.
Liver Tumors
193
29. Herfarth KK, Debus J, Lohr F, Bahner ML, Rhein B, Fritz P, et al. Stereotactic
single dose radiation therapy of liver tumors: results of a phase I/II trial. J Clin
Oncol 2001; 19:164170.
30. Wilson SM, Adson MA. Surgical treatment of hepatic metastases from
colorectal cancers. Arch Surg 1976; 111(4):330334.
31. Nordlinger B, Vaillant JC, Guiguet M, Balladur P, Paris F, Bachellier P, et al.
Survival benet of repeat liver resections for recurrent colorectal metastases: 143
cases. Association Francaise de Chirurgie. J Clin Oncol 1994; 12(7):14911496.
32. Femandez-Trigo V, Shamsa F, Sugarbaker PH. Repeat liver resections from
colorectal metastasis. Repeat Hepatic Metastases Registry. Surgery 1995;
117(3):296304.
33. Herfarth C, Heuschen UA, Lamade W, Lehnert T, Otto G. Rezidiv-Resektionen
an der Leber bei primaren und sekundaren Lebermalignomen. Chirurg 1995;
66(10):949958.
34. Solbiati L, Goldberg SM, Ierace T, Livraghi T, Meloni F, Dellanoce M, et al.
Hepatic metastases: percutaneous radio-frequency ablation with cooled-tip
electrodes. Radiology 1997; 205:367373.
35. Vogl TJ, Muller PK, Mack MG, Straub R, Engelmann K, Neuhaus P. Liver
metastases: interventional therapeutic techniques and results, state of the art.
Eur Radiol 1999; 9:675684.
36. Onik GM, Atkinson D, Zemel R, Weaver ML. Cryosurgery of liver cancer.
Semin Surg Oncol 1993; 9(4):309317.
37. Shirato H, Seppenwoolde Y, Kitamura K, Onimura R, Shimizu S. Intrafractional tumor motion: lung and liver. Semin Radiat Oncol 2004; 14(l):1018.
38. Rosu M, Dawson LA, Baiter JM, McShan DL, Lawrence TS, Ten Haken RK.
Alterations in normal liver doses due to organ motion. Int J Radiat Oncol Biol
Phys 2003; 57(5):14721479.
39. Antolak JA, Rosen, II. Planning target volumes for radiotherapy: how much
margin is needed? Int J Radiat Oncol Biol Phys 1999; 44(5):11651170.
40. Mageras GS, Yorke E. Deep inspiration breath hold and respiratory gating
strategies for reducing organ motion in radiation treatment. Semin Radiat
Oncol 2004; 14(l):6575.
41. Nakagawa K, Aoki Y, Tago M, Terahara A, Ohtomo K. Megavoltage
CT-assisted stereotactic radiosurgery for thoracic tumors: original research in
the treatment of thoracic neoplasms. Int J Radiat Oncol Biol Phys 2000;
48(2):449457.
42. Uematsu M, Shioda A, Suda A, Tahara K, Kojima T, Hama Y, et al. Intrafractional tumor position stability during computed tomography (CT)-guided
frameless stereotactic radiation therapy for lung or liver cancers with a fusion
of CT and linear accelerator (FOCAL) unit. Int J Radiat Oncol Biol Phys
2000; 48(2):443448.
43. Onishi H, Kuriyama K, Komiyama T, Tanaka S, Sano N, Aikawa Y, et al.
A new irradiation system for lung cancer combining linear accelerator, computed tomography, patient self-breath-holding, and patient-directed beamcontrol without respiratory monitoring devices. Int J Radiat Oncol Biol Phys
2003; 56(l):1420.
194
Herfarth and Fuss
44. Kim DJ, Murray BR, Halperin R, Roa WH. Held-breath self-gating technique
for radiotherapy of non-small-cell lung cancer: a feasibility study. Int J Radiat
Oncol Biol Phys 2001; 49(l):4349.
45. Mah D, Hanley J, Rosenzweig KE, Yorke E, Braban L, Ling CC, et al.
Technical aspects of the deep inspiration breath-hold technique in the treatment
of thoracic cancer. Int J Radiat Oncol Biol Phys 2000; 48(4):11751185.
46. Rosenzweig KE, Hanley J, Mah D, Mageras G, Hunt M, Toner S, et al. The
deep inspiration breath-hold technique in the treatment of inoperable
non-small-cell lung cancer. Int J Radiat Oncol Biol Phys 2000; 48(l):8187.
47. Hanley J, Debois MM, Mah D, Mageras GS, Raben A, Rosenzweig K, et al.
Deep inspiration breath-hold technique for lung tumors: the potential value
of target immobilization and reduced lung density in dose escalation. Int
48. Murphy MJ, Martin D, Whyte R, Hai J, Ozhasoglu C, Le QT. The effectiveness
of breath-holding to stabilize lung and pancreas tumors during radiosurgery.
Int J Radiat Oncol Biol Phys 2002; 53(2):475482.
49. Wong JW, Sharpe MB, Jaffray DA, Kini VR, Robertson JM, Stromberg JS,
et al. The use of active breathing control (ABC) to reduce margin for breathing
motion. Int J Radiat Oncol Biol Phys 1999; 44(4):911919.
50. Blomgren H, Lax I, Naslund I, Svanstrom R. Stereotactic high dose fraction
radiation therapy of extracranial tumors using an accelerator. Clinical experience of the rst thirty-one patients. Acta Oncol 1995; 34(6):861870.
51. Wulf J, Hadinger U, Oppitz U, Olshausen B, Flentje M. Stereotactic radiotherapy of extracranial targets: CT-simulation and accuracy of treatment in
the stereotactic body frame. Radiother Oncol 2000; 57(2):225236.
52. Herfarth KK, Debus J, Lohr F, Bahner ML, Fritz P, Hoss A, et al. Extracranial
stereotactic radiation therapy: set-up accuracy of patients treated for liver
metastases. Int J Radiat Oncol Biol Phys 2000; 46(2):329335.
53. Vedam SS, Kini VR, Keall PJ, Ramakrishnan V, Mostafavi H, Mohan R.
Quantifying the predictability of diaphragm motion during respiration with a
noninvasive external marker. Med Phys 2003; 30(4):505513.
54. Wagman R, Yorke E, Ford E, Giraud P, Mageras G, Minsky B, et al. Respiratory gating for liver tumors: use in dose escalation. Int J Radiat Oncol Biol Phys
2003; 55(3):659668.
55. Kubo HD, Hill BC. Respiration gated radiotherapy treatment: a technical
study. Phys Med Biol 1996; 41(l):8391.
56. Keall P. 4-dimensional computed tomography imaging and treatment planning.
Semin Radiat Oncol 2004; 14(1):8190.
57. Tada T, Minakuchi K, Fujioka T, Sakurai M, Koda M, Kawase I, et al. Lung
cancer: intermittent irradiation synchronized with respiratory motionresults
of a pilot study. Radiology 1998; 207(3):779783.
58. Chen GT, Kung JH, Beaudette KP. Artifacts in computed tomography scanning of moving objects. Semin Radiat Oncol 2004; 14(l):1926.
59. Balter JM, Ten Haken RK, Lawrence TS, Lam KL, Robertson JM. Uncertainties in CT-based radiation therapy treatment planning associated with patient
breathing. Int J Radiat Oncol Biol Phys 1996; 36(l):167174.
Liver Tumors
195
60. Balter JM, Lam KL, McGinn CJ, Lawrence TS, Ten Haken RK. Improvement
of CT-based treatment-planning models of abdominal targets using static
exhale imaging. Int J Radiat Oncol Biol Phys 1998; 41(4):939943.
61. Lawrence TS, Ten Haken RK, Kessler ML, Robertson JM, Lyman JT,
Lavigne ML, et al. The use of 3-D dose volume analysis to predict radiation
hepatitis. Int J Radiat Oncol Biol Phys 1992; 23:781788.
62. Dawson LA, Normolle D, Balter JM, McGinn CJ, Lawrence TS, Ten Haken RK.
Analysis of radiation-induced liver disease using the lyman NTCP model. Int
63. Jackson A, Ten Haken RK, Robertson JM, Kessler ML, Kutcher GJ,
Lawrence TS. Analysis of clinical complication data for radiation hepatitis
using a parallel architecture model. Int J Radiat Oncol Biol Phys 1995;
31(4):883891.
64. Seong J, Park HC, Han KH, Chon CY. Clinical results and prognostic factors
in radiotherapy for unresectable hepatocellular carcinoma: a retrospective study
of 158 patients. Int J Radiat Oncol Biol Phys 2003; 55(2):329336.
65. Park HC, Seong J, Han KH, Chon CY, Moon YM, Suh CO. Dose-response
relationship in local radiotherapy for hepatocellular carcinoma. Int J Radiat
Oncol Biol Phys 2002; 54(l):150155.
66. Pirzkall A, Debus J, Lohr F, Fuss M, Rhein B, Engenhart-Cabillic R, et al.
Radiosurgery alone or in combination with whole-brain radiotherapy for brain
metastases. J Clin Oncol 1998; 16(11):35633569.
67. Chen JC, ODay S, Morton D, Essner R, Cohen-Gadol A, MacPherson D,
et al. Stereotactic radiosurgery in the treatment of metastatic disease to the
brain. Stereotact Funct Neurosurg 1999; 73(l4):6063.
68. Herfarth KK, Debus J. Stereotactic radiation therapy of liver tumors. Radiother Oncol 2003; 68(S1):S45.
69. Herfarth KK, Hof H, Bahner ML, Lohr F, Hoss A, van Kaick G, Wannenmacher M, Debus J. Assessment of focal liver reaction by multiphasic CT after
stereotactic single-dose radiotherapy of liver tumors. Int J Radiat Oncol Biol
Phys 2003; 57:444451.
70. Willemart S, Nicaise N, Struyven J, van Gansbeke D. Acute radiation-induced
hepatic injury: evaluation by triphasic contrast enhanced helical CT. Br J
Radiol 2000; 73(869):544546.
71. Schefter TE, Kavanagh BD, Timmerman RD, Cardenes HR, Baron A, Gaspar
LE. A Phase I trial of stereotactic body radiation therapy (SBRT) for liver
metastases. Int J Radiat Oncol Biol Phys 2005; 62:13711378.
9
of Lung Tumors
Robert D. Timmerman
Department of Radiation Oncology, Indiana University School of Medicine,
Bloomington, Indiana, U.S.A.
Jorn Wulf
Department of Radiotherapy, University of Wurzburg,
Wurzburg, Germany
INTRODUCTION
The purpose of stereotactic irradiation of tumors in the lung is improvement
of local tumor control by escalating the radiation dose. Simultaneously,
acute and late radiation toxicity must be kept to an acceptable level despite
the increased dose. These almost contradictory intentions are matched
together by decreasing the irradiated volume, which is achieved by maximizing efforts to ensure precision of radiation delivery and minimizing breathing mobility of the targets. Therefore, stereotactic irradiation of lung tumors
is best suited for patients who will benet from increased local tumor
control probability achieved by dose escalation. These criteria are fullled
in patients with node negative non-small cell lung cancer (NSCLC) stage I
(cTl-2 cN0 cM0), and in selected cases of stage II (cT3 cN0 cM0 without
central disease but inltration of a small part of the peripheral thoracic wall)
or patients with solitary or few lung metastases, who usually are selected for
surgical treatment.
197
198
Timmerman and Wulf
Nevertheless a signicant number of patients will not be suited for

surgery due to confounding medical conditions, or will refuse surgery
because of its invasiveness. For these patients, a minimally invasive treatment approach leading to similar local control rates by surgery is required.
Unfortunately doses of 6070 Gy, usually used in conventional fractionated
3D-conformal radiotherapy, lead to local control rates of only 3050% for
stage I disease and therefore could not meet this demand. Retrospective analyses and early data from dose escalation studies support the evidence that
increasing the dose will lead to improved local control rates. Compared to
dose escalation by conventional 3D-conformal radiotherapy, stereotactic
irradiation is performed in one or few fractions, with the advantage that
the efforts to achieve maximal setup accuracy and decrease target (breathing)
mobility can be feasibly optimized. Furthermore problems such as tumor cell
repopulation during a prolonged treatment time, which is often associated
with escalating doses by increasing the number of fractions alone, are minimized due to the hypofractionated concept.
Indications
The purpose of stereotactic irradiation of pulmonary targets is local control
of circumscribed tumors achieved by very high fraction doses of 2030 Gy
(single dose) or 8 6 Gy to 3 20 Gy (hypofractionation). These tumor
ablative doses are ideally restricted to the tumor itself (planning target
volume, PTV). Prophylactic or therapeutic irradiation of the loco-regional
lymph nodes is not feasible within the stereotactic approach. Due to these
methodical restrictions rst choice-indications for stereotactic irradiation of
pulmonary targets are small primary NSCLC stage I (cTl/2 cN0 cM0) with
low risk for lymphatic spread and solitary (or very few) lung metastases
(Table 1). Additionally, cT3 cN0 cM0 (stage II) tumors can be considered
for stereotactic irradiation if the tumor is peripherally invading the pleura
Table 1 Suggested Indications for Stereotactic Irradiation of Pulmonary Tumors
First choice:
NSCLC stage I cTl-2 cNO cMO
Solitary or <4 lung metastases (with controlled extrapulmonary disease)
Second choice:
NSCLC stage II cT3 cNO cMO
(only cT3-tumors with inltration of the peripheral pleura at the thoracic wall,
no central tumors)
Local recurrences of previous irradiated tumors
(target amenable for localized and volume restricted irradiation)
Stereotactic boost to intrapulmonary primary tumors
(during conventional RT/RChT of higher stage NSCLC)
Lung Tumors
199
at the thoracic wall. Centrally growing cT3 tumors should be avoided due to
the adverse late radiation response of the mediastinal structures, as discussed
later. In general the practice of patient selection should follow the considerations of thoracic surgeons to choose patients for treatment who will benet
from local tumor control quo ad vitam or at least symptomatically (avoidance
of bleeding or treatment of pain due to inltration of the thoracic wall).
In patients with pulmonary metastases, the benet from local control
of a particular metastasis has to be balanced against the risk of further
dissemination. One of these benecial situations in metastasized disease
might be observed in patients with isolated lung metastasis after pneumonectomy. In these patients even growth of a single lung metastasis will
increase the risk for rapid impairment of lung function.
Second-choice indications might be stereotactic radiotherapy of locally
recurrent NSCLC in previously irradiated patients or stereotactic boost
irradiation to intrapulmonary tumors, e.g., during primary radio- or radiochemotherapy of advanced stage NSCLC. Under the precondition that
organs at risk, such as spinal cord, trachea, and main bronchi or esophagus,
can be spared from this additional dose according to the amount of the previous dose, these patients might have a second chance or an increased
chance for local tumor control due to a precise and volume sparing therapy.
Patients referred for stereotactic radiation may include individuals
with very poor pulmonary function. In these cases, the risk of tumor
progression must be carefully weighed against the risk of therapy related
pulmonary compromise. Up to now there are no consistent data available
on how much even stereotactic radiotherapy is limited by impaired lung
function. Theoretically the risk for damage of functional lung tissue should
be dependent on the size of the target, location of the tumor (central vs. peripheral), and the assessment of the irradiated volume. In the authors experience, even patients with a FeV1 of less than 1 L could be treated without
negative impact on lung function. This might be due to the fact that the irradiated volume is restricted to the tumor, which again is not contributing to
lung function anymore. Nevertheless there is a risk of about 4% of symptomatic pneumonitis (see below), which mainly affects the functional lung
tissue. Therefore some authors restrict stereotactic radiotherapy to patients
with a FeV1 of 1 L and treat patients with a FeV1 of less than 1 L only exceptionally for very small volumes.
Currently the published results of stereotactic radiotherapy of lung
tumors are still based on reports of single institutions with limited target
numbers (n 1766). Therefore at this time surgical treatment should be
considered as treatment of rst choice in operable patients until larger
patient numbers and data from phase-III studies are available. But in
patients not amenable to or refusing standard therapy for medical or personal reasons, the stereotactic approach can be offered as a promising
treatment modality.
200
Timmerman and Wulf
Standard Treatment for Pulmonary Tumors Amenable

for Stereotactic Irradiation NSCLC Stage I/II
In most countries lung cancer is the leading cause of cancer deaths in males
and one of the leading causes of cancer deaths in females. About 80% of
lung cancer is NSCLC. While the mortality rate is slightly decreasing in
males, it is continuously increasing in females. In Germany the standardized
mortality rate is 42 per 100,000 in males, and 11 per 100,000 in females
(year 2000). In the United States the standardized mortality rate is even
higher (58/100,000 in males and 25/100,000 in females); in Japan it is lower
(30/100,000 in males and 8/100,000 in females). Comparing the incidence
rates of 67/100,000 (males) and 9/100,000 (females) reveals that most
patients with lung cancer will not be cured (1). A detailed description of
incidence and mortality rates and of treatment strategies for different tumor
stages in European countries over a period from 1978 to 1997 has recently
been published by Janssen-Heijnen and Coebergh (2).
The unfortunate prognosis of patients is mainly due to advanced
disease: Most tumors are diagnosed with loco-regional lymph node involvement, inltration of relevant structures (cT4) or already with distant metastases. Only about 1530% of tumors are diagnosed with limited disease
stage I (cT1-2, cN0, and cM0) or II (cTl-2, cN1 or cT3 cN0, and cM0) (3).
Nevertheless these early tumors are treated under curative intention by surgical resection leading to ve-year overall survival of 65% for patients with stage
I and 41% for patients with stage II disease (4). Five-year overall survival with
this therapy ranges from 60% to 90%, with lower survival in the United States
and Europe and higher survival in Japan (57). Mountain (5) reported 5-year
survival rates of 67% for cTl cN0 cM0, 57% for cT2 cN0 cM0, and 38% for
cT3 cN0 cM0. But survival rates depend on the type of surgical resection.
In a study comparing limited (wedge or segmental) resection vs. lobectomy
or pneumonectomy the survival rates were 59% for limited but 77% for radical resection (8). A randomized trial performed by the Lung Cancer Study
Group (9) to compare lobectomy to limited wedge or segmental resection
in cTl cN0 cM0 patients revealed a signicant increase of recurrence rate of
75% in the limited surgery group. The overall ve-year survival dropped
from 65% to 45%. These differences appear to be related to resection line
recurrences from inadequate margins associated with less extensive and
non-anatomical resections. More recent reports of selected patients treated
with wedge resections from Japan, however, would indicate that with modern
techniques and staging, the likelihood of close margins is low even with a
wedge resection. Indeed, the event-free survival in modern series with wedge
resection appears to approach that of lobectomy (7).
Nevertheless in many patients with NSCLC of limited stage the treatment decision will depend on the individual medical condition represented
by age, performance status, lung function, and other medical diseases.
Lung Tumors
201
According to this assessment and the results of staging procedures including

history (weight loss is an important prognostic factor), physical examination,
a CT scan of the thorax, abdomen, brain, cardio-pulmonary function tests,
and (desirable) a PET scan to detect occult tumor spread, the most appropriate treatment will be dened. If the patient is assumed to be medically inoperable, radiotherapy will be the treatment of choice.
Unfortunately conventional fractionated radiotherapy or 3D-conformal
radiotherapy of limited dose <70 Gy has only reached local control and veyear survival rates that are inferior to those achieved by surgery. The cause
of this might be partially due to a selection bias choosing the more favorable
patients for surgery, but there additionally is evidence that the radiation dose
might have been too low. Jeremic et al. (10) published an overview on results
achieved by conventionally fractionated radiotherapy: doses of 3080 Gy
to stage I/II NSCLC led to initial/isolated local failure rates of 1155%.
Five-year overall survival ranged between 6% and 45%. The authors concluded that normofractionated doses of at least 65 Gy or equivalent doses of
other fractionations are needed to achieve improved local control. Locally
uncontrolled tumor was the predominant pattern of failure. A similar analysis
of Sibley et al. reviewing publications on tumors treated with a median dose of
6066 Gy revealed that on average only 15% of patients will be long-term
survivors. About 25% will die on intercurrent disease, 30% on distant metastases, and another 30% on locally uncontrolled tumor alone (11,12). From
that data the authors derived the importance of dose escalation for further
improvement of local tumor control of stage I/II lung cancer in medically
inoperable patients. Despite clinical plausibility from these retrospective analyses no clear dosevolume relationship on local control could be derived.
Nevertheless there was a trend to superior local control rates for smaller
tumors (cT1 vs. cT2).
Starting from that insight several dose escalation protocols have been
inaugurated during the last few years. An overview on these studies can be
found at Belderbos et al. (13). Because overall treatment time may be a relevant prognostic factor these studies try to increase fraction dose or the number of fractions per time (CHART) to achieve dose escalation up to
equivalent doses >80 Gy. While long-term clinical results of these studies
are still pending, in all of these approaches volume restriction to achieve
an acceptable rate of toxicity is an issue. Therefore in small tumors (stage I)
elective irradiation of regional lymph nodes is omitted as supported by reports
of Slotman et al. (14), Krol et al. (15), and Bradley et al. (16).
Both strategies of these dose escalation studiesincrease of fraction
dose to keep a short overall treatment time and decrease of the irradiated
volumeare inherent characteristics of the concept of stereotactic radiotherapy. The only restriction of the stereotactic approach compared to normofractionated dose escalation studies is the exclusion of tumors close or
adjacent to mediastinal organs at risk due to the very high fraction doses.
202
Timmerman and Wulf
Therefore stereotactic irradiation of pulmonary targets is just another

concept of dose escalation including all tools available in modem radiotherapy: patient support and external reference systems as stereotactic body
frames, 3D-conformal dose calculation and dose distributions achieved by
multiple elds shaped by multileaf collimators, breathing control devices
and CT verication prior to radiotherapy.
LUNG METASTASES
Oncologically, metastatic disease in the lung denes the systemic spread of
the disease and therefore limits the role of local treatment and local tumor
control. Nevertheless under certain circumstances patients will benet even
from local control of single metastasis as surgical data show. The clinical
results of a large study of the International Registry of Lung Metastases
treating 5206 cases from 18 departments in Europe, United States, and
Canada with metastasectomy were reported (17). The primary tumor was
epithelial in 2260 cases, sarcoma in 2173 cases, germ cell in 363 cases, and
melanoma in 328 cases. In 2383 cases single and in 2726 cases multiple
metastases were resected, in 88% complete. After a median follow-up of
46 months the 5-, 10-, and 15-year survival was 36%, 26%, and 22% after
complete and 13%, 7%, and 0% after incomplete resection. The multivariate
analysis showed better prognosis for patients with germ cell tumors, diseasefree survival of >36 months after treatment and single metastasis. Nevertheless even in patients with >3 metastases ve-year survival was 27% if
complete resection could be obtained (17). A recent overview of results of
different histologic subtypes is given by Davidson et al. (18), who also
describe the criteria for patient selection for metastasectomy:
1. the patient must be able to tolerate the planned procedure,
2. the patients pulmonary function tests indicate sufcient reserve to
compensate for resected lung tissue,
3. the site of the primary tumor must be controlled,
4. no evidence of extrapulmonary disease (or non-uncontrollable
extrapulmonary disease), and
5. no better therapy is available.
The last point is the rationale for stereotactic irradiation of lung metastases. It might be superior to metastasectomy because of its non-invasive character and, depending on the size of the target, the sparing of functional lung
tissue allowing treatment of patients even with impaired lung function as, for
example patients, with lung metastasis after pneumonectomy. Nevertheless
stereotactic radiotherapy of pulmonary metastases rst has to prove its
efcacy to achieve local control rates comparable to those of metastasectomy.
Appropriate thoracic targets are well demarcated and feasibly dened
on CT or MRI. Patients with malignant pleural or pericardial effusions or
Lung Tumors
203
diffuse miliary involvement of the lung (e.g., in some disseminated

presentations of broncheoalveolar cancer) are not appropriate candidates.
There is a controversy over the number of lesions capable of being treated
based on both oncological and technical reasons. While treatments may
be technically possible, patients with multifocal disease are at greater risk
for occult dissemination appearing after treatment, obviating any benet
of controlling existing lesions. Furthermore, dose fall-off from adjacent
lesions may overlap creating pockets of substantial unintended dose within
normal tissue. As such, treatment-related toxicity of multifocal disease may
be supra-additive as compared with toxicity associated with single lesion
treatments. With these limitations in mind, extracranial radiation treatment
seems best suited for solitary lesions, and few centers will treat more than
three to four lesions in total depending on the clinical circumstances.
In summary, as for lung metastases, surgical results for stage I NSCLC
underline the importance of local control for prognosis of these patients.
Local control can be achieved by sufciently high radiation doses to the
complete tumor volume. The surgical results have important implications
toward stereotactic body radiation therapy since tissue destruction for that
therapy more closely resembles a wedge resection than a lobectomy. If
meaningful doses are delivered, recurrence will be dependent on whether
the entire extent of the disease is encompassed at time of treatment. Therefore after careful selection of patients, hopefully the results from stereotactic
irradiation will reach equivalency to surgically treated patients.
RADIOBIOLOGY
Normal Tissue Considerations
The primary function of the lung is to exchange oxygen for carbon dioxide
between the terminal airways (alveoli) and the blood (respiration). The lung
serves to deliver the oxygen-rich air to the alveoli via a series of branching
airways. Airow within these branching airways (bronchi and bronchioles)
is powered by pressure gradients generated by the diaphragm and chest wall
musculature that are transmitted throughout the lung via the elastic structure of the lung parenchyma. At the level of the alveoli, the lung has a tremendous amount of inherent redundancy, with each neighboring alveoli/
blood capillary complex functioning independently and doing basically
the same activity (exchanging oxygen for carbon dioxide). The lung is a
large organ and most people have a great deal more respiratory capacity
than is actually required, constituting a reserve. Throughout life, this reserve
may be depleted especially by activities that diffusely damage the parenchyma, like cigarette smoking. Surgeons contemplating a lung resection
rst try to quantify the amount of reserve in a given patient by measuring
surrogate markers (e.g., pulmonary function tests). The surgeon will then
204
Timmerman and Wulf
determine whether removal of a certain fraction of lung will leave the patient
with enough respiratory function to carry on daily activities. All in all, these
considerations account for the inherent function of the lung (respiration), the
organizational structure of the lung (branching airways leading to terminal
alveoli), the redundancy of lung function (e.g., the left lung carries out the
same activity as the right lung), and appreciation and quantication of the
additional capability (reserve) inherent in the lung. These same considerations are paramount to understanding normal tissue and host responses after
irradiation of the lung.
In describing normal tissue changes after therapeutic radiation,
Wolbarst et al. (19) described a model where tissue is broken down into relatively small functional subunits (FSU). These FSUs are composed of an
organized population of differentiated cells and a smaller population of clonagenic (stem) cells capable of replenishing the differentiated cells. Wolbarst
divided FSUs into two general groups: (1) structurally dened units with discrete anatomical structure, and (2) structurally undened units characterized
by a monotonous structure without anatomical boundaries. In this model,
damage from radiation was related to cumulative damage of constituent
FSUs. According to this model, each alveolus/capillary complex within
the lung constitutes a structurally dened FSU. It is presumed that after
delivery of radiation, both differentiated and clonagenic cells are damaged,
some lethally. In order for surviving clonagens to rescue the damaged tissue, they must rst migrate to the damaged area and then divide into differentiated cells capable of performing the tissues function. The migration of
rescuing clonagens can occur within an alveolus but not between two adjacent alveoli, even though the two adjacent alveoli are in close proximity. As
such, if all clonagenic cells within a single alveolus are damaged, all functional capability of that alveolus will be lost. This is in contrast to a structurally undened FSU, like the mucosa of the esophagus, where clonagens are
free to migrate long distances to rescue damaged epithelium.
Again considering the Wolbarst model, one can identify a threshold dose
beyond which all clonagens within a particular structurally dened FSU are
incapable of rescue and the FSU will become totally dysfunctional. Moreover,
delivering an additional dose beyond the threshold dose within a particular
volume containing a dened number of FSUs will not increase the dysfunction
since all function is already lost at the threshold dose. For the lung, this threshold dose is probably quite low, in the range of 1520 Gy given in 2 Gy fractions. The fact that fairly large volumes of lung can be irradiated by these
doses without untoward consequences attests by the large functional reserve
inherent in the lung.
Tissues that are made up predominantly of structurally dened FSUs
are called parallel functioning tissues (including peripheral lung, peripheral
kidney, peripheral liver, etc.) and occur within organs that are characterized
by redundancy of function and large inherent reserves. In contrast, tissues
Lung Tumors
205
that are made up of predominantly structurally undened FSUs are called

serially functioning tissues (including the gastrointestinal tract, large airways,
spinal cord, etc.) and occur within organs that involve a chain of function.
Certainly, some tissues do not t well into either category (e.g., bone marrow).
In treating a lung cancer in the peripheral lung, a parallel functioning tissue,
high tumor doses are required to control the clonagenic capability inherent
to most lung cancers. Adjacent lung tissue will be exposed to relatively the
same dose as the tumor. According to the critical volume model proposed
by Yeas and Kalend (20), any dose beyond the threshold dose dened above
will not add additional toxicity to a given volume. For parallel functioning
tissues, the organ will become dysfunctional if a critical volume getting
the threshold dose is exceeded. Therefore, according to this model, organ
dysfunction is not avoided by limiting the magnitude of dose beyond the
threshold but rather by limiting the volume exposed to any dose beyond
the threshold. This critical importance of limiting volume, with more attention than limiting dose, in order to spare organ dysfunction is the hallmark
radiobiological principle of extracranial stereotactic radiation therapy.
The biggest shortcoming of both the Wolbarst and Yeas models
relates to the fact that an organ such as the lung is really composed of both
parallel and serial tissue entwined in proximity to each other. The actual
organization of the lung is similar to a tree or bush with a very large trunk
(trachea) branching into large main branches (mainstem bronchi), branching further into smaller branches and twigs (lobar bronchi and bronchioles),
and nally into terminal buds or leaves (alveoli/capillary complexes). All of
the airways described above are serially functioning tissues since air is being
directed along a single path as a chain of function and the clonagens within
the airways are situated in the epithelium without anatomical boundaries. In
contrast, the alveoli/capillary complexes are parallel functioning tissues
with basement membranes and septa separating one alveolus from another
limiting clonagen migration.
The type of anatomical arrangement seen in the lung may be referred to
as a branching tubular structure and is in contrast to organs of the GI tract in
which the lumen follows a single straight path known as linear tubular structure. An important consideration in relation to extracranial stereotactic
radiation therapy between these structures follows because if the dose is
intense enough to totally disrupt the function of the serial functioning component (e.g., the bronchus or esophagus), then all downstream functioning
tissue will be lost as well (even if they were not irradiated) via collapse of
the lumen. In a branching tubular structures, such damage will result in a
collapse of the particular branch affected, not the entire organ. The same
damage to a linear tubular structure will obstruct all downstream function
of the organ (e.g., complete bowel obstruction), a more catastrophic problem
for the patient. In the case of lung treatment, with potent doses of radiation
delivered to a bronchus or bronchiole, distal collapse or atelectasis will occur
206
Timmerman and Wulf
which may likely be permanent. As long as the volume lost is less than the
organ reserve for the particular individual, no signicant symptomatic
toxicity will result. If the lost volume is larger than the patients reserve,
the patient will have symptomatic respiratory decline. As such, with potent
treatment doses (e.g., ablative doses), the loss of functional lung tissue
may be larger than the actual volume irradiated beyond the threshold dose.
It may still be reasonable to use such a strategy in order to control tumor
proliferation; however, the treating physician must be aware of these two
components of lung dysfunction (direct radiation damage to the volume
irradiated and subsequent distal collapse of non-irradiated lung) when
formulating the treatment plan.
Tumor Control Considerations
According to the models of Douglas and Fowler (21), the logarithm of
tumor clonagenic survival as a function of dose may be approximated by
a truncated power series known as the linearquadratic model. Various
physical explanations have been offered as to why the curve would not be
linear, including that double strand DNA damage constitutes an irreparable
defect while single strand breaks may be repaired. But, at any rate, with
rather low doses per fraction (i.e., up to 6 Gy per fraction), tumors have
an enhanced ability to withstand the damaging effects of radiation. Beyond
this dose per fraction, tumor kill has an exponential relationship with dose,
implying that tumor repair mechanisms are overwhelmed.
In addition to DNA repair as a mechanism for poor local control,
radiobiologists have observed the ability of remaining viable tumor cells
to increase their rate of cell division after being exposed to radiation. This
accelerated repopulation is considered to be one of the most signicant
factors resulting in failure of treatment. Since it takes the cell some time
to initiate this response (i.e., many days to weeks), the most viable therapeutic counter to this inherent tumor defense is to deliver all of the radiation
very quickly before repopulation is initiated.
Inherent radioresistance is related to many factors. One of the most
difcult factors to overcome is tumor hypoxia. Oxygen is required to
x damage caused by especially photon radiation. In addition, poorly
oxygenated cells are probably not actively dividing, placing them in cell
cycle portions less sensitive to radiation. The general strategy in radiation
oncology for overcoming tumor hypoxia has been to protract the radiation
delivery. The basis for this was that tumors that have out-grown their
blood supply due to large size would shrink and effectively get closer to a
vascular supply. During the later portions of the protracted course, the
tumor would theoretically be well oxygenated. Certainly, regardless of the
theory behind this strategy, protracted fractionated radiation therapy has
not been particularly effective at controlling large necrotic epithelial tumors.
Lung Tumors
207
Furthermore, experiments using miniature oxygen probes have indicated

that areas of signicant tumor hypoxia are not necessarily in the central core
of the tumor and, in fact, migrate within the tumor as a result of dynamic
vascular changes (22). In this context, lengthy protraction of radiation
may not be a viable solution for hypoxia while still a detriment in overcoming repopulation.
Cell cycle effects may also inuence radiosensitivity. Certain portions
of the cell cycle, like mitosis and G2, are thought to be quite sensitive while
other portions, like late S-phase and G0, are considered radioresistant.
Again, the general response to overcoming cell cycle effects has been to
protract the fractionation of radiation therapy. It is assumed that cells will
be committed to moving through the cell cycle. Cells in resistant phases
will become sensitive later in the course of radiation according to this
theory. The optimal length of protraction of radiation is unclear. Cell cycle
times vary, even within a single tumor. At any rate, this feature of radioresistance has implications as to whether a stereotactic treatment regimen
should be single fraction vs. a few separated fractions. Further research is
needed to resolve this issue.
RADIATION TOLERANCE OF PULMONARY TISSUES
Within the lung itself, there are a variety of tissues that possess unique
radiation tolerance characteristics, namely, the airways (both large and
small functioning as serial structures), the alveoli/capillary complexes (functioning as parallel structures), and the arterial and venous network (likely
functioning as serial structures).
The classic tolerance-dening toxicity for conventional radiation
delivery in most prospective trials has been pneumonitis. Radiation pneumonitis is a subacute (weeks to months from treatment) inammation of
the end bronchioles and alveoli. The clinical picture may be very similar
to acute bacterial pneumonia with fatigue, fever, shortness of breath, nonproductive cough, and a pulmonary inltrate on chest X-ray. The inltrate
on chest X-ray should include the area treated to a high dose, but may
extend outside of these regions. The inltrates may be characteristically
geometric corresponding to the radiation portal, but may also be illdened. Pneumonitis is generally treated with corticosteroids. It may resolve
over time, but many progress into brosis with linear opacities appearing
on imaging studies. Based on the pattern of inltration during the initial
presentation of pneumonitis, it is most likely that pneumonitis is a toxicity
related to damage to end bronchioles and alveolar/capillary complexes. As
such, this damage is to parallel functioning tissue and is most likely more
volume dependent than dose dependent. Furthermore, the incidence of
pneumonitis likely occurs at a relatively low threshold dose, in the range
of the equivalent of 1520 Gy in conventional fractionation. As such, high
208
Timmerman and Wulf
dose per fraction stereotactic body radiation therapy likely causes pneumonitis within the high dose rim surrounding the tumor target in most cases.
Whether pneumonitis becomes symptomatic depends on the volume of
tissue that exceeds this threshold dose. Because normal lung volume is
relatively small in stereotactic treatments with little or no prophylactic irradiation, it would be potentially a less likely outcome as compared to conventional radiation therapy. Indeed, in the Indiana University phase I dose
escalation trial, symptomatic radiation pneumonitis occurred relatively
infrequently despite very potent effective dose delivery (23).
Conventional radiotherapy commonly causes large serially functioning
airway irritation, such as cough, but rarely dose limiting toxicity. In contrast, high dose stereotactic body radiation therapy treatment schemes
may cause signicant large airway damage by both mucosal injury and ultimate collapse of the airway. This loss of functional capacity results from
both mucosal sloughing, cartilage damage, and peribronchial brosis; all
effectively causing bronchial stenosis. In turn, bronchial stenosis will in
many cases lead to distal atelectasis of lung parenchyma downstream from
the obstruction. This loss of lung function appears to mostly affect oxygenation parameters including diffusing capacity for carbon monoxide (DLCO),
arterial oxygen tension (pressure) on room air (PO2), and supplemental
oxygen requirements (FIO2) (23). Because the degree of this airway injury
toxicity is related to the proximity of the target to proximal trunks of the
branching tubular lung structure, great care should be taken when considering treatment to tumors near the hilum or central chest. More protracted
fractionation schedules for central tumors may facilitate treatments in these
locations at the expense of potentially less effective tumor control.
Unexpected toxicity may also occur when treating central chest target
relating to toxicity to mediastinal structures, including esophagus and heart.
While acute and sometimes severe esophageal toxicity is commonly seen
after conventionally fractionated radiation for lung cancer, most of the
injury is self-limiting and resolves after treatment. After high dose stereotactic body radiation therapy, esophageal strictures may form as a late effect.
Another more unique toxicity from stereotactic body radiation therapy
relates to pericardial injury. In the Indiana University phase I study, several
patients with tumors adjacent to the heart had asymptomatic pericardial
effusions, while one patient treated at the highest dose level had a large
and symptomatic pericardial effusion that required surgical intervention
to resolve (unpublished data).
Most reports of stereotactic body radiation therapy do not include
long-term follow-up data. As such, there may be unexpected toxicities that
need to be recognized, monitored, and evaluated. Particularly with large
doses per fraction there may be unexpected injury related to nerve tissue
and vascular tissue. Ideally, the dose to brachial plexus, spinal cord,
phrenic nerves, and intercostal nerves will be kept low via prudent
Lung Tumors
209
treatment planning. Furthermore, avoiding large blood vessels in the

central chest may be reasonable as well. While there have not been reports
of vascular wall damage including aneurysms and stulas with hemoptysis
or internal bleeding, these events may only manifest after many years of
follow-up.
With a paucity of long-term data relating tissue effects after large dose
per fraction radiation, it is difcult and somewhat dangerous to identify
specic normal tissue tolerances. Nonetheless, in order for thoughtful investigation to proceed, a starting point must be established. The Radiation
Therapy Oncology Group in the United States has developed a protocol
for using stereotactic radiation to treat early stage lung cancer. The prescription dose to the margin of the planning target volume (PTV) for this protocol that treats tumors up to 5 cm in dimension is 60 Gy total over three
fractions (20 Gy per fraction). A committee of experienced radiation oncologists, physicists, and biologists has established organ dose limits for this
protocol based on limited institutional follow-up and linearquadratic
conversions of known dose tolerance parameters from conventional radiation fractionation schemes. The tolerances for this protocol are shown in
Table 2 for several critical organs. These are absolute limits relating to a
point rather than a volume. Obviously, a goal of radiation dosimetry for planning stereotactic lung radiation therapy should be to minimize the volume of
normal tissue, even getting lower doses than those listed in Table 2. It must
be emphasized that these tolerances gures have not been validated with
long-term follow-up.
A proposed tolerance of the lung itself is not identied in Table 2.
Based on the critical volume model of Yeas described above, it is assumed
that the lung within the PTV exceeds tolerance and is no longer functional
after high dose per fraction stereotactic radiation therapy. A dose fall-off
region exists outside of the PTV, the volume of which depends on the size
of the PTV, the location of the PTV within the chest, the quality of the radiation dosimetry (e.g., number of beams, beam arrangements, radiation
energy, etc.), and the type of radiation (e.g., photon vs. proton, etc.). This
Table 2 RTOG Proposed Radiation Tolerances of Thoracic Normal Tissue
Organ
Spinal cord
Esophagus
Ipsilateral brachial
plexus
Heart
Trachea and ipsilateral
bronchus
Volume
Dose
Any point
Any point
Any point
18 Gy total over three fractions

Any point
Any point

210
Timmerman and Wulf
dose fall-off region, also called the gradient region, constitutes unintended
radiation exposure and should be kept as small as possible. The lung tolerance criteria used in conventionally fractionated radiotherapy, such as the
percentage volume receiving 20 Gy (V20), do not lend themselves to stereotactic body radiation therapy since these volumes are already relatively small.
For the RTOG study, it is required that the ratio of the prescription isodose
to 50% of the prescription isodose (which occurs in normal tissue) be no
greater than 3.2. Considering then the maximum lesion size treated (5 cm),
these constraints would allow no more than 320 cc of normal lung (excluding
PTV) to exceed 30 Gy total over three fractions (10 Gy per fraction). It
should be possible to keep this volume of normal lung considerably less
for smaller lesions.
Treatment Delivery
Patient Immobilization and Target Reproducibility
The purpose of stereotactic radiotherapy of pulmonary targets is improved
local tumor control achieved by dose escalation and volume restriction.
Therefore setup inaccuracy and target mobility have to be minimized as
much as possible.
Several immobilization devices have been developed during the last
decade. All of them rely on a vacuum pillow, which is individually molded
to the patients body and xed to a stereotactic frame. The frame itself is
not only an immobilization device but also an external reference system,
which allows identication of the isocenter by 3D-stereotactic coordinates.
Although a stereotactic treatment might be performed without a dedicated
immobilization device (2428) the time for verication of the correct target
position and irradiation itself of fraction doses up to 30 Gy often lasts for
3060 min. During that time sufcient immobilization has to be ensured if
the efforts for treatment precision should not be diminished by (uncontrolled) patient motion during treatment.
The other important factor which has to be addressed is breathing
mobility. Uncontrolled breathing mobility of pulmonary tumors ranges up
to more than 20 mm (29,30). The amount of mobility is related to the target
location in the lung with larger mobility in the lower lobes. Breathing mobility can be evaluated by uoroscopy or by CT scans. The CT-evaluation
relies on dynamic examination of the tumor at the same couch position
during some breathing cycles. The change of the axial tumor shape represents
the amount of mobility and can be measured directly by comparing
the different slices during the breathing phases. The longitudinal mobility
can be measured by multi-slice technique or estimated by comparison
of the evaluated slice level to slices cranio-caudal of that level. The CT
evaluation has the advantage that even small targets and targets covered by
other structures such as heart or diaphragm in uoroscopy are clearly visible.
Lung Tumors
211
If breathing mobility of the target exceeds 5 mm in any direction

attempts to reduce these motions are performed. While some groups irradiate patients just by shallow breathing with oxygen support (25,26), most
authors use a mechanical breathing control device. It is attached to the
immobilization frame and consists of a template pushed into the patients
epigastrium to a reproducible amount. By the use of this very easy-to-handle
system, abdominal pressure is increased leading to decreased mobility of
the diaphragm and secondary reduced target mobility. This technique
allows reduction of breathing mobility down to 5 mm (3136). Although
most patients tolerate this technique very well, in cases such as patients with
very poor lung function (e.g., after pneumonectomy), the effect is limited. In
these patients breathing mobility has to be evaluated carefully and taken
into account for PTV denition with enlarged margins.
Because increasing volume due to increased security margins is clearly
not desirable, especially in patients with poor lung function, more advanced
techniques have been developed. While jet ventilation under general anesthesia (37) needs high logistical efforts and to some extent contradicts the
non-invasive character of the stereotactic approach, other authors have
inaugurated active breathing control systems or breathing-triggered radiotherapy. If performed properly, breathing mobility intrafractionally can be
reduced to almost zero (38,39). A still unresolved problem is inter-fractional
reproducibility, which can be again up to 5 mm (39). To resolve this problem
further evaluation supported by technical improvements, such as CT verication in the treatment room prior to irradiation or cone-beam CT, has to be
performed. Another disadvantage is prolongation of treatment time due to
restriction of irradiation to dened phases of the breathing cycle. An extreme
of this approach is irradiation of lung tumors by the CyberKnife. The tumor
is marked by intralesional gold seeds and irradiated by a robot-like accelerator, which calculates the actual target position from real-time at-panel
uoroscopy. Nevertheless the pneumothorax-rate from gold seed implantation was 15% (3/21 patients) and an irradiation session of a single fraction
treatment lasted for four hours at mean (26 hr) (40).
The accuracy of target reproducibility in a stereotactic body frame
using oxygen supported shallow breathing (26,28) or mechanical breathing
control (33,35) ranges from 3.1 to <5 mm in axial and 4.45.5 mm in longitudinal direction (median, for both 1 SD) evaluated by comparison of CT
data from treatment planning and CT verication prior to treatment.
According to these results most groups use security margins for PTV
denition of 5 mm in axial and 510 mm in the longitudinal direction added
to the CTV.
For denition of security margins the absolute target reproducibility
independent on its origin from setup-inaccuracy or breathing mobility
is the relevant parameter. Nevertheless for the process of minimizing
setup-inaccuracy and for improvement of skills, information on the
212
Timmerman and Wulf
contribution of both factors separately is necessary. Setup accuracy can be

measured by comparison of non-mobile bony structures relative to an external reference system, such as the stereotactic frame. This can be evaluated
from the two different CT scans performed for treatment planning and for
CT verication prior to irradiation. According to our own analysis this setup
inaccuracy is about 2 mm: SD in lateral direction was 1.9 mm, AP 1.8 mm, and
longitudinal 2.0 mm leading to a 3D-vector of 2.6 mm (35). These data were
conrmed by Yenice et al. (41), who evaluated a self-constructed, non-invasive
frame for stereotactic irradiation of paraspinal tumors.
Isocenter and Target Verification
The described data on target reproducibility are numbers representing the
median or 1 SD; therefore it will be representative for the majority of cases.
Nevertheless there will also be extremes of deviation, which will not be
covered by the usual security margins of 5 mm in axial and 510 mm in longitudinal direction. In our own analysis target deviation (including setup
inaccuracy and target mobility) exceeded 5 mm in 16% (AP), 12% (lateral),
and 9% (longitudinal) of targets. A deviation of more than 10 mm was
observed in 2% in the anteroposterior and lateral, respectively, and of 6%
in longitudinal direction (35). Target miss due to deviations of >5 mm might
not only lead to signicant underdose of the tumor but also to eventually
dangerous overdose to organs at risk due to the very high fraction doses used
in stereotactic radiotherapy. Therefore major target deviations should be
recognized and corrected prior to irradiation.
For that purpose three different approaches are used:
1. isocenter verication relative to bony structures by comparison of
DRRs to beam views or portal images,
2. CT simulation prior to irradiation at the CT unit with subsequent
transport of the patient to the linac,
3. CT verication on the treatment couch at the linac.
Isocenter verication relative to bony landmarks is the easiest and most
widespread method to control accuracy of radiotherapy. Nevertheless by using
this method it is assumed that the deviation of the (often invisible target) is
represented by deviation of bony reference structures. To analyze this assumption we evaluated the congruence of deviation of bony reference structures to
different types of targets, such as soft tissue targets xed to other structures or
mobile soft tissue targets, with or without use of the breathing control achieved
by abdominal pressure. Because the security margins used for PTV denition
were 5 mm at least in axial direction, congruence of the target to bony structures was assumed if the difference of deviation of both structures was less than
5 mm. This was the case in 80% of xed soft tissue targets but only in 33% of
mobile soft tissue targets (37.5% without breathing control and only 28.5%
Lung Tumors
213
with breathing control) (35). From these results it was concluded that
bony reference structures are not reliable to control correct target reproducibility within the security margins of 5 mm. Therefore, many centers use
CT verication prior to irradiation to control the correct isocenter position
in the target as is common practice in intracranial stereotactic radiotherapy.
While CT verication would be most appropriate directly at the treatment couch without subsequent transport of the patient from the CT unit
to the linac, this opportunity is not available at most institutions at this
time. Nevertheless even if it is performed outside the treatment room CTverication should allow for evaluation of target reproducibility.
To conrm target reproducibility after CT-simulation over the complete
target volume and the complete course of three treatment fractions we analyzed the data of 60 CT-verications in 22 pulmonary targets (42). For that
purpose the anatomically corresponding isocenter slices of the planning-CT
and the three verication-CTs of each patient were matched to each other
using a digital matching tool of the 3D-treatment planning system. The
CTV segmented in each verication CT was matched into the planning study
and a DVH for this volume was calculated using the original treatment plan.
Major deviations at any position of the CTV from the verication study
should result in a decrease of dose coverage of the simulated CTV, if the
deviation exceeds beyond the PTV-related reference isodose. As a result, in
only three of 60 CT verications (5%), the proportion to the CTV beyond
the reference isodose exceeded 5%. Two of these major deviations were
noticed in one patient treated for a lung metastasis in the left lower lobe after
pneumonectomy. It is concluded that target reproducibility conrmed and
eventually corrected due to CT-verication is accurate. Nevertheless for single
patients treated under difcult conditions eventually increased security margins or more advanced techniques for breathing control have to be used.
Target Definition
Most reports on clinical results of stereotactically irradiated lung tumors do
not focus on target denition in detail but give only information on security
margins for PTV denition added to either the GTV or CTV. Therefore the
practice of target denition might be potentially inhomogeneous among the
different groups working on stereotactic irradiation of lung tumors. From
surgical data on limited wedge or segmental resection it can be derived that
not only the macroscopic tumor should be treated to achieve high local control rates. Therefore the GTV should include the small spiculae often seen in
the periphery of the tumor and eventually the parts of inltrated pleura. For
this purpose target denition in the lung window (e.g., 1600, 400 HU) is
preferred. In targets close to mediastinal or hilar structures, i.v. contrast
eases the differentiation of tumor to blood vessels. To this GTV 23 mm
of potential microscopic disease may be added to achieve the CTV. Another
214
Timmerman and Wulf
Figure 1 CTV-denition and conformal dose distribution in a 43-year-old male

with primary lung cancer cT2 cN0 cM0 (adenocarcinoma grade II) in the left upper
lobe medically inoperable due to severe heart disease. The CTV was 45 cm3, the PTV
was 100 cm3. The tumor was treated by 3 10 Gy to the PTV-enclosing 100%-isodose
(the inner orange isodose) with normalization to 150% at the isocenter. For CTV
denition not only the macroscopic tumor but also the small tumor extensions into
the periphery have to be included into the target volume (the numbers in the coronal
and sagittal reconstruction show the point dose in percent to the prescribed fraction
dose of l0 Gy). (See color insert.)
5 mm in axial and 510 mm in longitudinal direction are added for PTV denition, depending on the results of individual evaluation, e.g., of breathing
mobility. An example for target denition with the consecutive 3D-dose distribution is shown in Figure 1.
Treatment Planning
For treatment planning usually CT slices of 35-mm thickness with or without
i.v. contrast medium are sufcient. In general and especially in patients with
impaired lung function the planning study should cover the complete lung
to allow for assessment of the amount of lung irradiated. The planning study
can be performed as an incremental or spiral scan, but it must be ensured that
the target is not randomly scanned in an extreme phase of the breathing cycle.
Therefore evaluation of breathing mobility and eventual use of breathing control techniques should be evaluated prior to the denite planning study.
The 3D-conformal treatment planning depends on dose prescription,
which differs considerably among the published results. Some groups
prescribe their dose to the isocenter (24,4345), others to the PTV-enclosing
isodose (23,25,27,31,32,36,37,40,46). Some groups use homogeneous
dose distributions (44), some allow slight inhomogeneity with the 80%isodose encompassing the PTV (23,25,26,37,40,46), and others prefer
Lung Tumors
215
increased inhomogeneity such as prescribing the dose to the PTV-enclosing

65%-isodose (which is equivalent to dose prescription to the PTV-enclosing
100%-isodose with normalizing 150% at the isocenter) (3136).
There are no detailed publications on treatment techniques and beam
arrangement. Most groups use 59 coplanar static elds to achieve a conformal dose distribution encompassing the PTV. Especially in larger volumes
rotational beams with leaf adaptation every 2040 are useful to avoid
triangulated high dose spikes from overlapping parts of static elds. Under
some circumstances non-coplanar beams or the use of wedges have to be
considered, but the benet of optimized dose distribution has to be balanced
against prolongation of irradiation time.
The quality of a dose distribution can be evaluated by comparison of
the volume of the reference isodose to the volume of the PTV (conformity
index 1, if these volumes are identical). Because the volume of the reference isodose does not necessarily cover the PTV but also potentially normal
tissue outside the PTV, the conformity index has to be related to the target
coverage (TC) of the PTV. The TC represents the amount of the PTV covered by the reference isodose. According to Vant Riet et al. (47), the product of TC and conformity index results in a conformation number (CN),
which optimally is one if both parameters are one. The evaluation of our
own treatment plans for 22 lung tumors achieved by static or rotational
beams revealed a TC of the PTV by the reference isodose of at mean
96% 2.3% (SD) and a conformity index of at mean 0.51 0.13 (SD).
The resulting CN was at mean 0.73 0.09 (SD) (48).
Tissue Heterogeneity Issues
Important for the quality and reliability of treatment plans is the dose
calculation algorithm. Recent studies revealed an energy dependent potential dose decrease at the tumor margin of up to 1020% if simple dose
calculation algorithms such as the widespread pencil-beam algorithm are
used instead of more sophisticated models such as the point kernel-based
collapsed cone algorithm (49,50,55).
The most obviously unique aspect of lung tissues vs. tissues elsewhere
in the body in regard to photon dosimetry is the spectrum of densities across
the organ. Within the lung and chest are areas of air density (large airways),
water density (mediastinum and tumor bearing tissue), bone density (chest
wall), and intermediate density (lung parenchyma). The density of the lung
parenchyma may vary between patients from 0.15 to 0.35 g/cm3, depending
on host factors like history of emphysema, etc. This heterogeneity of density
has signicant implications toward dose deposition, especially in areas of
inherent electronic dysequilibrium like the edge of a tumor.
As a beam of photon radiation passes through a patient toward a
tumor target, it rst is attenuated by the chest wall or perhaps mediastinum.
216
Timmerman and Wulf
The physics of this interaction are well described. However, once the beam
passes into lung parenchyma, there is considerably less energy loss through
attenuation. Effectively, more photon uence will be delivered to the edge
of the tumor resulting in higher central tumor doses than predicted by
algorithms considering all tissues to have water density (1.0 g/cm3). At the
edge of the tumor, a secondary buildup will occur resulting in relative underdosing as compared to what would be predicted by algorithms considering
all tissues to have water density. These effects have implications both to
tumor control and toxicity.
Beam energy also dramatically inuences the dose buildup characteristics
at the edge of a tumor. While higher energy beams will deliver more radiation
uence to the core of a tumor target relative to the skin dose compared to low
energy beams, the deeper location of achieving equilibrium (Dmax) may result
in signicant underdosing of the tumor margin. Rather than use high energy
beams to overcome problems with skin toxicity, it is probably more prudent
to add additional lower energy beams, thereby spreading out the entrance dose
among all beams.
Stereotactic body radiation therapy typically involves the use of many
beams with relatively small apertures. Heterogeneity effects are magnied by
such arrangements in that eld edges are very close to target edges. Strikingly
steep dose gradients result in the region, where accurate prescription dose must
be appreciated. The edge of a tumor can be effectively missed or underdosed
if these effects are improperly characterized. With stereotactic radiation,
greater care must be taken to commission beams with small apertures, especially toward the edge of the elds. Otherwise the minimum tumor dose, which
will nearly always occur at the edge of the tumor, will be mischaracterized.
Historically, treatment planning software systems did not account for
these heterogeneity effects. As such, knowledge of radiation response reects
an inaccurate characterization of dose both in terms of tumor control and
toxicity. Newer generation planning software makes approximations for
both attenuation and scatter effects in heterogeneous tissues. The attenuation algorithms from vendor to vendor consistently account for this effect.
However, the scattering corrections are not consistent and may lead again
to signicant differences in reported tumor doses from center to center,
especially at the edge of the target. Monte Carlo dosimetry will likely overcome these difculties, but is generally not available for treatment planning.
Until these obstacles are overcome, it is important that investigators report
the nature of their institutions calculation, including algorithms used for
calculating dose to the target margin.
TREATMENT OUTCOME
More published outcomes have been available for treating lung tumors with
stereotactic body radiation therapy than any other site. To the credit of the
Lung Tumors
217
investigators, much of this work describes prospective trials. The rst

authors inaugurating the stereotactic method for irradiation of extracranial
targets were Blomgren and Lax from Karolinska Hospital, Stockholm,
Sweden. They started with a hypofractionated treatment approach in the early
1990s irradiating pulmonary, liver, and abdominal targets with 215 Gy or
310 Gy, prescribed to the PTV-enclosing 65%-isodose and normalizing
the 100% dose to the isocenter (3134). While almost all groups treat
NSCLC of stage I/II and lung metastases, up to now no homogeneous
treatment concept has been established. The published data show a wide
variety of doses, fractionation, dose prescription, and normalization. In
general there are three groups of fractionation: single dose treatment
(37,40,43,46), hypofractionated treatment with 34 fractions (23,31,32,36,
44,51), and hypofractionated treatment in 515 fractions (24,25,27,45). An
overview on treatment concepts, tumor volumes and diameter, followup, and local control rates is given in Table 3.
Single Dose Irradiation
The prescribed dose ranges from 15 to 30 Gy. Some groups started with
lower doses and increased the dose over time in a dose escalation study
and some groups adapted to the tumor size and location. Dose prescription
varies from peripheral dose conventional fractionated irradiation (46),
prescription to the PTV-enclosing 80%-isodose (40) or the isocenter with
having the 80%-isodose enclosing the PTV (37), or as minimal dose to the
GTV (43).
Hypofractionation with 34 Fractions
In this group the Swedish concept of an inhomaogeneous dose distribution was
mainly used with 215 Gy or 310 Gy, prescribed to the PTV-enclosing 65%isodose and normalization to 100% at the isocenter (3136). This prescription
results in a peripheral fraction dose at the PTV of 15 Gy, respectively, l0 Gy and
isocenter doses of 22.5 Gy/15 Gy. (In terms of doses the prescription to
the PTV-enclosing 65%-isodose is identical to the otherwise used nomenclature
of prescribing the dose to the PTV-enclosing 100%-isodose with normalization
to 150% at the isocenter.) Nagata et al. (44) reported on 41012 Gy, prescribed to the isocenter. Lee et al. (51) prescribed 3410 Gy to the 90%-isodose.
Timmerman et al. (23) performed a phase I dose escalation study in patients
with medically inoperable NSCLC with three fractions using dose per fraction
ranging from 8 up to 20 Gy to generally the 80% isodose.
Hypofractionation with 515 Fractions
Uematsu et al. (25) treated their patients with more fractions ranging from
5 to 15 and doses ranging from 30 to 76 Gy prescribed to the 80%-isodose,
27
40
Prim. tumors/
metastases
Prim. tumors/
metastases
Wulf et al. (36)

Strahlenther
Onkol 2001
Nagata et al. (44)
IJROBP 2002
50
Prim. tumors
Uematsu
et al. (27)
IJROBP 2001
22
66
Prim. tumors/
metastases
Metastases
17
Prim. tumors/
metastases
Targets
(n)
Nakagawa
et al. (46)
IJROBP 2000
Blomgren
et al. (31)
J Radiosurg l998
Uematsu
et al. (25)
Cancer 1998
Study
Tumor
type
14.8 cm
tumor
diameter
0.555 cm3a
Chest wall
5126 (40)
Central lung
0.813 (4.5)
0.85.0 cm
tumor diam.
(median
3.2 cm)
3198 (15)a
Median (cm3)
510 fract.
5060 Gy Tu encl.
80%-isodose
(18 pts boost
after CFRT)
5277 (57)
310 Gy/100%isodose, norm.
150%
410 Gy to 412 Gy <4 cm diameter
isocenter
115124 Gy
peripheral dose
conv. fract. RT
515 fractions
3076
Gy/80%-isodose
310 to 215
Gy/65%-isodose
n/Normalization
19
(439)
8
(233)
36
(2266)
31/33
(94%)
23/27
(85%)
20/21
(95%)
64/66
(97%)
11
(331)
10
(282)
16/17
(94%)
16/31
evaluated
16/16
(100%)
11/12
(92%)
47/50
(96%)
1/1
(100%)
22/23
(96%)
3/3
(100%)
Local control Local control

(All targets)
prim. LC
8 (mean)
(3.525)
Minmax
(Month)
Table 3 Published Treatment Concepts and Results of Stereotactic Radiotherapy of Targets in the Lung
6/9
(66%)
8/11
(73%)
20/21
(95%)
42/43
(98%)
13/14
(93%)
Local
control
metastases
218
Timmerman and Wulf
37
34
10
Prim. NSCLC
stage I
Prim. tumors/
metastases
Prim. tumors
57
23
Prim. tumors/
metastases
Prim. tumors/
metastases
23
Prim. tumors/
metastases
4860 Gy/8fx
isocenter or PTVencl. 80%-isodose
11926 Gy/
isocenter, 80%isod. encl. PTV
3410 Gy/90%isodose
Dose escalation from
38 Gy/80% to
320 Gy/80%
isodose
12030 Gy to
minimal dose to
the GTV
115 Gy/80%isodose (?)
Cyberknife
Data have been recalculated from that given in the original publications.
Lee et al. (51)

Lung Cancer 2003
Timmerman
et al. (23)
Chest 2003
Onimaru
et al. (24)
IJROBP 2003
Hof et al. (37)
IJROBP 2003
Hara et al. (43)

Radiother
Oncol 2002
Whyte et al. (40)
Ann Thorac
Surg 2003
4.4230 (41)
(PTV)
1.5157
(22.5)
519 (12)
0.66 cm
(2.6 cm)
15 cm tumor
diameter
< 4 cm diameter
CTV: 116 (4)
18
(735)
15
(230)
15
(830)
18
(244)
7 (mean)
(126)
13
(324)
31/34
(91%)
50/57
(88%)
21/23
(91%)
19/23
(83%)
8/9
(89%)
31/37
(84%)
8/10
(80%)
n8
n 15
(no detailed
report on
local control)
20/25
(80%)
23/25
(92%)
18/20
(90%)
14/18
(78%)
5/5
(100%)
Lung Tumors
219
220
Timmerman and Wulf
Figure 2 (Caption on facing page)
respectively, 5060 Gy in 510 fractions with the 80%-isodose enclosing the

tumor. Within the latter concept stereotactic irradiation was given as
boost after conventional fractionated radiotherapy (CFRT; n 18) (27).
Fukumoto et al. (45) and Onimaru et al. (24) reported on eight fractions of
6 Gy with dose prescription to the 80%-isodose in peripheral and 6 Gy to
the isocenter in targets close to the mediastinum.
Target Volume
The reported target volumes again show a large variety. Some authors
report the tumor diameter; others the GTV, CTV, or the PTV. The tumor
diameters range from 0.6 to 6 cm. The CTV ranges from a minimum of
0.5 cm3 to a maximum of 277 cm3 (median 440 cm3) (Table 3).
LOCAL CONTROL AND SURVIVAL
Unfortunately, as expected for an emerging technology, follow-up on
most of these reports is relatively short. Median follow-up ranged from
7 to 19 months [36 months of surviving patients by Uematsu et al. (27)]
the reported maximum follow-up was 82 months (46). An overview on treatment parameter and results is given in Table 3, which encompasses 12
studies (2325,27,31,36,37,43,44,46,51). (The treatment results of two additional studiesBlomgren et al. (32) and Fukumoto et al. (45)were
Lung Tumors
221
Figure 2 CT follow-up over 61 months in a 62-year-old male with a medically

inoperable squamous cell lung cancer cT3 cN0 cM0 in the right lower lobe treated
by stereotactic irradiation of 3 10 Gy/PTV-enclosing 100%-isodose, normalization
to 150% with a CTV of 91 cm3 and a PTV of 174 cm3. After initial tumor regression
progressive brosis occurred with continuous changes of CT morphology during
follow-up. At 15 months a CT-assisted biopsy of the suspicious mass at the thoracic
wall revealed brous tissue and no evidence of malignant tumor. Typically the
appearance of the irradiated lung tissue changes rapidly over time without tumor
recurrence or clinical symptoms reported by the patient.
included in later publications by Blomgren et al. (31) and Onimaru et al.

(24). Only the data from the most recent papers appear in this overview.)
Most authors dene local control as complete or partial remission or stable
disease. Local failure is dened as progressive disease or regrowth after
initial response. A clinical example of a locally controlled tumor and
changes during follow-up is shown in Figure 2.
Summarizing the published data from the 12 studies, a total of 206
patients (ranging from 1 to 50 per study) with primary lung cancer have
been treated by stereotactic irradiation. Crude local control rates ranged
from 80% to 100%.
The largest cohort consisting of 50 patients with NSCLC stage I has
been published by Uematsu et al. (27). They reached a crude local control
rate of 96% with only three local failures. The actuarial cause specic survival
(overall survival) was 98% (98%) after six months, 96% (90%) after one year,
91% (77%) after two years, 88% (66%) after three years, and 81% (55%)
222
Timmerman and Wulf
after four years. These results are clearly superior to these achieved by CFRT
and reach equivalency to surgical results. Evaluating the overall survival of
29 patients with operable tumors, who had refused surgery and therefore
are comparable to surgically treated patients, the four-years OS of these
patients was 77%. Similar results with actuarial local control rates of
90100% after two to three years have been achieved by other groups
(36,44,51). Nevertheless DFS and OS differed considerably from only 11%/
27% to 73%/100% after two to three years, indicating the importance of
patient selection. While the groups treating patients with stage I disease only
achieve superior results, the authors also treating patients of stage II or even
initially metastasized patients (stage IV) report on inferior disease-free and
overall survival. A comparison of published treatment results on primary lung
cancer according to tumor stage is shown in Table 4.
Unfortunately, in most studies, no time-event analyses have been performed to describe treatment results of primary lung cancer and metastases
separately. Nine of the 12 papers presented in Table 3 report on treatment
results for pulmonary metastases. According to these publications a total of
169 lung metastases have been treated by stereotactic irradiation. The crude
local control rate ranged from 66% to 98% and was somewhat inferior to the
local control rates achieved for primary lung cancer (Table 4). Because of
the small number of local failures (a total of 30, ranging from 1 to 7 local
failures for the individual studies) no reliable evaluation of factors associated with locally uncontrolled failure could be performed up to now.
Additionally the treated patient groups, especially for metastases from different primaries, are inhomogeneous and therefore the role of tumor size
or histology on treatment results could not be evaluated sufciently.
Nevertheless some authors observed a dose dependence of local tumor
control without respect to differentiation in primary lung cancer or metastases. Hara et al. (43) reported an actuarial local control rate after 13 months
of 88% for tumors receiving a single dose 30 Gy, but only of 63% for dose
<30 Gy (p 0.102). Onimaru et al. (24) evaluated a three years local control
rate for pulmonary tumors (NSCLC and metastases) of 100% for doses
of 60 Gy compared to 70% for 48 Gy, both given in eight fractions
(p 0.0435). In our own data from Wuerzburg 65 lung tumors (24 primary
lung cancer and 41 metastases) were treated either by 310 Gy/PTVenclosing 100%-isodose, normalization 150% at the isocenter (n 27) or
31212.5 Gy (same dose prescription, n 19) and single dose irradiation
of 26 Gy/PTV-enclosing 80%-isodose (n 19). After a median follow-up of
10 months (261 months) the actuarial local control after one year and later
was 72% for the patients treated with 3l0 Gy compared to 100% for those
treated by 31212.5 Gy or single dose irradiation (log-rank test: 0.026; 56).
Evidence of a dose response relationship for primary lung cancer was also
observed in the Indiana University phase I dose escalation study. Although
this study was primarily a toxicity evaluation, six patients had local failure
Nagata
et al. (44)
IJROBP 2002
Hara et al. (43)
Radiother Oncol
2002
Wulf et al. (36)

Strahlenther
Onkol 2001
Blomgren
et al. (31)
J Radiosurg 1998
Nakagawa
et al. (46)
IJROBP 2000
Uematsu
et al. (27)
IJROBP 2001
Study
GTV 3,4,6,6,8 cm3
47/50 (96%)
cT1 cN0 cM0

(n 24)
cT2 cN0 cM0
(n 26)
cT2 cN0 cM0
(n 3)
cT3 cN0 cM0
(n 5)
cT2-3 cN0 cMl
(n 4)
cT1 cN0 cM0
5/5 (100%)
16/31 evaluated
16/16 (100%)
11/12 (92%)
No
1/1 (100%)
cTx cN0 cM1

(GTV 6 cm3)
No
Yes
Yes
(recalculated
from original
data)
Yes
No
3/3 (100%)
Actuarial
data
available
cTl-2 cN0 cM0
Tumor stage
Crude local
control
100/100/93
100/81/87
91/42/52
x/96/90
x/98/98
(cause specic
survival)
91/82/92
100/x/x
1 year
100/x/x
6 months
100/73/79
91/11/27
x/91/ 77
100/x/x
2 years
Local control/DFS/OS (%)
3 years
(Continued)
100/73/79
91/11/27
x/88/66
[4 years: x/
81/55]
Table 4 Actuarial Local Control Rates, Disease Free Survival (DFS) and Overall Survival (OS) for Primary Lung Cancer
Lung Tumors
223
I (n 19),
II (n 1),
III (n 1),
IV (n 4)
cT1 cN0 cM0
(n 2)
cT2 cN0
cM0 (n 8)
n.a. (7 medically
inop., 2 refused
surgery)
cT1 cN0 cM0
(n 19)
cT2cN0 cM0
(n 18) max
tumor 7 cm
Onimaru
et al. (24)
IJROBP 2003
31/37 (84%)
8/9 (89%)
8/10 (80%)
n 15
(no detailed
report on local
control of
primary lung
cancer)
20/25 (80%)
Yes (only at
15.2 months)
Yes
Yes
Yes
No
Actuarial
data
available
90/x/100
100/x/100
x/x/x
6 months
x/50/64 at
15.2 mo.
90/x/100
89/x/80
x/x/x
1 year
90/x/100
71/x/64
x/47/60
2 years
Local control/DFS/OS (%)
Some authors did not perform timeevent analyses or did not differentiate results of primary lung cancer to pulmonary metastases.
Timmerman
et al. (23)
Chest (in press)
Lee et al. (51)

Lung Cancer 2003
Hof et al. (37)

IJROBP 2003
n.a.
Tumor stage
Crude local
control
55/x/x
3 years
Actuarial Local Control Rates, Disease Free Survival (DFS) and Overall Survival (OS) for Primary Lung Cancer (Continued )
Whyte et al. (40)

Ann Thorac
Surg 2003
Study
Table 4
224
Timmerman and Wulf
Lung Tumors
225
at dose of 3 16 Gy or below while no patient had local failure at 3 18 Gy or

higher (23).
Toxicity
Published reports of stereotactic irradiation of pulmonary targets describe a
very low acute and late toxicity rate compared to conventional radiation
techniques to high dose used in medically impaired patients. Acutely, some
hours after irradiation, about 1022% of patients will exhibit pain, fever,
chills, nausea, or vomiting, which lasts for a few hours with spontaneous
remission and only rarely requires treatment by antipyretic pain killers
(31,32,36). The reported rate of radiation pneumonitis is low. It is diagnosed
without clinical symptoms by CT-scans during follow-up in 59% (27,36,46).
Symptomatic pneumonitis requiring steroid treatment is observed with a rate
of about 4% (24,43). Only two cases have been reported with impairment of
lung function due to stereotactic irradiation (40,43). Three cases of pneumothorax occurred due to seed insertion into the tumor for preparation of
CyberKnife-treatment, two of them requiring a chest tube (40).
The most obvious late toxicity is brosis of the lung in the high dose
area, potentially associated with atelectasis distantly. This brosis occurs in
almost every patient after stereotactic irradiation. Severe late complications
are very rare. Two cases of esophageal ulceration have been reported, one of
these with consecutive fatal bleeding (24,36). Another patient suffered from
grade 3 chronic cough after irradiation of a target close to the trachea by
210 Gy (31,32). At this time there are no data on radiosensitivity of the
wall of large vessels. One fatal bleeding from the pulmonary artery was
reported nine months after stereotactic treatment with 310 Gy to a tumor
recurrence progressively compressing the artery after preirradiation with
66 Gy a year before (36). Uematsu et al. (27) reported on two patients with
bones within the 80%-isodose, who suffered from a rib fracture or a vertebral compression fracture after irradiation.
The potential for dose escalation by stereotactic irradiation related to
toxicity has been evaluated in a phase I dose escalation toxicity study from
Timmerman et al. (23) in patients with medically inoperable early stage lung
cancer. The treatment population generally had very poor pulmonary function at the outset with mean FEV1 1.24 (46% predicted) and range 0.42.5
(1994% predicted). Despite their general frailty, dose escalation to dramatically high doses was accomplished from 3 8 Gy to 3 20 Gy prescribed
to the PTV-enclosing 80%-isodose in steps of 2 Gy per fraction. Sporadic
pneumonitis and hypoxia was observed irrespective of dose level. At the
higher doses, brosis in the treatment area with distal collapse of airway
was observed in all patients although rarely symptomatic. Pulmonary
function testing after treatment showed trends toward worsening of diffusing capacity and arterial oxygen tension. Spirometry indices showed no
226
Timmerman and Wulf
signicant decline after treatment; indeed, in many patients FEV1 and FVC
improved for unknown reasons. Overall, this study demonstrated that very
high biologically potent dose levels could be reached in a frail population
using stereotactic techniques likely owing to the care taken to exclude
uninvolved lung volume.
Summarizing these data, no relevant toxicity has been reported due to
stereotactic irradiation of peripheral lung tumors. Only targets close or
adjacent to the mediastinum with high dose spots to the organs at risk, such
as trachea, major bronchi, or esophagus, are associated with increased risk
toxicity.
CONCLUSION AND FUTURE DIRECTIONS
Summarizing the clinical data achieved by stereotactic irradiation of lung
tumors it can be concluded that high actuarial local control rates of
80100% after three years can be achieved for primary lung cancer and somewhat lower for pulmonary metastases. At the same time the treatment is associated with very low acute and late toxicity if high doses to organs at risk at the
mediastinum are avoided. The results of disease-free survival and overall
survival may reect the practice of patient selection for a new treatment
approach. Most authors report on case numbers of less than 30 targets, indicating publication of the treatment results of the very rst patients. Nevertheless the published results support the concept of dose escalation and volume
restriction due to the stereotactic approach with minimizing setup-inaccuracy
and target (breathing) mobility to achieve local tumor control. Therefore,
from the published data it seems justied to consider stereotactic irradiation
of lung tumors even for curative treatment. Despite the heterogeneity of treatment concepts all approaches seem to lead to very promising treatment results.
Nevertheless there is evidence on dose dependency of local control rates so
that doses that have been reported with inferior results should be avoided.
Stereotactic body radiotherapy will likely play an increasing role in
lung cancer treatment, particularly early stage NSCLC. Treatment toxicity
is related to the volume of normal tissue treated at or near the target dose.
The main challenge will be to better account for target motion and other
uncertainties allowing further eld reductions without missing the targets.
This will be of particular importance when treating larger tumors and
tumors closer to the central chest.
The Radiation Therapy Oncology Group in the United States is about
to embark on a phase II trial of stereotactic body radiation therapy in early
stage NSCLC using the Indiana University phase I study as the basis for
dose selection. This will be a multi-institutional trial assessing not only
patient outcome, but also feasibility of these treatments on a wider scale.
Comprehensive central review and quality assurance are incorporated into
this pilot trial. Eventually, it is envisioned that this therapy will be tested
Lung Tumors
227
against other therapies, including surgical resection, in a phase III trial.

Several institutions are studying the implementation of stereotactic body
radiation therapy as a boost only with more conventionally fractionated
radiation in locally advanced lung cancer. This approach is being used in
both initial treatment regimens as well as for patients with recurrent tumors.
Many of these types of patients also get chemotherapy as part of their treatment. Whether higher than conventional dose per fractions will alter the
chemotherapy response in tumor or normal tissues remains to be evaluated.
In future efforts to reduce the irradiated volume by further decrease of
security margins to compensate for target mobility and setup-inaccuracy
will be undertaken. In this direction 4D-image-guided radiotherapy,
e.g., by implementation of CT verication in the treatment room (cone
beam CT) or chased beam irradiation with the beam or treatment couch
following a moving target will be introduced (45). Other approaches will
focus on further reduction of breathing mobility by, e.g., breathingtriggered irradiation (5254).
REFERENCES
1. Becker N, Wahrendorf J. Atlas of Cancer Mortality in the Federal Republic of
Germany 19811990. 3rd. Berlin, Heidelberg, New York: Springer-Verlag,
1997.
2. Janssen-Heijnen MLG, Coebergh JWW. The changing epidemiology of lung
cancer in Europe. Lung Cancer 2003; 41:245258.
3. Hoelzel D, Klamert A, Schmidt M. Krebs: Haugkeiten, Befunde und Behandlungsergebnisse. W. Zuckschwerdt Verlag 1996:247261.
4. Nesbitt JC, Putnam JB Jr, Walsh GL, et al. Survival in early-stage non-small
cell lung cancer. Ann Thorac Surg 1995; 114:535543.
5. Mountain CF. Revisions in the international system for staging lung cancer.
Chest 1997; 111:17101717.
6. Adebonojo SA, Bowser AN, Moritz DM, et al. Impact of revised stage classication of lung cancer on survival: a military experience. Chest 1999;
115(6):15071513.
7. Kaseda S, Aoki T, Hangai N, et al. Better pulmonary function and prognosis
with videoassisted thorascopic surgery than with thoractomy. Ann Thorac
Surg 2000; 70:16441646.
8. Martini N, Bains MS, Burt ME, et al. Incidence of local recurrence and second
primary tumors in resected stage I lung cancer. J Thorac Cardiovasc Surg 1995;
109:120129.
9. Ginsberg RJ, Rubinstein LV. Lung Cancer Study Group: Randomized trial of
lobectomy versus limited resection for T1 N0 non-small cell lung cancer. Ann
Thorac Surg 1995; 60:615623.
10. Jeremic B, Classen J, Bamberg M. Radiotherapy alone in technically operable,
medically inoperable, early-stage (I/II) non-small cell lung cancer. Int J Radiat
228
Timmerman and Wulf
11. Sibley GS. Radiotherapy for patients with medically inoperable stage I nonsmall cell lung carcinoma. Smaller doses and higher doses. A review. Cancer
1998; 82:433438.
12. Sibley GS, Jamieson TA, Marks LB, et al. Radiotherapy alone for medically
inoperable stage I non-small cell lung cancer: the Duke experience. Int J Radiat
13. Belderbos JSA, De Jaeger K, Heemsbergen WD, et al. First results of a phase I/II
dose escalation trial in non-small cell lung cancer using three-dimensional conformal radiotherapy. Radiother Oncol 2003; 66:119126.
14. Slotman BJ, Antonisse IE, Njo KH. Limited eld irradiation in early stage
(T1-2N0) non-small cell lung cancer. Radiother Oncol 1996; 41:4144.
15. Krol ADG, Aussems P, Noordijk EM, Hermans J, Leer JWH. Local radiation
alone for peripheral stage I lung cancer: could we omit the elective regional
nodal irradiation? Int J Radiat Oncol Biol Phys 1996; 34(2):297302.
16. Bradley JD, Wahab S, Lockett MA, et al. Elective nodal failures are uncommon in
medically inoperable patients with stage I non-small cell lung carcinoma treated
with limited radiotherapy elds. Int J Radiat Oncol Biol Phys 2003; 56:342347.
17. Pastorino U, Buyse M, Friedel G, et al. Long-term results of lung metastasectomy:
prognostic analyses based on 5206 cases. J Thorac Cardiovasc Surg 1997; 113:3749.
18. Davidson RS, Nwogu CE, Brentjens MJ, Anderson TM. The surgical management of pulmonary metastasis: current concepts. Surg Oncol 2001; 10:3542.
19. Wolbarst AB, Chin LM, Svensson GK. Optimization of radiation therapy: integral-response of a model biological system. Int J Radiat Oncol Biol Phys 1982;
8:17611769.
20. Yeas RJ, Kalend A. Local stem cell depletion model for radiation myelitis. Int
21. Douglas BG, Fowler JF. The effects of multiple small doses of x-rays on skin
reactions in the mouse and a basic interpretation. Radiat Res 1976; 66:401426.
22. De Jaeger K, Merlo FM, Kavanagh MC, et al. Heterogeneity of tumor oxygenation: relationship to tumor necrosis, tumor size, and metastasis. Int J Radiat
Oncol Biol Phys 1998; 42(4):717721.
23. Timmerman RD, Papiez L, McGarry R, Likes L, DesRosiers C, Bank M, Frost
S, Randall M, Williams M. Extracranial stereotactic radioablation: results of a
phase I study in medically inoperable stage I non-small cell lung cancer. Chest
2003; 124(5):19461955.
24. Onimaru R, Shirato H, Shimizu S, Kitamura K, et al. Tolerance of organs at
risk in small-volume, hypofractionated, image-guided radiotherapy for primary
and metastatic lung cancers. Int J Radiat Oncol Biol Phys 2003; 56:126135.
25. Uematsu M, Shioda A, Tahara K, Fukui T, Yamamoto F, Tsumatori G, Ozeki Y,
Aoki T, Watanabe M, Kusano S. Focal, high dose, and fractionated modied
stereotactic radiation therapy for lung carcinoma patients. Cancer 1998; 82:
10621070.
26. Uematsu M, Shioda M, Suda A, Tahara K, Kojima T, Hama Y, Kojima T,
Kono M, Wong JR, Fukui T, Kusano S. Intrafractional tumor position
stability during computed tomography (CT)-guided frameless stereotactic
radiation therapy for lung or liver cancers with a fusion of CT and linear accelerator (focal) unit. Int J Radiat Oncol Biol Phys 2000; 48(2):443448.
Lung Tumors
229
27. Uematsu M, Shioda M, Suda A, et al. Computed tomography-guided frameless

stereotactic radiotherapy for stage I non-small cell lung cancer: a 5-year experience. Int J Radiat Oncol Biol Phys 2001; 51(3):666670.
28. Uematsu M, Sonderegger M, Shioda A, Tahara K, Fukui T, Hama Y, Kojima T,
Wong J, Kusano S. Daily positioning accuracy of frameless stereotactic radiation
therapy with a fusion of computed tomography and linear accelerator (focal)
unit: evaluation of z-axis with a z-marker. Radiother Oncol 1999; 50:337339.
29. Shimizu S, Shirato H, Ogura S, Akita-Dosaka H, et al. Detection of lung tumor
movement in real-time tumor-tracking radiotherapy. Int J Radiat Oncol Biol
Phys 2001; 51(2):304310.
30. Erridge SC, Seppenwoolde Y, Muller SH, van Herk M, et al. Portal imaging to
assess set-up errors, tumor motion and tumor shrinkage during conformal
radiotherapy of non-small cell lung cancer. Radiother Oncol 2003; 66:7585.
31. Blomgren H, Lax I, Goranson H, Krapelien T, Nilsson B, Naslund I,
Svanstrom R, Tilikidis A. Radiosurgery for tumors in the body: clinical experience using a new method. J Radiosurg 1998; 1(1):6374.
32. Blomgren H, Lax I, Naslund I, Svanstrom R. Stereotactic high dose fractionation radiation therapy of extracranial tumors using an accelerator. Acta Oncol
1995; 34:861870.
33. Lax I, Blomgren H, Larson D, Naslund D. Extracranial stereotactic radiosurgery of localized targets. J Radiosurg 1998; 1(2):135148.
34. Lax I, Blomgren H, Naslund I, Svanstrom R. Stereotactic radiotherapy of malignancies in the abdomen. Methodological aspects. Acta Oncol 1994; 33:677683.
35. Wulf J, Haedinger U, Oppitz U, Olshausen B, Flentje M. Stereotactic radiotherapy of extracranial targets: CT-simulation and accuracy of treatment in
the stereotactic body frame. Radiother Oncol 2000; 57:225236.
36. Wulf J, Haedinger U, Oppitz U, Thiele W, Ness-Dourdoumas R, Flentje M.
2001; 177:645655.
37. Hof H, Herfarth KK, Munter M, Hoess A, Motsch J, Wannenmacher M,
Debus J. Stereotactic single-dose radiotherapy of stage I non-small-cell lung
cancer (NSCLC). Int J Radiat Oncol Biol Phys 2003; 56(2):335341.
38. Wong JW, Sharpe MB, Jaffray DA, Kini VR, Robertson JM, Stromberg JS,
Martinez AA. The use of active breathing control (ABC) to reduce margin
for breathing motion. Int J Radiat Oncol Biol Phys 1999; 44(4):911919.
39. Dawson LA, Brock KK, Kazanjian S, Fitch D, et al. The reproducibility of
organ position using active breathing control (ABC) during liver radiotherapy.
40. Whyte RI, Crownover R, Murphy MJ, Martin DP, et al. Stereotactic radiosurgery for lung tumors: preliminary report of a phase I trial. Ann Thorac Surg
2003; 75:10971101.
41. Yenice KM, Lovelock DM, Hunt MA, et al. CT image-guided intensitymodulated therapy for paraspinal tumors using stereotactic immobilization.
42. Wulf J, Haedinger U, Oppitz U, Thiele W, Flentje M. Impact of target reproducibility on tumor dose in stereotactic radiotherapy of targets in the lung and
liver. Radiother Oncol 2003; 66:141150.
230
Timmerman and Wulf
43. Hara R, Itami J, Kondo T, Aruga T, Abe Y, et al. Stereotactic single high dose
irradiation of lung tumors under respiratory gating. Radiother Oncol 2002;
63:159163.
44. Nagata Y, Negoro Y, Aoki T, et al. Clinical outcomes of 3D conformal hypofractionated single high dose radiotherapy for one or two lung tumors using a
stereotactic body frame. Int J Radiat Oncol Biol Phys 2002; 52(4):10411046.
45. Fukumoto S, Shirato H, Shimizu S, Ogura S, Onimaru R, et al. Small volume
image-guided radiotherapy using hypofractionated, coplanar and noncoplanar
multiple elds for patients with inoperable stage I nonsmall cell lung carcinomas. Cancer 2002; 95:15461553.
46. Nakagawa K, Aoki Y, Tago M, Terahara A, Ohtomo K. Megavoltage CTassisted stereotactic radiosurgery for thoracic tumors: original research in the
treatment of thoracic neoplasms. Int J Radiat Oncol Biol Phys 2000;
48(2):449457.
47. Vant Riet A, Mak ACA, Moerland MA, Elders LH, Van der Zee W. A conformation number to quantify the degree of conformality in brachytherapy
and external beam irradiation: application to the prostate. Int J Radiat Oncol
Biol Phys 1997; 37(3):731736.
48. Haedinger U, Thiele W, Wulf J. Extracranial stereotactic radiotherapy: evaluation of PTV coverage and dose conformity. Z Med Phys 2002; 12:221229.
49. Engelsman M, Damen EMF, Koken PW, vant Veld AA, van Ingen KM,
Mijnheer BJ. Impact of simple tissue inhomogeneity correction algorithms on
conformal radiotherapy of lung tumors. Radiother Oncol 2001; 60:299309.
50. Engelsman M, Remeijer P, van Herk M, Lebesque JV, et al. Field size reduction
enable Iso-NTCP escalation of tumor control probability for irradiation of lung
tumors. Int J Radiat Oncol Biol Phys 2001; 51(5):12901298.
51. Lee S, Choi EK, Park HJ, et al. Stereotactic body frame based fractionated
radiosurgery on consecutive days for primary or metastatic tumors in the lung.
Lung Cancer 2003; 40:309315.
52. Kuriyama K, Onishi H, Sano N, et al. A new irradiation unit constructed of selfmoving gantry-CT and linac. Int J Radiat Oncol Biol Phys 2003; 55:428435.
53. Onishi H, Kuriyama K, Komiyama T, et al. CT evaluation of patient deep
inspiration self-breath-holding: how precisely can patients reproduce the tumor
position in the absence of respiratory monitoring devices? Med Phys 2003;
30(6):11831187.
54. Onishi H, Kuriyama K, Komiyama T, Tanaka S, et al. A new irradiation
system for lung cancer combining linear accelerator, computed tomography,
patient self-breath-holding and patient directed beam control without respiratory monitoring devices. Int J Radiat Oncol Biol Phys 2003; 56:1420.
55. Haedinger U, Krieger T, Flentje M, Wulf J. Inuence of calculation model on
dose distribution in stereotactic radiotherapy for pulmonary targets. Int J
56. Wulf J, Haedinger U, Oppitz U, Thiele W, Mueller G, Flentje M. Stereotactic
radiotherapy for primary lung cancer and pulmonary metastases: a noninvasive
treatment approach in medically inoperable patients. Int J Radiat Oncol Biol
Phys 2004; 60(1):186196.
10
Prostate Tumors
Raymond Miralbell
Servei de Radio-oncologia, Instituto Oncologico Teknon, Barcelona, Spain and
Service de Radio-oncologie, Hopitaux Universitaires, Geneva, Switzerland
Meritxell Molla` and Lluis Escude

Servei de Radio-oncologia, Instituto Oncologico Teknon, Barcelona, Spain
Dirk Verellen, Guy Soete, and Guy Storme

THE TEKNON EXPERIENCE

INTRODUCTION AND RATIONALE
Rationale for Dose Escalation in Curative Treatment of Localized
Prostate Cancer
Dose escalation above 70 Gy may prove benecial in prostate cancer
radiotherapy (RT) (13). The results of curative RT performed in the 1970s
and 1980s, assessed with sequential PSA tests, have shown a higher than
previously expected treatment failure rate (4050%) when delivering a dose
below 70 Gy (the recommended dose at that time) (1). Dose escalation studies performed at Memorial Sloan Kettering Cancer Center (2) and at M.D.
Anderson Hospital (3) have shown an improved outcome with doses 78 Gy
and above. Patients with poorly differentiated tumors (i.e., Gleason scores
710) and/or perineural invasion in the biopsy specimen are at higher than
average risk for local failure and may benet most from the administration
of a high dose to the tumor (46).
231
232
Miralbell et al.
Optimization of Target Repositioning Reproducibility May Lead to

Treatment Volume Reduction
In order to deliver high doses to the prostate, tight margins around the
target are necessary to spare critical organs such as the rectum and the
bladder. The urethra, though contained within the target volume, may also
be a dose-limiting organ if postirradiation stricture is to be considered (7).
A signicant reduction of the irradiated target volume may facilitate delivery of a higher, more effective dose, while at the same time reducing treatment side effects. However, suboptimal reproducibility of daily treatment
setup and random internal organ motion (811) may defeat any attempt
to reduce the usually generous safety margins around the clinical target
volume (CTV) (12). Improving patient repositioning and reducing random
prostate shifts, largely due to rectal volume changes, may help to tighten
the margins around the CTV enough to allow further dose reduction to
critical structures.
Non-Uniform Tumor Cell Distribution Within the Prostate
Largest tumor clonogen clusters are preferentially located in the peripheral
zone or in the central zone (base) of the prostate, especially in more advanced local tumor stages and/or in high (i.e., 710) Gleason scores (13). Thus,
a heterogeneous density distribution of tumor cells within the prostate supports the notion of an intentionally inhomogeneous dose distribution to
deliver a relatively higher dose to the high-density tumor bearing areas
and relatively lower doses to areas with smaller tumor foci (e.g., transitional
zone). Progress in imaging (e.g., endorectal spectroscopic magnetic resonance, 11C-choline or -acetate positron emission tomography) may help to
further improve denition of local tumor extent within the prostate (1416).
Radiobiology of Prostate Cancer: Hypofractionation
In RT, one of the main arguments for delivering a treatment in many
fractions is that late sequelae are generally more sensitive than tumor control to changes in fractionation. Thus, delivering the treatment in a large
number of small daily fractions (e.g., 1.82 Gy/fraction) generally spares
late-responding tissue (able to repair sublethal damage between fractions)
more than the tumor (unable to repair sublethal damage between fractions).
The a/b ratio is a parameter that purportedly denes the sensitivity of each
tissue (normal or tumoral) to changes in treatment fractionation (17). A low
a/b ratio (24 Gy) implies a high sensitivity to changes in fractionation. In
general, tumors are characterized by large a/b ratios (>8 Gy).
Prostate tumors, however, can be considered relatively exceptional
with regard to their fractionation sensitivity. Prostate cancer cells have a long
doubling time (42 days median potential doubling time, range 15170 days)
Prostate Tumors
233
and an effective repair of sublethal radiation damage at low dose per fraction
(18). Indeed, combined analysis of patient outcome after external beam
radiotherapy or brachytherapy has recently led to the conclusion that prostate cancer is characterized by a low a/b ratio (i.e., 1.5 Gy and 0.82.2 Gy
condence interval), lower than that of most tumors or even of the lateresponding normal tissues surrounding the tumor: the rectum and the
bladder (i.e., a/b ratio 4 Gy) (1820). Thus, large treatment fractions
(hypofractionation) may increase the tumor cell killing effect. Several authors have reported their respective experiences with doses per fraction above
2 Gy (2.5, 2.75, and 3.13 Gy) in prostate cancer. They all found the treatment
to be efcient and well tolerated (2123).
In summary, if the above observations are substantiated, hypofractionated treatments may have the potential to either increase tumor control for a
given level of late complications or decrease normal tissue complications for
a given level of tumor control. Hypofractionation in radiotherapy may not
only be biologically sound, but also economically advantageous (by increasing availability of treatment slots in each department) and may also improve
patient convenience (by reducing the number of treatment sessions).
HYPOFRACTIONATED BOOST WITH INTENSITY MODULATED

X-RAY BEAMS UNDER STEREOTACTIC CONDITIONS: A
PILOT STUDY
Preliminary Considerations
Clinical data suggest that 64 Gy in 32 daily standard 2 Gy fractions can
cure residual or relapsing microscopic local disease after postprostatectomy
biochemical failure (24). This notion can be extrapolated to the curative
treatment of gross disease, by prescribing a similar dose level (i.e., 64 Gy) to
areas of potentially microscopic foci in the transitional zone (near the urethra), while reserving higher dose levels to boost gross tumor-bearing regions (peripheral and/or central zones, and/or seminal vesicles) with >80 Gy
total dose to reach the desired local cure. Indeed, two fractions of 5, 6, 7, or
8 Gy to the tumor-bearing region in addition to 64 Gy in 2 Gy daily fractions
of standard RT to the whole prostate and seminal vesicles are equivalent to
82, 88, 96, and 104 Gy of standard fractionated RT, respectively (a/b for
prostate cancer 2 Gy). Thus, a local control probability well above 80%
might be expected (18).
Such a hypofractionated treatment approach has been applied since
June 2000 at the University Hospital of Geneva to treat patients with
non-metastatic, high-risk (e.g., perineural invasion and/or Gleason 810)
tumors. High dose rate (HDR) brachytherapy with temporary open MRIguided iridium implants have been used to escalate the dose in the boost
region. Through March 2004, 64 such patients have been implanted with
234
Miralbell et al.
a low incidence of moderate acute toxicity and no signicant severe late side
effects (25). About 19, 21, and 24 patients were treated with two fractions of
6, 7, and 8 Gy, respectively, to the boost volume. Unfortunately, the coverage
of the prostatic primary tumor volume with brachytherapy needles has not
been always optimal in the Geneva series, especially in very large prostates
or when the tumor inltrated the base of the gland or the seminal vesicles.
Intensity modulated radiotherapy (IMRT) under stereotactic conditions to boost exclusively the tumor-bearing region, as dened by endorectal
MRI, may be an interesting alternative to the above brachytherapy technique. Hypofractionated IMRT has the additional advantage of improving
tumor coverage and reducing costs, pain, and time compared with HDR
brachytherapy. Indeed, neither anesthesia, hospitalization, nor major pain
relievers (morphine) are necessary with IMRT.
We performed a study aiming to assess feasibility, tolerance, and
outcome of patients (with non-metastatic prostate cancer) treated according
to a dose escalation protocol to the boost region with IMRT under stereotactic conditions. The preliminary results of this study are presented
below.
Clinical Material
From June 2001 through May 2003, a dose escalation pilot study for
prostate cancer with high-precision RT was undertaken at Centro Medico
Teknon (CMT), Barcelona. Treatment was delivered with a commercially
available extracranial stereotactic repositioning system (ExacTrac, BrainLAB A.G., Heimstetten, Germany) and 6 MV X-ray beams IMRT with a
micromultileaf collimator-based linear accelerator (Novalis, BrainLAB
A.G., Heimstetten, Germany). Forty-three patients were included in the
study. The distribution of patients according to clinical stage, Gleason score,
and blood PSA level at diagnosis are presented in Table 1. The patients
have been followed for a median time of 19 months (range, 934 months).
Table 1 Patient Distribution According to Clinical Stage,
Gleason Score, and PSA at Diagnosis
Stage
T1c
T2ac
T3a,b
15 pts
11 pts
17 pts
Gleason
46
710
21 pts
22 pts
PSA at diagnosis
<10 ng/mL
1020 ng/mL
>20 ng/mL
22 pts
14 pts
07 pts
Prostate Tumors
235
Treatment Description
Neoadjuvant full androgen deprivation with leuprolide and bicalutamide
was given to 26 patients (those with a PSA at diagnosis >15 ng/mL and/or
those with a total Gleason score >7) for a duration of 624 months, with
radiotherapy starting one to three months after the rst day of hormonal
blockade. Three patients were referred for RT after orchidectomy.
The rst part of the radiation treatment included conventional fractionated 3D conformal external RT or IMRT (Fig. 1). The prostate and
seminal vesicles (CTV1) were to receive 64 Gy in 2 Gy daily fractions if the
risk of nodal involvement (according to Ref. 26) was <15% (22 patients).
If nodal risk was >15%, pelvic nodes, in addition to the prostate and seminal
vesicles, were treated with 50.4 Gy (in 1.8 Gy daily fractions) followed by a
volume reduction up to a total tumor dose of 64.4 Gy to CTV1 (21 patients).
Thirty-two patients were treated with 15 MV X-rays from a Clinac 23-EX
(Varian, Associates, Palo Alto, California, U.S.A.) in the supine position
without special immobilization devices. Eleven patients, all presenting with
low-risk disease, were treated to CTV1 with 6 MV X-rays from the Novalis
linear accelerator using IMRT or dynamic arc techniques. These patients
were immobilized in a customized vacuum body cast and repositioned with
ExacTrac as described below.
All patients received a nal boost to a reduced prostate volume using
IMRT under stereotactic conditions (Fig. 1). The reduced prostate boost
Figure 1 Dose distribution of an intensity modulation radiotherapy plan in the axial

central plane of the planning target volume (PTV). (A) 6464.4 Gy delivered to
the prostate and seminal vesicles (PTV1); (B) 1016 Gy boost delivered to a reduced
horseshoe-shaped volume (the prostate peripheral zone) (PTV2). Dose distribution
is given in percent values and is displayed in color bands. The yellow crosses in the gures
represent the treatment isocenters for PTV1 and PTV2, respectively. (See color insert.)
236
Miralbell et al.
volume (CTV2) included the peripheral and central zone tumor-bearing

regions, together with the seminal vesicles if involved (or the prostate
excluding the urethra if massive involvement of the gland). The CTV2
was expanded with a surrounding margin of 3 mm in order to dene a
PTV2. Two fractions of 5, 6, 7, and 8 Gy each were delivered with a time
interval of 3 to 7 days between fractions to 6, 7, 7, and 23 patients, respectively. This represents a biological equivalent tumor dose (in 2 Gy fractions
and a/b ratio 2 Gy) of 82, 88, 96, and 104 Gy, respectively.
Boost (PTV2) Treatment Planning
A standard single-isocenter seven-eld coplanar technique was most
frequently selected. This was the optimal technique for achieving the
doughnut or horseshoe shaped dose distributions around the urethra (localized with a urinary catheter at simulation) needed to correctly treat
the relatively complex PTV2 (Fig. 1). In order to reduce the overall treatment
time a dynamic MLC IMRT technique was preferred, rather than a static
step and shoot IMRT technique.
Dose constraints were established to optimize the dose distribution for
the PTV2 and the organs at risk (i.e., bladder, rectum, femoral heads, periprostatic neurovascular bundles, and, most of all, the urethra). The rectum
was set to receive a maximum dose of 80% of the prescribed dose but to no
more than 2050% of the volume. The urethra was set to receive a maximum
dose of 10%. Priority factors for all organs at risk, except for the urethra,
were set to 50%. The priority factor for the urethra was set to 100%. These
values were used as a starting point but often had to be modied according to the results of the calculation cycles in order to optimize the dose
distribution.
The femoral heads were distant from the PTV2 and did not present a
major dosimetric problem. Dosevolume histograms (DVHs) requirements
were thus easily met. Nor were bladder and rectum problematical, as only
small portions of their volumes received the highest doses (those lying close
to the PTV2). Both periprostatic neurovascular bundles and the urethra,
however, are so small that their volumes were included in the penumbra
surrounding PTV2. Hence, uncertainties in target repositioning might
have strongly inuenced overall dose distribution, potentially resulting in
a signicant overdosage to some organs at risk.
Furthermore, a volume within the prostate and around the urethra
was dened (the transition zone) in order to establish further dose constraints that could help to improve the dose gradient between the PTV2
and the urethra. Frequently, the size of the transition zone was so reduced
that such an optimal dose gradient was not obtained. In these cases a tradeoff between PTV2 dose distribution homogeneity and the dose to the
urethra had to be found.
Prostate Tumors
237
Treatment Planning Quality Assurance Procedure

After physician approval, treatment plans were transferred to the treatment
network in order to test the procedure on a plastic stacked slab phantom
(PMMA) with the selected beam uences, gantry angles, collimator angles,
and monitor units. Only the position of the isocenter and the monitor unit
scaling were modied. The aim was to compare dosimetric measurements
with calculations obtained by the treatment planning with the same phantom
system. The agreement between in-phantom measurement and calculation
was assumed to be applicable to patient treatment.
Two types of measurements were performed: a lm (Kodak EDR)
dose distribution on a coronal plane and a point dose measurement with
a small volume (0.125 cc) ionization chamber at a dened depth in the beam
central axis. In both scenarios, only the composite plan corresponding to the
sum of all the elds was checked.
The lm dose distribution was compared with the corresponding dose
distribution obtained with the treatment planning system and analyzed with
an in-house computer program. After automatic registration of both dose
distributions, isodose curves and dose proles were plotted and compared.
A gamma index was calculated according to Low et al. (27) and Depuydt
et al. (28) helping to standardize comparison criteria based on dose differences in low-dose gradient areas and distance to agreement (DTA) in
high-dose gradient regions. Tolerance values were established to 3% for
the dose difference and 2 mm for the DTA. For the point dose measurements the tolerance was set to 4%.
In an earlier phase of our prostate treatment program, the individual
elds at 0 incidence angle were also checked and measured. This helped
us to gain an insight into the dynamic behavior of the MLC measurements
of individual beams impinging perpendicularly on a at surface of an
homogeneous medium (less dependent on the patient or phantom-geometry).
Patient Immobilization
ExacTrac was used to reposition all patients for the nal high-dose boost
(CTV2) and 11 low-risk patients treated with Novalis to CTV1. The repositioning procedure with ExacTrac starts with immobilizing the patient in
a customized vacuum body cast. Five to seven metallic infrared (IR) reecting markers are asymmetrically taped to the skin of the abdomen (Fig. 2).
Starting from the planning CT, the position of the isocenter with regard
to the IR markers is calculated by the planning system. Before each treatment session the markers are placed back on the patient. Their spatial
arrangement is detected by a pair of IR cameras mounted to the treatment
rooms ceiling to reproduce the same coordinates when repositioning the
patients for daily treatment.
238
Miralbell et al.
Figure 2 Immobilization and stereotactic repositioning system (ExacTrac). The

patient is immobilized with a customized vacuum body cast and is repositioned with
several infrared reecting metallic markers taped to the skin of the abdomen.
In our study, patients were requested to void their bladders immediately before simulation and before each treatment fraction. To further limit
target motion and to help to improve le target dening process in the simulation CT, a magnetic resonance (MR)-based endorectal probe was used for
the CTV2 simulation and treatment (Figs. 3 and 4). Sodium phosphate enemas were used to evacuate the rectum the night before and again one to two
hours before each procedure. In order to reduce anxiety and prevent or alleviate potential painful rectal spasms during the simulation or treatment
intervals, alprazolam 0.5 mg per os was prescribed in later patients. After
introducing the probe in the rectum, 60 cc were introduced with a syringe.
The inated probe was then gently pulled toward the anus. Patients were
then tted in their immobilization casts, skin metallic markers were xed
on their respective spots, and the IR guided setup was undertaken.
Preliminary Results: Feasibility, Treatment Tolerance,
and Outcome
Patient compliance with treatment was optimal. All 43 patients completed
treatment as planned. Few patients complained of analrectal pain or
spasms while on treatment with the inated rectal balloon.
Urinary and lower gastro-intestinal (GI) acute effects were scored
according to the RTOG/EORTC scoring system (Table 2) (29). Tables 3
and 4 show the observed urinary and lower GI acute toxicity scores for
Prostate Tumors
239
Figure 3 MR endorectal probe used to reduce the internal organ motion of the
prostate during extracranial stereotactic radiotherapy and to help to optimize image
registration between the endorectal MR at diagnosis and the simulation CT.
all patients according to the delivered boost dose. Acute toxicity was scored
every week during treatment and ve to six weeks and again three months
after treatment completion.
Acute urinary toxicity was minimal (grade 1) to moderate (grade 2) for
most patients. A correlation between acute urinary toxicity score and dose
escalation was not observed. Acute lower GI toxicity was minimal (grade 1)
or moderate (grade 2) in more than one-third of patients with no relation
to dose escalation. It is noteworthy that no acute lower GI toxicity was
observed among the 11 patients treated with Novalis-IMRT to CTV1 and
receiving the highest boost dose (28 Gy) to CTV2. This compared favorably
with patients receiving the same 28 Gy-boost after being treated to CTV1
with a non-IMRT technique. Almost half of them presented grade 1 or 2 acute
lower GI toxicity.
Late urinary and lower GI toxicities were scored after a minimum
six-month post-treatment follow-up interval. Scoring was done according
to the SOMA and to the EORTC/RTOG systems (Table 5) in order to
grade urinary and lower GI toxicities, respectively (29,30). So far only moderate late toxicity scores have been obtained for both urinary and lower GI
morbidities. Five patients (12%) presented with late urinary grades 12 toxicity (Table 6) while 13 (30%) presented with late lower GI grades 12 toxicity (Table 7).
A systematic assessment of lower GI toxicity was designed as part of
the study protocol. At 1824 months postradiotherapy (the peak risk for late
240
Miralbell et al.
Figure 4 Dose distribution overlying the axial central plane of an endorectal MR

image of the prostate containing the PTV2 (bilateral prostatic peripheral zone).
Isodose bands of 100% or above (red), 90100% (yellow), and 8090% (green) are
displayed. (See color insert.)
rectal toxicity), patients are requested to undergo recto-sigmoidoscopy with

an assessment of the rectal mucosa and scoring of erythema and/or of telangiectasia. In addition, random biopsies of the normal rectal mucosa, or of
suspected areas of proctitis are undertaken. Functional testing with an
endorectal barostat is also performed in order to assess for potential passive
mechanical changes due to brosis and to evaluate changes in sensitivity to
rectal wall distension.
So far, 16 patients have undergone a rectosygmoidoscopy with 10
grade 0, three grade 1, and three grade 2 RTOG clinical toxicity scores
(see above). Rectal mucosa was normal in ve patients without late rectal
toxicity (grade 0). Erythema was observed in three and two patients with
grade 1 and grade 2 late rectal RTOG toxicity, respectively. Grade 1 telangiectasia (focal and isolated) was observed in ve patients with no clinical
late rectal toxicity. Grade 2 telangiectasia was observed in one patient
suffering from grade 2 (28 Gy boost) late rectal RTOG toxicity.
Prostate Tumors
241
Table 2 RTOG/EORTC Acute Radiation Morbidity Scoring System

Urinary
Grade 1
Grade 2
Grade 3
Lower GI
Grade 1
Grade 2
Frequency of urination twice pretreatment habit

Dysuria, urgency not requiring medication
Incontinence not requiring sanitary pads
Frequency of urination less frequent than every hour
Dysuria, urgency, bladder spasms requiring
non-narcotic medication
Incontinence requiring medication or pads
Frequency of urination hourly or more frequently
Dysuria, pain or spasms requiring narcotics
Gross hematuria
Increased frequency or change in quality of bowel

habits not requiring medication
Rectal or anal discomfort not requiring medication
Urgency not requiring medication
Diarrhea requiring medication
Mucous or blood discharge or soiling not
necessitating sanitary pads
Rectal or anal pain, urgency, requiring medication
Although biochemical relapse has not yet been observed in any of our
patients, much longer follow-up will be required before conclusions can be
drawn regarding the potential value of IMRT dose escalation in the curative
treatment for prostate cancer.
Quality Assessment on Patient and Target Repositioning
Reproducibility
In order to simulate repositioning reproducibility a second pelvic CT, under
simulation conditions, was performed before the last boost fraction (usually
1015 days after the rst CT used for boost simulation). Prostate repositioning
Table 3 Acute Urinary Toxicity According to Boost Dose (EORTC Score)
Dose (Gy)
G-0
G-1
G-2
G-3
5
6
7
8a
0
2
1
5(3)
4
2
2
7(3)
2
3
4
10(4)
0
0
0
1(1)
11/22 patients, in brackets, treated with IMRT.
242
Miralbell et al.
Table 4 Acute Rectal Toxicity According to Boost Dose (EORTC Score)

Dose (Gy)
5
6
7
8a
a
G-0
G-1
G-2
2
4
3
17(11)
2
2
2
4(0)
2
1
2
2(0)
11/22 patients, in brackets, treated with IMRT to CTV1.
was assessed, rst, after CT-to-CT registration of the stereotaxic IR reecting

metallic body markers (to simulate the setup reproducibility with ExacTrac)
(Fig. 5) and, second, after CT-to-CT registration of the pelvic bony structures
(to simulate for potential further improvement of patient repositioning with
pelvic bone registration as it is performed by the Novalis Body system from
BrainLAB, not used for patient treatment in our study). A study has been
recently reported assessing prostate and rectal probe repositioning in 22
Table 5 Late Radiation Morbidity Scoring System
Urinary (SOMA):
Grade 1
Grade 2
Grade 3
Lower GI (RTOG/EORTC):
Grade 1
Grade 2
Grade 3
Source: From Refs. 29, 30.
Occasional dysuria
Occasional hematuria
Microscopic hematuria
Occasional use of incontinence pads
Occasional medication for dysuria
Intermittent dysuria
Intermittent macroscopic hematuria
Intermittent use of incontinence pads
Regular non-narcotic medication for dysuria
Persistent or intense dysuria
Incomplete obstruction
Persistent macroscopic hematuria with clots
Regular use of incontinence pads
Regular narcotic medication for dysuria
Slight rectal discharge or bleeding

Urgency, anal pain, soiling
Excessive rectal mucus or bleeding requiring
medical treatment
Urgency, anal pain, or soiling requiring
medication
Bleeding requiring surgery or formolinization
Obstruction
Prostate Tumors
243
Table 6 Late Urinary Toxicity According to Boost Dose (SOMA)

Dose (Gy)
5
6
7
8
G-0
G-1
G-2
6
7
6
19
0
0
1
2
0
0
0
2
patients treated under optimal immobilization conditions according to the

present protocol (31).
The shifts between the center of mass of the prostate in the rst CT and
the center of mass of the prostate in the second CT were measured in the three
axes: latero-lateral (X), antero-posterior (Y), and cranio-caudal (Z). The
mean and standard deviation (SD) of the displacements in the three axes of
the prostates center of mass were calculated for all patients. This measurements helped to estimate margins around the CTV2 to dene an ideal PTV
using the model described by McKenzie et al. (32) with the following formula:
X

CTV ! PTV margin width 2:5
b s sr
in which S is the SD of the individual displacements of the prostate center
of mass; s the random error (takes in account organ motion and penumbra) and sp the SD describing the width of the beam penumbra (3.2 mm);
b is the penumbra correction coefcient that depends on the number of
beams used to treat a spherical volume in order to deliver at least 95% of
the dose to the CTV (0.52 for the transverse plane, and 1.64 in the cranio-caudal direction for a six non-parallel and opposed eld arrangement).
As mentioned above, 11 patients were also treated with Novalis for the
rst part of the treatment (64 Gy in 2 Gy daily fractions to CTV1). They
underwent weekly resimulation CT scans on treatment position (average,
six CTs for each patient). All of these patients were treated without an endorectal balloon during the rst treatment phase. Assessing the displacements of
the CTV1 center of mass with CT-to-CT fusion of the stereotaxic IR reecting metallic body markers and with a CT-to-CT registration of the pelvic
bony structures, allowed to assess the weekly repositioning reproducibility
Table 7 Late Rectal Toxicity According to Boost Dose (RTOG Score)
Dose (Gy)
5
6
7
8
G-0
G-1
G-2
5
5
4
16
1
1
3
5
0
1
0
2
244
Miralbell et al.
Figure 5 Digital volumetric reconstruction to simulate the setup reproducibility

with ExacTrac. Infrared marker based registration between the CT at simulation
(red) and the CT while on treatment (green) performed to assess target repositioning
quality. (See color insert.)
and to estimate the PTV1 margin width for these 11 patients. These results
were compared with the repositioning reproducibility and the estimated
PTV2 margin width for the 22 patients assessed during the CTV2 treatment
period (boost), treated with identical immobilization conditions but with an
endorectal balloon to optimally reduce the internal organ motion.
The estimated PTV margins around the prostate, for the 22 patients
evaluated under optimal setup conditions (i.e., bone registration and endorectal balloon), were 2.4, 4.3, and 6.4 mm in the latero-lateral, antero-posterior,
and cranio-caudal dimensions, respectively. According to the same model, the
estimated PTV1 margins around de CTV1 for the 11 patients evaluated under
setup conditions involving exclusively bone registration, but not a rectal
balloon, were 1.9, 7.6, and 7.4 mm in the latero-lateral, antero-posterior,
and cranio-caudal dimensions, respectively. Thus, an optimally positioned
rectal balloon signicantly reduced the estimated antero-posterior PTV
margins around the target from 7.6 mm (without balloon) to 4.3 mm (with
balloon), an improvement factor of 1.77.
FINAL COMMENTS AND CONCLUSIONS
Dose escalation above 70 Gy is necessary to improve curative treatment of
localized prostate cancer with RT. This is a challenge for radiation
Prostate Tumors
245
oncologists in their search for an optimal treatment of a movable, centrally

located, almost spherical target surrounded by dose-limiting critical organs
(the rectum, the bladder, and the femoral heads).
Optimal repositioning reproducibility is crucial if the goal is to deliver,
via external RT, the highest dose to the tumor while optimally decreasing
the treatment volume (i.e., reducing the safety margins around the target).
This may reduce the dose to the organs at risk and improve normal tissue
tolerance. In the present study, under optimal setup conditions (i.e., bone
registration and rectal balloon), estimated PTV safety margins around the
CTV were roughly 2, 4, and 6 mm in the latero-lateral, antero-posterior,
and cranio-caudal directions, respectively. This is a signicant reduction
compared to the usually recommended margins of >10 mm, with standard
immobilization approaches based on skin markers or tattoos. Furthermore,
even after optimal setup with bone registration, the use of a rectal balloon
improved repositioning in the antero-posterior axis by a factor of 1.77.
Signicant treatment volume reduction in the present study (compared
to more standard radiotherapy approaches for prostate cancer) may explain
the relative low incidence of acute or late toxicity at the urinary and/or
lower GI levels. Among the 43 patients fully evaluable for late lower GI
tolerance, 13 patients (30%) presented moderate grades 12 late toxicity.
This contrasts with a worse tolerance among 175 patients treated in Geneva
since 1999 with 7478.4 Gy standard fractionated conformal external RT
and with a minimum of eight-month follow-up. A 45% incidence of grades
12 late rectal toxicity was observed among those patients (unpublished
data).
The highest dose may be preferentially delivered in areas of the prostate with the largest number of tumor cells (e.g., the prostatic peripheral
zone). Thus, a non-uniform dose distribution within the prostate may well
be a treatment optimization tool if the highest dose is delivered to areas
at highest risk for failure. Progress in imaging (e.g., endorectal MRI with
or without spectroscopy and PET with C11 acetate or F18 choline) may help
to better dene these areas at risk within the prostate and improve targeting
strategies in the future. Furthermore, reducing the prostatic treatment
volume by purposely underdosing areas of potentially lower tumor burden
(e.g., transitional zone of the prostate) allows the risk of potential urethral
damage to be kept to a minimum level during the delivery of the hypofractionated boost.
Prostate cancer cells may have more chance of surviving low-dose
fractions (i.e., 2 Gy or less) than high-dose fractions (i.e., 4 Gy or more).
Thus, if fractionation schedules in prostate cancer are to be adapted to take
into account radiobiological considerations, large fractions are likely to play
a role in more effective curative strategies. Some, however, cautiously question this approach by suggesting that the benet derived from reoxygenation
of hypoxic cancer foci between fractions may be lost if few large-dose
246
Miralbell et al.
fractions are used (33). In the present study, the administration of the rst
6464.4 Gy in 3235 daily fractions of 1.82 Gy, preceded in the less favorable cases with androgen blockade, might have helped to decrease the
problem of hypoxia by reducing signicantly the tumor volume before the
delivery of the nal hypofractionated boost.
Although all patients in the present study have their disease under
biochemical control, longer follow-up will be required before cure rates can
be evaluated. Caution is called for, especially considering the tight PTV
boost margins used, that may have underdosed part of the CTV2 volume.
Only time will tell whether or not this underdosage has a negative inuence
on local control. The choice of the tight 3 mm margins, however, was based
on the need to protect the urethra, which might not have always been
possible with larger margins. The present quality assessment indicates that
margins of 4 mm minimum in the transverse plane are needed under optimal
setup conditions.
THE AZ-VUB EXPERIENCE
INTRODUCTION AND RATIONALE
Radiotherapy is one of the major treatment modalities for cancer of the
prostate. Increased dose levels are known to signicantly improve treatment
outcome but require specialized techniques to avoid important and sometimes permanent side effects to bladder and rectum. The AZ-VUB has
adopted three distinctive measures to avoid these complications with
the introduction of the Novalis system (BrainLAB, A.G., Heimstetten,
Germany) for stereotactic body radiation therapy (SBRT). Application of
laparoscopic lymphadenectomy allows for omission of prophylactic irradiation of the pelvic nodes for the majority of cases. Application of conformal
radiation therapy (CRT) techniques, such as dynamic eld shaping arc or
IMRT and accurate target positioning by image-guided radiation therapy
(IGRT) techniques, allow for a dose delivery with the highest possible
precision (3437). A preliminary follow-up based on the rst 100 patients
treated showed an almost complete absence of acute digestive complications. The use of reduced treatment volumes does not seem to compromise
outcome as illustrated by the two-year biochemical control rate of 96%.
TREATMENT PROTOCOL AND IRRADIATION TECHNIQUE
The current protocol is limited to patients that present with a localized
curative disease (T1-3N0M0) (38). For this patient cohort three prognostic
factors are important for the appropriate choice of treatment modality:
T-stage, PSA, and Gleason score (3942). Based on these criteria a classication into three groups is possible. Based on somewhat differing data from
Prostate Tumors
247
literature (2,3,7,43,44), these classes have been dened as follows at the AZVUB: good prognosis (combination of PSA 10 ng/mL, stage T1-2 and
Gleason score 2-6), bad prognosis (PSA 20 ng/mL combined with stage
T3-4 or Gleason score 7), and intermediate prognosis (all other cases).
For the cohort with good prognosis, local treatment, or in case of elderly
patients, a wait-and-see approach are considered adequate after discussion
with the patient provided a close follow-up schedule is applied. Those
patients from the intermediate cohort (in particular those with life expectancy of 10 years or more) will be proposed for a local treatment with curative intent. Patients with bad prognosis are treated with loco-regional RT.
T3-4 patients from the intermediate and bad prognosis groups will additionally receive hormone therapy (4547). Local treatment usually consists of
surgery or (external) RT at the AZ-VUB. Surgery has the psychological
advantage of removal of the tumor, but a recent meta-analysis showed
75% occurrence of impotence (50). Radiotherapy has the advantage of being
non-invasive, yet suffers from other disadvantages. Acute complications
might occur such as irritation of bladder, and (depending on the irradiate
bowel volume) abdominal cramps and diarrhea. In the long term, RT might
be the cause of chronic radiocystitis, erectile disfunction, and chronic radiorectitis. In recent years it has been shown that with radiation doses of up to
78 Gy a higher biochemical control can be achieved compared to conventional dose levels of 70 Gy, especially for the intermediate prognosis cohort
(3,43,4850). Needless to say that with these higher doses limitation of the
irradiated volume of healthy tissue becomes more important (3,5153).
With the introduction of CRT, and IGRT, the AZ-VUB has
attempted to optimize the irradiation technique for a complication-free
prostate treatment with curative intent. Based on the above rationale the
low-risk (PSA 10 and T1-2 and Gleason 26) and highrisk (PSA > 20
and T3-4 or Gleason 710) patients will be treated with 70 Gy, whereas
the intermediate cohort receives 78 Gy. Three distinctive measures can be
identied: limitation of the irradiated volume, application of CRT or
IMRT, and high accurate target positioning or IGRT.
a. Limitation of the irradiated volume. Previously, most patients
received a 50 Gy prophylactic irradiation of the pelvic nodes,
with known problems such as abdominal pain and diarrhea.
Additionally, a 20 Gy boost was administered to the prostate
and seminal vesicles. Currently, based on the threefold classication T-PSA-Gleason, the probability of microscopic disease of pelvic nodes and seminal vesicles can be assessed (54,55). For those
patients that present a low risk (i.e., 10% for the AZ-VUB) prophylactic irradiation will be omitted. A laparoscopic lymphadenectomy will be suggested for those patients with a risk that
exceeds 10%. When the No status is conrmed, the prophylactic
248
Miralbell et al.
pelvic irradiation is omitted and the seminal vesicles are not

included in the irradiated volume when the risk of microscopic
involvement is lower than 10%. When the latter is not the case,
only the proximal half of the seminal vesicles will be included in
the CTV most patients present with T1-2 tumors of which is
known that the possible involvement is limited to this area (56).
b. IGRT. Based on a three-step positioning study with the ExacTrac/Novalis BODY system (see previous chapter), the PTV
margins have been adapted to the precision that can be realized
with the different positioning techniques. All prostate patients
are positioned with the stereoscopic X-ray system and based on
the previous studies (34,35,37) the following margins are being
applied: (1) For patient positioning on bony structures (fusion
of X-ray images and DRRs) a 6 mm margin latero-lateral, and
10 mm antero-posterior and cranio-caudal. (2) For patients with
implanted radio-opaque markers a 5 mm margin is applied in the
antero-posterior and cranio-caudal direction, whereas a 3 mm
margin in leftright direction is used. Again, the IGRT will help
to reduce the irradiated volume.
c. Irradiation technique. To obtain a dose distribution true to the
shape of the PTV, two techniques possible with the Novalis
system are applied at the AZ-VUB. Depending on the absence
or presence of concavities, respectively, dynamic eld shaping
arc or IMRT are used (37). The former consists of one coplanar
arc of which the leaves dynamically adapt to the shape of the PTV
during rotation of the arc. In many cases the leaves showing overlap with the rectum at gantry position 90 and 270 are manually
retracted to the border of the rectum using beams-eye-view to fulll
the dose limitations to the rectum. The IMRT technique consisting
of ve non-opposing beams is explained in more detail in a previous
chapter on Novalis. In the AZ-VUB experience (actual 180 patients
treated, October 2003) only ve required IMRT to meet the
requirements with respect to toxic rectum dose. The rectal dose
constraints are a maximum of 74 Gy to the entire volume of the
rectum, less than 70 Gy to one-quarter and less than 65 Gy to half
of the rectum.
CLINICAL FOLLOW-UP
Presented are preliminary data on the rst 100 patients treated with dynamic
eld shaping arc and IGRT (based on fusion of bony structures, ExacTrac
2.0/Novalis BODY) from May 2000 to October 2002. Follow-up is still
short: median follow-up of 12 months (329 months). The patient cohort
consisted of patients with relative good prognosis (90% T1-2, 73% Gleason
Prostate Tumors
249
Table 8 Characteristics of the First 100 Patients Treated on the Novalis System at
the AZ-VUB
T-stage
1
2
3
4
Unknown
Gleason score
46
44
8
0
2
26
7
810
Unknown
PSA (ng/mL)
73
16
6
5
04
>410
>1020
>20
Unknown
10
51
29
9
1
26) (Table 8). Thirty-four patients underwent lymphadenectomy prior to

treatment. For 69 patients, radiotherapy was the unique treatment modality
and the 31 remaining cases received (neo-)adjuvant hormonal therapy (27)
or presented local recurrence after prostatectomy (4). Sixty-ve patients
received a 70 Gy total dose, whereas 35 patients (intermediate prognosis)
were treated with a 78 Gy total dose. Acute toxicity (occurring the rst three
months from the treatment start with radiotherapy) was scored using the
RTOG/EORTC classication (57). The questionnaire was presented to the
patient on the last treatment day, one and four months after treatment. Even
unique intake of symptomatic medication was classied as being grade 2.
The biochemical control was based on the ASTRO (American Association
for Therapeutic Radiology and Oncology) consensus panel (58), which
classies recurrence after three consecutive PSA rises. Due to applicability of
this denition for RT patients only, the biochemical analysis was limited to this
cohort only (69 patients), whereas the entire group was analyzed for toxicity.
A biochemical control of disease was observed for 96% of the patients
treated with radiotherapy alone for two years. PSA dropped from
8.9 4.8 ng/mL (1 SD) preradiotherapy to 1.0 0.5 ng/mL (1 SD) 18 months
after nalizing RT. Only a low number of acute digestive complications
have been observed and the 12 grade 1 toxicities observed might have been
attributed to other causes such as a period of enteritis in the family of the
patient or simultaneous use of potential diarrhea stimulating medicaments
(Table 9). Three cases of anal pain were observed in this group of 12
patients (a side effect that is not included explicitly in the RTOG.EORTC
scoring system). Acute urinary complaints have been reported by 54% of
the patients, yet medication was needed for a minority only. All acute side
effects had disappeared at control consult four months after termination
of RT. No causal relationship has been found between acute side effects
and treatment dose. The preliminary nature of the present data does not
allow for late side effects and a long-term follow-up is needed. So far,
one patient developed a chronic side effect: radiocystitis with macroscopic hematuria. This patient received coagulation and is currently free
of complaints.
250
Miralbell et al.
Table 9 Acute Toxicity Observed on 100 Patients Treated

on the Novalis System at the AZ-VUB
Grade
0
1
2
3
4
Digestive
Urinary
88
12
0
0
0
46
39
15
0
0
Although currently no clear causal relationship between biochemical

control and survival has been proven the predictive value with respect to
local control and metastasis has been shown (22,59), and PSA allows for
an early predictive evaluation of treatment for prostate carcinoma. The
96% biochemical control obtained in this study has to be interpreted as a
preliminary result and indeed it has been shown that with increasing follow-up time these gures have a tendency to decrease (60). Typically a slow
decreasing progression of the PSA is observed where the nadir is reached
after one year or later (61). No consensus has been reached as to which nadir
corresponds to disease-free status and values between 0.6 and 1.4 ng/mL
have been suggested (62). Nonetheless, a PSA value < 0.5 ng/mL ve-year
postradiotherapy may not warrant absolute cure (63). Concerning late side
effects follow-up is again too short to allow for valid conclusions. Side
effects in radiotherapy are generally related to dose and irradiate volume
of healthy tissue (64), comorbidity, and an increased risk seems to occur
after surgery for diabetics (65,66). Grade 2 acute toxicity (i.e., in need of
medication) has been observed in 4560% of the patients treated conventionally with prophylactic pelvic irradiation, whereas this fraction decreases
to 2530% with conformal irradiation techniques (48,67). Patients at the
AZ-VUB have not only been irradiated using CRT techniques but also
IGRT has been routinely applied on all patients, which might explain the
reported low gure of 15% acute grade 2 side effects. The use of implanted
radio-opaque markers for increased image-guided target localization (including internal organ movement) is being investigated at the AZ-VUB of
which clinical follow-up is currently on-going.
REFERENCES
1. Hanks GE, Martz KL, Diamond JJ. The effect of dose on local control of
prostate cancer. Int J Radiat Oncol Biol Phys 1988; 15:12991306.
2. Zelefsky MJ, Leibel SA, Gaudin PB, Kutcher GJ, Fleshner NE, Venkatramen
ES, Reuter VE, Fair WR, Ling CC, Fuks Z. Dose escalation with three-dimensional conformal radiation therapy affects the outcome in prostate cancer. Int
Prostate Tumors
251
3. Pollack A, Zagars GK, Starkschall G, Antolak JA, Lee JJ, Huang E, von
Eschenbach AC, Kuban DA, Rosen I. Prostate cancer radiation dose response:
results of the M.D. Anderson phase III randomized trial. Int J Radiat Oncol
Biol Phys 2002; 53:10971105.
4. Fiveash JB, Hanks G, Roach M, Wang S, Vigneault E, McLaughlin PW,
Sandler HM. 3D conformal radiation therapy (3DCRT) for high grade prostate
cancer: a multi-institutional review. Int J Radiat Oncol Biol Phys 2000; 47:
335342.
5. Vicini FA, Abner A, Baglan KL, Kestin LL, Martinez AA. Dening a doseresponse relationship with radiotherapy for prostate cancer: is more really
better? Int J Radiat Oncol Biol Phys 2001; 51:12001208.
6. Cheung R, Tucker SL, Dong L, Kuban D. Dose-response for biochemical
control among high-risk prostate cancer patients after external beam radiotherapy. Int J Radiat Oncol Biol Phys 2003; 56:12341240.
7. Zelefsky MJ, Hollister T, Raben A, Matthews S, Wallner KE. Five-year biochemical outcome and toxicity with transperineal CT-planned permanent
1125 prostate implantation for patients with localized prostate cancer. Int
8. Lebesque JV, Bruce AM, Kroes APG, Touw A, Shouman T, van Herk M.
Variation in volumes, dose-volume histograms, and estimated normal tissue
complication probabilities of rectum and bladder: implications for treatment
planning. Int J Radiat Oncol Biol Phys 1995; 33:11091119.
9. Roeske JC, Forman JD, Mesina CF, He T, Pelizzari CA, Fontenla E, Vijayakumar S, Chen GTY. Evaluation of changes in the size and location of the prostate, seminal vesicles, bladder, and rectum during a course of external beam
radiation therapy. Int J Radiat Oncol Biol Phys 1995; 33:13211329.
10. Rudat V, Schraube P, Oetzel D, Zierhuit D, Flentje M, Wannenmacher M.
Combined error of patient positioning variability and prostate motion uncertainty in 3D conformal radiotherapy of localized prostate cancer. Int J Radiat
Oncol Biol Phys 1996; 35:10271034.
11. Dawson LA, Man K, Franssen E, Morton G. Target position variability
throughout prostate radiotherapy. Int J Radiat Oncol Biol Phys 1998; 42:
11551161.
zsoy O, Pugliesi A, et al. Weekly CT-control in 3-D conformal
12. Miralbell R, O
radiotherapy of prostate cancer: dosimetric implications of changes in repositioning and organ motion. Radiother Oncol 2003; 66:197202.
13. Chen ME, Johnston DA, Tang K, Babaian RJ, Troncoso P. Detailed mapping
of prostate carcinoma foci. Biopsy strategy implications. Cancer 2000; 89:
18001809.
14. Coakley FV, Kurhanewicz J, Lu Y, Jones KD, Swanson MG, Chang SD,
Carroll PR, Hricak H. Prostate cancer tumor volume: measurement with endorectal MR and MR spectroscopic imaging. Radiology 2002; 223:9197.
15. Hara T, Kosaka N, Kishi H. PET imaging of prostate cancer using carbon-11
choline. J Nucl Med 1998; 39:990995.
16. Oyama N, Akino H, Kanamaru H, Suzuki Y, Muramoto S, Yonekura Y,
Sadato N, Yamamoto K, Okada K. 11C Acetate PET imaging of prostate cancer. J Nucl Med 2002; 43:181186.
252
Miralbell et al.
17. Price DT, Coleman E, Liao RP, Robertson CN, Polascik TJ, de Grado TR.
Comparison of 18F Fluorocholine and 18F Fluorodeoxyglucose for positron
emission tomography of androgen dependents and androgen independent prostate cancer. J Urol 2002; 168:273280.
18. Fowler J, Chappell R, Ritter M. Is alfa/beta for prostate tumors really low? Int
19. Brenner DJ, Martinez AA, Edmundson GK, Mitchell C, Thames HD, Armour
EP. Direct evidence that prostate tumors show high sensitivity to fractionation
(low DD ratio) similar to late-responding normal tissue. Int J Radiat Oncol Biol
Phys 2002; 52:613.
20. Brenner DJ. Hypofractionation for prostate cancer radiotherapywhat are the
issues. Int J Radiat Oncol Biol Phys 2003; 57:912914.
21. Logue JP, Hendry JH. Hypofractionation for prostate cancer. Int J Radiat
22. Kupelian PA, Reddy CA, Klein EA, Willouughby TR. Short-course intensitymodulated radiotherapy (70 Gy at 2.5 Gy per fraction) for localized prostate
cancer: preliminary results on late toxicity and quality of life. Int J Radiat
23. Yeoh EEK, Fraser RJ, McGowan RE, Botten RJ, Di Matteo AC, Roos DE,
Penniment MG, Borg MF. Evidence for efcacy without increased toxicity of
hypofractionated radiotherapy for prostate carcinoma: early results of a phase
III randomized trial. Int J Radiat Oncol Biol Phys 2003; 55:943955.
24. Parker C, Warde P, Catton C. Salvage radiotherapy for PSA failure after radical prostatectomy. Radiother Oncol 2001; 61:107116.
25. Popowski Y, Kebdani T, Taussky D, Rouzaud M, Nouet P, Miralbell R. Dose
escalation with high-dose rate (HDR) brachytherapy boost in prostate cancer:
preliminary results. 7th Annual Meeting of the Scientic Association of Swiss
Radiation Oncology, Geneva, Switzerland, Apr 37, 2003.
26. Roach M, Marquez C, Yuo H-S, Narayan P, Coleman L, Nseyo UO, Navvab Z,
Carroll PR. Predicting the risk of lymph node involvement using the pre-treatment prostate specic antigen and Gleason score in men with clinically localized
27. Low A, Harms W, Mutic S, Purdy J. A technique for the quantitative evaluation of dose distributions. Med Phys 1998; 25(5):656661.
28. Depuydt T, Van Esch A, Huyskens DP. A qualitative evaluation of IMRT distributions: renement and clinical assessment of the gamma evaluation. Radiother Oncol 2002; 62:309319.
29. Perez CA, Brady LW. Overview. Quantication of treatment toxicity. In:
Perez CA, Brady LW, eds. Principles and Practice of Radiation Oncology.
Philadelphia: Lippincott, 1992:5155.
30. Pavy J-J, Denekamp J, Letschert J, Littbrand B, Mornex F, Bernier J,
Gonzalez-Gonzalez D,Horiot J-C, Bola M, Bartelink H. Late effects toxicity
scoring: the SOMA scale. Int J Radiat Oncol Biol Phys 1995; 31:10431047.
31. Miralbell R, Molla M, Arnalte R, Canales S, Vargas E, Linero D, Waters S,
Nouet P, Rouzaud M, Escude L. Target repositioning optimization in prostate
cancer: is intensity modulation radiotherapy under stereotactic conditions
feasible? Int J Radiat Oncol Biol Phys 2004; 59:366371.
Prostate Tumors
253
32. McKenzie AL, van Herk M, Mijnheer B. The width of margins in radiotherapy
treatment plans. Phys Med Biol 2000; 45:33313342.
33. Nahun AE, Movsas B, Horwitz EM, Stobbe CC, Chapman JD. Incorporating
clinical measurements of hypoxia into tumor local control modeling of prostate
cancer: implications for the a/b ratio. Int J Radiat Oncol Biol Phys 2003;
53:391401.
34. Soete G, Van de Steene J, Verellen D, et al. Initial clinical experience with infrared reecting skin markers in the positioning of patients treated by conformal
radiotherapy for prostate cancer. Int J Radiat Oncol Biol Phys 2002; 52(3):
694698.
35. Soete G, Verellen D, Michielsen D, et al. Clinical use of stereoscopic X-ray
positioning of patients treated with conformal radiotherapy for prostate cancer.
36. Verellen D, Linthout N, Soete G, et al. Considerations on treatment efciency
of different conformal radiation therapy techniques for prostate cancer.
37. Verellen D, Soete G, Linthout N, et al. Quality assurance of a system for
improved target localization and patient set-up that combines real-time
infrared tracking and stereoscopic X-ray imaging. Radiother Oncol 2003; 67:
129141.
38. UICC International Union Against Cancer. TNM Classication of Malignant
Tumours. New York: Wiley-Liss, 1997.
39. DAmico AV, Whittington R, Malkowicz SB. Pretreatment nomogram for
prostate-specic antigen recurrence after radical prostatectomy or external
beam radiation therapy for clinically localized prostate cancer. J Clin Oncol
1999; 17:168172.
40. Gleason DF. Classication of prostatic carcinomas. Cancer Chemother Rep
1966; 50:125.
41. Kattan MW, Zelefsky MJ, Kupelian PA, et al. Pretreatment nomogram for
predicting the outcome of three-dimensional conformal radiotherapy in prostate cancer. J Clin Oncol 2000; 18:33523359.
42. Shipley WU, Thames HD, Sandler HM, et al. Radiation therapy for clinically
localized prostate cancer. A multi-institutional pooled analysis. JAMA 1999;
281:15981604.
43. Lyons JA, Kupelian PA, Mohan DS, et al. Importance of high radiation doses
(72 Gy or greater) in the treatment of stage T1T3 adenocarcinoma of the prostate. Urology 2000; 55:8590.
44. Hanks GE, Schultheiss TE, Hunt MA, et al. Factors inuencing incidence of
acute grade 2 morbidity in conformal and standard radiation treatment of
45. Bolla M, Collette L, Blank L, et al. Long-term results with immediate androgen
suppression and external irradiation in patients with locally advanced prostate
cancer (an EORTC study): a phase III randomised trial. Lancet 2002; 360:
103106.
46. Lawton CA, Winter K, Murray K, et al. Updated results of the phase III radiation therapy oncology group (RTOG) trial 8531 evaluating the potential benet
of androgen suppression following standard radiation therapy for unfavorable
254
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
Miralbell et al.
prognosis carcinoma of the prostate. Int J Radiat Oncol Biol Phys 2001; 49:
937946.
Pilepich MV, Winter K, John MJ, et al. Phase III radiation therapy oncology
group (RTOG) trial 8610 of androgen deprivation adjuvant to denitive radiotherapy in locally advanced carcinoma of the prostate. Int J Radiat Oncol Biol
Phys 2001; 50:12431252.
Hanks GE, Hanlon AL, Pinover WH, et al. Dose selection for prostate cancer
patients based on dose comparison and dose response studies. Int J Radiat
Pollack A, Smith LG, von Eschenbach AC. External beam radiotherapy dose
response characteristics of 1127 men with prostate cancer treated in the PSA
era. Int J Radiat Oncol Biol Phys 2000; 48:507512.
Zelefsky MJ, Hollister T, Raben A, Matthews S, Wallner KE. Five-year biochemical outcome and toxicity with transperineal CT-planned permanent
1125 prostate implantation for patients with localized prostate cancer. Int
Boersma LJ, van den Brink M, Bruce AM, et al. Estimation of the incidence of
late bladder and rectum complications after high-dose (7078 Gy) conformal
for prostate cancer, using dosevolume histograms. Int J Radiat Oncol Biol
Phys 1998; 41:8392.
Schultheiss TE, Lee WR, Hunt MA, et al. Late GI and GU complications
in the treatment of prostate cancer. Int J Radiat Oncol Biol Phys 1997;
37:311.
Storey MR, Pollack A, Zagars G, et al. Complications from radiotherapy doseescalation in prostate cancer: preliminary results of a randomized trial. Int
Pisansky TM, Zincke H, Suman VJ, et al. Correlation of pretherapy prostate
cancer characteristics with histologic ndings from pelvic lymphadenectomy
specimens. Int J Radiat Oncol Biol Phys 1996; 34:3339.
Pisansky TM, Blute ML, Suman VJ, et al. Correlation of pretherapy prostate
cancer characteristics with seminal vesicle invasion in radical prostatectomy
specimens. Int J Radiat Oncol Biol Phys 1995; 36:585591.
Kestin L, Goldstein N, Vicini F, et al. Treatment of prostate cancer with radiotherapy: should the entire seminal vesicles be included in the clinical target
volume? Int J Radiat Oncol Biol Phys 2002; 54:686697.
Cox JD, Stetz J, Pajak TF, et al. Toxicity criteria of the Radiation Therapy
Oncology Group (RTOG) and the European Organization for Research and
Treatment of Cancer (EORTC). Int J Radiat Oncol Biol Phys 1995; 31(5):
13411346.
ASTRO Consensus Panel. Consensus statement: guidelines for PSA following
radiation therapy. Int J Radiat Oncol Biol Phys 1997; 37:10351041.
Horwitz EM, Vicini FA, Ziaja EL, et al. The correlation between the ASTRO
consensus panel denition of biochemical failure and clinical outcome for patients with prostate cancer treated with external beam irradiation. Int J Radiat
Prostate Tumors
255
60. Vicini FA, Kestin LL, Martinez AA. The importance of adequate follow-up in
dening treatment success after external beam irradiation for prostate cancer.
61. Kabalin JN, Hodge KK, McNeal JE, et al. Identication of residual cancer in
prostate following radiation therapy: role of transrectal ultrasound guided
biopsy and prostate specic antigen. J Urol 1989; 142:326331.
62. Van Cangh PJ, Richard FJ. Prostate-specic antigen after denitive radiation
therapy. Curr Opin Urology 1994; 4:256.
63. Yock TI, Zietman AL, Shipley WU, et al. Long-term durability of PSA failurefree survival after radiotherapy for localized prostate cancer. Int J Radiat Oncol
Biol Phys 2002; 54:420426.
64. Emami B, Lyman J, Brown A, et al. Tolerance of normal tissue to therapeutic
irradiation. Int J Radiat Oncol Biol Phys 1991; 21:109122.
65. Douchez J, Allain YM, Cellier P, et al. Cancer de la prostate: intolerance et
morbidite de la radiotherapie externe. Bull Cancer 1985; 72:573577.
66. Herold DM, Hanlon AL, Hanks GE. Diabetes mellitus: a predictor for late
radiation morbidity. Int J Radiat Oncol Biol Phys 1999; 43:475479.
67. Vijayakumar S, Awan A, Karrison T, et al. Acute toxicity during external-beam
radiotherapy for localized prostate cancer: comparison of different techniques.
11
ECSRT for Spinal Tumors
Samuel Ryu
Department of Radiation Oncology and Neurosurgery, Henry Ford Hospital,
Detroit, Michigan, U.S.A.
Fang-Fang Yin
Department of Radiation Oncology, Duke Medical Center, Detroit,
Michigan, U.S.A.
INTRODUCTION AND CLINICAL INDICATIONS

Tumors of the Spine and the Spinal Cord
Primary tumors of the spine and the spinal cord are rare, and these include
the tumors arising from the vertebral bone, hematopoietic tumors like myeloma, or tumors arising from the cord itself. Metastatic tumors are the most
common involving the spine or spinal cord. Metastatic disease is a common
complication of cancer during the course of the disease in the majority of
cancer patients. Among these, the vertebral bones are the most common site
and may be involved in about 40% of the cancer patients (1). Up to 20% of
patients with neoplastic involvement of the vertebral column develop spinal
cord compression (2). It is estimated that more than 20,000 cancer patients
per year in the United States develop spinal cord or root compression as a
manifestation of their metastatic disease (35). Compression of the spinal
cord or nerve roots is second only to brain metastasis as the most frequent
neurologic complication of cancer (6,7). Usually, vertebral bone metastasis
occurs prior to developing compression to the spinal cord. The estimated
number of patients diagnosed with spinal metastases has increased with
the use of CT and MRI scans (3,4). The common presenting symptoms of
spine metastasis are back pain and spinal cord compression with weakness
and numbness below the level of spinal involvement. Many patients, if not
treated, become paraplegic or lose control of bowel and bladder, which
results in signicant morbidity and poor quality of life. Early treatment
257
258
Ryu and Yin
prior to the development of signicant neurologic decits improves the

chance for patients to remain ambulatory (8). Because cancer patients are
now experiencing longer survival, quality of life has become an important
factor in making the treatment decision and offering palliative treatment.
This is particularly true in the treatment of spine metastasis since pain
and neurologic symptoms further reduce the quality of life.
Treatment for spinal metastasis and/or cord compression has been
external beam radiotherapy and/or decompressive surgery. External beam
radiation therapy is offered, initially combined with steroids, in the vast
majority of patients. The effectiveness of external beam radiotherapy has
been well established. Varying degrees of pain relief were seen in two-thirds
of patients by three months after radiation (9,10). Surgery is usually offered
for a rapid neurologic change, spinal instability, or for tissue diagnosis
(11,12). Surgery has also been used for more aggressive and radioresistant
tumors, combined with radiation. Chemotherapy is offered for chemosensitive tumors such as small cell carcinomas or lymphomas. Whatever treatment is chosen, the intent of treatment for spine metastasis or cord
compression has been palliative. The main goals of the treatment are relief
or reduction of pain, improvement of neurological function, or to limit
tumor progression, if possible. Although there have been many efforts to
improve the treatment of these tumors by using multimodality treatments,
the clinical outcome of the available treatment for spinal tumors is still limited compared to the progress of treatments for the other tumor sites.
Radiosurgery of the Spine
Radiosurgery delivers a highly conformal large radiation dose to a localized
tumor by a stereotactic approach. This requires accurate targeting and
immobilization of the target organ during irradiation. The difculty of
applying radiosurgery to the extracranial site is mainly due to organ motion
associated with breathing and/or lack of immobilization techniques. Among
the extracranial organs, the spinal cord and vertebrae are the organs with
the least breathing-related organ movement. This makes the spinal cord
and vertebrae particularly suitable for stereotactic radiosurgery.
Stereotaxis has been used exclusively for cranial applications for several decades. However, interestingly, the rst attempt of stereotactic localization began, long before brain stereotaxy, with experimental sites of the
spinal cord by Dittmar and Worosciloff in the late 19th century, although
these systems were not based on a true coordinate system (13). The history of
the spinal stereotaxy is summarized in Table 1. The rst stereotactic instrument using Cartesian coordinate system for the spinal cord was reported
by Clarke in 1920 (14). Rand et al. performed the rst stereotactic procedure
on the human spinal cord for percutaneous cordotomy and cryosurgery in
1965 (15). All the spinal stereotaxy was limited to the cervicomedullary
Spinal Tumors
259
Table 1 History of Spine Stereotaxy

1873
1874
Dittmar
Woroschiloff
1920
Clarke
1947
1951
1965
Spiegel
Leksell
Rand
1972
Nadvornik
1982
1994
1995
2000
Betti
Lax
Hamilton
Device for directing probes to the spinal cord

Myelotome using clamps for skeletal xation in
animal
First spinal stereotactic device using Cartesian
coordinate system
First cranial stereotactic frame
First stereotactic radiosurgery for brain
First clinical stereotactic cryosurgery of
cervical cord
First spinal radiosurgery for lumbar vertebrae with
skeletal xation
Linear accelerator modied for radiosurgery of brain
Body frame with contour mold xation
Skeletal xation frame with bone-screw xation
Frameless spinal radiosurgery
region because the device was xed to the skull and open surgery was needed
to anchor the cradles to the bony structures of the spine. In 1972, Nadvornik
et al. expanded its clinical usefulness with an apparatus designed for lumbar
region (16). It also required xation to the vertebral arch in an open laminectomy wound. Since then, spinal radiosurgery has been used in limited numbers of cancer patients (17). The method used by Takacs and Hamilton
(18) and Hamilton et al. (19) was also an invasive procedure that required
anchoring of the stereotactic frame to the spinous process under general
anesthesia. Despite its invasiveness, the clinical outcome was encouraging.
More recently, image-guided frameless stereotactic technology of the
spine has been developed and used in the clinic to treat the patients with
spinal tumors. There are basically two different methods to achieve stereotaxy without frames by using the implanted seed as a point ducial, or using
the internal rigid bony structure (such as vertebral bone or target tumor
itself) as a volume ducial. The former system is used in CyberKnife unit
(Accuray, Sunnyvale, California, U.S.A.), in which mean total radial error
was 1.6 mm and the positioning error along each axis was 0.9 mm in the
phantom (20). This also has an invasive component with implantation of
a few metal markers in the patient to help determine the target. The later
system of frameless stereotaxy is the Novalis shaped beam radiosurgery
unit (BrainLAB, AG, Heimstetten, Germany). This system utilizes image
fusion of anatomical structures such as vertebral bone and infrared marker
technology for tumor localization and patient positioning. This procedure is
entirely non-invasive. The volume of the xed bony structure is registered at the
260
Ryu and Yin
time of simulation and used for image fusion. The precision of this system was
reported with less than 2 mm of isocenter variation for intensity-modulated
spinal radiosurgery (21). This radiosurgery procedure and clinical experience
with encouraging clinical results will be discussed in this chapter.
TECHNIQUES FOR SPINAL RADIOSURGERY
The challenge for spinal radiosurgery is how to accurately localize the
treatment target and deliver the prescribed dose to the treatment target
while keeping the dose to the normal tissues within the tolerances. The success of stereotactic radiosurgery for brain tumors is mainly attributed to its
precision in patient immobilization, target localization, and conformal dose
delivery. Less than 1 mm positioning error for the head xation is technically achievable because the head could be treated as a quasi-rigid body with
negligible organ motion (22). However, some of techniques used for the
brain radiosurgery will not be feasible for spinal tumors due to difculties
in immobilizing non-rigid patient body, localizing targets associated with
organ motion, and mechanical limits associated with allowable gantry and
couch rotations. Traditional arc techniques used for brain radiosurgery
may not be suitable to spinal radiosurgery. For spinal tumors, critical
organs such as spinal cord, kidney, and lung are always adjacent to the targets and present technical challenges for protection. Therefore, different
techniques should be adopted for spinal radiosurgery, including new
approaches for patient immobilization, target localization, and treatment
planning and delivery. The procedures developed at Henry Ford Hospital
are schematically illustrated in Figure 1.
Figure 1 Schematic illustration of clinical procedures for spinal radiosurgery.
Spinal Tumors
261
Immobilization and Simulation

Immobilization involves patient immobilization and organ/structure immobilization. Patient immobilization during spinal radiosurgery is the rst but
critical step, because a single or a few fractional high doses of radiation will
be delivered to spinal or para-spinal tumors surrounded by critical normal
organs such as the spine. Radiosurgery typically takes more time than conventional treatment and therefore requires the patient to be rigid and stable
throughout the treatment. Any patient movement during the treatment
could lead to the failure of the radiosurgery. Traditional immobilization
devices, such as alpha cradle and vacuum bags, may not be sufcient for
patient immobilization because they do not have constraints to limit patient
movement. The ideal immobilization devices should be able to provide some
type of constraints to the patient but also be feasible for all kinds of patients
to stay at the same position for a period of time up to 1 hr. Several approaches
have been used for immobilization during spinal radiosurgery. The early
approach used a metal frame to x the vertebral bodies because of the relatively xed relationship among spinal tumors, spinal cord, and vertebral
bodies (19). Therefore, accurate localization of vertebral bodies is equivalent
to the accurate localization of the target region. Although reasonable accuracy is achievable in terms of patient immobilization, it is an invasive surgical
procedure with anesthetic involvement. Several alternative non-invasive
immobilization devices are now available such as BodyFIX (Medical Intelligence, Schwabmunchen, Germany) and Stereotactic Body Frame (ELEKTA
AB, Stockholm, Sweden). The BodyFIX device combines a vacuum cushion
and a piece of special plastic foil (21,23). The vacuum cushion is used to stabilize the patient in a comfortable position while the plastic foil is used to constrain the patient by evacuating the region between the patient and plastic foil
so that the patient is constrained between the vacuum cushion and plastic foil.
The Body Frame uses a vacuum cushion to stabilize the patient in the supine
position inside a body frame and adds a diaphragm control to the patients
chest to minimize respiratory movement (24). Figure 2 illustrates immobilization setups using the BodyFIX and Body Frame devices.
The effect of organ motion is relatively minimal for spinal radiosurgery because the spinal tumors are typically located very close to vertebral
bodies. When the patient is in the supine position, the vertebral body motion
is limited to within 1 mm (21). This effect could be documented by uoroscopic imaging using a conventional simulator. However, if the patient is in
the prone position, the vertebral body motion could be more than 5 mm
due to breathing. Therefore, it is ideal to position the patient supinely for
spinal radiosurgery.
CT simulation has been the gold standard for conventional radiosurgery for brain tumors. It provides not only accurate patient geometry
but also three-dimensional (3D) anatomical information for dose calculation.
262
Ryu and Yin
Figure 2 Illustration of (A) BodyFIX and (B) Body Frame immobilization

devices.
Spinal Tumors
263
Similarly, CT simulation should be used for spinal radiosurgery. To accommodate complicated immobilization devices and variety of patient sizes, a
large bore CT simulator as supplied by both Philips Medical Systems (Andover, Maryland U.S.A.) and GE Healthcare Technologies (Waukesha, Wesconius, U.S.A.) would be preferable. Any localization markers should be
placed on or implanted in the patient prior to CT imaging. For example, if
infrared markers are used for patient setup and localization, infrared sensitive radiopaque markers should be placed on the patient skin so that they
are shown in the CT images for patient localization in the process of treatment planning (21). The relative location between the localization markers
and the planned isocenter could be established in the treatment planning
system using this technique.
Contrast agents such as Optiray 200 mg/mL organically bound iodine
are typically used to enhance tumors in the CT images. Enough patient body
portions should be scanned in simulation to assist accurate identication of
vertebral bodies and to provide sufcient anatomy for the use of non-coplanar
planning beams. A slice thickness of 3 mm or less without spacing should be
selected for scanning. When slice thickness is larger than 5 mm, the quality of
digitally reconstructed image (DRR) as calculated from CT images is shown to
be not good enough for accurate image fusion between DRRs and kV X-ray
images which are taken for the localization of isocenter in the treatment room
(25). To minimize the data transfer burden and potential errors, the simulation
DICOM 3.0 CT images should be electronically sent to the dedicated
treatment planning system through the internal network (26).
Treatment Planning
Treatment prescription involves both volume and dose. In addition to the
gross target volume (GTV), critical organs such as the spinal cord, kidneys,
and lungs are identied in CT images so that the radiation dose to these
organs can be minimized. The majority of GTVs involved one or two segments of the vertebral column. To accurately delineate the GTV for each
disease site, multiple imaging information is often referenced, especially contrast-enhanced MR images. Image fusion between CT and MR images needs
special caution and is often done manually to ensure the accurate anatomical
matching. Depending on the immobilization device used, proper margin
should be added to the GTV to generate the planning target volume (PTV).
When a BodyFIX device is used for immobilization, a margin of 23 mm
would be sufcient to accommodate patient positioning and target localization variations (21). Typically, this expansion is not extended into the critical
organs, especially not the spinal cord. Note that the dose distributions at the
joints between the target and the critical organs in an inverse plan are determined by their relative beam weights given in the prescription. Radiosurgery
dose is usually prescribed to the isodose line that encompasses the PTV.
264
Ryu and Yin
Treatment planning involves the construction of radiation beams to

deliver a conformal dose distribution. The treatment planning method is
inuenced by the delivery technique used. Non-coplanar beams are usually
used for treatment planning when a CyberKnife unit (Accuray, Sunnyvale,
California,U.S.A.) is applied fortreatment. Coplanar beams areusuallyapplied
for any tomotherapy treatment devices (NOMOS Corporation Chatsworth,
California, U.S.A., and TomoTherapy Inc., Madison, Wisconsin, U.S.A.).
When a conventional linac-based treatment unit such as Novalis Shaped
Beam unit (BrainLAB Inc., Westchester, Illinois, AG, Heimstetten, Germany)
or Varian Trilogy (Varian Medical Systems, Palo Alto, California, U.S.A.) is
used, the majority of treatment plans for the spinal radiosurgery use multiple
intensity-modulated beams to minimize the dose to critical organs. Sometimes, dynamic shaped conformal arcs may be used when the avoidance of
critical organs is not a high priority, such as intra-spinal tumors and tumors
involving the sacral portion of the vertebral column. Both coplanar and
non-coplanar intensity-modulated beams are used for spinal radiosurgery.
Typically, coplanar beams are used for tumors located close to the thoracic,
lumbar, and sacral sections of the vertebral column while non-coplanar beams
are sometime used for tumors near the cervical vertebral bodies.
Intensity-modulated beams are commonly used for spinal radiosurgery and are generated using an inverse treatment-planning algorithm, such
as the dynamically penalized likelihood method and a pencil beam dose calculation algorithm by BrainLAB (27,28). Dosevolume histograms (DVHs)
are traditionally used to specify the dose constraints for both target volumes
and critical organs.The DVH is also used to evaluate treatment plans. For a
given set of dose constraints specied by DVHs, the outcome of the planning results is inuenced by the setting of priority weights for both the target
and the critical organs. Typically, the higher the priority weight, the better
the dose coverage to each structure. An additional parameter, the conformity index (C), may be used to evaluate treatment plans. Here, the conformity index could be dened as C 1(Vn/Vt). Vn is the volume of the
normal tissue and Vt is the volume of the target receiving the indicated dose
(21,29). An acceptable plan is judged by the reasonable compromise
between the doses to the target and critical organs. This compromise is
determined by the setting of DVH constraints and priority weights.
When an inverse treatment plan is delivered using a multileaf collimator (MLC) during intensity-modulated radiosurgery (IMRS), the beam
number ranges from 5 to 9, with the majority being seven beams. The beam
orientation of a typical seven-coplanar intensity-modulated beam arrangement involves nearly equal gantry angles such as 150 , 100 , 50 , 0 , 310 ,
260 , and 210 with 0 at the anterior direction. Sometime, when the target
is not located too deeply, 5 to 6 IMRS beams could be set within a fan range
between 120 and 150 to avoid excessive dose to normal tissues. An example of
an intensity-modulated treatment plan with seven equally spaced beams is
Spinal Tumors
265
Figure 3 (A) An inverse plan using equally distributed seven beams and (B) corresponding DVHs. (See color insert.)
shown in Figure 3A and its corresponding dosevolume histogram is shown

in Figure 3B. An example of an intensity-modulated treatment plan with
six equally spaced fan beams is shown in Figure 4A and its corresponding
dosevolume histogram is shown in Figure 4B, where the dose is normalized
266
Ryu and Yin
Figure 4 (A) An inverse plan using ve beams and (B) corresponding DVHs.
(See color insert.)
Spinal Tumors
267
to the isocenter and the dose is prescribed to the volume included by the 90%
isodose line. The inhomogeneity corrections are included in dose calculation for
all treatment plans using IMRT. The number of the IMRT beams in a treatment
plan affects the dose distribution to the target as well as to the normal tissues.
Typically, dose distribution improves as beam number increases. However, this
trend becomes less obvious when the beam number is greater than nine.
After the radiation oncologist selects a suitable inverse treatment plan,
all treatment data should be electronically transferred to the treatment unit.
An electronic verication system, such as Varis, record-and-verify system
(Varian Oncology Systems) in the treatment unit will help to secure the
delivery accuracy. Similarly, when an imageguided system is used for
patient localization, such as Novalis Body system, the patient positioning
information could be electronically exported to the controlling system for
patient setup and target (or isocenter) localization (21,26).
Patient Setup and Target Localization
Accurate patient positioning and target localization are the keys for the success
of spinal radiosurgery. After the treatment plan is completed, the next step for
spinal radiosurgery is to reposition the patient in the treatment room as in the
simulation room and to align the planned isocenter to the treatment machine
isocenter. Since traditional methods such as skin marks are not quite capable
of providing high precision patient setup and target localization, various
image-guided techniques have been developed for patient setup, target localization, treatment monitoring, and verication. Among them are ultrasound
guided techniques, infrared camera imaging techniques, and kV X-ray imaging techniques (30,21,31). In-room kV X-ray imaging appears to be a very
promising approach for spinal radiosurgery. Some of the image-guided localization devices are the use of orthogonal dual kV imaging technique in the
CyberKnife unit (31) and a dual kV X-ray imaging system which is called
Novalis Body system in the Novalis unit (21). Differing from the
CyberKnife unit (as shown in Fig. 5A) in which the radiation beam orientation
as controlled by the robotic arm modies as the isocenter shifts as detected by
kV X-ray imaging, the Novalis unit (as shown in Fig. 5B) uses an image-guided
system to adjust patient positioning by moving the treatment couch.
The Novalis Body system as shown in Figure 5B consists of infrared and
video cameras and kV X-ray imaging system. The major functions for infrared
cameras are to detect infrared sensitive markers placed on the patient skin, to
automatically compare marker locations to the stored reference information,
and to instruct the treatment machine to move the patient to the preplanned
position by moving the treatment couch. A dedicated video camera system
is coupled to the infrared camera system to provide a visual check of
patient positioning. The two kV X-ray tubes and two amorphous silicon
(aSi) at panel digital detectors are controlled by an integrated computer
268
Ryu and Yin
Figure 5 Image-guided target localization systems in (A) CyberKnife system and

in (B) Novalis treatment unit.
Spinal Tumors
269
system with some image analysis software. Two digitally reconstructed

radiographs (DRRs) could be generated from the planning CT images at the
same orientations at which two kV X-ray images are taken. The system automatically compares internal structures such as vertebral bodies or implanted markers, in the kV X-ray images to those in the DRR images and indicates the
potential isocenter deviations between the DRRs and the X-ray images for
the patient setup. The deviations are evaluated by the radiation oncologist
and forwarded to the computer control system for recommended adjustment
of the isocenter position. It is an automated process because the Novalis Body
system is integrated with the treatment machine. This procedure could be
repeated during the treatment. Both anatomical structures and implanted
markers could be used for image fusion using either a manual or an automated
image fusion algorithm. Patient shift and rotational information could be
identied using a rigid body 3D to 2D image fusion technique in which DRRs
are generated iteratively with different angulations to simulate potential
patient deviations in all directions (32). Note that the accuracy of image fusion
based on anatomical markers using a rigid body 3D to 2D image fusion technique depends on liability of those structures used for matching in both types
of images. Regarding contrast, the image fusion based on the implanted markers is often more accurate and reliable (25,32). A cone-beam technology (33)
will allow three-dimensional comparison of treatment anatomy and should be
a very promising method for patient localization and treatment verication.
After the patient is immobilized on the treatment couch, the infrared
camera system guides the patients setup to the initial treatment position
based on the localization infrared sensitive markers placed on the patient
skin. Since the infrared markers only reect the patient surface information
and do not guarantee the accuracy of the planned isocenter, two kV X-ray
images are required to check the isocenter position. The internal structures
such as the vertebral bodies displayed in two X-ray images and those displayed in two corresponding DRRs are compared to indicate the relative
isocenter deviations from the planned isocenter position. After reviewing
the comparison result, the radiation oncologist decides whether an adjustment of the isocenter position is necessary. If so, the infrared camera system
will guide the patient to the adjusted position. With this image-guided
patient localization technique, accuracy of 1 mm is achievable as shown in
a study using a rigid-body phantom (25). However, depending on the immobilization technique used for the actual patient treatment, the localization
accuracy of 3 mm or less isocenter deviation is achievable. This accuracy
is considered to be acceptable if proper margin can be added.
Treatment Delivery
Accurate delivery of conformal dose as planned is another challenging step
for spinal radiosurgery. Traditional devices and arc techniques used
270
Ryu and Yin
for brain radiosurgery are not quite feasible for the majority of spinal
radiosurgeries. Various new delivery techniques are now available for
stereotactic radiosurgery of localized spinal tumors such as the use of a
CyberKnife technology (31), the use of a Novalis shaped beam surgery unit
(21), and the use of a linac with ne MLC (23). Other technologies such as
tomotherapy units may also be used for such treatment (34,35). The CyberKnife unit uses multiple circular cone beams and others use ne MLCs to
deliver intensity-modulated beams. For inverse plans delivered using different MLCs, the leaf width may potentially affect the dosimetry. However, the
dosimetric difference of intensity-modulated beams delivered using the
MLC leaf width of 5 mm or less is negligible (29).
The Novalis shaped beam unit is used for spinal radiosurgery at Henry
Ford Hospital and it is equipped with a built-in micromultileaf collimator
(mMLC) with a single 6 MV photon energy (36,37). There are 26 pairs of
leaves (14 pairs with a leaf width of 3 mm, six pairs with a leaf width of
4.5 mm, and six pairs with a leaf width of 5.5 mm) which form a maximum
eld size of 10 10 cm. It is capable of delivering radiation through circular
cone arcs, xed-shape conformal beams using mMLC, xed-shape conformal arcs using mMLC, dynamic shape conformal arcs using mMLC, and
xed-gantry with static and dynamic intensity-modulation beams. The dosimetric characteristics of this treatment unit are discussed in a separate
report (36). Intensity-modulated radiosurgery is capable of delivering conformal dose distribution to minimize radiation damage to the critical
organs. Typically, only a single isocenter is required for any kind of target
shape. Multiple isocenters may be required if the target size is larger than
the maximal eld size (for example, 10 cm 10 cm in the Novalis unit). Since
the entire process could be completed within a few hours, the procedure is
non-invasive, frameless, accurate, and efcient.
Either a sliding window (or dynamic MLC) or step-and-shoot technique
is used to deliver the intensity-modulated beams through the mMLC. Typically, there is no substantial difference between these two techniques if the
number of segments using the step-and-shoot technique is over 20 (21). Typically, a dose rate of 480 MU/min is used for delivery. The overall root-mean
square (RMS) of the leaf traveling accuracy by the use of this dose rate is
relatively, but not substantially, smaller than that by use of a dose rate of
800 MU/min. The analysis of RMS values in the MLC log les recorded in
the MLC workstation for a few typical IMRS plans shows this trend.
Precision delivery of high-dose radiation could be also achieved by
other means. The earlier reported spinal radiosurgery procedure combined
the surgical xation of patients with high-dose irradiation (19). In that
study, the procedure of surgically implanting the stereotactic xation device
to the vertebral body was necessary for both patient immobilization and
localization because the patient was positioned in the prone position.
High-dose radiation was planned and delivered based on the conventional
Spinal Tumors
271
techniques instead of IMRT. In an attempt to reduce the invasiveness of

spinal radiosurgery, the CyberKnife unit (31,38) uses a robotic arm to deliver 6 MV photon beam in a wide range of beam orientations, except at the
posterior region where the two X-ray detectors are located. The use of a different acceleration technique makes its treatment head much smaller than
the conventional accelerator gantry and, therefore, its head movement is
much more feasible using a robotic arm. The beam orientation is modied
and guided by two real-time imaging systems and the isocenter is modied
at any time when the imaging system indicates a deviation from the planned
isocenter. Therefore, delivery accuracy is largely dependent on the accuracy
of the detected signals and the accuracy of robotic arm movements. To
improve the accuracy, a few metal markers are implanted in the patient
so that the treatment target can be easily determined based on the markers
in X-ray images. This, however, adds an invasion component to this procedure. Since the treatment unit delivers radiation through various circular
cones, multiple isocenters and/or cones with different diameters have to
be used for irregular target volumes in a single treatment. Instead of using
intensity-modulated beams, the conformal dose distribution is delivered by
the use of numerous circular beams with variable isocenters.
Treatment Verification and Quality Assurance

Treatment verication involves verication of treatment plan, patient setup,
treatment isocenter, and dose calculation and delivery. Quality assurance
(QA) involves machine specic-QA and patient-specic QA. In the case
of intensity-modulated radiosurgery using MLCs, machine-specic QA
involves examination of MLC leaf sequence, leaf position, dosimetry, etc.
Patient-specic QA involves isodose measurement, intensity spectrum distribution, etc. The detail procedures of these methods are described in several
previous publications (39,40) and chapter 3. The imaging devices used for
patient setup and target localization should be properly calibrated based
on the requirements specied by the manufacturers (25,41).
Careful verication of the treatment plan involves verication of treatment parameters, such as isocenter position, beam geometry, monitor unit
calculation, beam intensity spectrum, and dose measurements. An independent dose calculation algorithm, based on traditional dose calculation methods such as modied Clarkson method (42) or Monte Carlo method (43), may
be used to calculate the isocenter dose contributed by each mMLC segment and its corresponding MUs for all segments of each intensitymodulated beam given by the planning system. Independent patient positioning
verication could be achieved by taking orthogonal images and compareing
them to simulation images. A pair of orthogonal portal lms could also be
used to document the nal patient positioning accuracy. Alternatively,
272
Ryu and Yin
this documentation could also be accomplished by comparing the two kV

X-ray images to the corresponding DRRs.
The intensity-modulated beams are delivered through the complicated
precalculated leaf sequences. Each intensity map needs to be checked by
delivering a given amount of radiation to a lm such as Kodak EDR2 (44),
an electronic imaging device such as Varian aSi500 portal imager (45),
and a gantry mounted device (46). The resulting intensity distribution
should be compared to the planned intensity map. To verify the absolute
point dose, the planned intensity maps are usually exported to a verication
phantom and the isocenter or any other point dose in the phantom is calculated. The phantom is then irradiated with all planned intensity-modulated
beams. The point dose in the phantom is measured using a micro ion-chamber such as Exradin T14 by Standard Imaging, Inc. (Middleton, Wisconsin, U.S.A.) and is compared to the planned point dose. Under normal
conditions, the deviation between two is less than 3%. Sometimes, the detector may not be able to accurately measure the dose when the eld is very
small due to the partial volume effect of the detector.
The accuracy of dose delivery could be examined by a phantom
study (21). An example of such procedure is described below. A hypothetical spinal tumor, a section of vertebral column, and the corresponding
critical structure, spinal cord, in the Rando phantom are identied as
the PTV and critical organ for treatment planning. To verify the dose distribution, a special lm dosimeter (for example, GAFCHROMIC2 Dosimetry Type MD 55) is placed at the isocenter slice and a 12 Gy dose is
delivered to the isocenter. For the purpose of calibration, a range of doses
from 2 to 14 Gy is delivered to a set of lms at the calibration distance. To
minimize the effect of non-uniform dispersal of the sensor medium of the
lms, a double-exposure measurement technique proposed by Zhu et al.
(47) can be used for this experiment. All lms are digitized using a microdensitometer so that the lm density could be converted to the dose. The
resulting dose image is compared to the planned isodose distributions. The
original CT image is shown in Figure 6A (the target and the spinal cord
are shown with solid contours) and the planned dose distributions (90%,
50%, and 30% normalized to the isocenter) are shown in Figure 6B. Figure
7A illustrates the planned isodose distributions in the region where the lm
is inserted. Solid curves represent planned isodose lines labeled 90%, 50%,
and 30% relative to the isocenter. Figure 7B shows the original lm dose
image with three corresponding isodose curves. The planned and the corresponding measured isodose distributions are then overlaid on the original CT image as shown in Figure 7C. Results indicate that isodoses
between the calculated and measured results match well up to the 50% isodose line. The discrepancies between two 90% isodose lines are negligible.
The majority of 50% isodose lines are matched within 1 mm discrepancy,
except some portions between the lung/soft tissue interfaces where the
Spinal Tumors
273
Figure 6 Phantom study images. (A) The original CT image with target and spinal
cord indicated. (B) The planned dose distributions for 90%, 50%, and 30% isodose
curves normalized to the isocenter. (See color insert.)
274
Ryu and Yin
Figure 7 Phantom study results. (A) The planned isodose distributions in the region
where the lm was inserted. Solid curves represent planned isodose lines labeled 90%,
50%, and 30% relative to the isocenter. (B) The original lm dose image with three
corresponding isodose curves. (C) Both planned and corresponding measured isodose distributions are overlaid on the original CT image. (See color insert.)
Spinal Tumors
275
discrepancy is up to 5%. This discrepancy is mainly related to the dose

calculation and display accuracy.
CLINICAL APPLICATIONS AND OUTCOMES
Clinical Feasibility Study
The rst clinical study was carried out to determine the precision and
accuracy of the spinal radiosurgery. This is particularly important because
the spinal cord is always the critical organ that is located in close contact with
the vertebral column. Ten patients were enrolled at Henry Ford Hospital
between April 2001 and December 2001. All patients had pathologically
proven primary malignant neoplasm and had single or two contiguous
vertebral metastasis with or without spinal cord compression that was seen
on CT or MRI. To test the accuracy of radiosurgery and to achieve the
desired palliation, standard external beam radiation therapy (25 Gy in 10
fractions) was given rst and then radiosurgery boost (68 Gy single dose)
to the site of the spine involvement or spinal cord compression. External
beam radiotherapy was given based on conventional method including
two vertebral bodies above and below the radiographic lesion. Following
the external beam radiotherapy, the patients received intensity-modulated
radiosurgery boost to the most involved site. During radiosurgery, orthogonal portal lms were obtained for the nal verication and to determine the
accuracy of the isocenter. The endpoint of the study was to measure the
isocenter variation in order to determine the accuracy and precision of the
spinal radiosurgery. It was measured by image fusion of simulation and
orthogonal portal lms taken during radiosurgery.
The accuracy of the spinal radiosurgery was dened as the degree of
variation (by distance) between the isocenters of the CT simulation and of
the portal lms at the time of radiation delivery. The deviation between
the simulation and actual treatment isocenter was 1.36 0.11 mm (48).
We also measured the radiation dose at the isocenter using the same positioning parameters of the individual patients in a phantom with a micro
ionization-chamber. The average deviation of the measured dose from the
estimated dose was 2%. This level of precision of the isocenter obtained in
this study was clinically acceptable for spinal radiosurgery.
Dosimetric Consideration of Spinal Cord Dose
The radiosurgery dose was invariably prescribed to the periphery of target
tumor volume encompassed by the 90% isodose line. The radiation dose
to the adjacent normal spinal cord was calculated using the computerized
isodose calculation program. Average distance between the 90% and 50%
isodose line at the spinal cord is 5.2 0.9 mm (as shown in Figs. 3A
and 4A). This represents the rapid dose fall-off of spinal radiosurgery. In a
276
Ryu and Yin
Figure 8 Dosevolume histograms (DVHs) for cord doses. (See color insert.)
transverse section at the isocenter level, 20% of the spinal cord volume
immediately adjacent to the diseased vertebra received higher than 50% of
the prescribed radiation dose (48). Figure 8 illustrates the average DVH
for spinal cord dose in 50 cases. In this gure, only the portion of the spinal
cord closer to the GTV is used for calculation. Also plotted in the gure is
the average maximum cord dose DVH in which only two slices with the
maximal cord dose were used for DVH calculation. The insert illustrated
the types of cord anatomy.
To determine the factors that may affect the dose to the spinal cord,
correlative analyses were performed from the dosimetry of 51 patients
who received 1016 Gy radiosurgery. The factors tested were target volume,
length and width of target, spinal cord volume, and the number of the intensity-modulated beams. Average tumor volume was 57.0 34.1 cc (range
3.4217.0 cc). Average tumor length was 49.1 15.3 mm (range 16.1
85.1 mm), and average tumor width was 45.5 10.7 mm (range 12.9
89.5 mm). Average spinal cord volume at the corresponding level of the
treatment was 5.9 2.2 cc (range 2.414.7 cc). In the lumbar and sacral
treatment, the vertebral canal was considered as spinal cord for volume analysis purpose. Dosevolume histograms were plotted and also the spinal
cord volumes that received 20%, 40%, 60%, 80%, and 100% of the prescribed
dose. Then, the best-t correlation curves were obtained. From this analysis,
total dose (above 14 Gy) and the number of intensity-modulated beams (less
than seven beams) appear to be the most important factors affecting the
spinal cord dose. The tumor volume and length did not affect the higher
(>60%) isodose regions of the spinal cord. However, the dose to peripheral
Spinal Tumors
277
regions (lower than 60% isodose lines) was inuenced by target volume
greater than 100 cc and tumor length greater than 6 cm (49). This is in contrast
to the experience of brain radiosurgery where the target volume is a major
determinant for radiosurgery dose selection. We believe that the independence
of the cord dose from the target volume or length is mainly due to the use of
intensity modulation that limited the dose to the critical organ at risk.
Clinical Outcome in Single Spinal Metastasis
Encouraged by the level of accuracy for spinal radiosurgery and the acceptable spinal cord dose, a subsequent study was carried out to determine
the dose and the clinical efcacy of radiosurgery for spine metastasis with
or without cord compression. A total of 49 patients (24 males and 25
females) with 61 lesions were treated with radiosurgery alone from May
2002 to May 2003. All patients had diagnosis of pathologically conrmed
malignant neoplasm and had either synchronous or metachronous metastasis to a single spine. Spinal metastases were diagnosed by radiologic studies with CT or MRI scans. Primary tumor sites were: breast 29.5%, lung
19.7%, prostate 9.8%, kidney 8.2%, and others 32.8%. The involved spines
were: cervical 13.1%, thoracic 54.1%, lumbar 29.5%, and sacral 3.3%
lesions. Patients had no previous radiotherapy to the involved spinal
lesion. Radiosurgery dose was a single dose in the range of 1016 Gy to
the involved spine. The dose was prescribed to the periphery of target
tumor volume encompassed by the 90% isodose line. The majority of
patients (70% of the lesions) experienced moderate to severe pain. Karnofskys status was above 70 in 75%. Since the goals of treatment for
spinal metastasis are pain control and preservation of neurologic function,
the endpoints of evaluation were the assessment of pain, neurologic status,
and radiologic studies for tumor control.
Pain Control
Pain was scored by using verbal/visual analogue scale; from 0 (no pain) to
10 (the worst imaginable pain). The scoring was performed before radiosurgery, four weeks, and eight weeks following radiosurgery. During the rst
four weeks the patients were evaluated by telephone for assessment of pain
status. Complete relief was dened as a complete absence of pain and no
need of analgesics. Partial relief was dened as a decrease of at least three
levels of the pain score or signicant reduction of analgesic medication such
as elimination of narcotics or breakthrough medication. The time to achieve
pain relief was recorded as the time span of pain relief/reduction from the
day of radiosurgery. The duration of pain relief was measured as lack of
progression of pain or without an increase in analgesic medication. Pain
progression was dened as an increase in the pain score by three levels
and/or increased pain medication.
278
Ryu and Yin
Rapid pain relief was achieved with the median time of 14 days (range
169 days). The earliest pain relief was seen within 24 hr. The change of
pain scores before and after radiosurgery showed a clear shift toward lower
pain scale following radiosurgery treatment. Four weeks after radiosurgery,
complete pain relief was achieved in 37.7% of cases involving pain, and partial pain relief in 47.6% of cases. At 8 weeks, complete relief was seen in 46%,
partial relief in 18.9%, and stable levels in 16.2% of cases. The estimated
median duration of pain relief was 16.6 months. Overall pain control rate
for one year was 84% (50).
Factors that may affect the pain relief were analyzed. Uni- and multivariate analyses did not reach statistical signicance with age, Karnofskys
performance status, primary tumor type, presence of neurologic symptoms
systemic metastases other than the spine, number of spinal lesions, dose of
radiosurgery, and chemotherapy. However, there was a strong trend of
increased pain control with higher radiation dose 14 Gy (50).
Neurological Improvement
There were 18 patients who presented either neurologic signs of motor
weakness and sensory changes of the extremities or radiologic epidural cord
compression. Twelve patients were treated with decompression surgery followed by postoperative radiosurgery. There were six patients with spinal
cord compression that were treated by radiosurgery alone. Follow-up
clinical and radiological examination was performed before radiosurgery,
one month, and then every 23 months following radiosurgery in patients
who were able to make the follow-up visits. Five of these patients continued
to be ambulatory or to have the full range of arm and nger motion (50).
Due to the small number of patients treated so far, dose response analysis
could not be established for neurologic and radiologic tumor control.
Treatment Failure and Complications
Four lesions (6.5%) had progressive pain at the treated site and required
stronger analgesic medication. All these patients had progressive systemic
metastases. Radiologic progression to the immediately adjacent vertebral
bodies was seen in three patients (4.9%) at six and nine months after radiosurgery. The status of the treated spine were stable (51). These patients had
progressive paraspinal soft tissue mass along the vertebral involvement.
Overall one-year survival rate was 74.3%. During this period, there
were no clinically detectable neurological signs that could be attributable
to the acute or subacute radiation-induced cord damage for a maximum follow-up of 24 months. The radiologic studies of six patients who have been
followed up for more than one year did not reveal any sign that was suggestive of spinal cord injury. No patient was admitted to the hospital as a result
of complications due to the radiosurgery treatment.
Spinal Tumors
279
Treatment of Recurrent Spinal Metastasis

The outcome of the radiosurgery for recurrent spinal metastasis that was
previously treated with external beam radiotherapy was similar to the new
spinal metastasis as described above. The radiosurgery dose was somewhat
limited due to the previous radiotherapy. Dose selection was individualized
based on the previous radiation dose, interval between the treatments, and
the patients general condition. In the interim analysis, complete pain relief
was also achieved in 40% of the patients and partial relief in 26%. The rapidity
of the pain relief was also rapid within two weeks from radiosurgery. Complete
and partial neurological recovery was also noted in six out of eight patients
Radiosurgery of Primary Spinal Tumors
Stereotactic radiosurgery is effective in the treatment of benign lesions of the
brain and the head and neck, as well as malignant tumors (52,53). The role
of radiosurgery for the treatment of benign and malignant spinal tumors has
been also dened (51,54). Spinal radiosurgery allows the use of a higher dose
needed to destroy the neoplastic tissue. Benign and malignant primary
spinal tumors were treated with a single dose or fractionated radiosurgery.
The primary spinal cord tumors include glioblastoma, anaplastic ependymoma, gliomas, chordoma, neurobroma, etc. The results show excellent
tumor response and these results are being analyzed. The primary osseous
tumors arising from the spine were also treated successfully (55).
Current Practice at Henry Ford Hospital
The spinal radiosurgery program at Henry Ford Hospital (HFH) is being carried out in a multi-disciplinary effort for spinal metastasis and primary spinal
tumors at the weekly spine tumor board. For metastatic disease, patients are
Table 2 HFH Spinal Radiosurgery for Spinal Metastasis
Group 1
Group 2
Group 3
Group 4
Focal disease with minimal neurologic decit

Radiosurgery alone to the involved spine
Focal disease with signicant neurologic decit
and spinal cord compression
Surgical decompression/stabilization followed by
radiosurgery
Diffuse metastatic disease with minimal neurologic decit
or symptoms
Radiosurgery alone to the most symptomatic site
Widespread metastatic disease
External beam radiation therapy boost radiosurgery
280
Ryu and Yin
placed into one of four cohorts as shown in Table 2. All patients are prescreened for evidence of spinal instability. Following treatment, the followup program includes questionnaires and clinical and radiologic tests. Follow-up clinical evaluation and neurologic examination are performed every
two months. A baseline MRI prior to radiosurgical treatment is followed
by repeat MRI examination at 2, 6, and 12 months. Patient questionnaires
consist of entry demographic information and monthly quality of life assessments (completed by the patient or a family member). The endpoints for clinical investigation are patient quality of life, neurologic and pain status, and
radiographic spinal cord abnormality and tumor control.
The use of spinal radiosurgery is not limited as a single modality to
new or recurrent lesions. It is also effective as an adjuvant to decompression
surgery, or in combination with vertebroplasty, or as the boost treatment to
external beam radiotherapy, or in combination with ongoing chemotherapy.
Multidisciplinary treatment efforts can improve the patients quality of life
and are, therefore, a signicant addition to the armamentarium for the management of spinal tumors. Our extracranial radiosurgery program was also
extended to other organ sites such as head and neck cancers with excellent
tumor control (56).
REFERENCES
1. Wong DA, Fornasier VL, MacNab I. Spinal metastasis: the obvious, the occult,
the imposters. Spine 1990; 15:14.
2. Siegal T, Siegal T. Current considerations in the management of neoplastic
spinal cord compression. Spine 1989; 14:223.
3. Young JM, Funk FJ Jr. Incidence of tumor metastases to the lumbar spine.
A comparative study of roentgenographic changes and gross lesions. J Bone
Joint Surg Am 1953; 35:5564.
4. Lada R, Kaminski HJ, Ruff RL. Metastatic spinal cord compression. In:
Vecht C, ed. Neuro-oncology, Part III. Neurological disorders in systemic cancer. Amsterdam: Elsevier Biomedical Publishers, 1997:167189.
5. Black P. Spinal metastases: current status and recommended guidelines for
management. Neurosurgery 1979; 5:726746.
6. Posner JB. Management of central nervous system metastasis. Semin Oncol
1997; 4:81.
7. Barron KD, Hirano A, Haraki S, Terry RD. Experiences with metastatic neoplasms involving the spinal cord. Neurology 1959; 9:91.
8. Helweg-Larsen S. Clinical outcome in metastatic spinal cord compression: a
prospective study of 153 patients. Acta Neurol Scand 1996; 94:269275.
9. Gilbert RW, Kim JH, Posner JB. Epidural spinal cord compression from metastatic tumor: diagnosis and treatment. Ann Neurol 1978; 3:4051.
10. Greenberg HS, Kim JH, Posner JB. Epidural spinal cord compression from
metastatic tumor: results with a new treatment protocol. Ann Neurol 1980;
8:361366.
Spinal Tumors
281
11. Loblaw DA, Laperriere NJ. Emergency treatment of malignant extradural

spinal cord compression: an evidence-based guideline. J Clin Oncol 1998;
16:16131624.
12. Siegal T, Siegal T. Surgical decompression of anterior and posterior malignant
epidural tumors compressing the spinal cord: a prospective study. Neurosurgery
1985; 17:424432.
13. Al-Rodham NR, Kelly PJ. Pioneers of stereotactic neurosurgery. Stereotact
Funct Neurosurg 1992; 58:6066.
14. Clarke RH. The Johns Hopkins Hopsital Reports. In: Investigation of Central
Nervous System, Part I. Methods and Instruments. Baltimore: The Johns Hopkins Press, 1920:1112111124.
15. Rand RW, Bauer RO, Smart CR, et al. Experiences with percutaneous steretoactic cryocordotomy. Bull LA Neurol Soc 1965; 30:142147.
16. Nadvomik P, Frolich J, Jezek V, et al. New apparatus for spinal cord stereotaxis and its use in the microsurgery of lumbar enlargement. Conn Neurol
1972; 34:311314.
17. Gabriel EM, Nashold BS. History of spinal cord stereotaxy. J Neurosurg 1996;
85:725731.
18. Takacs I, Hamilton AJ. Extracranial stereotactic radiosurgery. Neurosurg Clin
N Am 1999; 10:257269.
19. Hamilton AJ, Lulu BA, Fosmire H, Stea B, Cassady JR. Preliminary clinical
experience with linear accelerator-based spinal stereotactic radiosurgery. Neurosurgery 1995; 36:311319.
20. Murphy MJ, Cox RS. The accuracy of dose localization for an image-guided
frameless radiosurgery system. Med Phys 1996; 23:20432049.
21. Yin FF, Ryu S, Ajlouni M, Zhu J, Yan H, Guan H, Faber K, Rock J, Abdulhak M, Rogers L, Rosenblum M, Kim J. A technique of intensity-modulated
radiosurgery (IMRS) for spinal tumors. Med Phys 2002; 29:28152822.
22. Schell MC, Bova FJ, Larson DA, et al. Stereotactic Radiosurgery: Report of
Task Group 42 Radiation Therapy Committee. College Park, MD: American
Institute of Physics, 1995:68.
23. Shiu AS, Change EL, Ye JS, Lii M, Rhines LD, Mendel E, Weinberg J, Singh S,
Maor M, Mohan R, Cox JD. Near simultaneous computed tomography imageguided stereotactic spinal radiosurgery: an emerging paradigm for achieving
true stereotaxy. Int J Radiat Oncol Biol Phys 2003; 57:605613.
extracranial targets. Z Med Phys 1994; 4:112113.
25. Yan H, Yin FF, Kim JH. A phantom study on the positioning accuracy of the
Novalis Body system. Med Phys 2003; 30:30523060.
26. Yin FF, Rubin P. Imaging for radiation oncology treatment process. In: Bragg
DG, Rubin P, Hricak H, eds. Chapter 8 in Oncologic Imaging. 2nd edn. Philadelphia: WB Saunders Company, 2002.
27. Llacer J. Inverse radiation treatment planning using the dynamically penalized
likelihood method. Med Phys 1997; 24:17511764.
28. Llacer J, Solberg TD, Promberger C. Comparative behavior of the dynamically
penalized likelihood algorithm in inverse radiation therapy planning. Phys Med
Biol 2001; 46:26372663.
282
Ryu and Yin
29. Fiveash JB, Murshed H, Duan J, Hyatt M, Caranto J, Bonner JA, Popple RA.
Effect of multileaf collimator leaf width on physical dose distributions in the
treatment of CNS and head and neck neoplasms with intensity modulated
radiation therapy. Med Phys 2002; 29:11161119.
30. Timothy CR, Meeks SL, Traynelis V, Haller J, Bouchet LG, Bova F, Pennington
EC, Buatti JM. Ultrasonographic guidance for spinal extracranial radiosurgery;
technique and application for metastatic spinal lesions. Neuiosurg Focus 2001;
11(6).
31. Murphy MJ, Change S, Gibbs L, Le QT, Martin D, Kim D. Image-guided
radiosurgery in the treatment of spinal metastases. Neurosurg Focus 2003:11.
32. Kim JK, Yin FF, Kim JH. Characteristics of a CT/Dual X-ray image registration method using 2D texture map based DRR, gradient ascent, and mutual
information. ICCR 2004.
33. Jaffray DA, Siewerdsen JH, et al. Flat-panel cone-beam computed tomography
for image-guided radiation therapy. Int J Radiat Oncol Biol Phys 2002; 53:
13371349.
34. Mackie TR, Balog J, Ruchala K, Shepard D, Aldridge S, Fitchard E, Reckwerdt P, Olivera G, McNutt T, Mehta M. Tomotherapy. Semin Radiat Oncol
1999; 9:108117.
35. Kuo JV, Cabebe E, Al-Ghazi M, Yakoob I, Ramsinghani NS, Sanford R.
Intensity-modulated radiation therapy for the spine at the University of California, Irvine. Med Dosim 2002; 27:137145.
36. Yin FF, Zhu JH, Yan H, Guan H, Hammoud R, Ryu S, Kim JH. Dosimetric
characteristics of Novalis shaped beam surgery unit. Med Phys 2002; 29:
17291738.
37. Cosgrove VP, Jahn U, Pfaender M, Bauer S, Budach V, Wurm R. Commissioning of a micro multi-leaf collimator and planning system for stereotactic radiosurgery. Radiother Oncol 1999; 50:325336.
38. Ryu SI, Chang SD, Kim DH, Murphy MJ, Le QT, Martin DP, Adler JR.
Image-guided hypo-fractionated stereotactic radiosurgery to spinal lesions.
Neurosurgery 2001; 49:838846.
39. LoSasso T, Chui CS, Ling CC. Physical and dosimetric aspects of a multileaf
collimation system used in the dynamic mode for implementing intensity modulated radiotherapy. Med Phys 1998; 25:19191927.
40. LoSasso T, Chui CS, Ling CC. Comprehensive quality assurance for the delivery of intensity-modulated radiotherapy with a multileaf collimator used in the
dynamic mode. Med Phys 2001; 28:22092219.
41. Phillips MH, Singer K, Miller E, Stelzer K. Commissioning in image-guided
localization system for radiotherapy. Int J Radia Oncol Biol Phys 2000;
48:267276.
42. Zhu J, Yin FF, Kim JH. Point dose verication for intensity-modulated radiosurgery using Clarksons method. Med Phys 2003; 30:22182221.
43. Siebers VJ, Mohan R, Monte Carlo. Intensity-modulated radiation therapy: the
state of the art. In: Jatinder R Palta, Rockwell Mackie T, AAPM, eds. 2003.
44. Zhu XR, Jursinic PA, Grimm DF, Lopez F, Rownd JJ, Gillin MT. Evaluation of
Kodak EDR2 lm for dose verication of intensity modulated radiation therapy
delivered by a static multileaf collimator. Med Phys 2002; 29:16871692.
Spinal Tumors
283
45. Warkentin B, Steciw S, Rathee S, Fallone BG. Dosimetric IMRT verication

with a at-panel EPID. Med Phys 2003; 30:31433155.
46. Ma L, Phaisangittisakul N, Yu CX, Sarfaraz M. A quality assurance method
for analyzing and verifying intensity-modulated elds. Med Phys 2003;
30:20822088.
47. Zhu Y, Sirov AS, Mishra V, Meigoomi AS, Williamson IF. Quantitative evaluation of radiochromic lm response for two-dimensional dosimetry. Med
Phys 1997; 24:223229.
48. Ryu S, Yin FF, Rock J, Zhu J, Chu A, Kagan E, Rogers L, Ajlouni M, Rosemblum M, Kim J. Image-guided and intensity-modulated radiosurgery for
patients with spinal metastases. Cancer 2003; 97:20132018.
49. Ryu S, Sharif A, Yin FF, Ajlouni M, Kim JH. Tolerance of human spinal cord
to single dose radiosurgery. Proceedings of International Congress of Radiation
Research, 2003.
50. Ryu S, Rock J, Yin FF, Ajlouni M, Rosenblum M, Kim JH. Image-guided single dose radiosurgery for single spinal metastasis. Proceeding of American
Society of Clinical Oncology (ASCO) 2004.
51. Ryu S, Rock J, Rosenblum M, Kim JH. Pattern of failure after single dose
radiosurgery for single spinal metastasis. Journal of Neurosurgery 2004; 101:
402405.
52. Chin LS, Szerlip NJ, Regine WF. Stereotactic radiosurgery for meningiomas.
Neurosurg Focus 2003; 14:1.
53. Foote RL, Pollock BE, Gorman DA, et al. Glomus jugulare tumor: tumor control and complications after stereotactic radiosurgery. Head Neck 2002; 24:332
338.
54. Gertzen PC, Ozhasoglu C, Burton SA, Vogel WJ, Atkins BA, Kalnicki S,
Welch WC. Cyberknife frameless single-fraction stereotactic radiosurgery for
benignm tumors of the spine. Neurosurg Focus 2003; 14:15.
55. Rock J, Kole M, Yin FF, Ryu S, Guittierez J, Rosenblum M. Radiosurgical
treatment for Ewings sarcoma of the lumbar spine. Spine 2002; 27:471475.
56. Ryu S, Khan M, Yin FF, Ajlouni M, Concus A, Benninger MS, Kim JH.
Image-guided Radiosurgery of Head and Neck cancers. Otolaryngology Head
and Neck Surg 2004; 130:690697.
12
Stereotactic Radiotherapy of Head and
Neck Tumors
Robert Smee
Department of Radiation Oncology, Prince of Wales Hospital, Randwick,
New South Wales, Australia
Reinhard Wurm
Abteilurg Strahlentherapie, Universitat Klinikum, Charite, Berlin, Germany
INTRODUCTION
Highly conformal radiotherapy is now becoming the norm for treatment
delivery. This principle has added signicance for head and neck (HN) sites
because more typically we are dealing with malignancies (typically carcinomas)
that require higher doses for control in the context of nearby dose-sensitive
normal structures. The latter ranges from critical structures, such as the ocular
apparatus and the spinal cord, to dose important organs such as the major salivary glands. The latter structures impact very much upon the quality of life of
those treated, importantly the long-term survivors. The three-dimensional (3D)
conformal treatment approach including intensity modulated radiotherapy
(IMRT) is able to set particular structures as a dose-dening structure, and
using either forward or inverse planning to limit the dose to these structures.
It is important obviously that the adequate dose be given to the tumor being
treated. This can create a circumstance of competing priorities, with concern
that the large areas of intervening normal tissue may receive a higher dose.
In the above example of sparing normal tissues, the major salivary glands
may receive an acceptable dose for maintenance of salivary function, but
the minor salivary glands more typically excluded in the parallel-opposed
285
286
Smee and Wurm
technique may receive a high dose that could signicantly impair their function. The end result may thus be that the patient experiences the same side
effect but caused by a different end organ.
Stereotactic irradiation offers many possibilities for HN tumors. This
may be used as denitive treatment for the primary lesion, as a boost after
prior wide eld treatment, or for recurrence. These listed situations apply
equally to stereotactic radiosurgery (SRS), as well as to fractionated stereotactic treatment. The impetus for this consideration relates to the necessity
to give higher doses for local control in a circumstance of adjacent dose
critical structures. The concept of making this dose delivery stereotactic
relates to an extra dimension of treatment delivery accuracy over and above
that achievable by some form of conventional setup.
Broadly, stereotactic irradiation is divided into two: single-dose SRS
and stereotactic fractionated treatment. Machine limitations and tumor
location will inuence the treatment method chosen for those with available
technology. The literature, however, supports both methods with acceptable
outcomes in disease-specic circumstances. The aim of this chapter is thus to
bring to readers attention the techniques available, how they have been
used, their value, and thus whether the results justify continuing with this
approach. In addition, the treatment approach utilized at Prince of Wales
Hospital will be demonstrated; a number of case histories are used to
demonstrate the capabilities of these treatment methods.
BACKGROUND
Stereotactic Radiosurgery
This approach was the genesis of all stereotactic irradiation procedures. It
involves the xed attachment of a head ring device for coordinate localization purposes and delivery of the irradiation on the same day. Given that a
high single dose is delivered, there is a limit on the size of the lesion that can
be treated. As the head ring is typically attached to the frontal bone (some
publications have raised the concept of the needle insertion into the maxilla
for lower ring placement) the conditions treated are limited to those at the
base of skull or nasopharynx.
Three methods available are:
GammaKnife
Linear accelerator (linac)
Charged particles
The radiobiological effect in tissue of each of these methods is very
similar. Typically, a high dose is given to a small target using the conformality of the treatment approach to limit the dose to adjacent structures. It
was rst used in recurrent malignancies and as its value was demonstrated,
Head and Neck Tumors
287
an expanded role saw SRS being used as boost treatment in primary

management of malignancies, and for more localized relatively radioresistant tumors. Table 1A (17) demonstrates the role that this approach has
had for nasopharyngeal carcinomas in the context of recurrent disease.
Although the patient numbers are small, a reasonably consistent local control
gure of 5070% is apparent, where radiosurgery is given either alone or combined with fractionated radiotherapy, for recurrent malignancies. This is for a
low overall complication rate. It is recognized that this is a highly select group
with relatively short follow-up. However, for patients who do receive this
approach, it is likely that an improved and meaningful survival is obtained.
Chordoma is another skull base tumor treated by SRS. While
histologically benign, this tumor assumes a locally progressive nature, which
would limit responsiveness to conventional radiotherapy. Its typical site
involves the clivus, which makes it surgically irresectable. Conventional dose
radiotherapy with up to 60 Gy being delivered reports low local control
rates of 2030% (8). Hence, a reasonable approach was to consider giving
a high single dose. Muthukumar published the University of Pittsburgh (9)
experience of 15 patients (13 of whom had SRS plus surgery, two SRS alone).
A mean marginal dose of 18 Gy was given, with only one patient having an
in-eld failure. Two patients progressed outside the treatment volume with a
mean follow-up of four-year post-SRS. Whilst this is relatively short followup for this tumor, high local control rates are apparent for a single-dose
procedure. It is sufcient that such treatment should be considered for localized tumors postsurgery, or indeed even de novo for small tumors.
Chemodectomas, principally, are a benign tumor arising from neuroendocrine tissue in the HN area, in the temporal bone, and in the carotid
artery wall. This rare tumor is a slowly growing lesion with its presentation
relating to its proximity to the neurovascular bundle. Pulsatile tinnitus is a
common presenting event for temporal bone lesions, and a palpable mass
being apparent for carotid lesions. A number of the lower cranial nerves
(including the vagus) are adjacent to these tumors, hence surgical resection
of these tumors places these nerves at high risk of being resected with
subsequent decit. As can be seen in Table 2 (1016), with moderate
single-dose radiosurgery very high local control rates can be achieved,
within six of seven series there were no long-term complications. Patients
rarely die from these tumors although their site can cause signicant morbidity. Thus, the treatment that has high local control, low morbidity,
and can be done on an outpatient basis should be regarded as a treatment
of choice. It should be noted that these lesions need to be of a size suitable
for SRS and it may thus be a selection criterion for these lesions. A fractionated radiotherapy approach needs to be considered for the larger lesions.
Only one other HN site has been reported for SRS with Habermann
noting its use as a boost after surgical resection of nasal cavity and/or
paranasal sinus carcinomas (17). Eight patients were so treated with the
288
Smee and Wurm
GammaKnife, with seven patients being locally controlled with no adverse

effects attributable to the SRS. Given that, this was combination treatment,
it is difcult to dene the true impact of the SRS in achieving local control.
Its use, however, indicates that it can be considered for small volume
residual macroscopic or microscopic disease.
Stereotactic RadiotherapyFractionated
While smaller tumors may be suitable for SRS, within the HN area there are
a number of tumors, which either on the basis of the size of the macroscopic
component or the necessity to include a signicant volume of normal tissue
for microscopic disease, dictate that if the treatment is to be delivered stereotactically, it has to be fractionated. This behooves the use of some form of
relocatable xation device. This device should allow extension of the treatment volume down into the neck. The GammaKnife would be excluded in
this treatment device (on the basis of being a single-dose procedure, plus
the inability to extend the treated area well down into the neck), and the common approach would be with a linear accelerator (Linac), although some
centers have used charged particles particularly for skull base chordomas.
Whilst re-irradiation is an accepted method of managing locally recurrent nasopharyngeal carcinoma (1821), only two centers have reported any
results for fractionated stereotactic radiotherapy in this setting. Ahn
reported 19 patients (4) utilizing the Gill Thomas Cosman (GTC) relocatable head ring, delivering 4550 Gy in 1820 fractions, to the radiologically
abnormal areaTable 1B (4,22,23). Three patients died soon after, of
distant disease with no local recurrence. Follow-up time was too short following this treatment to know whether long-term benet would have been
achieved, or indeed to report any complication. Dhanachai reported on 19
patients (22), using a similar approach, treated over a 30-month time frame.
For 11 patients, this was for persistent/recurrent disease, for eight it was a
boost. Doses delivered varied between the two situations, with the boost
patients typically receiving only 46 fractions of 47 Gy each, whereas the
retreated patients received up to 55 Gy in up to 28 fractions. In terms of
results, only 5 of 11 retreated patients are alive with local control at the time
of reporting, while all eight boost patients are alive, median follow-up of
eight months, with local control. No complications were reported.
For both series, follow-up is short; however, it would be reasonable to
consider SRT as a means of safe dose escalation in denitive management.
Recurrent malignancy is still a major challenge for any treatment approach.
Irresectable chordomas have long been a radiotherapy challenge,
predominantly because of the necessity to give higher doses with radiotherapy
sensitive normal structures adjacent. Hence, charged particles have been used
with the Bragg peak effect giving a dose distribution advantage. Utilizing
either carbon ions or protons, local control rates at ve years are a very
289
respectable 7090% (24,25). However, from the Harvard experience (26), it

would seem that these results drop off to provide only 44% non-progression
at 10 years. Against this background, Debus (27) reported on 45 patients
treated postoperatively with median doses at the isocenter of 66 Gy for both
chordomas and chondrosarcomas, with stereotactically delivered photons,
and a local control of 82% at two years. However, similar to the Harvard
experience this dropped to 50% at ve years. The greatest success rate with
SRS was almost certainly a reection of the size factor, with much better local
control (90%) in the proton series for smaller tumors.
Conventional radiotherapy has been used as part of the treatment
approach for paranasal sinus carcinomas for many years. Treatment results
with radiotherapy alone have been disappointing, but given postsurgery,
high local control rates are evident (28). However, there has been no series
reporting this approach with a stereotactic methodology.
Similarly with chemodectomas, despite the very good local control
gures reported for the use of conventional radiotherapy alone with Hinerman
(29) reporting 94% local control rate with many years of follow-up, there are no
series reporting the use of fractionated stereotactic radiotherapy. It should be
noted that the dose initially used (4550 Gy) is the one tolerated by surrounding
structures with low likelihood of morbidity (30), thus there may be less need for
a highly conformal approach. It is the tumor, however, that does lend itself to
this method, being usually well dened on imaging, without a need to consider
covering microscopic disease.
The Prince of Wales Hospital approach has been to consider a stereotactic technique for appropriate situations. This has been to include both
benign and malignant lesions, with the necessary volume expansion to
encompass both macroscopic and microscopic components for the malignant
lesions. Conditions treated are listed in Table 3. Fractionated stereotactic
treatment became available in 1995, with, up to mid-2003, 380 patients
treated, 51 being for extracranial or extra-intracranial location.
TECHNIQUE AND INDICATIONS
Fractionated Stereotactic Radiotherapy
All treatments have been with the LINC (Siemens MD2, and Primus; Seemens AE, Berlen, Germany) with a couch-mounted attachment. From
1995 to mid-2000, all treatments were delivered with cone-type collimation.
The cone size and number of isocenters used were heavily dependent upon
the lesions size. The largest cone has a 4.5 cm diameter with gradations of
cone size of 0.25 cm down to 0.5 cm in diameter. Thus, the shape of the
lesion to be treated was covered by:

varying the size of the cone,

using multiple isocenters,
290
Smee and Wurm
preferential weighting of any of the arcs to give a more elliptical

shape,
using the primary jaws of the linac collimator system from either
(or both) sides for a large lesion to atten the spherical dose
shape of the cone.
Once the 4-mm leaf width Radionics (Burlington, Massachetts, U.S.A.)
MMLC became available, this became the main means of beam shaping. This
is an add-on device to our linac tting into the shadow-tray position. Attachment takes 1520 min, which needs to be programmed into the machine timetable each day, with removal taking 10 min. Consequently, all patients are
treated in one time block, usually the middle of the day, the start time being
dependent upon whether there are any children having general anesthetics
for this treatment. At any one time, there are 810 patients having this therapy. Thus, the dominant lesion for this technique remains intracranial tumors.
The MMLC weighs 20 kg and the attachment to the head of the machine
needs to be considered in terms of gantry movement. Counterbalancing this
weight was considered but after considerable testing, since the machine is
not used in arc mode, this was not felt necessary. After three years of use,
the linac gantry remains quite stable for stereotactic and non-stereotactic
use. The eld size of the MMLC is 10 12 cm; thus no stereotactic treatment
is possible to lesions larger than this. Clearance of the MMLC from the patient
is 45 cm, with the collimator set at 90 to avoid collisions. Thus, gantry angles
of 0360% are possible for beam entry.
Planning for all fractionated stereotactic treatments is done on the
Radionics X-Plan Planning System (Version 4). The process involves, as
with any 3D conformal technique critical structure outlining (both volumes
to be treated, and avoided or minimized), evaluation of which beam entry
portals are to be used and beams eye view (BEV) used to select the beam
portals with consideration given to target coverage for the coverage of
critical structures. For xed eld treatment, evaluation of dosevolume
histogram (DVH) is used to make the nal assessment of the volume to
be covered and structures to be avoided. A plan is then generated, printed,
and evaluated. If accepted, as part of the QA process, the physics staff
veries all beam and dose parameters, and treatment characteristics are
incorporated into the linac Record and Verify System (RVS).
For the IMRT patients, the inverse planning system (Konrad) requires
particular dose characteristics to be set including maximum and minimum
dose for the tumor to be treated as well as for any nominated critical structure. Priorities and penalties are set for exceeding the specied dose for any
of these structures. Thus, after the outlining component and nomination of
beam numbers and entry direction, as for the xed eld treatment, all the
material is imported into Konrad and the IMRT planning begins. Generation of DVHs require export back into XPlan where the uence maps are
291
generated. This process takes only a few minutes and thus any change can be
incorporated and evaluated quickly. Prior to the actual treatment, the whole
treatment process is delivered to X-ray lm, and the developed uence maps
are compared qualitatively and quantitatively with the planned uence
maps.
Head and neck xation was done initially by the Radionics GTC
xation device for awake patients and the Radionics TLC device for
anesthetized children. The GTC xation is a bite block-based device using
maxillary dental attachment, to which is then attached the stereotactic ring,
which also serves for attachment and xation, to the couch mount system.
Edentulous patients can be accommodated by adding more bulk to the dental plate. Velcro-backed strips are then pulled upwards across the side of the
head applying upward pressure against the gum margins. A depth helmet is
then attached to the ring, and initially 20 depth measurements were taken
for each treatment with a variation of no greater than 2 mm allowed before
continuing with treatment. After thousands of measurements were taken
it was apparent that the only variation was in six of the sites, and thus
measuring all sites was unnecessary. The QA process could thus be shortened without reducing its effect, by dispensing with redundant parts of
the process. A similar concept is used for the HN localizer (HNL). The base
board is attached at xed points to the treatment couch, and to this is then
attached the stereotactic localizer, and depth readings are done to make sure
the patient is being setup in the same position. Skin marks are used as a nal
verication process for patient setup.
Prior to the rst treatment, the actual treatment process is simulated
with the gantry and couch position for each treatment position veried to
ensure no potential collision. An isocenter phantom is used for GTC setup.
The treatment sequence is then delivered, with the head ring or HNL xed in
position on to the couch, to ensure that all leaf sequences proceed as determined by the planning computer and loaded into the R and V system. For
the actual treatment, the patient is lying on the couch with the head ring
attached, which is then docked to the xation device on the couch. The
patient is made comfortable, and a nal QA process begins. With a clear
Perspex box attached to the head ring, and clear paper sheets placed on each
of three surfaces (anterior and two laterals) the machine laser lights are used
to verify centering of the head ring or HNL, and the entry portal for each
beam direction is set accurately. Once all these steps are complete, treatment
can take place. All gantry positions can be set at the linac console, and
movement of the gantry takes place from there. Couch movements are done
inside the room. The total treatment time each day, including the QA steps
for the rst session is less than the time it takes to treat a 4-eld breast
patient. For the IMRT procedures, treatment time is about 30 min, this
however being dependent upon the number of segments treated. This
includes head ring application, with QA about 5 min longer than for
292
Smee and Wurm
multiple xed elds with the MMLC. No treatments are done with the
native 1-cm leaf width MLC.
The advantage of this is that a better-collimated treatment is delivered
but the disadvantage is the limited eld size (10 12 cm).
The practice for HN cancers requiring large eld irradiation has been
to use conventional wide eld coverage, to say 5056 Gy. The stereotactic
approach can then be used to take the primary up to the specied dose
(7074 Gy) with any neck nodes boosted to the required doses. Those malignancies in which the pattern of failure is more predominantly local have all
their treatment delivered stereotactically.
For the benign tumors (e.g., chemodectoma), all treatment is delivered
stereotactically.
Nasopharyngeal Carcinoma
Ample evidence now exists to indicate that to achieve optimum local control
for T2-4 carcinomas, doses of 7080 Gy are required, even with concurrent
chemotherapy. These doses are beyond the tolerance of adjacent mandible,
temporal lobes, and brain stem, if delivered with conventional radiotherapy.
For this treatment, a typical 3-eld non-salivary gland sparing approach was
given to 5056 Gy, with electrons to the posterior neck after the spinal cord
received 40 Gy. The neck was boosted with electrons as required, dependent
upon the bulk of the nodal disease. The primary site, typically all of the nasopharynx and adjacent paranasopharyngeal area, was then boosted stereotactically with a minimum of 20 Gy. The contralateral paranasopharyngeal was
excluded if the carcinoma was lateralized. Throughout the treatment 2 Gy
fractions were used. Limits on the dose to the temporal lobes and mandible
were set at a maximum of 10 Gy over that given by phase 1 treatment.
Three case histories represent the capabilities of this approach.
Patient 1A 14-year-old teenager, who presented with left neck disease with presumed nasopharyngeal primary, was given conventional eld
treatment to 26 Gy, and then stereotactic IMRT with the 1-cm leaf width
MLC to 66 Gy with major salivary gland sparing (Fig. 1). This particular
patient, whilst satisfactorily treated, raised concern regarding our QA process for the larger width MLC to such an extent that we did not continue
with that approach. All this treatment was thus done as one eld with the
primary site receiving a dose of 1.8 Gy per fraction, while the neck was
treated bilaterally at 1.4 Gy per fraction over the same number of fractions
using the intensity-modulated component to decrease the dose per fraction
to areas of lower risk disease.
Patient 2An Asian male aged 45 years presented a 12-month history
of intermittent epistaxis and neck pain. As Figure 2A indicates, there was
extensive spread of his carcinoma through the base of skull into sphenoid
sinus and clivus, resulting in a signicant parasellar and prepontine mass,
293
Figure 1 Fourteen-year-old teenager with nasopharyngeal carcinoma and involved

left neck nodes. Stereotactic MLC treatment. DVHs demonstrate dose separation
between malignancy and adjacent critical structures.
the latter encircling the basal artery and displacing the pons posteriorly. In
addition, there was a dural plaque of carcinoma extending along the clival
surface to the foramen magnum. Superiorly, its extension was up toward
the optic chiasm although there was 3 mm separation between these two
structures. Thus, in this situation the dose-limiting structures were: temporal
lobes, optic chiasm, and brain stem, with the maximum of 50 Gy set for both
phases 1 and 2. Despite the bulk of the primary there was no nodal disease,
and thus phase 1 with a conventional 3-eld technique went to 46 Gy, the
remainder was taken to 70 Gy, this being the extensive primary site and bilateral nasopharyngeal areas, delivered with concurrent cis-platinum and 5FU.
The stereotactic component of his treatment was given with multiple xed
elds using the MLC for beam shaping. The pituitary was enveloped by
the primary mass and could not be spared receiving full dose.
In terms of outcome, two-year post-treatment, this patient remains
disease free with a normal functioning pituitary, mild xerostomia, and
294
Smee and Wurm
Figure 2 (A) MRI: Extensive nasopharyngeal carcinoma extending through the

sphenoid sinus and clivus to pre-pontine location with dural plaque of malignancy
extending to foramen magnum. (B) MRI 18-months post-treatment: No evident
malignancy.
normal taste (Fig. 2C). The only area of brosis is in the left neck, where he
received 46 Gy the right neck, which received the same dose, has minimal
long-term effect.
Patient 3A 65-year-old male presented a right nasopharyngeal primary on a background of 10 years previously having had a laryngectomy
and postoperative radiotherapy to bilateral neck and pharyngeal areas to
50 Gy for an advanced larynx cancer. This new primary was thus his second
signicant cancer. His neck had considerable brosis and thus would not
tolerate any signicant extra dose. The primary site (whole nasopharynx
295
and right paranasopharyngeal area up to and including the base of skull)

was thus treated with stereotactic IMRT abutting up to the top edge of
the previous treatment eld. The typical dose-limiting structures within
the eld such as brain stem and chiasm could be regarded as untreated
structures. Total dose was 66 Gy in 33 fractions at 2 Gy per fraction. In this
circumstance, however, the intensity modulation was used to increase the
dose per fraction to the radiologically evident abnormality (GTV) such that
this area was receiving 2.4 Gy per fraction as a concurrent relative hypofractionated boost. There was signicant acute mucositis associated with this.
Unfortunately, this patient had two adverse outcomes. First, he failed in
the contralateral neck with nodal disease, and second he developed a radionecrotic ulcer at the site of the boost area. Subsequent investigations
indicated markedly abnormal bilateral carotid vasculature, probably subsequent to his initial radiotherapy plus the predisposing history of cigarette
smoking prior to his laryngectomy. In this circumstance presumably
the mucosa could not cope with the further radiotherapy from a vascular
healing effect. The increased dose per fraction, be it to malignancy, resulted
in a non-healing situation. Thus, this patients subsequent quality and quantity of life was mainly inuenced by the constraints of his previous disease
and treatment. Hyperbaric oxygen, the appropriate treatment for his radionecrotic ulcer, would have been inappropriate in the circumstance of
progressive malignancy.
Chemodectoma
This is a benign lesion readily identiable on imaging (best with MRInot all
patients today would have angiography) either at the skull base or projecting
into the neck. The signicance of it being benign is that there is no necessity
to add a margin for microscopic disease. Thus what is dened on scans is all that
needs to be treated. Given that there can be signicant cranial/caudal growth, it
is unlikely that a head base xation and the stereotactic coordinate system
would sufce for the majority of lesions. The smaller glomus tympanicum
lesions can be adequately treated by SRS by ensuring that the head ring is placed
as low as possible with marginal doses of 1416 Gy sufcing.
With the larger glomus tympanicum, and certainly all of the glomus
jugulare lesions a fractionated approach is more reasonable, the standard
being 1.82.0 Gy fractions to 4550 Gy. This follows the outstanding results
so far demonstrated for conventional radiotherapy. The purpose in delivering
this treatment stereotactically is the ability to mainly lower the dose to
surrounding normal structures.
The neck xation system is either by bite block or customized mask
attachment to a carbon ber backboard around which ts the stereotactic
coordinate system. Both components t on to a diagnostic CT couch top
and the treatment couch. Accurate positioning of this board in relationship
296
Smee and Wurm
Figure 3 Chemodectomaglomus tympanicum. (A) Volume rendered structures

treatment options. (B) Field arrangement14 xed beams simulating three conformal arcs. (C) Isodose display demonstrating conformality of various options, with
IMRT as the best option.
297
to the isocenter is important to ensure the day-to-day reproducibility of the

stereotactic reference system. This component of each patients treatment is
longer, although the treatment times with multiple xed elds may not vary,
compared to the GTC head xation system. In addition adjacent normal
structures, although important, are less critical, particularly given the doses
being used so there is little advantage to using IMRT. Multiple xed elds
usually give adequate coverage. Entry and exit directions can be dened
using BEV, with MMLC leaf placement used to shape the eld to the lesion
margins allowing a 3-mm expansion around the tumor as a PTV (Fig. 3A).
Although there can be some organ movement within the pharynx and larynx
(with swallowing and breathing, respectively), given the lateral location for
these lesions this is unlikely to have much effect. Hypofractionating the
dose has the advantage of convenience, but with an uncertain track record.
Conventional fractionation for the larger lesions would seem to be the
current standard of care.
Figure 3B demonstrates the dose shell display, eld arrangement, and
DVH for a large glomus jugulare tumor with multiple static elds using the
MMLC. The criteria of benet, as with most benign tumors, are lack of
progression (achieved in about 90% of cases) with most patients enjoying
considerable improvement in symptoms.
Paranasal Sinus Carcinomas
Although a number of centers have reported the use of radiotherapy alone
for these malignancies, this is also usually inferior to the gures generated
where major surgical resection and postoperative radiotherapy are delivered. Nodal spread is uncommon unless there is local extension through
the anterior wall of the maxilla into the skin of the cheek. Thus, for the
majority of patients, local control relates very much to the cancer-specic
survival. In the postoperative setting, it is typically microscopic disease that
is being addressed. Adjacent critical structures include bilateral globes (unilateral if an orbital exenteration has been performed), optic nerves, optic
chiasm, frontal lobes, and brain stem. Typical postoperative doses of
56 Gy plus are required, this being above the tolerance of these structures
even with the conventional fractionation. Published series report signicant
ocular and neurological morbidity rates using these types of doses. Two of
the cases treated with chondrosarcomas required higher doses (70 Gy).
There are a number of unique circumstances with these types of cancers:
Use of the GTC head ring needs modication to allow most if

not all of the maxilla to be included in the CTV. A spacer can
be added to the dental plate to allow for an extra margin of inferior
extension.
In the postoperative setting (except where orbital exenteration is
done in combination with a craniofacial resection and a large
298
Smee and Wurm
microvascular free ap is used to ll in the defect), a large air cavity is

created leading to some dose uncertainty due to the build up and
build down effect on the residual sinus walls. It is important that
the dose algorithm in the planning system be able to cope with this.
Given the above critical structures, malignancies in this site lend themselves appropriately to the use of IMRT. Figure 4 depicts the dose display,
Figure 4 (A) Paranasal chondrosarcoma demonstrating postsurgical air cavity.

Field arrangement and dose display on axial MRI. (B) DVH display of comparative
dose to tumor and adjacent structures.
299
and DVH for a patient treated postoperatively for a chondrosarcoma where

the mean dose delivered was 70 Gy at 2 Gy per fraction. Head xation was
via the GTC head ring.
Skin CarcinomaPerineural Spread
Skin carcinomas are the commonest malignancy affecting the Australian
population. A small proportion has perineural spread (this can occur with
both basal and squamous cell carcinomas), more typically this is evident histologically and is usually an indication for postoperative radiotherapy.
Some patients present overt features of perineural spread, typically as a consequence of involvement of nerves within the cavernous sinus. Thus, diplopia or altered facial sensation is a typical presenting feature. Figure 5A
demonstrates the extent of spread for one patient there was extension back
along the maxillary branch of the trigeminal nerve centripetally to involve
the nerve within the cavernous sinus, trigeminal ganglion, and then along
the main trunk toward the brain stem (MRI was not possible as this patient
had a cardiac pacemaker in situ).
In this situation, a high dose of 6066 Gy at 2 Gy fractions is necessary
to provide reasonable chance of control of the macroscopic component (surgery and/or chemotherapy are not viable options) with brain stem being the
dose critical structure. This dose has signicant risk of damage to those
areas within the cavernous sinus, as well as to the carotid artery. However,
with informed consent this risk is acceptable as the alternative of giving a
lower (and safer) dose would be a far higher risk of failure. Microscopic
Figure 5 (A) Axial CT indicates enlarged right cavernous sinus and enhancing mass
with central necrosis adjacent to brainstem. (B) Axial CT two-years post-treatment
indicating complete resolution of disease.
300
Smee and Wurm
extension into the brain stem toward the trigeminal nerve nucleus cannot be
excluded.
Thus, a dose limit of 50 Gy was set on the brain stem for the treatment
given via multiple xed elds. Although IMRT was considered, the alternative
plan is quite acceptable. Eighteen months after treatment, while the patient
still had altered facial sensation a subsequent CT scan (Fig. 5B) indicates complete clearance of macroscopic tumor, without any additional neurological
decit.
For all the situations an IMRT plan has to be clearly superior to a xed
eld plan to be preferred; for this patient there was no added advantage.
Chordomas
These are regarded as being radioresistant, although some conventionally
fractionated series do report reasonable results, be it with short-term follow-up. The charged particle series, however, demonstrate that at least
510-year post-treatment control rates of 50% can be achieved. The advantage
of charged particles is the BraggPeek effect whereby the dose to adjacent
structures can be signicantly reduced allowing dose escalation to the
tumor, in this situation delivering 70 Gy. There is no major radiobiological
advantage to protons versus photons.
A 45-year-old male, engaged in active sport, had presented three years
prior to his stereotactic treatment a 3 to 4-month history of altered sensation
down both arms and bilateral shoulder weakness. Investigations indicated
the presence of a C3 chordoma. Debulking surgery was done and he was
given postoperative radiotherapy via a parallel-opposed technique receiving
45 Gy in 25 fractions. He progressed 12 months later, and thus had three
further surgical procedures and was subsequently referred (Fig. 6A). Stereotactic IMRT using the MMLC was delivered; the dose given was a mean
dose of 69 Gy in 35 fractions. His spinal cord was set as a dose-limiting
structure as it had already received 45 Gy. Figure 6B demonstrates the
DVH for the dose-dening structures noting that the curve for the spinal
cord is shifted well to the left, with only a small volume of cord receiving
a higher dose. The uence map demonstrates that all eld angles were chosen to avoid chord. In follow-up, he remained well for 21 months with no
treatment-related abnormality, but unfortunately his MRI at that time indicated further disease progression. This outcome demonstrates that highdose treatment can be given, even as a retreatment, although it does not
ensure success.
Retinoblastoma
Retinoblastoma is an uncommon pediatric tumor, which in a majority of
circumstances is a local disease. It, typically manifests in infants, is usually
sporadic and it unilateral, but in at least 30% of cases there is a family
301
Figure 6 (A) Volume display of target volume wrapping around dose critical
structure, the spinal cord. (B) Dose display, DVH, and uence maps indicating the
sparing of the spinal cord and high dose to chordoma.
history and it is bilateral. Enucleation is the preferred method of addressing

the unilateral case, and it is appropriate for the bilateral cases to the worse
affected eye. For many years, radiotherapy to the affected eye was a frequent event with high local control rates; however, its effect upon normal
302
Smee and Wurm
tissue (including the globe and surrounding orbital bone) and the risk of signicant induction of malignancy prompted consideration of other options
to control the local disease. This is a chemotherapy-sensitive malignancy
demonstrating benet in the advanced and metastatic situation. Thus, it
has been utilized in infants for local disease along with laser ablation of
more discrete foci of malignancy. Radioactive plaques, as for melanoma,
can also be used for localized disease.
When, however, the malignancy becomes more extensive within the
globe, such as the presence of vitreous seeding, whole globe treatment
becomes a necessity. Published treatment methods include a single direct
lateral eld for unilateral disease (directed away from the contralateral globe
if this is unaffected) or parallel opposed for bilateral affected eyes. Some
published series have also looked at specialized electron treatments. However, in all these situations a major part of the orbit is still treated. Typical
doses used are 4045 Gy at 1.51.8 Gy fractionsa dose that would mainly
retard bone development in young children.
Since the globe is spherical and stationary (all affected children are
treated under general anesthetic), a stereotactic approach is appropriate.
Head xation and stereotactic localization is via the TLC head ring, allowing free access to the anesthetist for all anesthetic procedures. All treatment
methods can be considered (arcing with cones, xed MMLC shape eld, and
IMRT with an MMLC), depending upon dose coverage of all relevant structures. In this circumstance, adjacent bone becomes a dose-dening structure.
These children are at increased genetic risk of developing a second malignancy, with sarcomas in adjacent bone being a reported event occurring
because of the susceptibility plus the known long-term effects of radiotherapy. Sparing of the lens would be a desired goal; however, since the cases
now treated have extensive local disease this is usually not possible.
Our treatment approach has evolved with initially only cones available.
Figure 7A demonstrates a 15-month-old female with bilateral disease in
which the malignancy was progressing in one eye only, thus unilateral treatment was given, the contralateral orbit and globe dose being kept very low (as
a principle in a young child, plus also in case contralateral treatment was
required). This treatment cleared the vitreous component and most of the
retinal disease, although laser oblation was required to a small residual focus;
functioning vision remained. Eighteen months later, malignancy progressed
in the opposite globe by which stage the MMLC was being used for routine
treatment. While a comparable dose was given to the globe, a lower dose was
possible on adjacent bone. Similarly, good tumor control was achieved with
seemingly good vision initially. In the second eye, there was more extensive
anterior globe disease and thus the lens dose was higher resulting in cataract
formation at two-year post-treatment. Intra-ocular lens replacement signicantly improved vision. Both eyes and the child remained well controlled of
malignancy with adequate vision preservation.
303
Figure 7 (A) MRI: Axial, coronal, and sagittal display of dose distribution using
cones for conformality. (B) MRI: Axial display of three different treatment methods:
(i) arcs with cones, (ii) multiple xed with MMLC, and (iii) IMRT with MMLC, the
last one giving lower dose to adjacent bone.
304
Smee and Wurm
Subsequently, a 15-month-old boy with bilateral disease was referred

having had enucleation of the worst affected eye, and progressive disease
in the remaining eye. IMRT was now an option, thus all three treatment
methods were compared to evaluate dose on adjacent bone, with IMRT the
preferred option, although the difference between each of the three methods
was not great (Fig. 7B). For an anesthetized child, the treatment time may be
a signicant factor. Fortunately, all three methods are relatively short, and
thus the nal decision rests on the therapeutic advantage of one method over
another. Unfortunately for this young boy, despite the advantage of IMRT
progressive intra-ocular disease occurred because of prolonged treatment
time due to social circumstances.
Currently, IMRT is the preferred option for locally progressive retinoblastoma. For more bulky disease, localized dose escalation is always possible.
HN Tumors
Most HN cancers require high dose (6670 Gy) for the best chance of local
control as well as coverage of adjacent lymph nodes to moderate doses if
uninvolved for best regional control. There can also be considerable organ
movement with swallowing during the time frame of each individual treatment. It is an option for more localized cancers such as a unilateral glottic
carcinoma although not an appropriate choice. It could be considered as a
boost for a small volume supraglottic carcinoma or tongue base carcinoma
to try to avoid high dose to the arytenoids for voice maintenance. Although
the stereotactic localization becomes less critical (given the organ movement) as a means of lateralizing treatment and compensating for organ
movement within the CTV for infratemporal fossa diseases, a viable treatment approach thus becomes available. Figure 8 demonstrates such an
approach.
For any malignancy involving the infratemporal fossa, stereotactic irradiation could be considered. A 65-year-old female presented an adenocarcinoma ex pleomorphic adenoma arising 20 years after initial supercial
parotidectomy for a pleomorphic adenoma. Twelve months after the initial
surgery, recurrence occurred secondary to capsule rupture at initial surgery.
This was treated by further surgery and postoperative radiotherapy. The
carcinoma arose from the deep lobe of the parotid, and thus presented a bulky
local disease not surgically resectable. The laterally placed mandible thus
became the dose-dening structure having already received a signicant dose
from the previous radiotherapy. The aim in this situation was to give 66 Gy at
2-Gy fractions. Very little morbidity was experienced in this situation. Nodal
volumes were not included as the pattern of failure is either local or distant for
adenocarcinoma. Perhaps a higher dose should have been given, for despite
seemingly good initial control and radiological improvement, ultimately, local
progression occurred.
305
Figure 8 MRI: Axial slices demonstrating volume to be treated.
For smaller lesions (maximum diameter less than 3.5 cm) situated at the
skull base, single-dose SRS is an appropriate treatment method. All
treatments were delivered with head xation and stereotactic localization
via a BRW head ring. Frontal bone pin insertion was as low as possible
to allow the ring to be placed below the dened lesion. Planning took place
with XKnife with the common theme being to individualize each patients
treatment. Cones were used with the linac in arc mode to deliver the
treatment. Over a 12-year time frame, nine patients were treated, three with
recurrent nasopharyngeal carcinoma, four glomus tympanicum as primary
treatment, and two chordomas also as primary treatment. One of the
patients with nasopharyngeal carcinoma demonstrates the utility of stereotactic treatment. He presented in June 1994 a T2N3 nasopharyngeal carcinoma and received conventional radiotherapy (prior to the availability of
fractionated stereotactic treatment) to a dose of 68 Gy in 34 fractions
obtaining a complete remission. Fifteen months later, he recurred on the
right lateral nasopharyngeal wall as a localized area. This was treated
with SRS receiving a dose of 20 Gy to the 100% isodose curve within the
lesion. Once again, there was complete clearance of tumor only to recur
at the same site 3.5 years later. Since this was still a localized disease, he
had nasopharyngeal resection with microvascular free ap reconstruction.
There were no untoward healing problems with this, given the vascularized
graft. Surgical margins were very close, and thus he went on to have fractionated stereotactic radiotherapy receiving via a cone and arc technique a
306
Smee and Wurm
further 60 Gy in 30 fractions utilizing the relocatable head ring. Unfortunately, two years later he recurred laterally out in the ipsilateral infratemporal fossa and was given via a stereotactic fractionated approach 35 Gy
in 15 fractions as a palliative procedure. At no stage was there any evidence
of bone or soft tissue necrosis attributable to his treatment. Also, he did not
develop recurrent nodal disease despite having at presentation bulky nodal
malignancy. While not advocating this approach for all recurrent nasopharyngeal cases, it demonstrates that with small volume disease retreatments
are possible with normal tissue tolerance expectation. All three patients ultimately failed locally however, prolonged local control was achieved for a
day-only-procedure with no signicant morbidity.
For the chemodectoma patients, the diameter range of the tumors was
1.83.2 cm with dose varying according to the size. For the two larger
tumors (3.1 and 3.2 cm in diameter), a dose of 20 Gy was given, whereas
for the two smaller lesions it was 14 Gy as the marginal dose. This differential was on the premise that a higher dose may be required to control a
larger tumor. There was no signicant morbidity associated with treatment,
and all four patients had controlled tumor with the pulsatile tinnitus
decreased in all four.
The two chordoma patients had relatively localized tumor (maximum
diameter being 2.5 and 3.2 cm). Cones and arcs were used to deliver 20 and
18 Gy, respectively. The smaller lesion slightly increased in size, while the
larger lesion that received a slightly lower dose remains controlled eight
years after treatment.
The decision to use a stereotactic fractionated procedure is very much
based on the clinical circumstances, age, and the proximity of dose-limiting
structures. In a pediatric setting, there is an almost universal use of this
approach provided it ts into a eld size determinant. The necessity to limit
the dose to adjacent growing normal brain with an anticipated many years
of life is the driver for this approach. Although dose-escalation studies and
thus doseresponse relationships have not been done in many of the suitable
pediatric tumors, intuitively, a safe increase of dose, particularly if there is
macroscopic disease being treated, would seem to provide a greater likelihood of local control. It is worth noting that previous attempts at dose
increase would have been mainly limited by the inability to deliver high
doses with conventional treatment approaches safely. More wide eld treatment, such as paranasal sinus malignancy or nasopharyngeal carcinoma in a
child or teenager, can be performed using the stereotactic approach by using
the native 1-cm leaf width MLC for the linac, or by abutting the eldsa
traditional concept in conventional radiotherapy. In this approach, say to
treat the neck, conventional radiotherapy can be used for this component
and then the stereotactic approach for the local site (e.g., the sinus
or nasopharynx). The limiting feature is going to be not having the neck
treatment inuenced by the stereotactic localizing device. The GTC head
307
ring typically ts around the chin using a dental plate device for head
ring application; this may then inuence treatment of the neck disease.
The alternative is the use of a plastic mask device through which treatment
can be delivered although this may slightly compromise the precision of the
stereotactic approach. The reproducible accuracy for the bite block xation
device is 12 mm (of error), with a mask device having an error rate of
23 mm. The planning approach has to be with co-planer elds for the
stereotactic approach, thus placing some limitation upon beam direction
and normal organ avoidance. Scatter off the leaves needs to be considered
as a contributor to the abutted neck elds. The dosimetric aspects of this
would need to be tested in a phantom situation during the QA process for
dose verication.
Clinical circumstances dictating the stereotactic approach include those
tumors for which there is a readily dened tumor, well demonstrated on
imaging procedures, in which there is no necessity to encompass microscopic
disease. Chemodectomas t into this circumstance, although it could be
argued, however, that since moderate dose treatment (50 Gy) provides 90%
or greater chance of local control, conventional or even non-stereotactic
3D conformal approaches could be used. Other clinical circumstances
include the requirements to dose escalation, such as for nasopharyngeal
and paranasal sinus carcinomas. For nasopharyngeal carcinomas, numerous
studies demonstrate a benet in terms of local control with dose escalation.
Shrinking elds provides the opportunity to encompass microscopic
disease with sequential dose escalation. IMRT approaches allow the opportunity to simultaneously dose escalate, although the effect of a higher dose
per fraction upon normal tissue contained within the malignancy needs to
be considered.
Different normal tissues have varying tolerance levels. The most
sensitive structure is hair, although this is not typically regarded as a
dose-determining structure for treatment. A stereotactic approach even in
this situation may provide an advantage. Multiple elds (even 610) can
be used and thus the entry (and to lesser extent exit) dose for each beam
portal will consequently be less. Thus patchy alopecia, rather than large
volume hair loss may occur. More importantly, however, it is internal structures that dene sensitivity. The optic structures, starting with the lens
through to retina, optic nerve, and chiasm, are the dominant organs, progressing through to the brain stem, and increasingly the temporal lobe.
These can be regarded as absolute determinants of dose and dose per fraction. The relative determinants are now salivary glands, with preservation of
function impacting upon quality of life. Long-term follow-up has now
demonstrated that cranial nerve dysfunction (e.g., hypoglossal) can develop
even 1015 years following treatment. As local control becomes a more
achievable aim, factors such as this may inuence how we address the
neurovascular structures within or adjacent to treated tumors.
308
Smee and Wurm
For SRS, given that a high single dose is used, the indications are more
specic. Size is by far the greatest determinant, with this approach unable to
deliver meaningful doses to lesions larger than 3.5 cm maximum diameter,
even with multiple isocenters. This would exclude the opportunity to encompass microscopic disease around any malignancy. Organ movement can be
important, and thus make this approach difcult to extend below the skull
base even if a suitable stereotactic device were developed. Thus, the effect of
laryngeal movement with breathing and swallowing would make it such that
the high dose given as a single fraction with SRS would create signicant
risk of damage to normal tissues during the time frame the tumor moves
out of the treatment beam and normal tissue into it. For skull base tumors,
this is less of a problem as there is little to no organ movement at this site.
There are few dose-limiting structures at this level other than the spinal
cord. Most centers will use a separation of 3 mm from the optic chiasm
for the treatment of suprasellar tumors, and it would seem reasonable to
use the same approach for spinal cord, e.g., for foramen magnum lesions.
Whilst frameless stereotactic systems are being developed that can be
used for SRS, their denition of accuracy is of the order of 24 mm error
allowable, compared to less than 1 mm for a xed stereotactic system.
Specic tumors that are likely to be treated include: meningiomas (more
typically extending from intra to extracranial), chemodectomas, schwannomas, nasopharyngeal carcinomas (as a boost at primary presentation or a
relapse), or other localized carcinomas (including in the pediatric age group).
INCIDENCE
The number and types of tumors that will be treated are very much
inuenced by the referral pattern for the particular department. Obviously,
where a department has a relatively large referral practice for, say, nasopharyngeal carcinomas, this malignancy will feature prominently in the
number of patients treated. Although infrequent, pediatric tumors typically
require very conformal treatment approaches including dose escalation for
optimum local control whilst limiting dose to adjacent structures. Thus,
normal tissue sparing has an added dimension, given the impact of
chemotherapy (typically used in all pediatric malignancies), and for the
survivors the longevity of life that they are exposed to after their cure.
The type of tumors treated is reected by a co-operative multidisciplinary interaction with HN surgeons, neurosurgeons, and pediatricians.
The typical HN cancers are unlikely to be treated in this fashion, except as a
boost to macroscopic disease. With these cases excluded, it is the infrequent
benign tumors that would be more likely appropriate for this approach as
a denitive treatment. The number of these tumor types referred to any
HN unit is always going to be few; hence, the tumor experience would be
small.
309
Neurosurgical tumors may indeed constitute the majority of lesions

treated in this region. The types of tumors treated by this approach could
include:

schwannomas
neurobromas
cervical metastases
chordomas
hemangioblastomas
There is certainly evidence in the intracranial tumor population that

stereotactic approaches (specically SRS) have a track record of benet for
these types of lesions. While some altered movement would occur in the neck
region, the proximity of these lesions to the vertebral bodies is going to limit
that as a variable. Thus, a hypofractionated or even single-dose approach
could be considered. There is evidence for intracranial vestibular schwannomas that hypofractionated methods have high control rates, and authors
stated benet in terms of hearing preservation. Although there is no literature
evidence to support it, this philosophy could be applied in the cervical vertebral region where the dose-limiting structure, the spinal cord, is crucial. While
conventional wisdom would see the use of 1.82.0-Gy fractions, the parallel
of the vestibular schwannoma in apposition to the brain stem, in tolerating
the hypofractionated approach, would make this method worth considering.
This approach has certainly been used with the CyberKnife and could be
used for other systems that have an extracranial technique.
ORGAN SITE-SPECIFIC DIFFICULTIES
Recognition of the dose-limiting structures in any region of the body is the
beginning to determining which tumor types can be treated in that region.
Moving outside the head and considering relatively small volumes, the dominant structure in the neck is going to be the cervical spinal cord. This structure is going to have varying sensitivities depending upon whether single
dose, hypofractionated, or conventionally fractionated approaches are used.
From the experience of vertebral metastases, it is known that at least 23
vertebral segments of the cervical cord will tolerate a single dose of 8 Gy,
20 Gy in ve fractions, and 30 Gy in 10 fractions. The limiting feature to carte
blanche use of these doses is that the majority of so-treated patients do not
survive many years after treatment. Thus, our current statement of tolerance
may be an over statement. Certainly, the large worldwide experience with
HN cancers would have 50 Gy conventionally fractionated as well tolerated
by the spinal cord. The tolerance of laryngeal cartilages to large single doses
or hypofractionated doses is not well known in the long-term situation. This
should be considered cautiously, particularly if large volumes of the cartilage
are to be included in the treatment elds.
310
Smee and Wurm
The main sites of specic difculty would then relate to specic tumors
and their shape. Foramen magnum meningiomas adopting a horse-shoe
shape that have intra- and extracranial components represent a specic
challenge. At the beginning of this process, the xation has to allow
treatment across the skull base boundaries. This would exclude any headorientated xation device, since the ring-base system would not be able to
be extended sufciently down into the neck while maintaining the head
attachment. The HNL xation device copes with this application being able
to be applied for coordinate referencing in the head and neck without altering patient position. This can be used with a bite block chin positioning
system or thermoplastic mask. It must be recognized that both devices have
reproducible errors of setup at best 12 mm, worst 24 mm.
For any xation device in the neck to work well, a patient with a long
neck is ideal. The patient of stocky build with very short neck can be taxing
for any device. Tolerance of an extended neck posture for longer time can be
difcult in this circumstance. This is obviously exacerbated for all patients
by any degree of kyphosis. Providing extended support for the head may
allow too much movement within the stereotactic xation device such as
to negate this as a treatment approach.
Those tumors adjacent to the spinal cord provide some challenge in
terms of dose delivery. Fortunately, most of the benign tumors (e.g., meningiomas, schwannomas, etc.) have dose-control parameters that t within
tolerance levels of the spinal cord. As indicated above, however, for those
tumors that abut up to the cord on multiple surfaces (such as a horse-shoe
type shape), respecting cord tolerance in this situation is difcult. It is for
this reason that parallel-opposed eld arrangements are frequently used.
This, however, regards the spinal cord as a target structure to the same
extent as the tumor. This circumstance is handled well with stereotactic
IMRT using the MMLC. Figure 9A demonstrates such a case with a foramen magnum meningioma, the DVH contrasting the dose received by the
cord for 1-cm leaf width MLC (Fig. 9B) versus the 4-mm leaf width with
MMLC (Fig. 9C). Although the differences are not large, the concept of
lowering normal tissue dose is paramount.
There are malignancies arising within the cervical region for which
control with radiotherapy can be expected; however, the doses required
are those that would ordinarily exceed spinal cord tolerance. Cervical cord
chordomas and neurobrosarcomas (malignant schwannomas) are two such
malignancies, fortunately quite rare. Doses of the order of 70 Gy are required
for control. Conventionally planned treatment cannot provide this.
There are also situations in which retreatment is considered. This
could apply to both benign and malignant lesions, provided reasonable benet had been gained from the initial treatment. Any distance away from the
cord makes sparing this structure easier. However, proximity to the cord
does not exclude this option, particularly where there is no other treatment
311
Figure 9 Foramen magnum meningeoma. (A) Volume shell demonstrating proximity of structures. (B) DVH for 1-cm leaf width MLC. (C) DVH for 4-mm leaf width
MMLC.
method available. The cervical chordoma patient detailed above demonstrates that retreatment is still possible even for a relatively radio-resistant
malignancy.
FRACTIONATED DOSE SELECTION

Results with conventional radiotherapy give us a background for determining
appropriate doses. For most of the benign tumors (e.g., schwannomas, chemodectomas, etc.), doses of 4550 Gy provide local control rates of 8595%.
A more sophisticated approach is unlikely to improve on this. Even for
312
Smee and Wurm
retinoblastomas, high local control rates have been recorded with these doses.
Thus, initial consideration is going to be to avoid adjacent structures.
Malignancies require higher doses, macroscopic disease typically
requiring doses of 6070 Gy in 2-Gy fractions, microscopic disease greater
than 55 Gy. Lower doses can be considered where the treatment is hypofractionated; however, dose-limiting structures such as the spinal cord need then
to be duly considered. Simultaneous dose escalation within the malignancy
is feasible with IMRT using comparative hypofractionation while normally
treating a large area.
For benign tumor control such as with vestibular schwannomas, there
is ample evidence supporting SRS, with that marginal doses of 1214 Gy
sufcing, with follow-up for many series up to about ve years. A similar
dose could be used for the same tumor in other sites. Chemodectomas would
have a comparable dose response prole, given that moderate doses control
these tumors in a conventionally fractionated series.
It is with malignancies that higher doses are required, particularly
where this is given as a sole treatment. Where this treatment is given as a
boost, the delivered dose would be inuenced by the initial wide eld treatment plan. The volume to be treated is also obviously going to be a determinant. Thus, marginal doses of 1824 Gy are appropriate in these situations
for nasopharyngeal carcinomas, infratemporal fossa malignancies, and
chordoma/chondrosarcomas.
Within the tumor dose, heterogeneity has been previously considered
to be a signicant feature, a reection of the different dosing practices for
GammaKnife versus linac treatment. The 5060% isodose is typically used
as a dose parameter for a GammaKnife procedure, hence, part of the tumor
volume would be receiving up to double the prescribed dose. The concern
with this approach is that if a critical structure such as a cranial nerve is
at the site of dose maximum, then this would increase the risk of treatment
related side effects. This does not appear to be a problem with currently used
doses for benign tumors as the cranial nerve decit rate, say in treating
vestibular schwannomas, is low. However, when higher marginal doses,
1820 Gy, were used to treat these tumors, cranial nerve palsy rates of the
order of 1020% were encountered. There is no reason to believe that the
same effect would not happen in the extracranial situation. The high local
control rates of nasopharyngeal carcinoma with low morbidity indicates
that these doses can be tolerated provided what is treated is the macroscopically evident malignancy, without allowing for microscopic extension.
Unless some form of control can be achieved obviously, progressive malignancy will cause the same decit with certainty.
There is less heterogeneity with the linac and charged particle
treatment methods as the dosing is to the 80100% isodose. Hence, the dose
maximum is fortunately not that much greater than the prescribed dose.
Although there are fewer published linac series, comparable results are
313
reported suggesting that dose inhomogeneity may not be a fact or in achieving tumor control. As indicated above, it may have an impact, however, on
complication rates.
TREATMENT DELIVERY
These patients are treated similar to any patient with an intracranial lesion
who is having SRS. An MRI is performed on the day, or within a few days
prior to the procedure, the BRW headring is applied, and then the non-contrast CT scan is performed. The two images sets are fused, and all outlining is
performed on the MR images, and in whatever planes that are necessary
axial, coronal, or sagittal. Utilizing BEV various beam directions are chosen,
and a beam shaping method selected. If the lesion is spherical/elliptical then
cones are used; thus the planning process will require selection of appropriate
arc start and stop points, each patients treatment being individualized. There
is no formal library of plans. An isocenter is thus determined. The dose contribution from each arc is evaluated with the accompanying number of monitor units for each arc. After approval of the best plan, a paper printout is then
generated with all dose parameters being veried by an independent physicist.
Once this step has been accepted, all treatment parameters are delivered to the
linac control and Record and Verify System (RVS). The phantom base is
attached to the couch mount system, and the isocenter location is further veried, and all arcs go through a dummy run to ensure there is no collision prospect. The patient is then placed on the couch, and the head ring docked to
the couch mount system. A simple verication is then performed to ensure
that there has been no movement of the head ring during the day, as all measurements are in relationship to the head ring, not the patients head. A port
lm, AP and lateral, is taken with the angiogram localizer box attached. Subsequently, the ducial points on this are digitized and a coordinate determined. This is a verication that the intended isocenter is in fact the point
treated. The localizer box is removed and treatment proceeds. Gantry movements are directed from outside the room at the console, couch movements
are done inside the room. Following this the head ring is removed and the
patient goes home, all these treatment being done as an outpatient. For irregular shaped lesions, not suitable for treatment via cones, the MMLC can be
used for eld shaping, but the dose is delivered as a single fraction. In this
situation, 1015 different beam directions are usually used as static elds. This
enables slightly larger lesions to be treated. The same planning process takes
place with static eld directions being the only differential. The MMLC is
directly linked to the linac RVS and control system. Isocenter verication is
performed in the same manner.
For neck lesions, a different xation system is used with depth readings
replacing the phantom localizer. Although single-dose treatments could be
314
Smee and Wurm
performed, the slightly greater error (23 mm) on setup would create
concern about their use.
Fractionated Stereotactic
Irrespective of how many fractions are being used, a relocatable device is
still required. The planning process is essentially the same, the same treatment being delivered each day, same amount of monitoring, same verication process, and input to the RVS. For all patients an isocenter verication
port lm is done with the localizer box, the ducial points being digitized,
and compared to the intended isocenter. For the patients having only a
few treatments, only one port lm (AP and lateral) is taken, but for those
patients having many weeks of treatment this is done once per week.
For the rst ve years, this treatment was done with cones (if necessary
multiple isocenters, and arcs, using the same xation and setup devices).
There were more restrictions on the amount of arc travel possible because
of the shoulders compared to intracranial lesions. Hence to begin with, there
were few patients treated for extracranial HN lesions during this time frame.
However, with the MMLC multiple xed elds became possible. Treatment
planning was the same, as was delivery. In terms of scheduling, all MMLC
patients were grouped together for convenience. Given the long background
with fractionated stereotactic treatments, there is no real difference in
delivery process for cone versus MMLC.
LOGISTICS
In terms of treatment delivery devices, this can be divided into dedicated and
semidedicated. It can be argued that a cyclotron delivering charged particles
is a dedicated device. Other than chordomas, there are few reports of extracranial tumors in the HN region being treated by protons. Thus, for the
dedicated facilities, this can be further divided into GammaKnife and linac.
Because of the limitation of extending the GammaKnife into the neck, its
role is always going to be limited to the tumors at the immediate skull base
and only to radiosurgery. Whilst a linac can be designed to function as a
dedicated machine for stereotactic treatment, in most facilities it would
work as a semidedicated device. BrainLAB has developed the Novalis as
a designated device capable of providing: single-dose and fractionated treatment, multiple xed elds or arcing, and dynamic or segmental IMRT.
For the semidedicated facilities, many of the reported series have used
1-cm leaf width MLC for beam shaping for their treatment. Any of the three
major manufacturers (Varian, Siemens, and ELEKTA) can provide this
method. For smaller irregular eld treatment, the smaller leaf width MMLC
can provide greater dose conformality (Fig. 10). Varian has an accelerator
model with a smaller leaf width capability at the middle of its MLC that
315
Figure 10 Fluence maps generated as part of IMRT quality assurance indicating

better target shaping with the smaller leaf width MMLC. MLC: 1-cm leaf width;
MMLC: 4-mm leaf width.
is capable of providing small and large eld capability. Virtual MMLC

treatment may be possible with MLC devices by rotating the gantry head
through 90 for part of the treatment, and thus delivering the divided
dose from 0 to 180 and 90 to 270 angulation. True MMLC treatment
typically requires an add-on device, and in this situation the weight of
the MMLC and its effect upon gantry rotation needs to be considered.
BrainLAB has their own MMLC as an integral part of the Novalis.
The Radionics MMLC can be added to both Varian and Siemens linac. Collimator clearance from the patient with Varian-may place some limitation
with an add-on MMLC for particular gantry positions.
Helical tomotherapy using a small linac imbedded in a CT gantry can
also be considered a dedicated facility although there are no reports of this
being used stereotactically.
Treatment and planning systems require a suitable planning platform
that provides both hardware and software. Usually a workstation is required
for the former with the software having full 3D visualization processing,
image fusion, and a fast inverse planning algorithm to allow for IMRT. Many
of the commercially available planning systems that incorporate an IMRT
component can be utilized in a stereotactic mode although not specically
designed for it. More dedicated software approaches exist with BRAINLAB
and Radionics for stereotactic treatment. These modules also allow for
SRS. There are also non-commercial packages developed by individual
316
Smee and Wurm
institutions of high quality. The GammaKnife has its own planning system,
Gamma Plan.
In terms of localization devices for SRS, the GammaKnife centers use
the Leksell headframe. The majority of linac centers whose results are
reported here will use the RadionicsBRW headframe. For the fractionated
series various commercial models are now available. For the patients
reported in this article, the authors have used the Radionics GTC, and
HNL xation and localizing systems.
FUTURE USE AND RESEARCH
Given the relative ease that treatments can be delivered stereotactically in the
HN region, its use can be extended to greater patient numbers. Certainly,
any department that has a large number of patients with nasopharyngeal
carcinoma to treat should have fractionated stereotactic approaches available as a routine to boost the primary site to doses of 7080 Gy. This dose
may need to be adjusted if given with concurrent chemotherapy. For small
primaries, or where signicant regression of the primary has occurred during treatment, SRS may be an appropriate boost method. Stereotactic
IMRT may become a preferred method for the total duration of the radiotherapy in the same context that many centers will now use IMRT as their
standard approach for nasopharyngeal carcinoma. Retreat situations
require high doses to be given to imaging evident lesions, an ideal situation
for stereotactic treatment.
Other malignancies that can be considered for stereotactic approach
are paranasal sinus carcinomas and chordomas. There would seem to be
little role for SRS in the primary situation except as a boost. With the
appropriate head xation device this site lends itself to this approach with
high doses required and critical normal structures immediately adjacent to
at risk areas. Given the ability to dose escalate, it may be possible to consider denitive radiotherapy as an alternative to major surgery and postoperative radiotherapy. As there would not be an option for surgical
salvage, this would require truly informed consent to have this as a standard method. An appropriate circumstance is obviously the patient with
surgically resectable disease who has a medical contraindication to such a
procedure.
Gating techniques are now being developed to enable higher dose delivery
for lung cancers. A similar approach could be applied to laryngeal malignancies
that are localized in nature. The clinical circumstance, however, is going to be
limited, as it will only relate to those malignancies that have low likelihood of
developing nodal disease. This relates to the fact that while the central larynx
may move vertically with swallowing, the adjacent nodal areas remain stationary. There would have to be differential allowance for movement in larynx
structure while another target remained stationary.
317
Salivary gland sparing is now a major challenge for radiotherapy techniques to improve the quality of life for the survivors. This has spawned the
use of IMRT in the HN region particularly for nasopharyngeal carcinomas.
Greater use of this approach is now possible with stereotactic xation.
A comparison of this method with non-stereotactic delivery is a worthy
research project. It may come down to a difference only in the ease of reproducibility of patient setup.
The number of benign tumors available to be treated in this region is
always going to be small, and since the doses required for control are usually
well tolerated, demonstrating a clear advantage for a stereotactic approach
may be difcult. There could be an advantage in hypofractionating the treatment, delivering the radiotherapy over 510 fractions. This would lead to less
disruption to the patient in terms of treatment visits. For the larger lesions, a
negative to this approach would be the length of the cranial nerve (usually the
vagus) that would be exposed to the hypofractionated dose (e.g., 5 4 Gy).
A crucial aspect of the process that makes all of this possible is the
quality assurance program. For SRS, doses are being used that aim to
damage the tissue being treated. It is the small size of the lesion being treated
and the accuracy of delivery that makes this approach safe. While not compromising this aspect, streamlining the process may shorten overall treatment time, thus making it more comfortable for the patient. Similarly,
with fractionated treatments evaluating the extent of the QA process may
enable better use of departmental, including staff, resources.
REFERENCES
1. Dhanachai M, Kraiphibul P, Pochanugool L, Dangprasert S, Sitathanee C,
Laothamatas J, Kulapraditharom B, Theerapancharoen V, Sirisinha T,
Yongvithisatid P, Assavaprathuangkul P, Boonpitak K. Stereotactic radiotherapy in nasopharyngeal carcinomapreliminary results. In: Kondziolka D, ed.
Radiosurgery. Vol. 4. Basel: Karger, 2002:162166.
2. Len Gyel E, Baricza K, Somo Gyi A, Olajos J, Papai Z, Godney M, Nemett G,
Esik O. Reirradiation of locally recurrent nasopharyngeal carcinoma.
Strahlenther Onkel 2003; 179(5):298305.
3. Lee AW, Foo W, Law SC, Peters LJ, Poon YF, Chapell R, Sze WM, Tu SY,
Lau WH, Ho JH. Total biological effect on late reactive tissues following
reirradiation for recurrent nasopharyngeal carcinoma. Int J Radiat Oncol Biol
Phys 2000; 46(4):865872.
4. De Crevoisier R, Bourhis J, Domenge C, Wibault P, Koscielny S, Lusine A,
Mamelle G, Janot F, Julieron M, Leridant AM, Marandas P, Armano JP,
Schwaab G, Luboinski B, Eschwege F. Full-dose reirradiation for unresectable
head and neck carcinoma: experience at the Gustave-Roussy Institute in a series
of 169 patients. J Clin Oncol 1998; 16(11):35563562.
5. Hwang JM, Fu KK, Phillips TL. Results and prognostic factors in the retreatment of locally recurrent nasopharyngeal carcinoma. Int J Radiat Oncol Biol
Phys 1998; 41(5):10991111.
318
Smee and Wurm
6. Chau DT, Sham JS, Kwong DL, Wei WI, Au GK, Choy D. Locally recurrent
nasopharyngeal carcinoma: treatment results for patients with computed tomographyassessment. Int J Radiat Oncol Biol Phys 1998; 41(2):379386.
7. Bari ME, Kemeny AA, Forster DM, Radatz MW. Radiosurgery for the control
of glomus jugulare tumors. J Pak Med Assoc 2003; 53(4):147151.
8. Maarouf M, Voges J, Landwehr P, Bramer R, Treuer H, Kocher M, Muller RP,
Sturm V. Stereotactic linear accelerator-based radiosurgery for the treatment of
patients with glomus jugulare tumors. Cancer 2003; 97(4):10931098.
9. Eustacchio S, Trummer M, Unger F, Schrottner O, Sutter B, Pendl G. The role
of Gamma Knife radiosurgery in the management of glomus jugular tumors.
Acta Neurochir Suppl 2002; 84:9197.
10. Elshakkh MA, Mahmoud-Ahmed AS, Kinney SE, Wood BG, Lee JH,
Barnett GH, Suh JH. Recurrent head-and-neck chemodectomas: a comparison
of surgical and radiotherapeutic results. Int J Radiat Oncol Biol Phys 2002;
52(4):953956.
11. Foote RL, Pollock BE, Gorman DA, Schomberg PJ, Stafford SL, Link M,
Kline RW, Strome SE, Kasperbauer JL, Olsen KD. Glomus jugulare tumor:
tumor control and complications after stereotactic radiosurgery. Head Neck
2002; 24(4):332328; discussion 338339.
12. Leber KA, Eustacchio S, Pendl G. Radiosurgery of glomus tumors: midterm
results. Stereotact Funct Neurosurg 1999; 72(suppl 1):5359.
13. Jordan JA, Roland PS, McManus C, Weiner RL, Giller CA. Stereotactic radiosurgery for glomus jugulare tumors. Laryngoscope 2000; 110(1):3538.
14. Liscak R, Vladyka V, Wowra B, Kemeny A, Forster D, Burzaco JA, Martinex R,
Eustacchio S, Pendl G, Regis J, Pellet W. Gamma Knife radiosurgery of the
glomus jugulare tumora multicentre experience. Acta Neurochir (Wien) 1999;
141(11):11411146.
15. Liscak R, Vladyka V, Simonova G, Vymazal J, Janouskova L. LekselI Gamma
Knife radiosurgery of the tumor glomus jugulare and tympanicum. Stereotact
Funct Neurosurg 1998; 70(suppl 1):152160.
16. Zorlu F, Gurkaynak M, Yildiz F, Oge K, Atahan IL. Conventional external
radiotherapy in the management of clivus chordomas with overt residual
disease. Neurol Sci 2000; 21(4):203207.
17. Debus J, Schulz-Ertner D, Schad L, Essig M, Rhein B, Thillmann CO,
Wannenmacher M. Stereotactic fractionated radiotherapy for chordomas and
chondrosarcomas of the skull base. Int J Radiat Oncol Biol Phys 2000; 47(3):
591596.
18. Crockard A, Macaulay E, Plowman PN. Stereotactic radiosurgery. VI.
Posterior displacement of the brainstem facilitates safer high dose radiosurgery
for clival chordoma. Br J Neurosurg 1999; 13(1):6570.
19. Muthukumar N, Kondziolka D, Lunsford LD, Flickinger JC. Stereotactic
radiosurgery tor chordoma and chondrosarcoma: further experiences. Int
20. Kocher M, Voges J, Staar S, Treuer H, Sturm V, Mueller RP. Linear accelerator radiosurgery for recurrent malignant tumors of the skull base. Am J Clin
Oncol 1998; 21(1):1822.
319
21. Chua DT, Sham JS, Kwong PW, Hung KN, Leung LH. Linear acceleratorbased stereotactic radiosurgery for limited, locally persistent, and recurrent
nasopharyngeal carcinoma: efcacy and complications. Int J Radiat Oncol Biol
Phys 2003; 56(1):177183.
22. Ahn YC, Lee KC, Kim DY, Huh SJ, Yeo IH, Lim DH, Kim MK, Shin K,
Park S, Chang SH. Fractionated stereotactic radiation therapy for extracranial
head and neck tumors. Int J Radiat Oncol Biol Phys 2000; 489(2):501505.
23. Chang JT, See LC, Liao CT, Ng SH, Wang CH, Chen IH, Tsand NM,
Tseng CK, Tang SG, Hong JH. Locally recurrent nasopharyngeal carcinoma.
Radiother Oncol 2000; 54(2):135142.
24. Tate DJ, Adler JR Jr, Chang SD, Marquez S, Eulau SM, Fee WE, Pinto H,
Gofnet OR. Stereotactic radiosurgical boost following radiotherapy in primary nasopharyngeal carcinoma. Int J Radiat Oncol Biol Phys 1999; 45:
915921.
25. Kocher M, Voges J, Staar S, Treuer H, Sturm V, Mueller RP. Linear accelerator
radiosurgery tor recurrent malignant tumors of the skull base. Am J Clin Oncol
1998; 21(1):1822.
26. Cmelak AJ, Cos RS, Adler JR, Fee WE Jr, Gofnet OR. Radiosurgery tor skull
base malignancies and nasopharyngeal carcinoma. Int J Radiat Oncol Biol Phys
1997; 37(5):9971003.
27. Kondziolka D, Lunsford LD. Stereotactic radiosurgery for squamous cell
carcinoma of the nasopharynx. Laryngoscope 1991; 101(5):51922.
28. Habermann W, Aznarotti U, Groell R, Wolf G, Stammberger H, Sutter B ,
Pendl G. Combination of surgery and gamma knife radiosurgerya therapeutic
option for patients with tumors of nasal cavity or paranasal sinuses inltrating
the skull base. Acta Otorhinolaryngol Ital 2002; 22(2):7479.
29. Schulz-Ertner D, Haberer T, Jakel O, Thilmann C, Kramer M, Enghardt W,
Kraft G, Wannenmacher M, Debus J. Radiotherapy for chordomas and lowgrade chondrosarcomas of the skull base with carbon ions. Int J Radiat Oncol
Biol Phys 2002; 53(1):3642.
30. Noel G, Habrand JL, Mammar H, Pontvert D, Hair-Meder C, Hasboun D,
Moisson P, Ferrand R, Beaudre A, Boisserie G, Gaboriaud G, Mazal A,
Kerody K, Schlienger M, Mazeron JJ. Combination of photon and proton
radiation therapy for chordomas and chondrosarcomas of the skull base: the
Centre d Prontontherapie DOrsay experience. Int J Radiat Oncol Biol Phys
2001; 51(2):392398.
31. Debus J, Schulz-Ertner D, Schad L, Essig M, Rhein B, Thillmann CO,
Wannenmacher M. Stereotactic fractionated radiotherapy for chordomas and
chondrosarcomas of the skull base. Int J Radiat Oncol Biol Phys 2000; 47(3):
591596.
32. Ahn YC, Lee KC, Kim DY, Huh SJ, Yeo IH, Lim DH, Kim MK. Fractionated
stereotactic radiation therapy for extracranial head and neck tumors. Int J
33. Lohr F, Debus J, Frank C, Herfarth K, Pastyr O, Rhein B, Bahner ML, Schlegel W, Wannenmacher M. Noninvasive patient xation for extracranial stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 1999; 45(2):521527.
320
Smee and Wurm
34. Pai PC, Chuang CC, Wei KC, Tsang NM, Tseng CK, Chang CN. Stereotactic
radiosurgery for locally recurrent nasopharyngeal carcinoma. Head Neck 2002;
24(8):748753.
35. Feigenberg SJ, Mendenhall WM, Hinerman RW, Amdur RJ, Friedman WA,
Antonelli PJ. Radiosurgery for paraganglioma of the temporal bone. Head
Neck 2002; 24(4):384389.
36. Foote RL, Pollock BE, Gorman DA, Schomberg PJ, Stafford SL, Link MJ,
Kline RW, Strome SE, Kasperbauer JL, Olsen KD. Glomus jugulare tumor:
tumor control and complications after stereotactic radiosurgery. Head Neck
2002; 24(4):332338; discussion 338339.
37. Chau DT, Sham JS, Hung KN, Leung LH, Cheng PW, Kwong PW. Salvage
treatment for persistent and recurrent T1-2 nasopharyngeal carcinoma by
stereotactic radiosurgery. Head Neck 2001; 23(9):791798.
38. Hinerman RW, Mendenhall WM, Amdur RJ, Stringer SP, Antonelli PJ, Cassisi
NJ. Denitive radiotherapy in the management of chemodectomas arising in
the temporal bone, carotid body, and glomus vagale. Head Neck 2001;
23(5):363371.
39. Chau DT, Sham JS, Hung KN, Kwong DL, Dwong PW, Leung LH. Stereotactic radiosurgery as a salvage treatment for locally persistent and recurrent
nasopharyngeal carcinoma. Head Neck 1999; 21(7):620626.
40. Ahn YC, Kim DY, Huh SJ, Baek CH, Park K. Fractionated stereotactic radiation therapy for locally recurrent nasopharynx cancer: report of three cases.
Head Neck 1999; 21(4):338345.
41. Firlik KS, Kondziolka D, Lunstord LD, Janecka LP, Flickinger JC. Radiosurgery for recurrent cranial base cancer arising from the head and neck. Head
Neck 1996; 18(2):160165; discussion 166.
42. Buatti JM, Friedman WA, Bova FJ, Mendenhall WM. Linac radiosurgery for
locally recurrent nasopharyngeal carcinoma: rationale and technique. Head
Neck 1995; 17(1):1419.
Index
Anatomical imaging, 109122

AutoCrane system, for serial
tomotherapy, 92
AZ-VUB approach to treatment
verication, 4052
class solution analysis, 4749
dose distribution tool, 50
evaluation of functionality,
4346
prefraction verication, 52
pre-treatment verication, 4950
prostate tumors, 244248
quality assurance method, 4346
quality assurance test pattern, 46
verication plans, 4849
Beam weighting, with CyberKnife ,

7778
Biological imaging, 109122
Biologically effective dose (BED),

141143
Body Frame , spine, radiosurgery of,
261
BodyFIX system
serial tomotherapy, 93
spine, radiosurgery of, 261
Breathing control, pentagonal
template for, 78, 14
Cholangiocellular carcinoma (CCC),

176177, 184
Cell survival curves, in
fractionation radiobiology,
136140
Charged particle method, for head
and neck tumors, 285
Chemodectomas, radiotherapy of,
287, 289,307,308,311,312
Italics indicates pages with illustrations.
321
322
Cholangiocellular carcinoma (CCC),
178
Chordoma, radiotherapy of,
287,300,301
Clincal target volume (CTV),
calculation of, 12
Conformal radiation therapy
(CRT), 19
Corvus Inverse Treatment Planning
System, in serial
tomotherapy, 95
CT scans
in serial tomotherapy, 114119
slow, for tumor mobility
characterization, 115116
for target verication, in treatment
planning, 1516, 16
for tumor mobility characterization,
115
two-phase, for tumor
mobility characterization,
115116
CyberKnife
beam weighting, 7778
clinical trials, 8485
components of, 76
inverse planning with, 7578
limitations of, 72
node selection, 77
overview of, 75
respiration tracking with, 7884
spine radiosurgery and,
258, 261
whole-body radiosurgery, 7186
Debulking, of tumors, SRS vs. SRT,

165166
Digitally reconstructed radiographs
(DRRs), 2728, 32
X-rays fusion of images, residual
set-up errors, 32
Dose distribution, heterogeneous,
149158
Index
Dose volume histogram (DVH), 12,
149151
for spinal cord, 276
Double-trouble effect, in
radiobiology, 146148
Dynamic arc treatment delivery,
3637, 36, 37
Dynamic eld shaping, 31
ECSRT, site-specic aspects
of, 120122
ELEKTA Stereotactic Body Frame ,
517, 7
EORTC score
radiation toxicity, 239
rectal toxicity, 240
urinary toxicity, 239
Equivalent uniform biologically
equivalent dose (EUBED),
163164
Equivalent uniform dose (EUD),
162163
ExacTrac/Novalis Body System
graphical interface, 24
positioning procedure, 2627
prostate cancer studies, 3034
UCLA patient motion
tracking, 5760
evaluation of the technique,
6465
gated control software, 5960
target localization example, 63
See also Novalis BODY
System
Fluoroscopy, for tumor mobility

characterization, 115
Fractionated dose selection,
head and neck
tumors, 311313
Fractionated radiotherapy
head and neck, 287
techniques, 286, 316, 317
Index
Fractionation radiobiology, 133134
biologically effective does (BED),
141143
cell survival curves, 136140
empirical power laws, 134136
incomplete repair model,
143144
mechanistic models, 136140
normal tissue complication
probability (NTCP),
145146
redistribution, 134
reoxygenation, 133
repopulation, 133134
single-hit, multitarget killing model,
137138
tumor control probability
(TCP), 145
Fractionation
unconventional, for SRT,
166167
vs. single dose treatment, 133
See also Fractionated
radiobiology; Fractionated
radiotherapy.
GammaKnife , 72
for head and neck tumors, 285
Guardian slider, 39
Head and neck tumors,

fractionated dose delivery, 310
fractionated dose selection,
311313
fractionated radiotherapy
of, 287
IMRT uence maps, 315
incidence of, 308309
logistics of treatment delivery,
314316
radiotherapy of, 285317
323
[Head and neck tumors,]
charged particle method, 286
chemodectomas,
287, 289, 307, 308,
311, 312
chordoma, 287301
GammaKnife method, 286
linac method, 286
nasopharyngeal carcinoma, 287,
292295, 294
paranasal sinus carcinoma,
287289, 297299, 298
dosing, 312
retinoblastoma, 300304, 303
site-specic difculties, 309311
skin carcinoma with perineural
spread, 299300
stereotactic radiosurgery,
305308
future use, 316317
treatment delivery, 313314
Hepatic malignancies, role
of imaging in patient
selection, 112
Hepatitis C, and liver tumors, 166
Heterogeneous dose distribution,
149158
and NTCP, 151152
and TCP, 152158
Hidden target test (HTT), for
verication of data, 29
Hypofractionation
for liver tumors, 185187
for lung tumors, 217218
of prostate tumors, 232233
See also Hypofrationation boost,
for prostate tumors.
Hypofractionation boost, for prostate
tumors, 232243
patient immobilization, 237238
PTV2 treatment planning, 236
quality assurance procedure, 237
results, 238241
treatment description, 235236
324
Image-guided radiation therapy
(GRT), 20
Image-guided robotic radiosurgery,
rationale for, 73
IMRS treatment delivery, 3740
Novalis Body System
and, 3839
IMRT targets, in serial
tomotherapy, 96
Incomplete repair model, 143144
Infrared tracking. See IR.
Intensity modulated X-ray beams,
prostate tumors and,
233244
Intensity-modulated radiation therapy
(IMRT), 19, 20
ExacTrac/Novalis Body System ,
20, 22
IR tracking, real-time, 2224
in Novalis Body System, 23
Isocenter
illustration of measurements,
prostate cancer
study, 32
in lung tumors, 211213
verication, in treatment planning,
1012, 1617
Late radiation toxicity score

(EORTC), 241
Linac method
head and neck tumors, 286
alignment parameters, 1011
Linear accelerator. See Linac.
Liver tumors, 177191
and hepatitis C, 178
hypofractionation, 185186
incidence of, 178179
radiosurgery for, 186189
and SBRT, 182, 183, 190, 191
dose escalation, 183185
doses and outcome, 183
organ motion, 183185
Index
[Liver tumors]
partial radiation, 185189
target doses, 190
treatment planning,
182183
Lung cancer, role of
imaging in patient
selection, 111112
Lung metastases, stereotactic
raditation for, 202203
Lung tumors
radiation tolerance of healthy
tissues, 207216
normal tissue considerations,
203206
and SBRT, 197227
hypofractionation 217220
indications for, 198199
isocenter, 212213
local control and survival,
220226
patient immobilization,
210213
single dose irradiation, 217
standard treatments, 200202
target denition, 213214
target reproducibility,
210213
target verication, 212213
target volume, 220
tissue heterogeneity issues,
215216
toxicity, 225226
treatment outcome, 216220
treatment planning, 214215
target volumes in multislice CT
scans, 118
tumor control considerations,
206207
Lungs, radiation tolerance of,
207216
RTOG criteria, 209
Index
325
Nasopharyngeal carcinoma,
radiotherapy of,
287, 292295, 294
Node selection, with
CyberKnife , 77
NOMOS Peacock serial
tomotherapy approach, 94
Normal tissue complication
probability (NTCP)
fractionation radiobiology,
145146
heterogeneous dose distribution,
151152
Novalis Body System, 1965
description of, 20
features of, 2136
IR tracking, real-time, 2224
patient positioning, 2128
plan optimization window, 41
planning wizard, 39
prostate cancer studies, 3034
spine radiosurgery and,
258260
stereoscopic X-ray imaging,
2528
verication testing, 29
Pentagonal template, for breathing

control, 78, 14
Phantom choice, 49
Phantom measurements, as verication
test, 2830
Physical dose vs. prescribed dose,
146
Planning target volume (PTV),
calculation of, 12
Power-law relationship, in
fractionation
radiobiology, 135
Prostate cancer studies,
uence treatment maps, 52
and Novalis Body System,
3035
treatment verication, 4057
AZ-VUB approach, 4052
UCLA approach, 5357
Prostate tumors, and SBRT,
231250
AZ-VUB experience, 246250
clinical follow-up, 248250
irradiation technique, 246248
treatment protocol, 246248
hypofractionation, 232233
hypofractionation boost via X-ray
beams, 233244
Teknon experience, 231233
See also Hypofractionation boost,
for prostate tumors.
Paranasal sinus carcinoma,

radiotherapy for, 288289,
297299, 298
Partial organ irradiation, 149158
Patient xation, stereotactic body
frame and, 6, 8
Patient immobilization, lung tumors,
and SBRT, 210212
Patient selection, for ECSRT,
110112
Radiobiology, as 4-dimensional
problem, 131132, 160164
equivalent uniform biologically
equivalent dose (EUBED),
163164
equivalent uniform dose (EUD),
162163
of the lung, 202206
Radiosurgery
Leksells denition of, 7273
for liver tumors, 186189
Moving targets, treatment of,

5763
serial tomotherapy, 104105
326
Radiotherapy
head and neck. See Head and neck
tumors.
spine. See Spine.
tumor tracking, 118119
Rectal toxicity score (EORTC), 239
Redistribution, in fractionation
radiobiology,
Reoxygenation, in fractionated
radiobiology, 133
Repopulation, in fractionation
Respiration tracking, with
CyberKnife , 7884
Respiratory gating, 117118
Respiratory motion, restricting,
116117
Retinoblastoma, 300304, 303
RTOG/EORTC radiation
morbidity scoring
system, 241
SBRT, and lung tumors, 197227

Schwarzschilds law of
photochemistry, 135
Sereotactic radiotherapy (SRT),
development of, 24
Serial tomotherapy, 89107
AutoCrane system, 92
BodyFIX system, 93
Corvus software, 95
descrition of, 8991
future directions, 107108
IMRT targets, 96
logistics of, 106
patient immobilization/alignment
and, 9194
planning CT scans, 114119
treatment of moving targets,
104105
Index
Setup errors, estimated, in prostate
cancer treatment, 33
Single dose irradiation, lung
tumors, 217
Single-hit, multitarget killing model,
137138
Skin carcinoma, perineural spread,
299300
Slow CT scans, for tumor
115116
Spinal cord dose
dose-volume histogram
(DVH), 276
dosimetric considerations, 275277
Spinal metastasis
pain control, 277278
recurrent, 279
single, clinical outcome, 277,278
Spinal tumors, stereotactic
radiotherapy for, 257280
Spine, radiosurgery of, 258260
beam construction, 264
Body Frame , 262
BodyFIX , 262
clinical applications and outcome,
275280
clinical procedures diagram, 260
CT simulation 261
CyberKnife and, 259, 264, 268,
269271
dose distribution, 273
at Henry Ford Hospital,
279280
immobilization techniques,
261263
Novalis Body System and, 26,
267270
patient setup, 267269
phantom images, 273, 274
primary tumors, 278
quality assurance, 271275
simulation techniques, 261263
spinal cord dose, 275277
Index
[Spine]
techniques for, 260275
treatment failure, 278
treatment localization, 267269
inverse plan, 265, 266
SRS (stereotactic radiosurgery)
vs. SRT, general guidelines,
164165
treatment eld margins,
165166
tumor debulking and, 165166
SRT (stereotactic radiotherapy)
hypofractionated, 119120
liver tumors. See Liver tumors.
vs. SRS, general guidelines,
164165
treatment eld margins,
165166
tumor debulking and, 165166
unconventional fractionation,
166167
Stereoscopic X-ray imaging, in
Novalis Body System,
2528
Stereotactic body frame (SBF)
breathing control, 78
description of, 69
development of, 5
dose absorption of, 89
patient xation, 6, 8
reference system, 67
target accuracy, 912
serial tomotherapeutic approaches,
89107
Stereotactic radiosurgery (SRS),
history of, 14.
See also SRS.
Stereotactic radiotherapy.
See SRT.
Stereotactic reference system,
67, 9
327
Target accuracy, factors in, with
stereotactic body
frame, 912
Target denition, in treatment
planning, 1415
Target verication, in treatment
planning,1517
3D treatment planning, 15
Tissue organization, 158160
exible vs. hierarchical,
158159
parallel vs. series, 159160
Tomotherapy. See Serial
tomotherapy.
TPS BrainSCAN, 37, 38
parameters, in serial
tomotherapy, 96
Treatment eld margins, SRS vs. SRT,
165166
Treatment planning, 1317
CT for, 14
patient positioning, 1314
target verication, 1517
3D, 15
Tumor control probability (TCP)
fractionation radiobiology, 145
heterogeneous dose distribution,
152158
Tumor debulking, with SRS vs.
radical eld coverage, 165
Tumor mobility, intrafractional
characterization of, 115116
CT scans for, 115
uoroscopy for, 115
Tumor mobility, respiratory gating,
117118
Tumor mobility, restricting respiratory
motion, 116117
Tumor tracking radiotherapy,
118119
Two-phase CT scans, for tumor
115116
328
Index
UCLA approach to treatment

verication, 5357
dose distribution map, 55
uence maps, 56
IMRT phantoms, 5354
IMRT plans, 5354
schwannoma dose distribution
map, 56
Urinary toxicity score (EORTC), 241
late, (EORTC), 241
Vertebral metastases, role of imaging

in patient selection, 112
Volume effect, in radiobiology,
148160
Vacuum pillow, in patient positioning,

9, 10
Verication tests, 2830
X-rays and DRRs, fusion of

images, 27
residual set-up errors, 34
Whole-body radiosurgery, via

CyberKnife , 7186
Figure 3-2 Left: Patient with IR reflective marker. (Note that the camera system has identified the markers indicated by the circles, and the coincidence with the planned position
indicated by the small crosses.) Right: CT-image showing IR marker (localized by
software), contours of CT, PTV, and rectum, and position of the treatment isocenter.
Figure 3-3 Illustration of the graphical interface of ExacTrac 3.0/NOVALIS BODY with
the patient on the treatment couch prior to treatment setup. Note the detection of transversal and rotational patient position at the right side and bottom of the image. The circles
indicate the actual position and the crosses indicate the planned position where the patient
will be moved.
Figure 3-4 Flowchart illustrating the different steps in the positioning procedure using
ExacTrac 3.0/NOVALIS BODY. From top to bottom: (A) Patient on the treatment couch
with IR reflective markers. (B) Acquisition of X-rays (only one shown). (CD) Calculation of 3D correction vector based on either automated fusion of X-ray images with
DRRs representing the ideal position (left) or matching of implanted radio-opaque
markers (right). (E) Automated patient positioning.
2
Figure 3-5 Illustration of the distances taken to define the position of the treatment
isocenter with respect to bony structures for verification with portal film. The distance of
the isocenter to the midline and to the lines tangential to the superior and ventral border of
the os pubis are measured according to the dotted line.
Figure 3-7 Beams eye views of a dynamic arc (10 gantry steps). The yellow line
surrounding the target (green) indicates the conformal field shape created by the multileaf
collimator. The spinal cord (magenta) is shown running vertically through each frame.
Parameters such as arc length, dose, and the margin between the field edge and the tumor
are specified by the user. Dose distributions may be customized using software tools that
allow for preferential sparing of organs at risk (OARs) and for graphical editing of field
shapes in any BEV.
Figure 3-8 Beams eye view of a dynamic arc with an organ at risk (OAR). The field
shape (yellow) intentionally blocks part of the target (purple) in order to minimize the dose
to the OAR behind it.
Figure 3-11 Plan optimization window.
Figure 3-14 The importance of the choice of an appropriate phantom when mapping
treatment parameters from a patient treatment to a phantom for verification of dose distribution is illustrated here. The left-hand pane shows the result from mapping into an anthropomorphic phantom; the white erased pixels indicate regions where the gamma tolerance (4% DD, 4mm DTA) is not met between measured and calculated dose distributions.
The right-hand pane shows the same treatment mapped into a homogeneous cubic phantom; the bold pixels indicate regions where the gamma tolerance is not met. The homogeneous mapping does not show possible errors due to tissue heterogeneities and might
give a false sense of confidence.
Figure 3-15 Illustration of in-house developed tool for verification of dose distribution
(measured and calculated) at the AZ-VUB showing percent difference, absolute difference, and gamma map overlayed with both dose distributions and a cumulative dose
histogram.
5
Figure 3-17 Eight-field IMRS dose distribution for T1l metastasis on fused CT/MR. The
30%, 50%, 80%, 90%, and 105% isodose lines are displayed. The maximum dose is 105%.
Figure 3-18 Eight-field IMRT plan mapped to a MEDTEC benchmark phantom. The
30%, 50%, 80%, and 90% isodose lines are displayed.
Figure 3-19 Eight-field IMRT plan mapped to a CIRS thorax phantom. The 30%, 50%,
80%, 90%, and 105% isodose lines are displayed.
Figure 3-20 Comparison of a film measurement to calculation using gamma-index

analysis for an IMRT dose distribution. Solid black lines are the calculated isodose lines;
colorwash is the corresponding isodose lines from film. Dark green are the area where the
dose criterian was exceeded.
7
Figure 3-22 Dynamic arc dose distribution for L2 schwannoma on fused MR/CT. The
Figure 3-23 Conformal field dose distribution for L5 metastasis on fused MR/CT. The
Figure 3-24 Vertical isocenter displacement as a function of time for three markers
attached to the anterior surface of one patient. Though data are shown only for the first
60 sec, the pattern repeats over the entire 20 min.
Figure 3-25 Marker displacement as a function of time for a single marker attached to
the anterior surface of one patient. The stereophotogrammetry system is capable of reporting marker coordinates at a frequency of up to 10 Hz.
9
Figure 3-26 The gating control software monitors respiration through the ExacTrac,
calculates and displays the Baroni F-function, establishes gating windows or thresholds,
and triggers the NOVALIS via the MHOLDOFF/status bit.
Figure 3-27 2D data were measured using the amorphous silicon device. Gating frequencies of 0.2, 0.5, and 1.0 Hz were used in addition to non-gated conditions. Profiles are
shown for an open 10 10 cm2 square field (top left), a dynamic wedge (top right), and
arbitrary intensity map (bottom right).
10
Figure 3-30 Example of target localization for a lung lesion using co-registration information from PET and CT imaging, with cumulative dosevolume histogram resulting
from a coplanar dynamic conformal arc treatment.
11
Figure 3-31 Illustration of a four noncoplanar dynamic conformal arc technique for a
meningioma. The 30%, 50%, 90%, 98%, and 100% isodose lines are shown.
12
Figure 4-3 (A) Manual delineation of the target. Beam directions for optimized treatment
of specific tumor shape (up to 1200, lower left corner) are computed automatically by the
inverse planning system. (C) Both beam directions and beam weights are computed automatically, once target and critical regions have been delineated, and upper/lower dose
thresholds have been entered.
13
Figure 5-5 Isodose distribution for the case report example. Shown are the CTV (bright
red), PTV (darker red), esophagus protection region (green), spinal cord protection region
(blue), and isodose lines: 100% (dark blue) 90% (red), 70% (yellow), and 50% (green).
14
Figure 5-7 Cumulative dose volume histogram for the case report example. Shown (from
back to front) are the CTV, PTV, esophagus, and cord.
15
16
Figure 5-8 Depicting five of the

64 different intensity patterns used
for the case report example (gantry
angles 340, 260, 180, 100, and
20). Note that the patient was
treated with a straight couch
(i.e., 180 Varian), and four couch
increments (or treatment slices).
The shaded squares represent pencil beam intensities utilized, with
brighter shades indicating higher
intensities. The central region of
the figure graphically depicts pencil beams used, as dots on the surface of the patient. Note that the
final arc of the treatment uses only
the cranially located row of
MIMiC vanes because inferior row
of pencil beams does not see the
target volume.
Figure 6-2 Generating target volumes for a stage I lung tumor. The ITV (orange contour
on left panel) encompasses all GTVs contoured on six consecutive multi-slice CT scans
(light yellow contours on left and right panel). The ITV was expanded with a 3 mm
margin to derive the PTV (red contour on right panel).
Figure 6-4 Changes in ITVs seen on weekly CT scans in two patients with peripheral
lung tumors when five fractions of stereotactic radiotherapy were delivered in 5 weeks
(12 Gy/fraction).
17
Figure 9-1 CTV-definition and conformal dose distribution in a 43-year-old male with
primary lung cancer cT2 cN0 cM0 (adenocarcinoma grade II) in the left upper lobe
medically inoperable due to severe heart disease. The CTV was 45 cm3, the PTV was 100
cm3. The tumor was treated by 3 10 Gy to the PTV-enclosing 100%-isodose (the inner
orange isodose) with normalization to 150% at the isocenter. For CTV definition not only
the macroscopic tumor but also the small tumor extensions into the periphery have to be
included into the target volume (the numbers in the coronal and sagittal reconstruction
show the point dose in percent to the prescribed fraction dose of l0 Gy).
Figure 10-1 Dose distribution of an intensity modulation radiotherapy plan in the axial
central plane of the planning target volume (PTV). (A) 6464.4 Gy delivered to
the prostate and seminal vesicles (PTV1); (B) 1016 Gy boost delivered to a reduced
horseshoe-shaped volume (the prostate peripheral zone) (PTV2). Dose distribution is
given in percent values and is displayed in color bands. The yellow crosses in the figures
represent the treatment isocenters for PTV1 and PTV2, respectively.
18
Figure 10-4 Dose distribution overlying the axial central plane of an endorectal MR
image of the prostate containing the PTV2 (bilateral prostatic peripheral zone). Isodose
bands of 100% or above (red), 90100% (yellow), and 8090% (green) are displayed.
Figure 10-5 Digital volumetric reconstruction to simulate the setup reproducibility with
ExacTrac. Infrared marker based registration between the CT at simulation (red) and the
CT while on treatment (green) performed to assess target repositioning quality.
19
Figure 11-3
(A) An inverse plan using equally distributed seven beams and (B)
corresponding DVHs.
20
Figure 11-4 (A) An inverse plan using five beams and (B) corresponding DVHs.
21
Figure 11-6 Phantom study images. (A) The original CT image with target and spinal
cord indicated. (B) The planned dose distributions for 90%, 50%, and 30% isodose curves
normalized to the isocenter.
22
Figure 11-7 Phantom study results. (A) The planned isodose distributions in the region
where the film was inserted. Solid curves represent planned isodose lines labeled 90%,
50%, and 30% relative to the isocenter. (B) The original film dose image with three corresponding isodose curves. (C) Both planned and corresponding measured isodose distributions are overlaid on the original CT image.
23
Figure 11-8 Dosevolume histograms (DVHs) for cord doses.
24

Extracranial Stereotactic Radiotherapy and Radiosurgery

Uploaded by

Copyright:

Available Formats

Extracranial Stereotactic Radiotherapy and Radiosurgery

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Extracranial Stereotactic Radiotherapy and Radiosurgery

Uploaded by

Copyright:

Available Formats

Extracranial

DK2966_Discl.fm Page 1 Monday, August 29, 2005 11:02 AM

Library of Congress Cataloging-in-Publication Data

Visit the Taylor & Francis Web site at

Treatment of Moving Targets . . . . 57

6. Anatomical and Biological Imaging . . . . . . . . . . . . . . . . 109

7. Radiobiological Considerations of Stereotactic Body

8. Stereotactic Radiation Therapy of Liver Tumors . . . . . . . 177

10. Prostate Tumors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231

12. Stereotactic Radiotherapy of Head and Neck Tumors . . .

John R. Adler Stanford University Medical Center, Stanford,

Meritxell Molla` Servei de Radio-oncologia, Instituto Oncologico Teknon,

Luebeck University, Luebeck, Germany

Suresh Senan Department of Radiation Oncology, VU University

Osaka University Hospital, Osaka, Japan

Ben J. Slotman Department of Radiation Oncology, VU University

Department of Radiotherapy, AZV-UB, Brussels, Belgium

Timothy Solberg Department of Radiation Oncology, University of

Department of Radiotherapy, AZV-UB, Brussels, Belgium

H. Rodney Withers Department of Radiation Oncology, UCLA

Reinhard Wurm Abteilung Strahlentherapie, Universitat Klinikum,

The concept of stereotactic radiosurgery was rst described by Lars Leksell

sources focused at one locus. The clinical usefulness of the GammaKnife

ELEKTA Stereotactic Body Frame

longitudinal scale in millimeters, which is bilaterally xed to the outside of

Dose Absorption of the SBF

ELEKTA Stereotactic Body Frame

included into the calculation model, the difference increased by additional

ACCURACY OF PATIENT AND TARGET

Figure 4 Accuracy of patient repositioning in the SBF is dependent on the quality

Accuracy of treatment in the SBF can be evaluated at three different

ELEKTA Stereotactic Body Frame

ELEKTA Stereotactic Body Frame

TREATMENT PLANNING AND DELIVERY

The CT study for treatment planning starts with obtaining a reference

ELEKTA Stereotactic Body Frame

ELEKTA Stereotactic Body Frame

can be performed parallel to evaluation of the CT verication to shorten the

Medin and Verellen

implementation of CRT is only feasible when two important requirements are

Medin and Verellen

Figure 1 Room with Novalis system.

Medin and Verellen

Figure 3 Illustration of the graphical interface of ExacTrac 3.0/Novalis Body with

Stereoscopic X-Ray Imaging Device

Figure 4 (Caption on facing page)

Medin and Verellen

Medin and Verellen

transferred to the positioning system on the treatment machine. Due to the

Medin and Verellen

with respect to the bony structures used for positioning purposes. An

positioning and alignment of the treatment beamsetup margin or SM) (2) a

Medin and Verellen

Figure 6 An estimate of the distribution of setup errors for prostate treatments

Medin and Verellen

Large errors (10 mm)

monitoring of patient positioning is not limited to patient observation, but

Medin and Verellen

TREATMENT PLANNING AND TREATMENT DELIVERY