optimization of collimator parameters to reduce … · course of treatment, but also the dose that...

DltnAtdathpiisp

mhrdcdtr

OC

Medical Dosimetry, Vol. 30, No. 4, pp. 205-212, 2005Copyright © 2005 American Association of Medical Dosimetrists

doi:10.1016/j.meddos.2005.06.002

OPTIMIZATION OF COLLIMATOR PARAMETERS TO REDUCE RECTALDOSE IN INTENSITY-MODULATED PROSTATE TREATMENT PLANNING

JULIE CHAPEK, C.M.D., MATT TOBLER, C.M.D., BEAU J. TOY, M.D.,CHRISTOPHER M. LEE, M.D., DENNIS D. LEAVITT, PH.D.

Huntsman Cancer Hospital, University of Utah Health Science Center, Salt Lake City, UTRadiation Oncology Centers of Las Vegas, Las Vegas, NV

(Received 1 March 2005; accepted 1 June 2005)

Abstract—The inability to avoid rectal wall irradiation has been a limiting factor in prostate cancer treatmentplanning. Treatment planners must not only consider the maximum dose that the rectum receives throughout acourse of treatment, but also the dose that any volume of the rectum receives. As treatment planning techniqueshave evolved and prescription doses have escalated, limitations of rectal dose have remained an area of focus.External pelvic immobilization devices have been incorporated to aid in daily reproducibility and lessen concernfor daily patient motion. Internal immobilization devices (such as the intrarectal balloon) and visualizationtechniques (including daily ultrasound or placement of fiducial markers) have been utilized to reduce theuncertainty of intrafractional prostate positional variation, thus allowing for minimization of treatment volumes.Despite these efforts, prostate volumes continue to abut portions of the rectum, and the necessary volumeexpansions continue to include portions of the anterior rectal wall within high-dose regions. The addition ofcollimator parameter optimization (both collimator angle and primary jaw settings) to intensity-modulatedradiotherapy (IMRT) allows greater rectal sparing compared to the use of IMRT alone. We use multiple patientexamples to illustrate the positive effects seen when utilizing collimator parameter optimization in conjunctionwith IMRT to further reduce rectal doses. © 2005 American Association of Medical Dosimetrists.

Printed in the USA. All rights reserved0958-3947/05/$–see front matter

Key Words: Optimization, Collimator parameters, Prostate radiation therapy, IMRT.

iWc5atrtiat7aawocicA(mitsbr

INTRODUCTION

evelopments in treatment planning systems have al-owed a shift from conventional, nonconformal radiationherapy to highly conformal 3-dimensional (3D) plan-ing and intensity-modulated radiation therapy (IMRT).s dose is made to be more conformal to the planning

arget volume (PTV), there can be a reduction of doseelivered to normal tissues. Improved dose conformalityllows for the consideration of prescription dose escala-ion. In turn, the ability to safely achieve higher dosesas lead to important milestones in the treatment ofrostate cancer, where data support that dose escalationn the treatment of prostate cancer results in significantlymproved prostate-specific antigen (PSA) relapse-freeurvival in patients with intermediate and unfavorablerognosis.1

Along with dose escalation, it is essential to mini-ize the volume of normal tissues that are treated to

igher doses to be able to achieve a higher therapeuticatio. In treating the prostate, the main limiting factor isose delivered to the rectum. This dose must not only beonsidered as a maximum point dose, but the entireose-volume histogram (DVH) must be evaluated. Whenreating patients to doses of 70.2–75.6 Gy, it has beeneported that rectal bleeding correlated to significant

Reprint requests to: J. Chapek, C.M.D., Department of Radiation

rncology, Huntsman Cancer Hospital, 1950 Circle of Hope, Salt Lakeity, UT 84112-5560. E-mail: [email protected]

205

ncrease in irradiated rectal volume.2 In a study byachter et al., it was reported that grade 2 late rectal

omplications occurred in patients who had more than7% of the rectum irradiated to 60 Gy.3 Boersma et al.lso showed, through analysis of irradiated volumes, thathere was a significant increase in the incidence of severeectal bleeding in patients where more than 40–50% ofhe rectal wall received at least 65 Gy.4 In their prelim-nary report of a randomized 3D-CRT (conformal radi-tion therapy) dose escalation trial, Storey et al. reportedhat patients with � 25% of rectal volume irradiated to �

0 Gy developed grade 2 or higher complications.5 Overlonger follow-up period, Huang et al. found that dose

nd volume were continuously interrelated variablesith a significant volume effect observed at rectal dosesf 60, 70, 75.6, and 78 Gy, with the risk of rectalomplications increasing exponentially with increasedrradiated volume.6 This dose-volume relationship be-omes increasingly significant as dose escalation occurs.rm 2 of the current Radiation Therapy Oncology Group

RTOG) protocol for prostate cancer (#0126) recom-ends prostate prescription doses of 79.2 Gy (with dose

nhomogeneity � 7%) and provides strict guidelines forhe evaluation of the DVH. With these escalated pre-cription doses, the need to reduce dose to rectal volumesecomes more critical. In this RTOG protocol, it isecommended that less than 15% of the rectal volume
eceive 75 Gy, less than 25% receive 70 Gy, less than

3r

ibtsPcotbttmlulicap

pdviIoinpttoemtewir

P

wtInmimttl

lrfo

D

turftPct

FdCcmtmb

Medical Dosimetry Volume 30, Number 4, 2005206

5% receive 65 Gy, and that no more than 50% volumeeceive more than 60 Gy.

In an effort to reduce the volume of rectal wall,rradiated innovative techniques have been employed toetter localize or stabilize the target volume at eachreatment. These include intrarectal balloons, daily ultra-ound, and fiducial markers implanted in the prostate.atel et al. have reported on the use of a rectal balloonatheter resulting in a significant decrease in the volumef rectal wall receiving high dose (� 60 Gy).7 Localizinghe prostate using transabdominal ultrasound on a dailyasis provides a rapid way of adjusting fields to matchhe target volume prior to each treatment.8,9 Many au-hors have evaluated the feasibility of gold-seed fiducialarkers implanted into the prostate and utilized daily to

ocalize the prostate prior to treatment.10 –13 Better vol-me stabilization and/or daily volume localization willead to increased confidence in daily treatment reproduc-bility. As a result of this increased confidence, physi-ians may decrease the amount of expansion they requireround the clinical target volumes (CTVs) to create thelanning target volumes (PTVs).

Despite gains made in daily localization and theossible reduction in size of PTVs, prostate volumesirectly abut portions of the rectum and when theseolumes are expanded to create PTVs, they may actuallynclude portions of the rectum in the high-dose regions.t has been shown that utilizing 3D-conformal planningr IMRT reduces rectal and bladder doses and thereforemproves the risk-to-benefit ratio.14 These planning tech-iques have also led to clinically acceptable escalation ofrescription doses for prostate cancer treatment. It isherefore essential for dosimetrists to continually inves-igate techniques that will result in the further reductionf doses to the rectum. In this study, we examined theffect of adjusting collimator parameters. Primary colli-ator settings and collimator angles were adjusted rela-

ive to the geometry of the rectum, making these param-ters more conformal to the rectum. We evaluatedhether the adjustments of these parameters, when used

n conjunction with IMRT planning, can result in aeduction of dose delivered to the rectum.

METHOD

atient selectionThree patients were selected for this study. Patients

ere chosen based on the angle of the rectum comparedo a neutral collimator angle (i.e., 0°, 90°, 180° or 270°).t was felt that a large angle between the rectum and theeutral collimator angle would be the condition when theaximum amount of rectum would be unshielded. This

s partially due to the stair-stepped appearance of theultileaf collimator (MLC) leaves along the length of

he rectum as they create the field shape required to coverhe PTV (Fig. 1a). This figure also shows that there is a
arge volume of rectum that is shielded only by the MLC r
eaves and not the primary collimators. This underlyingectal volume would receive dose during treatment in theorm of transmission through the MLC leaves (which isn the order of 2.4%).

efinition of rectum and rectal wallFor the purpose of our study, the rectum was con-

oured from the inferior portion of the ischial tuberositiesp to the superior aspect of the rectosigmoid flexure. Theectal wall was defined as a 2-mm-thick wall extendingrom 2.5 mm inferior to the most inferior aspect of PTVo 2.5 mm superior to the most superior aspect of theTV. The rectal wall was evaluated in this study, be-ause it extended such a small distance beyond the PTV;herefore; this volume benefited most from the study,

ig. 1. Currently, the standard IMRT collimator setup proce-ure utilizes no optimization of collimator parameters. (A)ollimator angle is set at a neutral angle (180°) and the primaryollimators are set to the pre-set default settings outside theaximum MLC aperture. (B) Full optimization of the collima-

or parameters utilizes optimization of both collimator angle toatch the angle of the rectum and of the primary collimators to

ring jaws in as close as possible to the edge of the MLCaperture.

esulting in greater reduction to rectal wall dose.

D

aTamda

E

w

1

2

3

d com

Optimization of collimator parameters ● J. CHAPEK et al. 207

efinition of target volumesIn each case, the CTV was defined as the prostate

nd proximal 1 cm of seminal vesicles without a margin.he PTV was an expansion of 1 cm beyond the CTV inll directions except posterior, where expansion was 6m. The dosimetric coverage of the target volumes was

efined as the 100% isodose line to cover the entire CTVnd the 95% isodose line to cover the entire PTV.

valuated variablesFor each patient studied the following variables

ere applied:

. Collimator angle set at a neutral angle of 180°. Pri-mary collimators were set to preset default of 3 mm

Fig. 2. DVHs for the CTV of (A) patient A, (B) patienprimary collimator setting optimization only (triangle)

optimized collimator parameter (cross) plan delivere

(in Y direction) and 5 mm (in X direction) outside the

MLC leaves. (Fig. 1a). This is currently the mostcommonly accepted collimator parameter setup pro-cedure.

. Primary collimator settings optimized to bring thejaws in as close as possible to the edge of MLCswithout impinging on the treatment field. This willreduce the amount of extra dose delivered as a resultof transmission through the MLC leaves, as the pri-mary collimators shield as much of the MLC area aspossible.

. Collimator angle optimized so that the angle of thecollimator followed a best fit of the angle of therectum. Primary collimator settings were allowedto follow the preset default defined above in point

d (C) patient C show that no optimization (diamond),mator angle optimization only (circle), and the fullyparable results. Plans done with 1-cm MLC leaves.

t B, an, colli

no. 1. Collimator angle was determined by visual

4

mA

cvafppmtobctzwsppocmats

Fnrot(tcl

Ffft


adjustment until it best followed the slope of therectum. This resulted in a minimized stair-stepeffect of the MLCs, as the angle of incidence of theleaves was more conformal to the angle of therectum. This more conformal collimator angle alsoresulted in the primary collimators shielding anincreased area of the MLCs, therefore reducing

ig. 3. Patient A. (A) DVH for rectal wall shows that theon-optimized plan (diamond) delivers the highest dose to theectal wall compared to the plans with primary collimatorsptimized (triangle), collimator angle optimized (circle), andhe fully optimized collimator and primary collimator positionscross). (B) DVH comparison for the rectal wall showing onlyhe non-optimized results (diamond) and the fully optimizedollimator and primary collimator positions results (cross) il-ustrate a dramatic improvement in rectal sparing with the fully

optimized technique. Plans done with 1-cm MLC leaves.

transmission through the leaves to the patient.

. Collimator angle and primary collimator settingswere fully optimized as described in point nos. 2 and3 above, to achieve the maximum amount of shieldingon the rectum while still covering the PTV (Fig. 1b).

Plans were evaluated using the Varian Eclipse treat-ent planning system. (Varian Medical Systems, Palolto, CA).

It was essential to ensure that all plans were exe-uted in a similar manner. Only the evaluated individualariables were altered for each plan. Due to the computerlgorithm applied within the planning system as it per-orms an optimization on an IMRT plan, it is neverossible to achieve an identical optimization on the sameatient with the same plan. To reduce this variation to ainimum, an initial plan was run in which both collima-

or angle and primary collimator position were fullyptimized. This plan was the first one to be executedecause we felt that this situation of fully optimizedollimator parameters would give the best end results. Inhe IMRT optimization process for this plan, the optimi-ation parameters (dose constraints and priority settings)ere determined. This developed optimization parameter

et was saved as an optimal constraint template. Fluenceatterns were then deleted from the initial plan and thelan was re-optimized and calculated using the savedptimal constraint template. The IMRT optimization pro-ess was allowed to continue until no further improve-ents were registered in the planning system with each

dditional iteration. In the same manner, this constraintemplate was then applied to all subsequent planningituations.

ig. 4. Patient A. The DVH for the rectal wall showing resultsor the non-optimized collimator parameters (diamond) and theully optimized collimator angle and primary collimator set-ings (cross) with the 0.5-cm MLC leaves. Results show an
improvement in rectal sparing with the fully optimized plan.

ifGdea

0c

rcHctrctcwmmtnsstwta

dmstr

Fttot(tclu

Ffft


Each variable was evaluated using geometricallydentical 7 field plans. As per arm 2 of RTOG protocolor prostate cancer (#0126), a prescription dose of 79.2y was used for all plans in this study. Within ourepartment, we utilize linear accelerators equipped withither 1-cm leaves or 0.5-cm leaves. It was decided that

ig. 5. Patient B. (A) DVH for the rectal wall again shows thathe non-optimized plan (diamond) delivers the highest dose tohe rectal wall compared to the plans with primary collimatorsptimized (triangle), collimator angle optimized (circle), andhe fully optimized collimator and primary collimator positionscross). (B) DVH comparison for the rectal wall showing onlyhe non-optimized results (diamond) and the fully optimizedollimator and primary collimator positions results (cross) il-ustrates, as for patient A, the reduction in rectal dose whentilizing the fully optimized plan. Plans done with 1-cm MLC

leaves.

ll plans would be evaluated using both the 1- and

.5-cm MLC leaves, and the resulting DVHs would beompared.

RESULTS

DVH evaluation of doses delivered to the CTVevealed that for all 3 patient cases, equivalent volumeoverage was achieved for each planning situation.ence, the application of the studied variables in no way

ompromised CTV coverage (Fig. 2a– c). Evaluation ofhe DVH for the rectal wall, however, shows gains inectal sparing as a result of the optimization of theollimator parameters. For patient A, Fig. 3a illustrateshat the non-optimized technique (currently the mostommonly used parameter setup procedure) achieved theorst result for dose delivered to the rectal wall. Opti-ization of the primary collimator setting alone or opti-ization of the collimator angle alone achieved results

hat were a slight improvement in rectal wall dose overo optimization. Combination of both primary collimatoretting and collimator angle optimization achieved re-ults that were a dramatic improvement over the otherechniques evaluated. This improvement was also notedith the fully optimized 0.5-cm MLC leaves, although

he improvement in rectal wall dose was not as dramatics that achieved with the 1-cm leaves (Fig. 4).

For both patients A and B, the angle of the rectumeviated from the neutral collimator angle by approxi-ately 14°. DVHs for patient B show results that are

imilar to those of patient A (Fig. 5a and 5b, Fig. 6). Forhe fully optimized technique, results show a greatereduction in rectal wall dose compared to the partially

ig. 6. Patient B. The DVH for the rectal wall showing resultsor the non-optimized collimator parameters (diamond) and theully optimized collimator angle and primary collimator set-ings (cross) with the 0.5-cm MLC leaves. Results show an
improvement in rectal sparing with the fully optimized plan.

octip

ftaaomfet1(

tpap

Fnrotpom(fpr

curve. Plans done with 1-cm MLC leaves.

Ffft

Plans done with 1-cm MLC leaves.


ptimized situations, even with this small amount ofollimator angle introduced into the plan. For patient A,he 0.5-cm MLC leaves also show the greatest reductionn rectal wall dose when the fully optimized collimatorarameters were applied to patient B.

The reduction in rectal wall dose is most impressiveor patient C. The anatomy of this patient was such thathe rectal wall angle deviated from the neutral collimatorngle by approximately 24°. This large angle allowed forlarge change in both shape and conformity (as a resultf the collimator angle optimization) and in MLC trans-ission (due to primary collimator optimization). DVHs

or patient C show reductions in rectal wall dose thatxtend well into the high-dose regions of the curve withhe fully optimized collimator parameters for both the-cm MLC leaves (Fig. 7) and the 0.5-cm MLC leavesFig. 8).

Table 1 shows that the dose to the rectal wall athe 50% volume point on the DVHs is reduced in thelans with fully optimized collimator parameters forll patients. The most commonly used collimator setuprocedure (non-optimized situation) provided the

ig. 8. Patient C. The DVH for the rectal wall showing resultsor the non-optimized collimator parameters (diamond) and theully optimized collimator angle and primary collimator set-ings (cross) with the 0.5-cm leaves. Results show an improve-

ment in rectal sparing with the fully optimized plan.

all at 50% volume point of DVH

Volume of Rectal Wall (Gy)ull Optimization

Change in Dose to 50% Volumeof Rectal Wall

44.0 8% less dose with optimization42.0 3.3% less dose with optimization44.8 3.8% less dose with optimization

ig. 7. Patient C. (A) DVH for the rectal wall shows that theon-optimized (diamond) plan delivers the highest dose to theectal wall compared to the plans with primary collimatorsptimized (triangle), collimator angle optimized (circle), andhe fully optimized collimator and primary collimator (cross)ositions. (B) DVH comparison for the rectal wall showingnly the non-optimized results (diamond) and the fully opti-ized collimator and primary collimator positions results

cross) again show the benefit gained in reducing rectal dose inully optimized plan. The separation of the DVH curves for thisatient is greater than that for patients A or B. Reductions ofectal wall dose extend well into the high-dose regions for the

Table 1. Comparison of dose to rectal w

PatientDose to 50% Volume of Rectal Wall (Gy)

No OptimizationDose to 50%

F

A 47.7B 43.4C 46.5

wsrgCrat

smortnimtoaasnicpro

ttctcmfstM

ltIbii

iact

1

P


orst result in all cases analyzed. Similarly, Table 2hows that for the fully optimized scenario there is aeduction in the volume of rectal wall receiving dosesreater than 60 Gy. This is most noticeable in patient. As stated in these previously sited studies, these

eductions in rectal dose would be predicted to lead todecrease in the probability of both long- and short-

erm rectal complications.

DISCUSSION

This study reveals that collimator optimizationtrategies such as utilizing either collimator angle opti-ization, primary collimator position optimization, or

ptimization of both of these parameters simultaneouslyesulted in a reduction of rectal wall volume dose overhe non-optimized technique. The non-optimized tech-ique is currently the most common technique employedn radiotherapy departments, although the fully opti-ized situation demonstrated the best results in its ability

o reduce the rectal wall dose. The implementation of anyne of these optimized techniques does not require theddition of any hardware or software to most linearccelerators or treatment planning systems. These areimple techniques that utilize currently available tech-ology and are therefore techniques that can easily bemplemented in any radiotherapy department. The appli-ation of these techniques need not be limited to IMRTlans, but can be applied to 3D planning, and shouldesult in similar reductions of dose to the rectum or anyther structure of interest.

CONCLUSIONS

This study has shown that through the optimiza-ion of collimator angle and primary collimator posi-ions, in conjunction with IMRT planning for prostateancer, it is possible to reduce the dose delivered tohe volume of the rectal wall. Optimization of theollimator angle allows improvement of the confor-ality of the MLC leaves to the rectal volumes, there-

ore providing increased shielding to this normaltructure. Optimizing the primary collimators so thathey do not extend beyond the maximum extent of the

Table 2. Comparison of volume o

PatientVolume Receiving 60 Gy (cm3)

No OptimizationVolume

F

A 4.2

B 6.4

C 3.1

lans done with 1-cm MLC leaves.

LC opening ensures that transmission through1

eaves to the patient is kept to a minimum. This MLCransmission reduction is especially important inMRT planning, where commonly, an increased num-er of monitor units is delivered daily and therefore anncreased dose is delivered to the patient throughnterleaf and intraleaf leakage.

The application of collimator and jaw optimizations effective in reducing dose to the rectal wall volumend could be equally applied to other radiosensitiveritical structures near target volumes in various ana-omic sites.

REFERENCES

1. Zelefsky, M.J.; Leibel, S.A.; Gaudin, P.B.; et al. Dose escalationwith three-dimensional conformal radiation therapy affects theoutcome in prostate cancer. Int. J. Radiat. Oncol. Biol. Phys.41:491–500; 1998.

2. Jackson, A.; Skwarchuk, M.W.; Zelefsky, M.J.; et al. Late rectalbleeding after conformal radiotherapy of prostate cancer. II. Vol-ume effects and dose-volume histograms. Int. J. Radiat. Oncol.Biol. Phys. 49:685–98; 2001.

3. Wachter, S.; Gerstner, N.; Goldner, G.; et al. Rectal sequelae afterconformal radiotherapy of prostate cancer: Dose-volume histo-grams as predictive factors. Radiother. Oncol. 59:65–70; 2001.

4. Boersma, L.J.; van den Brink M., Bruce, A.M.; et al. Estimation ofthe incidence of late bladder and rectum complications after high-dose (70–78 Gy) conformal radiotherapy for prostate cancer, usingdose-volume histograms. Int. J. Radiat. Oncol. Biol. Phys. 41:83–92; 1998.

5. Storey, M.R.; Pollack, A.; Zagars G.; et al. Complications fromradiotherapy dose escalation in prostate cancer: Preliminary resultsof a randomized trial. Int. J. Radiat. Oncol. Biol. Phys. 48:635–42;2000.

6. Huang, E.H.; Pollack, A.; Levy, L.; et al. Late rectal toxicity:Dose-volume effects of conformal radiotherapy for prostate can-cer. Int. J. Radiat. Oncol. Biol. Phys. 54:1314–21; 2002.

7. Patel, R.R.; Orton, N.; Tome, W.A.; et al. Rectal dose sparing with aballoon catheter and ultrasound localization in conformal radiationtherapy for prostate cancer. Radiother. Oncol. 67:285–94; 2003.

8. Trichter, F.; Ennis, R.D. Prostate localization using transabdominalultrasound imaging. Int. J. Radiat. Oncol. Biol. Phys. 56:1225–33;2003.

9. Little, D.J.; Dong, L.; Levy, L.B.; et al. Use of portal images andBAT ultrasonography to measure setup error and organ motion forprostate IMRT: Implications for treatment margins. Int. J. Radiat.Oncol. Biol. Phys. 56:1218–24; 2003.

0. Dehnad, H.; Nederveen, A.J.; van der Heide, U.A.; et al. Clinicalfeasibility study for the use of implanted gold seeds in the prostateas reliable positioning markers during megavoltage irradiation.Radiother. Oncol. 67:295–302;2003.

l wall receiving more than 60 Gy

ng 60 Gy (cm3)mization

Change in Volume of Rectal WallReceiving 60 Gy

4 4.7% reduction in rectal wall irradiated to� 60 Gy with optimization



f recta

Receiviull Opti

4.

6.

3.

1. Welsh, J.S.; Berta, C.; Borzillary, S.; et al. Fiducial markersimplanted during prostate brachytherapy for guiding conformal

1

1

1


external beam radiation therapy. Technol. Cancer Res. Treat.3:359–64;2004.

2. Poggi, M.M.; Gant, D.A.; Sewchand, W.; et al. Marker seedmigration in prostate localization. Int. J. Radiat. Oncol. Biol. Phys.56:1248–51;2003.

3. Kitamura, K.; Shirato, H.; Shimizu, S.; et al. Registration accu-
racy and possible migration of internal fiducial gold marker
implanted in prostate and liver treated with real-timetumor-tracking radiation therapy (RTRT). Radiother. Oncol.62:275– 81;2002.

4. Zelefsky, M.J.; Fuks, Z.; Hunt, M.; et al. High-dose intensitymodulated radiation therapy for prostate cancer: Early toxicity andbiochemical outcome in 772 patients. Int. J. Radiat. Oncol. Biol.
Phys. 53:1111–6; 2002.

optimization of collimator parameters to reduce … · course of treatment, but also the dose that...

Documents