Beam shaping for conformal fractionated stereotactic radiotherapy: A modeling study

Int. J. Radiation

Oncology

Biol.

Phys., Vol. 38. No. 5. pp. 1113-1121. 1997 Copyright 0 1997 Elsevier Science Inc. Printed in the USA. All rights reserved 0360.3016/97 $17.00 + .OO

PI1 SO360-3016(97)00151-X

0 Physics Contribution BEAM

SHAPING

FRED L. HACKER,

FOR CONFORMAL FRACTIONATED STEREOTACTIC RADIOTHERAPY: A MODELING STUDY PH.D.,

+ HANNE M. KOOY,

PH.D.,

+ MARC

R. BELLERIVE,

M.Sc

JOSEPHH. KILLORAN, PH.D., t ZACHARY H. LEBER, M.S., * DENNIS C. SHRIEVE, M.D., NANCY J. TARBELL, M.D.+ ANDJAY S.LOEFFLER, M.D.’ ‘Joint Center for Radiation

Therapy, Department of Radiation Oncology, Harvard Medical *Radionics and RSA Inc., Burlington, MA

., ’ PH.D.,’

School, Boston, MA; and

Purpose: The patient population treated with fractionated stereotactic radiotherapy (SRT) is significantly different than that treated with stereotactic radiosurgery (SRS). Generally, lesions treated with SRT are larger, less spherical, and located within critical regions of the central nervous system; hence, they offer new challenges to the treatment planner. Here a simple, cost effective, beam shaping system has been evaluated relative to both circular collimators and an ideal dynamically conforming system for effectiveness in providing conformal therapy for these lesions. Methods and Materials: We have modeled a simple system for conformal arc therapy using four independent jaws. The jaw positions and collimator angle are changed between arcs but held fixed for the duration of each arc. Eleven previously treated SRT cases have been replanned using this system. The rectangular jaw plans were then compared to the original treatment plans which used circular collimators. The plans were evaluated with respect to tissue sparing at lOO%, 80%, 50%, and 20% of the prescription dose. A plan was also done for each tumor in which the beam aperture was continuously conformed to the beams eye view projection of the tumor. This was used as an ideal standard for conformal therapy in the absence of fluence modulation. Results: For tumors with a maximum extent of over 3.5 cm the rectangular jaw plans reduced the mean volume of healthy tissue involved at the prescription dose by 57% relative to the circular collimator plans. The ideal conformal plans offered no significant further improvement at the prescription dose. The relative advantage of the rectangular jaw plans decreased at lower isodoses so that at 20% of the prescription dose tissue involvement for the rectangular jaw plans was equivalent to that for the circular collimator plans. At these isodoses the ideal conformal plans gave substantially better tissue sparing. Conclusion: A simple and economical field shaping device has been shown to provide all of the beam shaping advantage of a hypothetical ideal dynamically conforming system at the prescription level. This system may be immediately implemented in the clinic. It offers a substantial advantage over the currently used circular collimators in the high dose region with equivalent performance in the low dose region. 0 1997 Elsevier Science Inc. Stereotactic

radiosurgery,

Stereotactic

radiotherapy,

Arc therapy,

INTRODUCTION

Conformal

therapy.

With the advent of fractionated stereotactic radiotherapy (SRT) the patient population treated using stereotactic techniques has significantly changed. In stereotactic radiosurgery ( SRS ) a single large dose of radiation is delivered. In SRS toxicity has been shown to increase with tumor volume ( 13, 15 ) To avoid complications the dose delivered is decreased as tumor size increases ( 11, 12). This limits the size of tumors which can be treated with SRS to approximately 10 cc or 3 cm ( 3, 15) . Radiosurgery

is not appropriate for initial treatment of tumors located adjacent to or within sensitive structures such as the optic chiasm or brain stem (3, 12) . Fractionated stereotactic radiotherapy combines stereotactic localization techniques with fractionated dosedelivery (3, 20). The normal tissue sparing provided by fractionated dose delivery allows tumors which could not be treated with SRS, either due to location near a critical structure or due to size, to be treated with SRT (3, 20). In general, tumors treated with SRT are both larger and less spherical than those treated with SRS. In these cases

Reprint requests to: Fred L. Hacker, Ph.D., Joint Center for Radiation Therapy, Department of Radiation Oncology, Harvard Medical School, 330 Brookline Avenue, Boston, MA 02215. email: hacker @gog.jcrt.harvard.edu. Acknowledgements-This research was supported in part by Ra-

dionics Inc., Burlington, MA. Thanks to Goran Svensson, Robert Cormack and Edward Holupka for their efforts in reviewing this paper. Thanks to Edward Mannarino and Jonathan Unger for their assistance in this project. Accepted for publication 20 March 1997. 1113

1114

I. J. Radiation Oncology 0 Biology 0 Physics

Volume 38, Number 5, 1997

Table 1. Summary of patient population Tumor type Meningioma Pituitary adenoma Low grade astrocytoma Optic glioma Totals

used in this study

Number of cases

Maximum dimension cm, mean (range)

Volume cc, mean (raw)

4

3.9 (3.6-4.2)

8.7 (3.3-17.8)

2.2 (1.8-3.0)

5220

2

4.3 (4.1-4.5)

6.0 (5.5-6.6)

3.7 (3.2-4.1)

4500

1 4 11

4:: (4.6-7.3) 4.6 (3.6-7.3)

9.4 16.3 (6.5-22.1) 11.1 (3.3-22.1)

1.9 1.8 (1.3-2.4) 2.3 (1.3-4.1)

the fixed circular apertures currently in use are clearly not ideal. The presence of critical structures within the treated volume for many SRT casesalso necessitatesa more homogeneousdose distribution than is used for SRS. While in general our SRS casesare prescribed to the 80% isodose surface (normalized to isocenter dose) our SRT casesare prescribed to the 95% isodose surface. This requirement of dose homogeneity precludes the use of multiple isocenters to provide conformal field shaping such as is done in SRS (4). To provide conformal dosedistributions using arc therapy some form of beam shaping is required. Several groups have proposed devices for SRS/SRT beam shaping (8, 10, 16, 18, 19). The ideal system for conformal arc therapy, in the absenceof fluence modulation, would dynamically conform the beam aperture to the beam’s eye view (BEV) projection of the tumor throughout each treatment arc. This ideal system is not currently feasible, although in time it may be approximated using devices such as the miniature multileaf collimator ( 19). With any beam shaping device, however, a compromise must be reached between complexity, reliability and beam shaping ability. In order to choose an appropriate device it is necessary to evaluate the relative gain which is achieved for each increase in complexity. For SRS casesthis was done by Nedzi et al. ( 16). Five beam shaping systems were compared for beam shaping ability. All of the systems used a circular collimator for final collimation. The systems compared were a circular collimator alone, two independent jaws, four independent jaws, a Leavitt collimator (10) and an ideal conformal collimator as described above. The same arc selections and isocenter location were used for each collimation system. All of the systems, with the exception of the circular collimator conformed dynamically to the BEV projection of the tumor throughout each arc. Beam shaping ability was quantified in terms of treatment volume ratio which Nedzi defined as target volume divided by the treatment volume. In this case it was found that simple beam shaping devices such as independent jaws provided roughly half the beam shaping advantage at the prescription level of the ideal system. In this work we have evaluated a simple beam shaping system consisting of four independent jaws with respect to healthy tissue sparing for 11 tumors previously treated

Aspect ratio, mean @we)

Cumulative prescription dose cGy

5940 5220-5400 NA

with SRT at the Joint Center for Radiation Therapy. To maximize simplicity and hence facilitate clinical implementation of this system the jaw positions and collimator angle were held fixed for the duration of each arc. The treatment plans produced with this system were compared to both the original treatment plans which utilized a circular collimator and to treatment plans produced for the ideal dynamically conforming system described above.

METHODS

AND MATERIALS

Patient population A sampling of eleven tumors previously treated with SRT at the Joint Center for Radiation Therapy were selected. A cross-section of tumor types was chosen. These are summarized in Table 1. Each tumor was characterized in terms of volume, maximum tumor extent and aspect ratio. Aspect ratio is an indicator of the tumor shape, and is determined by finding the dimensions (length, width and height) of a box which would enclosethe tumor when oriented so as to minimize its size. Aspect ratio is defined as the ratio of the maximum dimension to the minimum dimension of this box. A spherical tumor would therefore have an aspect ratio of 1.0. The orientation of the lesions was dependent on tumor type. Due to their growth pattern the optic gliomas were primarily elongated in the A-P direction. The pituitary adenomaswere extended in both the A-P and lateral orientation but were narrow in the inferiorsuperior direction. The meningiomas did not have a predominant orientation. Table 1 also gives the cumulative prescription dosesused. All caseswere treated with conventional fractionation using a fraction size of 180 cGy/ day. The number of fractions ranged between 25-33. All plans were normalized to the 95% isodose. Treatment plan development Three treatment plans utilizing different beam shaping devices were compared for each tumor. These devices were a circular collimator, a simple rectangular jaw system and the ideal conformal system described above. The plans for each beam shaping device were prescribed to the 95% isodose surface where the normalization is done relative to the isocenter dose. The average ratios of the max-

Beam shaping for conformal

imum dose to the prescription dose was 1.10, 1.07, and 1.08 for the circular collimator, fixed rectangular and ideal conformal plans, respectively. The plans utilizing a circular collimator were the original plans used to treat the patient. Collimator diameters ranged from 4.0 cm to 6.0 cm. The original circular collimator plan was generated using the XKnife ’ treatment planning system. In all cases a single isocenter was used. Arc selection was used to better conform the dose distribution to the planning target volume. In addition, one of the circular collimator plans utilized beam weighting. The two other plans were generated using a developmental 3 D treatment planning system. This system is an extension of the XKnife system. As with XKnife, a fast TMR dose algorithm is used. Unlike XKnife, an equivalent square formalism is used to determine output factor and TMR. The penumbra is modeled by fitted penumbra functions. This algorithm has been shown to produce outputs which agree with measured values to within 0.7%. The system did not have a built-in arc mode and each treatment arc was modeled by placing static beams at 15” intervals along the arc. This interval was found to be adequate to simulate a continuous arc. The rectangular jaw plan utilized four independent jaws. Dosimetric data for a linac with standard jaws was used. The jaw positions and collimator angles were held fixed within a single arc. For the ideal conformal plan each beam aperture which made up an arc was individually conformed to the beam’s eye view (BEV) projection of the tumor. Figure 1 shows an example of a BEV tumor projection along with the generated aperture for the ideal conformal plan. Dosimetrically each aperture of the ideal conformal plan was treated as if it were formed by a block mounted in the linac block tray. The rectangular jaw plan was made independently of the original circular collimator plan with new arcs selected to optimize the available beam shaping advantage. To maintain the clinical relevance of this comparison the sameiterate planning processwas used for the rectangular jaw plans as for the circular collimator plans. The initial arcs and beam apertures were selected, keeping the jaw positions and collimator angle fixed throughout each arc. A volume dose envelope was generated and evaluated for conformity to the tumor. Modifications were made to the arcs and beam apertures by the planner and then the processwas repeated. The rectangular jaw plan arcs were also used for the ideal conformal plan unless an obvious advantage could be gained by further optimization. The aperture for each beam of the ideal conformal plan was automatically generated to conform to the BEV projection of the tumor on a digitally reconstructed radiograph (DRR). The aperture would also include a margin around the tumor input by the planner. The automatic contouring of the aperture was based on the DRR and hence the pre-

’ Radionics

Software Applications,

Burlington,

MA.

SRT

0 F. L. HACKER

et al.

Fig. 1. The beam’seye view projectionof a 17 cc optic glioma on a digitally reconstructed radiograph, along with the associated beam aperture at a single gantry angle generated for the ideal conformal plan.

cision with which the margin could be set was dependent upon the resolution of the DRR and was typically 1.5 mm. The planning process for this mode consisted of determining the initial arcs and margin, generating a volume dose distribution and evaluating it for tumor coverage. Based on that evaluation the margin would be adjusted, and if necessary, the arc selections modified. New apertures would then be generated and the process repeated. Plan evaluation criteria The goal in generating the rectangular jaw and ideal conformal plans was to optimize healthy tissue sparing while maintaining equivalent or superior tumor coverage to that in the circular collimator plans. Sparing of critical structures was also examined. Unlike SRS, however, the critical structures are frequently within the high dose region in SRT. This is possible because of the therapeutic advantage provided through fractionation. Critical structure sparing, while still important, is not of the sameoverriding concern that it is in SRS. In producing the rectangular jaw and ideal conformal plans the attempt was made to minimize critical structure involvement aswell as overall healthy tissue involvement. The minimum standardfor critical structure sparing was to equal the circular collimator plans. Tumor coverage was evaluated using two independent techniques. The first was minimum dose to the planning target volume (PTV) (6) surface as contoured on the CT slices. The second was percent of the PTV receiving prescription dose or higher as determined by dose volume histograms (DVH ) . The conformity of each plan and relative healthy tissue sparing was also determined from the DVHs. A modified version of the treatment volume ratio

I. J. Radiation Oncology 0 Biology 0 Physics

1116

Volume 38, Number 5, 1997

Table 2. Summarv of tumor coverage

Treatment planning

systems

Mean percentage of PTV receiving prescription dose (%) 99.6 99.7 99.9

Circular collimator Rectangular jaw Ideal conformal

Standard deviation in the mean (%)

Mean value for minimum dose to the PTV surface (% of prescription)

Standard deviation in the mean (%)

0.6 0.2 0.2

96.3 100.3 100.6

5.9 0.6 2.3

introduced by Nedzi ( 16) was used to determine beam conformation. The target volume ratio (TVR) is defined as the treated volume divided by the planning target volume (PTV) where both the PTV and treated volume are as defined by the ICRU (6). The TVR was determined for each case. Since essentially full coverage was achieved in each modeled case the treated volume completely encloses the PTV. The TVR may thus be used to compare the conformation of the prescription dose envelope to the planning target volume. The theoretical minimum value for TVR is 1.0. The relative healthy tissue sparing was determined at lOO%, 80%, 50%, and 20% of the prescription dose. This was defined as a percent decrease(or increase) in volume of healthy tissue involved at each dose level relative to the original circular collimator plan.

RESULTS The tumor coverage results are summarized in Table 2. The three treatment modesgive essentially identical tumor coverage in terms of the percent of the PTV receiving prescription dose. In terms of minimum dose to the PTV surface the rectangular jaw and ideal conformal plans are somewhat superior to the circular collimator plans. The minimum surface dose is also more consistent from one case to the next as indicated by the standard deviations given in Table 2. The slightly larger standard deviation for the ideal conformal plan relative to the rectangular jaw plan is due to the finite adjustment increment (see above) which could be made in the margin used to generate the conforming apertures. Table 3 gives a summary of TVR data and relative healthy tissue sparing at the prescription level for the 11 cases examined. No significant difference was seen between the rectangular jaw and ideal conformal plans with regard

to conformity

of the prescription

increment in the margin for the ideal conformal plan and the fact that the jaw settings for the rectangular jaw plan were manually adjusted to optimize the dose envelope while the ideal conformal plan apertures were geometrically determined. In no caseswas the difference between the rectangular jaw and ideal conformal plan TVRs significant. When healthy tissue involvement was examined as a function of dose level the advantage of ideal conformal beam shaping became apparent. Table 4 gives the mean reduction in healthy tissueinvolvement relative to the circular collimator plans at lOO%, 80%, 50%, and 20% of the prescription dose. The ideal conformal beam shaping provides a substantial reduction in healthy tissue involvement down to 20% of the prescription dose. The rectangular jaw plans tended to converge with the circular collimator plans in terms of healthy tissue involvement at lower dose levels. This is shown graphically in Fig. 3 where the mean involved volume for each case has been plotted versus dose level and normalized to the mean circular collimator volume at that dose level.

DISCUSSION It is instructive to examine a single case in detail and see how these three treatment planning systems compare in terms of dose distribution. Figure 4 showsreconstructed CT images of the coronal, sagittal and transverse slice through isocenter for a 17 cc optic glioma which is 5.4 cm in maximum extent. This was treated using a 6 cm

Table 3. Summary of TVR and tissue sparing data for the cases examined

dose to the tumor

as indicated by TVR. In all of these casesboth the rectangular jaw and ideal conformal plans provided substantial tissue sparing relative to the original circular collimator plans. Figure 2 shows the spread of TVR values for each plan. As can be seenin the figure the TVRs were not only lower for the rectangular jaw and ideal conformal plans but also more consistent from one caseto another. In some casesthe TVR for a rectangular jaw plan was found to be slightly lower than that of the corresponding ideal conformal plan. This was attributed to the finite adjustment

Treatment

planning

system

Circular collimator Rectangular jaw Ideal conformal

Mean TVR

Mean reduction in healthy tissue (%I

5.1 t 2.2 2.5 2 0.7 2.5 t 0.7

NA 57 -c 17 59 ? 17

TVR is defined as the ratio of the treatment volume to the PTV as defined in ICRU 50. The mean reduction is determined by calculating the percent reduction in healthy tissue involvement relative to the circular collimator plan for each case examined and then taking the mean. (2 refers to one standard deviation from the mean).

Beam

shaping

for conformal

SRT 0 F. L. HACKER

ef al.

1117

10.0

9.0 8.0

q

El

7.0 p: 6.0 z

5.0 -

--k

4.0 -

0 0 D

3.0 -

d8

2.0 -

of TVRs for each case modeled. The horizontal

diameter circular collimator. Superimposed on the slices are the prescription isodoselineand isodosesat 90%, 80%, and 50% of prescription dose for each treatment plan. As would be expected, the dose distribution at the prescription level for the circular collimator plan is nearly spherical. In contrast, the dose distributions for both the rectangular jaw and ideal conformal plans conform much more closely to the tumor shape.The ideal conformal plan dose distribution can be seen to mirror even small scale variations in the tumor shape.The rectangular jaw plan matches the overall shapewell but does not match the small scale details of the tumor shape. Despite this there is no signif-

Table 4. Reduction in healthy tissue involvement as a function of dose level Mean reduction in healthy tissue involvement relative to the circular collimator plans (%)

100 80 50 20

Rectangular plans 57 k 17 32 -+ 20 9 + 25 -1530

jaw

Ideal conformal plans 59% 57 It 51 + 43 !I

;

Rectangular JaW3

Circular CoUima83r

Dose (% of prescription)

+

!I

1.o

Fig. 2. Distribution

+ + $

8

17 17 16 18

The mean reduction is determined by calculating the percent reduction in healthy tissue involvement relative to the circular collimator plan for each case examined and then taking the mean. Since the percent reductions are averaged equal weighting is given to each case regardless of the actual volume of involved healthy tissue. (2 refers to one standard deviation from the mean)

Ideal confomlel lines indicate the mean value for each case.

icant difference in the volume of healthy tissue involved at the prescription level between the rectangular jaw and ideal conformal plans. This is typical of the observed differences between thesethree techniques at the prescription level. It should also be noted that spatial separation of the prescription, 90%, and 80% isodoselinesis greater for the rectangular jaw plan than either the circular collimator or ideal conformal plans. This is characteristic of all the rectangular jaw plans and results in the reduced advantage at lower dose levels reported above. This may also be seen in Fig. 5 which shows the dose volume histogram for all involved tissue for each plan. The percent relative advantage of the rectangular jaw plan compared to the circular collimator plan drops off between the prescription level and approximately 70% of the prescription. For this particular case the relative tissue sparing of the rectangular jaw plan remains approximately constant at lower dose levels. This was found to be characteristic of the largest tumors modeled; however, for the group as a whole the tissue involvement for the rectangular jaw and circular collimator plans tended to converge at lower dose levels.

Geometrical

penumbra

The lower dose gradient for the rectangular jaw plans in comparison to the circular collimator and ideal conforma1plans is explained by examining the geometry of the three systemsin a treatment plan. A treatment plan using any of these collimation systemsconsists of a number of non-coplanar arcs. Intersecting arcs create a high dose region, the shapeof which dependson the beam collimation. A circular collimator with a circular BEV dose distribu-

1118

I. J. Radiation

Oncology

0 Biology

20

0 Physics

50

Volume

38, Number

80

5, 1997

100

Dose (9%of Prescription) Fig. 3. Mean volume of healthy tissue involved as a function of dose level. Volumes are normalized to the mean circular collimator volumes at each dose level. Since the volumes are averaged, cases with a higher volume of involved healthy tissue are weighted more heavily than those with a smaller volume of involved healthy tissue.

tion always creates a spherical or ellipsoidal high dose region around isocenter. The rectangular collimator has a rectangular BEV dose distribution. When arcs of this kind intersect, the high dose regions of the arcs will in general not coincide completely;

Circular Collimator Plan

specifically, the corners of the rectangular apertures for individual arcs will tend not to overlap. The dose distribution from intersecting arcs using rectangular collimators will therefore consist of a central region where the high dose from each arc has been added, and a region where

Rectangular Jaw Plan

Ideal Conformal Plan

Sagital

Transverse

Fig. 4. Isodose lines showing lOO%, 90%, 80%, and 50% of prescription dose on the coronal, sagittal and transverse slices through isocenter for a 17 cc optic glioma. The FTV is shown in dark gray.

1119

Beam shaping for conformal SRT 0 F. L. HACKER et al. 17 cc Optic Glioma Involved Tissue DW

t

\i

I

\

I

0

20

In the ideal conformal case, the beam aperture is constantly adjusted to conform to the BEV projection of the target throughout each arc. The primary portion of the beam will generally be restricted to the region defined by the BEV projection of the target volume. The volume of high dose will therefore follow the surface of the target volume. In addition, the volume of partially overlapping field comers are reduced, resulting in a minimal penumbra enlargement, approaching the caseof a circular collimator.

40

60

.

I

.

80

100

Normalized Dose (prescription dose = 100) Fig. 5. Involved tissue dose volume histogram for a 17 cc optic glioma. The rectangular jaw and ideal conformal plans are equivalent at the prescription level.

the high dose from one arc is added to a low dose from another arc, thus creating an enlarged “geometric penumbra” region typically between 50- 100% of the prescription dose. The shape of the high dose region and the enlarged geometric penumbra will depend on, and can be controlled by, the jaw positions, collimator/table angle and the gantry position throughout the arcs. For conformal therapy using rectangular jaws, these parameters are chosen to optimize the conformation of the prescription dose envelope to the tumor shape.This allows a very conformal plan to be produced; however, since the corners of the rectangular fields do not generally overlap, an enlarged penumbra is produced in the dose region from 50% to 100% of prescription. This could present a serious disadvantage for the rectangular jaw system if used in stereotactic radiosurgery where a very sharp dose gradient is required to spare adjacent critical structures. In stereotactic radiotherapy, however, the beam shaping advantage of the rectangular jaws, when compared to the circular collimator, provides a substantial advantage despite the decrease in dose gradient. It should also be noted that for stereotactic radiosurgery a hybrid system using jaws and a circular collimator may provide the advantages of the rectangular jaws in healthy tissue sparing along with the required sharp penumbra near critical structures.

Table 5. Optic chiasm involvement

Critical structure sparing It is difficult to quantify the ability of a beam shaping system to spare critical structures because of the dependence on case specific parameters such as tumor size, shape, and proximity to the relevant critical structure. For this study the critical structure sparing for the rectangular jaw and ideal conformal plans was required to at least equal the original circular collimator treatment plan which had been approved by the attending physician. In many casesthe new plans significantly reduced the critical structure involvement. An example of this is given in Table 5, which shows the maximum and average dose to the optic chiasm for the three meningiomas and one pituitary adenoma in which it was outlined as a critical structure. For reference, the tolerance level for the optic chiasm is approximately 5000-6000 cGy (17) with the fractionation schedule used. While Table 5 is not a quantitative comparison of the relative utility of these systems for critical structure sparing it does at least provide empirical examples of the gains which might be expected. It should be noted that critical structure proximity was not used in the generation of the beam apertures for the ideal conformal plans. It is likely that further critical structure sparing could be achieved by adjusting the beam apertures around the critical structures. This may, however, result in either compromising tumor coverage or increasing overall healthy tissue involvement. Treatment planning considerations The basic principles and methods for treatment planning with the rectangular jaws are the sameas for circular collimators. The beam’s eye view is used extensively to choose optimum approaches for the arcs, as well as jaw settings and collimator angles. The jaw settings and collimator angle are initially selected to minimize aperture

for four of the cases modeled

Maximum doseto chiasm(cGy)

Average dose to chiasm (cGy)

Case

Circular collimator

Rectangular jaws

Ideal conformal

Circular collimator

Rectangular jaws

Ideal conformal

Meningioma 1 Meningioma 2 Meningioma 3 Pituitary adenoma 1

5658 5558 5587 4890

4583 5534 5443 4023

5037 5543 5559 3164

3685 4357 4363 4865

3215 3870 3919 3821

2323 3219 3835 3033

1120

I. J. Radiation

Oncology

0 Biology

0 Physics

size while still enclosing the tumor volume. This is a purely geometrical problem. Beyond this, the dose envelope around the tumor is calculated and evaluated. The jaws, collimator angle, and arcs are then reset to optimize the dose envelope. Optimal planning in this way requires fast dose calculation and accurate 3D rendering of structures and dose. For this study individual beams had to be placed to model an arc. For this reason the time required to generate a rectangular jaw plan was longer than that for a corresponding circular collimator plan. In a system with an integrated arc mode the planning time for a rectangular jaw plan should be comparable or, in many cases, less than that for a circular collimator plan. Time required to choose and optimize settings for the rectangular jaws is offset by a reduced sensitivity to arc selection. Further requirements for a treatment planning system for conformal therapy have been discussed by Killoran et al. (7). It was expected that aspect ratio would be an indicator of the advantage which would be gained by using conforma1 beam shaping; however, this was not found to be the case. The ideal and rectangular jaw plan TVRs were quite constant for all aspect ratios. The circular collimator TVRs, while substantially different from one case to the next, did not vary in any regular way with aspect ratio. Performance was comparable for all the tumor types examined. All benefited greatly from beam shaping with no significant performance difference between the rectangular jaw and ideal conformal plans. A preliminary examination of three cases with maximum tumor dimensions ranging from 2.8 cm to 3.2 cm has indicated poorer performance for the rectangular jaws when compared to the ideal conformal plans. For these smaller cases it was also found that the rectangular jaw plans had increased healthy tissue involvement for doses at and below 80% of the prescription level when compared to the circular collimator plans. This indicates that for tumors smaller than 3.5 cm in maximum extent the circular collimator may be superior to rectangular jaws. For some individual tumors other treatment methods such as static conformal beams (7, 9) or fluence modulation (2) may give superior tissue sparing. For this reason it is important for the treatment planner to have a variety of treatment modalities to select from when developing the optimum plan. For all of the tumors modeled in this study and, we believe, for the majority of clinically significant tumor shapes in the range from 3.5 cm to approximately 7 cm the use of rectangular jaws for beam shaping is a highly effective modality. Comparison to other conformal treatment options In this study we have restricted ourselves to systems which retain the arc delivery method characteristic of conventional linac based stereotactic radiosurgerylradiotherapy. There is substantial interest in the use of either multiple static conformal fields or intensity modulation to treat SRT and SRS patients. While these systems have not been

Volume

38, Number

5, 1997

included within this study it is necessary to comment on how they compare to our system. The use of multiple conformal static fields to treat SRS and SRT cases has been studied at several centers ( 1, 5, 9, 14, 18). If a sufficient number of static conformal fields ( 15-20 fields) are used the dose distributions from our ideal conformal arc model may be reproduced; however, the treatment time involved in the use of this number of fields would be prohibitive. It takes approximately the same amount of time to treat a single static field as it does to treat a single arc. To be competitive, a static field plan should use approximately the same number of treatment fields as an arc plan, i.e., 5-6 fields. Studies have been done (5, 9) comparing conformal static field plans to single isocenter circular collimator plans. These studies indicate a reduction in normal tissue involvement, using 4-12 static fields, comparable to the reduction we see using rectangular jaw arc treatments. We have also done static field plans for some of the cases used in our study. These plans utilized five conformal static fields and provided similar healthy tissue sparing to the rectangular jaw arc plans. In all cases, however, the most healthy tissue sparing was achieved with the ideal conformal arc plans. While the fixed rectangular arc treatments and conformal static field treatments provided similar performance there are important differences between them. Even with identical DVHs an arc treatment and a static field treatment will have different characteristic dose distributions. Static fields tend to place regions of higher dose in concentrated areas of beam entrance. An arc treatment tends to smear these regions over a volume surrounding the target volume. Which distribution is clinically superior has not been proven, yet it is important to recognize the difference. We also feel that a fixed rectangular jaw arc system is simpler to implement, as it does not require any additional field shaping apparatus such as custom blocks or an MLC. This reduces both the cost of implementation and the QA requirements. We have not compared these shaping systems to a fluence modulated system, such as discussed by Carol et al. (2). A study by Woo et al. (2 1) did compare the performance of a fluence modulated system to conventional radiosurgery using circular collimators. The relative improvement observed in dose conformation at the prescription level for that system compared to circular collimation was no larger than what we achieved through simple rectangular jaw collimation. Any fluence modulation system will also, by its nature, be substantially more complex and costly than the simple system we have presented. CONCLUSION A simple beam shaping method utilizing four independent jaws has been evaluated relative both to current circular collimator technology and to a hypothetical ide-

Beam shaping for conformal SRT 0 F. L.

ally conforming beam shaping system for use in fractionated stereotactic radiotherapy. For tumors greater than 3.5 cm in maximum dimension this system was found to provide equivalent tissue sparing to the ideal case at the prescription level. Relative to plans done with a circular collimator the rectangular jaw system reduced healthy tissue involvement at the prescription level by a mean value of 57%. While the dose fall-off is less rapid with this system, for tumors larger than 3.5 cm, a substantial advantage is maintained at 80% of the prescription dose when compared to the circular collimator plans. At lower isodoses this system becomes

HACKER

et al.

1121

equivalent to the circular collimator on average. An ideal conformation system is clearly superior at doses below the prescription level; however, it will be a substantial period of time before such a system can be developed and approved for clinical use. In contrast, all of the required hardware to implement the rectangular jaw system is currently in clinical use. The use of rectangular jaws thus offers a system which is simple and cost effective and can be implemented for the immediate benefit of patients being treated with fractionated stereotactic radiotherapy. Such a system will soon be in operation at the Joint Center for Radiation Therapy.

REFERENCES 1. Bourland, J. D.; McCollough, K. P. Static field conformal stereotactic radiosurgery: Physical techniques. Int. J. Radiat. Oncol. Biol. Phys. 28:471-479; 1993. 2. Carol, M.; Grossman, R. ; Selker, R.; Polk, D.; Wu, J.; Butler, B.; Woo, S.; Wisenberg, M.; Figura, J.; Onufrey, V.; Wazer, D.; Ling, M.; Grant, W. H. III; Pavord, D.; Turner, A.; Berta, C.; Engler, M.; Tsai, J.; Curran, B.; Kania, A.; Larson, S.; Tracy, L. Preliminary clinical experience with the peacock intensity modulation 3-D conformal radiation therapy system. Radiosurgery 1:327-335; 1996. 3. Dunbar, S. F.; Tarbell, N. J.; Kooy, H. M.; Alexander, E. III; Black, P.; Barnes, P. D. ; Goumnerova, L. ; Scott, R. M.; Pomeroy, S. L.; LaVally, B.; Sallan, S. E.; Loeffler, J. S. Stereotactic radiotherapy for pediatric and adult brain tumors: Preliminary report. Int. J. Radiat. Oncol. Biol. Phys. 30:531539; 1994. 4. Fhckinger, J. C.; Lunsford, L. D.; Wu, A.; Maitz, A. H.; Kalend, A. M. Treatment planning for gamma knife radiosurgery with multiple isocenters. Int. J. Radiat. Oncol. Biol. Phys. 18:1495-1501; 1990. 5. Hamilton, R. J.; Kuchnir, F. T.; Sweeney, P.; Rubin, S. J.; Dujovny, M.; Pelizzari, C. A.; Chen, G. T. Y. Comparison of static conformal field with multiple noncoplanar arc techniques for stereotactic radiosurgery or stereotactic radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 33: 1221- 1228; 1995. 6. ICRU. Report 50. Prescribing, recording, and reporting photon beam therapy. Bethesda, MD: ICRU; 1993. 7. Killoran, J. H.; Kooy, H. M.; Bellerive, M. R.; Hacker, F. L.; Mannarino, E. G.; Shiu, A. S.; Shrieve, D. C.; Tarbell, N. T.; Loeffler, J. S. Treatment planning support for dynamic linac stereotactic delivery. Radiosurgery 1:288-297; 1996. 8. Kooy, H. M.; Cosman, E. R.; Harlan, R. B.; Ledoux, R. J.; Bellerive, M. R.; Hacker, F. L.; Killoran, J. H.; Mannarino, E. G.; Tarbell, N. J.; Alexander, E. III; Shrieve, D. C.; Loeffler, J. S. A dynamic collimator system for conformal stereotactic radiosurgery and radiotherapy. Radiosurgery 1:276-287; 1996. 9. Laing, R. W.; Bentley, R. E.; Nahum, A. E.; Warrington, A. P.; Brada, M. Stereotactic radiotherapy of irregular targets: A comparison between static conformal beams and non-coplanar arcs. Radiother. Oncol. 28:241-246; 1993. 10. Leavitt, D. D.; Gibbs, F. A.; Heilbrun, M. P.; Moeller, J. H.; Takach, G. A., Jr. Dynamic field shaping to optimize stereotactic radiosurgery. Int. J. Radiat. Oncol. Biol. Phys. 21:1247-1255; 1991. 11. Loeffler, J. S.; Alexander, E. III; Siddon, R. L.; Saunders, W. M.; Coleman, C. N.; Winston, K. R. Stereotactic radio-

12.

13. 14.

15.

16.

17.

18.

19.

20.

21

surgery for intracranial arteriovenous malformations using a standard linear accelerator. Int. J. Radiat. Oncol. Biol. Phys. 17:673-677; 1989. Loeffler, J. S.; Siddon, R. L.; Wen, P. Y.; Nedzi, L. A.; Alexander, E. III. Stereotactic radiosurgery of the brain using a standard linear accelerator: a study of early and late effects. Radiother. Oncol. 17:311-321; 1990. Marks, L. B.; Spencer, D. P. The influence of volume on the tolerance of the brain to radiosurgery. J. Neurosurg. 75:177180; 1991. Marks, L. B .; Sherouse, G. W.; Das, S.; Bentel, G. C.; Spencer, D. P.; Turner, D. Conformal radiation therapy with fixed shaped coplanar or noncoplanar radiation beam bouquets: A possible alternative to radiosurgery. Int. J. Radiat. Oncol. Biol. Phys. 33:1209-1219; 1995. Nedzi, L. A.; Kooy, H.; Alexander, E. III; Gelman, R. S.; Loeffler, J. S. Variables associated with the development of complications from radiosurgery of intracranial tumors. Int. J. Radiat. Oncol. Biol. Phys. 21:591-599; 1991. Nedzi, L. A.; Kooy, H. M.; Alexander, E. III; Svensson, G. K.; Loeffler, J. S. Dynamic field shaping for stereotactic radiosurgery: A modeling study. Int. J. Radiat. Oncol. Biol. Phys. 25:859-869; 1993. Parsons, J. T.; Bova, F. J.; Fitzgerald, C. R.; Mendenhall, W. M.; Million, R. R. Radiation optic neuropathy after megavoltage external-beam irradiation: Analysis of timedose factors. Int. J. Radiat. Oncol. Biol. Phys. 30:755-763; 1994. Schlegel, W.; Pastyr, 0.; Bortfeld, T.; Gademann, G.; Menke, M.; Maier-Borst, W. Stereotactically guided fractionated radiotherapy: Technical aspects. Radiother. Oncol. 29:197-204; 1993. Shiu, A. S.; Kooy, H. M.; Ewton, J. R.; Tung, S. S.; Wong, J.; Antes, K.; Maor, M. H. Comparison of miniature multileaf collimation (MMLC) with circular collimation for stereotactic treatment. Int. J. Radiat. Oncol. Biol. Phys. 37:679688; 1997. Shrieve, D. C.; Kooy, H. M.; Tarbell, N. J.; Loeffler, J. S. Fractionated stereotactic radiotherapy. In: DeVita, V., Hellman, S., Rosenberg, S., eds. Important Advances in Oncology 1996. Philadelphia, PA: Lippincott-Raven; 1996:205224. Woo, S. Y.; Grant, W. H.; Bellezza, D.; Grossman, R.; Gildenberg, P.; Carpenter, L. S.; Carol, M.; Butler, E. B. A comparison of intensity modulated conformal therapy with a conventional external beam stereotactic radiosurgery system for the treatment of single and multiple intracranial lesions. Int. J. Radiat. Oncol. Biol. Phys. 35:593-597; 1996.