proton therapy coverage for prostate cancer treatment

Int. J. Radiation Oncology Biol. Phys., Vol. 70, No. 5, pp. 1492–1501, 2008Copyright � 2008 Elsevier Inc.

Printed in the USA. All rights reserved0360-3016/08/$–see front matter

doi:10.1016/j.ijrobp.2007.09.001

CLINICAL INVESTIGATION Prostate

PROTON THERAPY COVERAGE FOR PROSTATE CANCER TREATMENT

CARLOS VARGAS, M.D.,*y MARCUS WAGNER, M.D.,* CHAITALI MAHAJAN, M.D.,*

DANIEL INDELICATO, M.D.,y AMBER FRYER, B.S.,* AARON FALCHOOK, B.S.,* DAVID HORNE, C.M.D.,*

ANGELA CHELLINI, B.S.,* CRAIG MCKENZIE, C.M.D.,* PAULA LAWLOR, C.M.D.,* ZUOFENG LI, D.SC.,*

LIYONG LIN, PH.D.,* AND SAMEER KEOLE, M.D.*

*University of Florida Proton Therapy Institute, Jacksonville, FL; and yDepartment of Radiation Oncology,University of Florida College of Medicine, Gainesville, FL

Purpose: To determine the impact of prostate motion on dose coverage in proton therapy.Methods and Materials: A total of 120 prostate positions were analyzed on 10 treatment plans for 10 prostatepatients treated using our low-risk proton therapy prostate protocol (University of Florida Proton Therapy Insti-tute 001). Computed tomography and magnetic resonance imaging T2-weighted turbo spin-echo scans wereregistered for all cases. The planning target volume included the prostate with a 5-mm axial and 8-mm superoin-ferior expansion. The prostate was repositioned using 5- and 10-mm one-dimensional vectors and 10-mm multidi-mensional vectors (Points A–D). The beam was realigned for the 5- and 10-mm displacements. The prescriptiondose was 78 Gy equivalent (GE).Results: The mean percentage of rectum receiving 70 Gy (V70) was 7.9%, the bladder V70 was 14.0%, and thefemoral head/neck V50 was 0.1%, and the mean pelvic dose was 4.6 GE. The percentage of prostate receiving78 Gy (V78) with the 5-mm movements changed by �0.2% (range, 0.006–0.5%, p > 0.7). However, the prostateV78 after a 10-mm displacement changed significantly (p < 0.003) with different movements: 3.4% (superior),�5.6% (inferior), and �10.2% (posterior). The corresponding minimal doses were also reduced: 4.5 GE,�4.7 GE, and �11.7 GE (p # 0.003). For displacement points A–D, the clinical target volume V78 coverage hada large and significant reduction of 17.4% (range, 13.5–17.4%, p < 0.001) in V78 coverage of the clinical targetvolume. The minimal prostate dose was reduced 33% (25.8 GE), on average, for Points A–D. The prostate minimaldose improved from 69.3 GE to 78.2 GE (p < 0.001) with realignment for 10-mm movements.Conclusion: The good dose coverage and low normal doses achieved for the initial plan was maintained with move-ments of #5 mm. Beam realignment improved coverage for 10-mm displacements. � 2008 Elsevier Inc.

Protons, Dose–volume, Prostate cancer, High dose, Motion.

INTRODUCTION

Interfraction prostate motion has been an area of major con-

cern for the past few years in prostate cancer therapy (1–6).

Large variations in daily prostate position have been de-

scribed in various studies (1, 2, 4, 6). Furthermore, studies

have found a poor correlation between the location of the

prostate and skin marks or bony anatomy (5, 7). Also, ultra-

sound systems cannot reliably define the daily prostate posi-

tion (4, 5). As a result, when daily treatment setup is based

on skin marks, bony anatomy, or ultrasound guidance, large

margins are necessary to ensure accurate delivery of the pre-

scribed dose to the prostate (2, 4, 5).

Several different approaches have been used in an attempt

to solve the problem of prostate motion and reduce the treat-

ment margins. Yan et al. (6) used multiple computed tomog-

raphy (CT) scans to define prostate motion and electronic

Reprint requests to: Carlos Vargas, M.D., University of FloridaProton Therapy Institute, 2015 N. Jefferson St., Jacksonville, FL32206. Tel: (904) 588-1800; Fax: (904) 588-1300; E-mail: [email protected]

14

daily images to quantify setup inaccuracies. They were able

to derive patient-specific margins through an off-line adap-

tive process. The planning target volume (PTV) was greatly

reduced compared with the margin definition using a class so-

lution. A second approach, on-line image guidance, relies on

daily information to derive the daily prostate position (2, 4, 5,

7, 8). Different approaches have been used, including cone-

beam CT, CT-on-rails, fiducial markers, and beacon tran-

sponders (1, 2, 4, 5, 7–11). The accuracy of these systems

relies on the accuracy of the measurement and the appropriate

patient translation adjustments in relation to the treatment

machine isocenter. None of these approaches eliminates the

problem of intrafraction motion, which must still be ad-

dressed.

Changes in prostate dose with defined changes in prostate

position are important in developing treatment strategies for

Conflict of interest: none.Received May 25, 2007, and in revised form Aug 15, 2007.

Accepted for publication Sept 8, 2007.

92

mailto:[email protected]

mailto:[email protected]

Proton therapy coverage for prostate cancer treatment d C. VARGAS et al. 1493

proton therapy. For our treatments, we prescribed a minimal

dose to the PTV of 74.1 Gy equivalent (GE) and directed that

78 GE cover 95% of the PTV. Thus, the doses within the PTV

were not completely homogeneous, and different clinical

target volume (CTV) positions even within the PTV had dif-

ferent minimal doses and prescribed dose coverage. To fully

evaluate the differences in dose distribution inside and out-

side the PTV, we repositioned the prostate using unidirec-

tional vectors of 5 and 10 mm and multidirectional vectors

of 10 mm.

In the present analysis, we simulated prostate displacement

in multiple dimensions and magnitudes in 10 different pa-

tients. By characterizing the variations in prostate dose cov-

erage of the prostate with displacement, we hoped to increase

the understanding of the effect of daily variations in prostate

location on the actual doses delivered to the target according

to historical setup techniques that use skin marks and/or bony

anatomy as a surrogate for prostate position. These findings

are critical to understanding the optimal margins or PTV

expansions and the role of image guidance.

METHODS AND MATERIALS

Between August and December 2006, the data from 10 patients

treated on our institutional review board–approved Phase II proton

protocol for low-risk prostate cancer were analyzed (University of

Florida Proton Therapy Institute 001). One of us (C.V.) was respon-

sible for reviewing all the plans and treating the patients. Subsequent

reviews were performed by our physics staff. The plans met all

protocol requirements, with no major deviations.

The patient characteristics were defined according to the National

Comprehensive Cancer Network low-risk definition: clinical Stage

T2a or less, prostate-specific antigen level of <10 ng/mL, and Glea-

son score #6. The staging workup for all patients included 10–12-

core biopsies, pelvic magnetic resonance imaging (MRI), bone scan,

chest X-ray, and blood work, including determination of prostate-

specific antigen, free prostate-specific antigen, alkaline phosphatase,

complete blood count, and testosterone levels.

SimulationAll patients underwent treatment simulation at the University

of Florida Proton Therapy Institute using a 16-slice large-bore

(40-cm) helical CT scanner and 0.23-T open MRI scanner (Phillips,

Eindhoven, The Netherlands). A custom-indexed vacuum-lock

bag was used for immobilization. A rectal balloon with 100 mL of

saline was used in all cases. Laser marks were used for visual inspec-

tion and to correct the alignment in all degrees of freedom. To mea-

sure the rotation of the patient at the level of the femoral heads, we

obtained a scout CT scan and created a horizontal guide line. A ver-

tical line was used to assess the vertical position and yaw. The pa-

tient was then scanned from 5 cm below the ischial tuberosities to

the L5–S1 interspace in 1-mm slices. For the MRI scan, T2-weighted

turbo spin-echo sequences were acquired with 3-mm spacing and

3-mm interpolation.

Volume definitionThe images were imported into Syntegra (Pinnacle, ADAC, Mil-

pitas, CA), and the two data sets were automatically fused. Land-

marks, including the symphysis pubis and femoral heads, were

used to define the appropriate bony alignment in manual correction.

Soft-tissue landmarks, including the rectal wall–prostate interface

and bladder–prostate interface were used to define the soft-tissue

alignment, and additional manual corrections were performed as

necessary to optimize the fusion of the MRI and CT data. The pros-

tate volume alone, using the sagittal-, coronal-, and transverse-reg-

istered information, defined the CTV and was used for contouring.

The seminal vesicles were not included in low-risk cases (12). The

CTV was expanded 5 mm in the axial and 8 mm in the craniocaudal

dimensions to create the PTV. A greater margin was used in the

superoinferior axis to counter large uncertainties when identifying

the prostate apex and base that affected changes in PTV coverage.

The rectum was contoured from the ischial tuberosities to the

sigmoid flexure for all cases (13). The rectal wall was defined with

a 3-mm internal wall extraction from the rectal contour to create

a rectal wall volume. The entire bladder was contoured, and the

bladder wall volume was also defined by a 3-mm extraction to create

an inner wall.

Proton planningThe treatments were planned with Eclipse software (Varian Med-

ical Systems, Palo Alto, CA), and all necessary beam information

from our gantries was downloaded into the planning software pro-

gram. We used double-scattered proton beams. The proton beam

angles were optimized for each patient. All cases were treated with

posterior oblique beams with angles between 4–12� and 168–176�.The energy for all cases was 232 MeV, and the range was 25.6–28.0

cm. The modulation varied from to 8.5 to 10.5 cm. Distal and prox-

imal margins were added to the PTV and were 5–8 mm minimum

from the edge of the PTV to the 98% isodose line. Smearing was

calculated according to the quadratic summation of the different

preparation and execution errors and was defined as 1.9 cm. The

plans were designed using two beams centered on the PTV. Each

beam was optimized independently with respect to PTV coverage

and to avoid the bladder and rectum. The beam angles were also

individually optimized to improve the PTV dose distribution and

minimize the rectal and bladder dose. The distance to the block

edge was 1 cm, except for around the rectum and bladder, where

the aperture was manually customized to minimize the dose to the

normal surrounding structures while maintaining the PTV coverage.

Specifically, the block edge distances were 7–8 mm in the posterior

aspect and 8–10 mm in the superior/bladder aspect. A dose of 78 GE

was prescribed in two GE daily fractions to the PTV. The target cov-

erage requirements were that 95% of the PTV would receive 100%

of the dose and that the minimal dose to the prostate (CTV) would

be 99% of the prescribed dose. The doses were calculated using a

relative biologic effectiveness of 1.1.

Position changesFor each case, a second image data set was created from the plan-

ning CT information. Using the bony anatomy and soft tissue as

guides, both identical data sets were fused to the MRI scan using

Eclipse software (Varian Medical Systems). Identical isocenter po-

sitions for the prostate were verified. In each case, the prostate vol-

ume was then repositioned relative to the original data set and to the

remainder of the pelvic anatomy. In total, 11 new prostate positions

were generated for each patient using each of the following unidirec-

tional shifts: 5 mm anterior, 5 mm inferior, 5 mm posterior, 5 mm

superior, 10 mm inferior, 10 mm posterior, and 10 mm superior.

Four additional prostate (CTV) positions were defined by tridirec-

tional shifts: Point A, 10 mm superiorly, 10 mm posteriorly, and

10 mm left; Point B, 10 mm superiorly, 10 mm anteriorly, and 10

mm right; Point C, 10 mm inferiorly, 10 mm posteriorly, and 10

1494 I. J. Radiation Oncology d Biology d Physics Volume 70, Number 5, 2008

mm right; and Point D, 10 mm inferiorly, 10 mm anteriorly, and 10

mm left. These 11 positions, determined according to the unidirec-

tional shifts, were surrogates for the prostate movements that occur

during and between treatment fractions. The beam was realigned in

the verification mode to correspond to prostate displacements of 5

and 10 mm.

Statistical analysisThe relative prostate coverage every 5 GE between 5 and 75 GE

and every 2 GE between 78 and 82 GE were used to completely de-

fine the dose–volume coverage for each prostate position. The min-

imal dose, maximal dose, and mean dose to the prostate for each

position was also analyzed. The rectal, bladder, femoral head, and

total pelvic doses were used to calculate the dose to the surrounding

normal structures. Analysis of variance was used to test the differ-

ences in coverage between the different treatment positions. Line

graph representations with standard errors were used for graphic

representation of the differences. Post hoc analyses were done using

least significant difference pair-wise multiple comparison tests.

Least significant difference pair-wise multiple comparison was

used to test the significance between the different prostate positions

and the initial prostate position. Beam realignment was done for all

5- and 10-mm displacements. The Student paired t test was used to

compare the prostate coverage with or without beam realignment. A

two-tailed p value of #0.05 was considered statistically significant.

Statistical analysis was performed with SYSTAT, version 11.0

(SPSS, Chicago, IL).

Table 1. Patient characteristics

Characteristic Value

Prostate volume (cm3)Mean volume 72.7Median 78.795% CI 60.9–84.5Range 45–95

Prostate length (cm)Mean 5.0Median 5.1

PTV (cm3)Mean 151.1Median 161.095% CI 131.6–168.4Range 106–181

PTV length (cm)Mean 7.0Median 7.0

Rectal volume (cm3)Mean 180Median 18895% CI 160–201Range 79–222

Rectal length (cm)Mean 13Median 13

Bladder volume (cm3)Mean 236.1Median 214.595% CI 188–284Range 88–442

Abbreviations: CI = confidence interval; PTV = planning targetvolume.

RESULTS

The patient characteristics are given in Table 1. The PTV

was, on average, twice the size of the prostate volume and

one-half the rectal length. Small changes in the PTV margins

significantly increased the PTV and, therefore, the irradiated

volume. For the initial prostate position, coverage to the pre-

scribed dose (percentage of prostate receiving 78 Gy [V78])

was >99.9% � 0.008%, the minimal dose was 78.17 �0.11 GE, the maximal dose was 81.30 � 0.96 GE, and the

mean dose was 79.60 � 0.35 GE.

Dose to normal structuresThe doses to the surrounding pelvic structures are given in

Table 2. Low doses were delivered to the rectum, bladder,

femoral heads, and pelvic tissue.

Prostate coverage with motionWithin the PTV. The dose–volume curves can be seen for

different prostate positions in Figs. 1–3. Excellent coverage

of the prostate was achieved when the prostate position

shifted only 5 mm in every direction (Fig. 1). The histogram

curve for the initial prostate location overlapped the curves

for the four 5-mm prostate shifts. The prostate (CTV) cover-

age only decreased after 78 GE, at which point the plot shoul-

ders began to curve downward. When the prostate target was

moved #5 mm, we observed >99% coverage to the pre-

scribed dose, and no statistically significant difference com-

pared with the initial prostate position (Table 3 and Fig. 4).

The minimal, maximal, and mean dose were similar to the

values obtained with the initial prostate position (Table 4

and Figs. 5 and 6).

Table 2. Percentage of volume of rectum and bladderreceiving doses of 10–80 GE and mean dose (n = 20 plans)

Relative volume SD

RectumV10 (%) 29.8 5.6V30 (%) 20.7 3.9V50 (%) 14.6 3.0V70 (%) 7.9 1.8V78 (%) 2.9 1.2V80 (%) 0.1 0.3Mean dose (GE) 14.2 3.7

BladderV10 (%) 36.4 13.2V30 (%) 27.7 11.1V50 (%) 21.5 9.1V70 (%) 14.0 7.2V78 (%) 7.6 5.1V80 (%) 2.1 3.0Mean dose (GE) 18.4 6.2

Femoral head (%)V35 11.1 10V50 0.1 0.3

Mean pelvic dose (GE) 4.6 0.9

Abbreviations: SD = standard deviation; GE = Gray equivalent;V10, V30, V35, V50, V70, V78, V80 = percentage of target volume re-ceiving 10, 30, 35, 35, 50, 70, 78, 80 Gy, respectively.


Fig. 1. Dose–volume curve for initial prostate position and prostate positions at 5 mm for 1 case.

Outside the PTV. Inferior prostate coverage was seen

when the prostate position moved 10 mm (Fig. 2). Compared

with the initial prostate position, the volume receiving 78 GE

was moderately reduced (Table 3). Slightly larger changes

were seen for the prostate when it moved posteriorly rather

than superiorly or inferiorly. The minimal dose to the prostate

volume was also considerably decreased with these shifts

(Table 4). However, the maximal and mean doses were not

significantly altered, suggesting that these two parameters

might not be ideal to assess adequate CTV coverage (Table

4 and Fig. 6).

Combined motion. With multidirectional displacements

>10 mm, as defined by Points A–D, significant changes in

dose coverage were seen for the prostate (Tables 3 and 4

and Figs. 4–6). The best coverage achieved with such dis-

placement resulted in only 85% of the prostate receiving

the prescribed dose. Furthermore, the minimal dose de-

creased by 27 GE in the worst-case scenario (Fig. 5). As

seen with the 10-mm unidirectional shifts, the maximal and

mean doses poorly reflected the degree to which CTV cover-

age was compromised (Table 4 and Fig. 6).

Dose profilesThe dose profiles are illustrated in Fig. 7. The right–left

beam profile shows the beam dose distribution in the beam

direction. Given uncertainties in the correlation between the

Hounsfield units and stopping power (mean energy loss per

traveled path length [de/dx]), a correction curve was applied

to our planning system. Furthermore, the variability of the

correction de/dx curve and machine energy output were taken

into account when defining the range and modulation. As a

result, we extended our margins in the RL direction outside

the PTV by an additional 5–9 mm. The beam margins were

widest in this direction, which included the periprostatic

soft tissue (Fig. 7). The superoinferior profile shows the

changes with respect to the base and apex. A very sharp

dose fall off, only a few millimeters wide, was seen from

the 90% to 10% isodose lines. As the right side of the curve

Fig. 2. Dose–volume curves for initial prostate position and prostate positions at 10 mm for 1 case.


Fig. 3. Dose–volume curves for initial prostate position and prostate positions A–D for 1 case.

suggests, we commonly saw a small increase of 3–4% over

the prescribed dose in the CTV (Fig. 7, arrow). The third

line represents the anteroposterior direction, with the right

side of the curve correlating with the posterior direction.

This illustrates how the aperture margin can be reduced in

the posterior aspect to achieve greater rectal sparing. The

beam fall off was 8–9 mm at the depth of the prostate poste-

riorly toward the rectum, and the distance between 95% and

50% was 9 mm.

Beam realignmentThe beam position was realigned with the prostate move-

ment for 5- and 10-mn displacements (Table 5). For the 5-

mm shifts, minimal differences were seen in coverage with

or without beam realignment. However, small statistically

significant improvements in the minimal target dose were

Table 3. Relative volume of prostate covered by prescriptiondose (78 GE) and absolute changes for different positions

Mean (Median) 95% CI Change (%) p*

Initial 100 (100.0) 99.99–100.0 NA5 mm

Anterior 99.6 (99.7) 99.2–100 �0.4 0.7Inferior 99.95 (100.0) 99.8–100 �0.05 0.9Posterior 99.5 (99.8) 98.9–100 �0.5 0.6Superior 99.99 (100.0) 99.9–100 �0.01 0.9

10 mmInferior 96.6 (97.0) 95.8–97.4 �3.4 0.003Posterior 89.8 (89.5) 87.1–91.4 �10.2 <0.001Superior 94.4 (94.9) 93.1–95.8 �5.6 <0.001

PointA 83.8 (84.8) 80.5–86.9 �16.2 <0.001B 85.7 (86.7) 83.4–87.9 �14.3 <0.001C 82.6 (82.0) 79.8–85.4 �17.4 <0.001D 86.5 (86.2) 83.9–89.2 �13.5 <0.001

Abbreviations: GE = Gray equivalent; CI = confidence interval.* Step-wise comparisons and Fisher’s least-significant-difference

test for different pairs between initial position and new prostateposition.

seen with realignment in the anterior (p = 0.001) or posterior

(p = 0.008) direction. For 10-mm shifts, larger, significant

improvements were seen for beam realignment. In the infe-

rior, posterior, and superior positions, the prostate minimal

V78 (p < 0.001) and the minimal dose (p < 0.001) were sig-

nificantly improved for all cases.

DISCUSSION

Interfraction changes in prostate position within the pelvis

are considered the major source of dosimetric misses for

prostate irradiation (1, 2, 4–8, 11, 14–19); therefore, margin

Fig. 4. Relative prostate volume receiving prescription dose (78 Gyequivalent) in several positions: 1, initial; 2, 5-mm anterior; 3, 5-mminferior; 4, 5-mm posterior; 5, 5-mm superior; 6, 10-mm inferior; 7,10-mm posterior; 8, 10-mm superior; 9, Point A; 10, Point B; 11,Point C; and 12, Point D.


formulas have been heavily weighted to account for this error

(2, 6, 16). For proton therapy, dosimetric variation with pros-

tate displacement has not been adequately described. It is crit-

ical to understand how prostate dose coverage is altered by

motion along clearly defined vectors for (1) patient-specific

intra- and inter-fraction prostate motion profiles; and (2)the impact of the actual dose distribution in the CTV with

prostate motion to quantify the potential benefit of image-

Table 4. Minimal, maximal, and average dose to prostatewith absolute changes for different positions

Dose Mean (Median) 95% CI Change p*

MinimalInitial 78.2 (78.1) 77.9–78.4 0 NA5 mm

Anterior 76.3 (76.4) 75.3–77.2 �1.88 0.7Inferior 78.4 (78.1) 77.5–79.4 0.26 0.9Posterior 77.3 (77.3) 75.7–78.8 �0.89 0.6Superior 78.5 (78.3) 77.6–79.4 0.31 0.9

10 mmInferior 73.4 (72.6) 71.2–75.7 �4.74 0.003Posterior 66.4 (66.3) 60.9–72.0 �11.72 <0.001Superior 73.6 (72.9) 71.6–75.7 �4.53 <0.001

PointA 52.5 (53.9) 49.0–55.9 �25.72 <0.001B 53.6 (52.9) 50.0–57.‘ �24.61 <0.001C 50.7 (52.4) 45.7–55.7 �27.47 <0.001D 53.9 (53.0) 50.1–57.7 �24.22 <0.001

MaximumInitial 81.3 (81.0) 80.6–82.0 0 NA5 mm

Anterior 81.3 (81.0) 80.6–81.9 �0.020 0.9Inferior 81.2 (81.0) 80.5–81.9 �0.11 0.8Posterior 81.1 (80.9) 80.4–81.7 �0.245 0.5Superior 81.0 (80.9) 80.3–81.7 �0.29 0.5

10 mmInferior 81.2 (81.0) 80.5–81.9 �0.106 0.8Posterior 80.8 (80.8) 80.1–81.4 �0.525 0.2Superior 80.9 (80.8) 80.2–81.5 �0.431 0.3

PointA 80.6 (80.7) 80.2–81.1 �0.649 0.1B 81.1 (80.9) 80.5–81.6 �0.23 0.6C 81.2 (80.9) 80.6–81.8 �0.103 0.8D 81.3 (81.2) 80.7–82.0 0.022 0.9

MeanInitial 79.6 (79.6) 79.4–79.9 0 NA5 mm

Anterior 79.6 (79.6) 79.4–79.9 0.010 0.9Inferior 79.0 (79.5) 77.8–80.2 �0.58 0.6Posterior 79.2 (79.5) 78.5–79.8 �0.43 0.7Superior 78.8 (79.5) 77.1–80.4 �0.84 0.5

10 mmInferior 76.9 (79.4) 71.2–82.6 �2.68 0.02Posterior 78.1 (78.8) 76.3–80.0 �1.48 0.2Superior 79.2 (79.3) 79.1–79.4 �0.35 0.8

PointA 78.5 (78.6) 78.2–78,8 �1.07 0.4B 78.7 (78.6) 78.4–78.9 �0.95 0.4C 78.4 (78.3) 78.1–78.7 �1.22 0.3D 78.7 (78.9) 78.5–79.0 �0.86 0.5

Abbreviations: CI = confidence interval; NA = not applicable.* Step-wise comparisons and Fisher’s least-significant-difference

test for different pairs.

guided therapy and the minimal PTV expansion necessary

to maximally benefit from the dose distribution advantages

of proton therapy.

As a result of the use of two fields and lateral oblique beam

angles, the dose distribution shapes for our treatments vary

Fig. 5. Minimal prostate dose in several positions: 1, initial; 2, 5-mm anterior; 3, 5-mm inferior; 4, 5-mm posterior; 5, 5-mm superior;6, 10-mm inferior; 7, 10-mm posterior; 8, 10-mm superior; 9, PointA; 10, Point B; 11, Point C; and 12, Point D.

Fig. 6. Mean prostate dose in several positions: 1, initial; 2, 5-mmanterior; 3, 5-mm inferior; 4, 5-mm posterior; 5, 5-mm superior; 6,10-mm inferior; 7, 10-mm posterior; 8, 10-mm superior; 9, Point A;10, Point B; 11, Point C; and 12, Point D.


Fig. 7. Dose profiles in different beam directions at beam isocenter: 1, superoinferior (SI); 2, anteroposterior (AP); and3, right–left (RL).

from the traditionally seen isodose lines for intensity-modu-

lated radiotherapy. For proton therapy, the dose line distribu-

tions will cover the PTV with a more square-shaped dose.

Within the PTV, dose homogeneity will be great, with

a dose variation of 95–104% at any point. However, a 9%

dose variation seen in the PTV, as well as a sharp fall off

from the 80% to 20% isodose line, can affect CTV coverage

in different positions. To fully evaluate the coverage of the

CTV for different positions inside and outside the PTV, it

is insufficient to evaluate the isodose lines within the PTV.

Thus, the CTV needs to be repositioned and the coverage an-

alyzed for the different motion vectors.

Motion within our PTVZhang et al. (20) was able to show inferior CTV coverage

without an image-guided approach for proton therapy that re-

sulted from prostate movement. Reductions of $10% in the

volume treated to the prescribed dose were seen in 30% of

cases. Quantification of the reduction in terms of absolute

changes in position, however, was not available. In our study,

we were able to show that prostate motion of #5 mm, or

within our PTV margin, allowed coverage of our prescription

dose to 99.8% of the volume (Table 3). Furthermore, as dem-

onstrated in Fig. 1, our PTV dose–volume histogram curves

for prostate motion overlapped with the curve for our original

position CTV. As a result, the minimal doses to the CTV

were, on average, decreased only 0.6 GE. Thus, systematic

motion of #5 mm will have a small effect on the dose deliv-

ered. This was because of the shape of the isodose lines as

they conform to the PTV. As seen in Fig. 1, the CTV dose

curve was very steep with a small shoulder. As a result, the

minimal dose for the initial CTV was $78 GE, which wais

maintained with motion #5 mm. However, it is important

to realize that movements of the CTV of 8 mm in the super-

oinferior dimension—still within the PTV—will result in

larger reductions in coverage than the isodose lines falling in-

side the PTV in this axis. This information could also be use-

ful to quantify the coverage of the prostate after daily image

guidance in which the intrafraction error will likely be within

Table 5. Prostate coverage with and without beamrealignment for different prostate positions

Beam realignment (SD)

Coverage No Yes

5-mm AnteriorProstate V78 (%) 99.6 (0.5) 100 (0.03)Prostate minimal dose (GE) 76.52 (1.17) 78.15 (0.27)

5-mm InferiorProstate V78 (%) 99.6 (0.5) 100 (0.03)Prostate minimal dose (GE) 78.03 (0.34) 78.19 (0.23)

5-mm PosteriorProstate V78 (%) 99.4 (0.8) 100 (0.007)Prostate minimal dose (GE) 76.75 (1.49) 78.29 (0.30)

5-mm SuperiorProstate V78 (%) 99.9 (0.02) 100 (0.004)Prostate minimal dose (GE) 78.10 (0.30) 78.27 (0.32)

10-mm InferiorProstate V78 (%) 96.5 (1.2) 100 (0.1)Prostate minimal dose (GE) 72.47 (0.90) 78.07 (0.27)

10-mm PosteriorProstate V78 (%) 89.8 (3.9%) 100 (0.1)Prostate minimal dose (GE) 64.75 (5.90) 78.31(0.53)

10-mm SuperiorProstate V78 (%) 94.4 (2.0) 100 (0.3)Prostate minimal dose (GE) 72.78 (0.70) 78.28 (0.41)

Abbreviations as in Table 2.


5 mm. However, as mentioned by Litzenberg et al. (2), mar-

gins of $5 mm are needed to account for inter-setup error.

An alternate option to deliver the prescribed dose for

interfraction motion >5 mm would be to increase the PTV

margins. However, increasing the PTV margins would com-

promise the low dose delivered to the rectum and bladder

(Table 2). Expanded margins will increase the normal tissue

dose and the risk of toxicity (13, 21, 22). As seen in Table 5,

the beam realignment can have small, but significant, bene-

fits, even for 5-mm movements.

Motion outside our PTVMargins of approximately 10 mm are necessary for the

treatment of prostate cancer without off-line or on-line image

guidance (2, 4–6). This distance served as the basis of our

theoretical prostate shifts. After a 10-mm displacement of

the CTV, 93.6% of the volume received the target dose (Ta-

ble 2). The largest changes were seen with posterior move-

ment, likely attributable to our reduced (7–8-mm) posterior

margin (Fig. 7). Because tissue densities vary across the

PTV, and we had a relatively broader region uncovered in

the posterior aspect, we could selectively decrease our poste-

rior aperture margin distance. This allowed us to achieve very

good coverage with a small (#5 mm) posterior shift toward

the edge of the PTV. However, posterior displacements

>5 mm would place a large percentage of the CTV in the

low-dose region, substantially decreasing the coverage. As

a result, when we moved the CTV 10 mm in the superior

and inferior directions, the minimal dose was only decreased

by 4.6 GE or 5.9% of the prescribed dose (78 GE), but the

CTV coverage was reduced by 11.7 GE or 15% with poste-

rior shifts of 10 mm.

When a proton target is displaced in the beam direction, the

effect is different than that seen with intensity-modulated ra-

diotherapy. Theoretically, prostate motion into the beam will

not compromise the coverage in the same manner as would

setup errors. However, changes in the position of the prostate

can affect target coverage by altering the amount of tissue

a proton beam must transverse. As a result, motion in-line

with the beam direction was only defined when we analyzed

the combined inter- and intrafraction prostate motions (23).

Because movement along the anteroposterior vector is seen

in prostate cancer, it is likely that the underdosing found in

our study with posterior displacements would be reflected

in decreased coverage during a course of proton radiotherapy

(24). A large benefit could be realized by using image guid-

ance for movements outside the PTV. However, it is neces-

sary to account for variations in stopping power in beam

realignment to optimize prostate coverage in an image-

guided environment.

Inter- and intrafraction motionPoints A–D reflect motion due to interfraction error com-

pounded by intrafraction motion. All four points were used

to evaluate the results of the 10-mm movements in the three

axes. This displacement distance was chosen on the basis of

the prostate motion analysis performed by Litzenberg et al.

(2). Kupelian et al. (9) demonstrated that intrafraction dis-

placement >5 mm and lasting $30 s occurred during the

course of therapy in 15% of fractions. As a result, the treated

volume covered by the prescription dose dramatically de-

creased (range, 82.6–85.7%). By accounting for potentially

large interfraction errors with on-line corrections, we can

minimize the PTV expansion required to account for intra-

fraction motion. Therefore, prostate coverage is influenced

largely by the degree of the displacement. The minimal

dose to the CTV was also greatly reduced (range, 24.2–

27.4 GE).

The average and maximal CTV doses are poor reflections

of the varying target coverage. The minimal change in either

descriptive measure corresponded to dramatic changes in our

dose–volume histograms. This raises questions about the

utility of average or maximal dose reporting when attempting

to describe the relationship between target motion and CTV

coverage.

Although real benefits have been obtained through dose

escalation, we have continued to see target underdosage

when patients were treated without image guidance. This

highlights two different issues: (1) image guidance is pivotal

for external radiotherapy delivery, including proton therapy,

if truly high doses are to be delivered; and (2) the true dose

necessary to treat prostate cancer patients with different

risk factors remains unknown for image-guided treatment.

The Radiation Therapy Oncology Group recently opened

a key study comparing 75.6 Gy in 1.8-Gy fractions and 70

Gy in 2.5-Gy fractions (biologically effective dose, 84.5

Gy [a/b = 1.5 in 1.8 Gy]) with image guidance. At our insti-

tution, we have adopted an image guidance protocol that

includes intraprostatic fiducial markers and daily X-ray local-

ization in the true beam direction before treating each field.

Strategies for proton therapyOrgan motion has been a major concern in the clinical ap-

plication of proton therapy. Because the high-dose area shape

and depth are determined by the calculation of the cumulative

stopping power through the tissue in the beam path, the var-

iations in stopping power with beam realignment is of con-

cern. Two major treatment philosophies can be used. First,

if changes in the stopping power are to be avoided to maintain

conformality in a static image, the margins should be in-

creased to account for all prostate motion, at the cost of

greater normal tissue doses. However, if the goal is to reduce

the normal tissue doses by using tighter margins and daily

confirmation of pretreatment prostate position, the dosimetry

should be optimized for coverage of the different prostate po-

sitions through dynamic position representation. In this situ-

ation, changes in the stopping power relative to the beam path

should be accounted for. We have used the latter philosophy.

The dose distribution in proton therapy is determined by

the different tissues along the beam path. Thus, a beam actu-

ally traversing a longer path in bone than modeled will de-

posit its energy short of the target. Furthermore, a beam in

air will travel longer and deposit its energy beyond the target.

Given this uncertainty in range, the beam orientation must be


optimized to limit delivering the dose to normal structures be-

fore or after the target. Therefore, the ideal beam orientation

will be determined by a path to the target that minimizes

interaction with dose-limiting normal structures.

Image guidance for proton therapy will only determine the

x and y axes, and the depth at which the dose will be delivered

is determined by the range calculated during the treatment

planning process. More specifically, the depth of the high-

dose regions is determined at planning by the range and

modulation. Three different options are then possible: (1) de-

termine the depth of the target on a daily basis and change the

range and modulation accordingly; (2) reduce prostate mo-

tion; and (3) determine the motion in the beam direction and

quantify the changes in radiologic path length and incorporate

this information into the treatment plan. The first option will

require smaller proximal and distal margins around the PTV

that are not feasible. Daily cone-beam CT scans will be nec-

essary to determine the changes in the radiologic path length,

in addition to the target position. The second option, although

ideal, has certain limitations. Rectal balloons have not been

found to decrease interfraction motion, the greatest source of

error (25). Intrafraction motion with a rectal balloon needs

further study, preferably incorporating cine-MRI informa-

tion. The third option is feasible as long as the variation in

depth of the target is relatively small. The prostate has rela-

tively large vectors of anteroposterior and superoinferior

motion but very little motion in the right–left direction cor-

responding to the beam orientation (7, 26, 27). At our institu-

tion, we have implemented the third strategy.

The second problem for image guidance results from

changes in the prostate position in relation to the bony anat-

omy and other pelvic structures. Differences in prostate

position will determine different radiologic path lengths for

the beam (28). Thus, the changes in density of the surround-

ing structures also must be determined and incorporated

into the compensator design (28, 29). A relatively simple

approach consists of analyzing all sources of preparation

and execution error in a defined plane perpendicular to

the beam direction. Thus, a quadratic summation of all

sources of error in a given direction should determine the

smearing. We have been using daily image guidance before

each beam with an orthogonal pair of X-rays, including

a true beam’s eye view, which decreases the sources of

execution error. A smearing of 1.9 cm was determined to

be necessary. This decreased plan conformality in a single

static image. However, coverage was maintained with this

strategy, with prostate positional changes of 5 and 10 mm

(Table 5).

CONCLUSION

Excellent PTV coverage with minimal normal structure

exposure can be achieved with planning using on a single

prostate position. Optimal dose coverage was maintained

for small prostate shifts; however, larger movements dramat-

ically decreased the prostate dose. Off-line or on-line image-

guided approaches are necessary for proton treatment of

prostate cancer if genuinely greater doses are to be delivered

to the tumor, as illustrated by our data.

REFERENCES

1. Letourneau D, Martinez AA, Lockman D, et al. Assessment of

residual error for online cone-beam CT-guided treatment of

prostate cancer patients. Int J Radiat Oncol Biol Phys 2005;

62:1239–1246.2. Litzenberg DW, Balter JM, Hadley SW, et al. Influence of intra-

fraction motion on margins for prostate radiotherapy. Int JRadiat Oncol Biol Phys 2006;65:548–553.

3. Rudat V, Schraube P, Oetzel D, et al. Combined error of patient

positioning variability and prostate motion uncertainty in 3D

conformal radiotherapy of localized prostate cancer. Int J RadiatOncol Biol Phys 1996;35:1027–1034.

4. Scarbrough TJ, Golden NM, Ting JY, et al. Comparison of

ultrasound and implanted seed marker prostate localization

methods: Implications for image-guided radiotherapy. Int JRadiat Oncol Biol Phys 2006;65:378–387.

5. Serago CF, Buskirk SJ, Igel TC, et al. Comparison of daily meg-

avoltage electronic portal imaging or kilovoltage imaging with

marker seeds to ultrasound imaging or skin marks for prostate

localization and treatment positioning in patients with prostate

cancer. Int J Radiat Oncol Biol Phys 2006;65:1585–1592.6. Yan D, Lockman D, Brabbins D, et al. An off-line strategy for

constructing a patient-specific planning target volume in adap-

tive treatment process for prostate cancer. Int J Radiat OncolBiol Phys 2000;48:289–302.

7. Schallenkamp JM, Herman MG, Kruse JJ, et al. Prostate posi-

tion relative to pelvic bony anatomy based on intraprostatic

gold markers and electronic portal imaging. Int J Radiat OncolBiol Phys 2005;63:800–811.

8. de Boer HC, van Os MJ, Jansen PP, et al. Application of the no

action level (NAL) protocol to correct for prostate motion based

on electronic portal imaging of implanted markers. Int J RadiatOncol Biol Phys 2005;61:969–983.

9. Kupelian P, Willoughby T, Mahadevan A, et al. Multi-institu-

tional clinical experience with the Calypso system in localiza-

tion and continuous, real-time monitoring of the prostate

gland during external radiotherapy. Int J Radiat Oncol BiolPhys 2007;67:1088–1098.

10. Madsen BL, Hsi RA, Pham HT, et al. Intrafractional stability of

the prostate using a stereotactic radiotherapy technique. Int JRadiat Oncol Biol Phys 2003;57:1285–1291.

11. Serago CF, Chungbin SJ, Buskirk SJ, et al. Initial experience

with ultrasound localization for positioning prostate cancer pa-

tients for external beam radiotherapy. Int J Radiat Oncol BiolPhys 2002;53:1130–1138.

12. Kestin L, Goldstein N, Vicini F, et al. Treatment of prostate can-

cer with radiotherapy: Should the entire seminal vesicles be

included in the clinical target volume? Int J Radiat Oncol BiolPhys 2002;54:686–697.

13. Vargas C, Yan D, Kestin LL, et al. Phase II dose escalation

study of image-guided adaptive radiotherapy for prostate can-

cer: use of dose–volume constraints to achieve rectal isotoxic-

ity. Int J Radiat Oncol Biol Phys 2005;63:141–149.14. Wu J, Haycocks T, Alasti H, et al. Positioning errors and pros-

tate motion during conformal prostate radiotherapy using

on-line isocentre set-up verification and implanted prostate

markers. Radiother Oncol 2001;61:127–133.


15. Willoughby TR, Kupelian PA, Pouliot J, et al. Target localiza-tion and real-time tracking using the Calypso 4D localizationsystem in patients with localized prostate cancer. Int J RadiatOncol Biol Phys 2006;65:528–534.

16. Van Herk M, Remeijer P, Rasch C, et al. The probability of cor-rect target dosage: Dose–population histograms for derivingtreatment margins in radiotherapy. Int J Radiat Oncol BiolPhys 2000;47:1121–1135.

17. van der Heide UA, Kotte AN, Dehnad H, et al. Analysis of fi-ducial marker-based position verification in the external beamradiotherapy of patients with prostate cancer. Radiother Oncol2007;82:38–45.

18. Van den HF, Powell T, Seppi E, et al. Independent verificationof ultrasound based image-guided radiation treatment, usingelectronic portal imaging and implanted gold markers. MedPhys 2003;30:2878–2887.

19. Trichter F, Ennis RD. Prostate localization using transabdomi-nal ultrasound imaging. Int J Radiat Oncol Biol Phys 2003;56:1225–1233.

20. Zhang X, Dong L, Lee AK, et al. Effect of anatomic motion onproton therapy dose distributions in prostate cancer treatment.Int J Radiat Oncol Biol Phys 2007;67:620–629.

21. Vargas C, Martinez A, Kestin LL, et al. Dose–volume analysisof predictors for chronic rectal toxicity after treatment of pros-tate cancer with adaptive image-guided radiotherapy. Int JRadiat Oncol Biol Phys 2005;62:1297–1308.

22. Sohn M, Yan D, Liang J, et al. Incidence of late rectal bleedingin high-dose conformal radiotherapy of prostate cancer using

equivalent uniform dose-based and dose–volume-based normaltissue complication probability models. Int J Radiat Oncol BiolPhys 2007;67:1066–1073.

23. Thomas SJ. Margins for treatment planning of proton therapy.Phys Med Biol 2006;51:1491–1501.

24. Padhani AR, Khoo VS, Suckling J, et al. Evaluating the effectof rectal distension and rectal movement on prostate gland po-sition using cine MRI. Int J Radiat Oncol Biol Phys 1999;44:525–533.

25. van Lin ENJT, van der Vight LP, Witjes JA, et al. The effect ofan endorectal balloon and off-line correction on the interfractionsystematic and random prostate position variations: A compar-ative study. Int J Radiat Oncol Biol Phys 2005;61:278–288.

26. Nederveen AJ, Dehnad H, van der Heide UA, et al. Comparisonof megavoltage position verification for prostate irradiationbased on bony anatomy and implanted fiducials. RadiotherOncol 2003;68:81–88.

27. Van Herk M, Bruce A, Kroes AP, et al. Quantification of organmotion during conformal radiotherapy of the prostate by threedimensional image registration. Int J Radiat Oncol Biol Phys1995;33:1311–1320.

28. Moyers MF, Miller DW, Bush DA, et al. Methodologies andtools for proton beam design for lung tumors. Int J Radiat OncolBiol Phys 2001;49:1429–1438.

29. Moyers MF, Miller DW. Range, range modulation, and field ra-dius requirements for proton therapy of prostate cancer. TechnolCancer Res Treat 2003;2:445–447.

proton therapy coverage for prostate cancer treatment

Documents