Localized volume effects for late rectal and anal toxicity after radiotherapy for prostate cancer

Int. J. Radiation Oncology Biol. Phys., Vol. 64, No. 4, pp. 1151–1161, 2006 Copyright © 2006 Elsevier Inc. Printed in the USA. All rights reserved 0360-3016/06/$–see front matter

doi:10.1016/j.ijrobp.2005.10.002

CLINICAL INVESTIGATION

Prostate

LOCALIZED VOLUME EFFECTS FOR LATE RECTAL AND ANAL TOXICITY AFTER RADIOTHERAPY FOR PROSTATE CANCER STEPHANIE T. H. PEETERS, M.D.,* JOOS V. LEBESQUE, M.D., PH.D.,* WILMA D. HEEMSBERGEN, M.SC.,* WIM L. J. VAN PUTTEN, M.SC.,† ANNERIE SLOT, M.D.,‡ MICHEL F. H. DIELWART, M.D.,§ AND PETER C. M. KOPER, M.D., PH.D.† *Department of Radiation Oncology, The Netherlands Cancer Institute–Antoni van Leeuwenhoek Hospital, Amsterdam, The Netherlands; †Department of Radiation Oncology, Erasmus Medical Centre, Rotterdam, The Netherlands; ‡Radiotherapeutic Institute Friesland, Leeuwarden, The Netherlands; §Zeeuws Radiotherapeutic Institute, Vlissingen, The Netherlands Purpose: To identify dosimetric parameters derived from anorectal, rectal, and anal wall dose distributions that correlate with different late gastrointestinal (GI) complications after three-dimensional conformal radiotherapy for prostate cancer. Methods and Materials: In this analysis, 641 patients from a randomized trial (68 Gy vs. 78 Gy) were included. Toxicity was scored with adapted Radiation Therapy Oncology Group/European Organization for the Research and Treatment of Cancer (RTOG/EORTC) criteria and five specific complications. The variables derived from dose–volume histogram of anorectal, rectal, and anal wall were as follows: % receiving >5–70 Gy (V5–V70), maximum dose (Dmax), and mean dose (Dmean). The anus was defined as the most caudal 3 cm of the anorectum. Statistics were done with multivariate Cox regression models. Median follow-up was 44 months. Results: Anal dosimetric variables were associated with RTOG/EORTC Grade >2 (V5–V40, Dmean) and incontinence (V5–V70, Dmean). Bleeding correlated most strongly with anorectal V55–V65, and stool frequency with anorectal V40 and Dmean. Use of steroids was weakly related to anal variables. No volume effect was seen for RTOG/EORTC Grade >3 and pain/cramps/tenesmus. Conclusion: Different volume effects were found for various late GI complications. Therefore, to evaluate the risk of late GI toxicity, not only intermediate and high doses to the anorectal wall volume should be taken into account, but also the dose to the anal wall. © 2006 Elsevier Inc. Prostate cancer, Radiotherapy, Toxicity, Volume effects, Dose–volume histograms.

We performed a multi-institutional, randomized phase III trial to investigate whether an additional dose of 10 Gy results in a better local control and survival in prostate cancer patients (1). A higher dose to the prostate implies that the surrounding organs at risk may also receive higher doses. In a first report we analyzed acute and late gastrointestinal (GI) and genitourinary complications in relation to general treatment factors and patient-related factors. Patients treated to 78 Gy had higher incidences of complications compared with those who received 68 Gy, but for only late rectal bleeding and late nocturia were these differences statistically significant. Other factors were more

important: History of abdominal surgery and the presence of pretreatment GI symptoms were strongly related to late GI toxicity. The main goal of the present analysis was to study dose–volume effects for late rectal toxicity in prostate cancer patients treated with three-dimensional conformal radiotherapy. The relationship between the dose to the (ano)rectum and Radiation Therapy Oncology Group/European Organization for the Research and Treatment of Cancer (RTOG/EORTC) Grade ⱖ2 end points and/or rectal bleeding has been studied extensively (2–19). In most of these studies, significant volume effects were found for the highdose regions; in some studies, significant effects were found for the intermediate-dose regions, as well.

Reprint requests to: Joos V. Lebesque, M.D., Ph.D., Department of Radiation Oncology, The Netherlands Cancer Institute, Antoni van Leeuwenhoek Hospital, Plesmanlaan 121, 1066 CX, Amsterdam, The Netherlands. Tel: (⫹31) 20-5122118; Fax: (⫹31) 206691101; E-mail: [email protected] Presented at the Forty-Seventh Annual Meeting of the American Society for Therapeutic Radiology and Oncology (ASTRO), October 16 –20, 2005, Denver, CO.

This project was supported by the Dutch Cancer Society (NKB Grant NKI 98 –1830 and CKTO 96 –10). Acknowledgments—The authors are grateful to all the data managers for their valuable contributions to this project, and want to thank H. Bartelink and M. Van Herk for critically reading the manuscript. Received Aug 11, 2005, and in revised form Oct 5, 2005. Accepted for publication Oct 6, 2005.

INTRODUCTION

1151

1152

I. J. Radiation Oncology

● Biology ● Physics

However, it has been shown that, besides rectal bleeding, other complications may be as bothersome or even more bothersome for the patients. According to Denham et al. (20), fecal urgency and bleeding have the highest impact on daily life. Koper et al. (10), on the other hand, have shown that patients were more bothered by symptoms such as soiling, fecal loss, and mucus discharge rather than blood loss, urge, and bowel cramps. Therefore, in the present analysis, we studied dose–volume effects not only for RTOG/EORTC Grade ⱖ2 and rectal bleeding, but also for anal incontinence, stool frequency, proctitis requiring the use of steroids, and pain, cramps, or tenesmus. Furthermore, a previous study done by our group on a different patient population (21) showed that different symptoms originated from different regions of the irradiated anorectal tract. We therefore analyzed also whether all these complications were more related to the exposure of particular regions of the lower GI tract, i.e., to the anorectal, rectal, or anal wall dose distributions. METHODS AND MATERIALS Patients and treatment Between June 1997 and February 2003, 669 patients with a localized adenocarcinoma of the prostate were entered in a multicenter phase III trial that randomized patients between 68 Gy and 78 Gy. The details of the study have been described extensively before (1, 22). Stratification was done by hospital (A, B, C, D), hormonal therapy (yes vs. no), age, and four treatment groups. These treatment groups were defined according to the risk of involvement of the seminal vesicles, according to Partin et al. (23). In previous reports, these treatment groups were called “dose– volume groups” (1, 22). Patients with a T1b, T1c, and T2a prostate cancer with a risk of involvement of ⬍10%, 10 –25%, and ⬎25% were included in treatment Groups I, II, and III, respectively. All patients with a T2b and T3a were treated in Group III. Group IV comprised all patients with a T3b or T4. For each treatment group, specific planning target volumes were defined. For treatment Groups I and IV, the clinical target volumes (CTVs) were defined as the prostate only and as the prostate with seminal vesicles, respectively. In treatment Groups II and III, the seminal vesicles were excluded from the CTV after 50 Gy and 68 Gy, respectively. For all patients, a margin of 10 mm was added to the CTV to obtain the planning target volume for the first 68 Gy, and this margin was reduced to 5 mm (except toward the rectum where no margin was taken) for the last 10 Gy in the 78 Gy arm. The prescription of neoadjuvant hormonal therapy was left to the discretion of the treating physician and was done mainly for patients of treatment Groups III and IV. It was started 0 –7 months before start of radiotherapy and prescribed for a total duration of 6 months or 3 years, depending on the hospital’s policy. All patients were scanned in treatment position (supine) with a slice thickness of 3 to 5 mm. Different three-dimensional conformal treatment techniques were used: 4-field technique (hospital C), 3-field techniques using 1 anterior and 2 lateral wedged fields (hospitals B and D) or 2 posterior oblique wedged fields (hospital A), and intensity-modulated radiation therapy (hospital B). For the boost of 10 Gy in the high-dose treatment arm, similar techniques were used. The boost was given sequentially, except for 41 patients in hospital B, where a simultaneous integrated boost was

Volume 64, Number 4, 2006

given using intensity modulated radiation therapy (24). The dose was specified to the ICRU reference point and was delivered in fractions of 2 Gy. The participating centers were instructed to delineate the anorectum from the ischial tuberosities until the level of the inferior border of the sacroiliac joints, or when the rectum was no longer adjacent to the sacrum (25). The treatment plans (including CT scan, delineated structures, and dose distributions) of all patients from the 4 hospitals were transferred to a dosimetric and volumetric database. This database has been developed especially to perform extensive analyses of dose–volume relationships in large radiotherapy studies. The coordinating center (hospital B) reviewed all delineated structures for consistency with the protocol to reduce interobserver variability and achieve a more homogeneous delineation. When a deviation was found for the organs at risk, the delineated structure was adapted according to the guidelines. When the anal verge was situated below the inferior part of the ischial tuberosities, the anorectum was extended in caudal direction. The inner contour of the anorectal wall was then automatically constructed from the delineated outer wall surface, using the method of Meijer et al. (26) (Fig. 1). After that, the anal canal was obtained by taking the caudal 3 cm of the anorectum (measured in craniocaudal direction) (21, 27, 28). The remaining cranial part of the anorectum was defined as the rectum. All dose–volume histograms (DVHs) were recalculated by the database-analysis package with a grid size of 1 mm2. Relative DVHs of the anorectal wall, the rectal wall, and the anal wall were constructed for each patient, in dose bins of 0.5 Gy. According to the protocol, the percentage of the (ano)rectum receiving 74 Gy or more should not exceed 40%, and the dose to the small bowel should not be higher than 68 Gy. Therefore, 19 patients in the high-dose treatment arm received a lower dose. Sixteen more patients in the high-dose treatment arm also were treated with a lower dose (range, 68 –76 Gy) for various reasons (1): because of toxicity, because a patient requested it, or because of a technical problem. All dosimetric variables were derived from the actually delivered treatment, but comparisons of both treatment arms were based on intention to treat. Of the 669 randomized patients, 5 were excluded because they were not eligible or not treated. Follow-up data were available for 656 patients. Analyses of volume effects were done with 641 patients, because DVH data were missing for 15 patients.

Scoring of toxicity, end points, and analyzed variables Late gastrointestinal symptoms were classified according to a slightly modified RTOG/EORTC scoring system (1). More specific symptoms, called indicators, were also scored: (1) Rectal bleeding requiring laser treatment or transfusion of packed cells; (2) Use of incontinence pads for rectal loss of blood, mucus, or stools (requiring the use of pads more than twice a week); (3) Stool frequency (ⱖ6 times a day); (4) Proctitis requiring the use of steroids; (5) Pain, cramps, or tenesmus requiring medication. Scoring for at least one indicator resulted by definition in RTOG/ EORTC Grade ⱖ2. The five indicators, together with RTOG/ EORTC Grade ⱖ2 and ⱖ3, were chosen as the seven end points in this analysis on late GI toxicity. Late toxicity was scored at each follow-up visit from 120 days after start of radiotherapy. Follow-up visits were scheduled once every 3 months in the first year of follow-up, every 4 months in the second year, biannually in the following 3 years, and yearly thereafter. When a recurrence was diagnosed, further assessment of complications was omitted from

Localized volume effects after RT for prostate cancer

● S. T. H. PEETERS et al.

1153

The analyses with dose–volume parameters were performed on both treatment arms together.

Statistical analyses The complication rates were determined using Kaplan–Meier estimates, and the log–rank test was applied to compare subgroups. Patients were censored at relapse or last contact if without complications at that time point. The Cox proportional hazards regression (PHR) model was performed in a stepwise manner (based on the likelihood ratio statistic) to analyze whether the clinical variables and the dosimetric variables were significantly associated with late GI toxicity. First the clinical variables for each end point were tested in multivariate Cox PHR analysis. After that, the dosimetric parameters were all tested in Cox PHR analyses adjusted for the variable “hospital.” The dosimetric parameters were all tested separately because of their strong interdependence. Finally, the significant clinical variables were added to the multivariate models with dosimetric variables, but only for the part of the lower GI tract where the dosimetric parameters were significant. Because of the multiple tests that were performed, a two-tailed p value of ⱕ0.01 was considered to be statistically significant. For statistical analyses we used the software package SPSS for Windows software, release 10.0 (SPSS Inc., Chicago, IL).

RESULTS Late GI toxicity Patients’ characteristics are shown in Table 1. About 50% of the patients were treated in treatment Group III. Acute GI toxicity was seen in 48% of the patients. Of the 181 patients with a history of abdominal surgery, 82% had surgery of the lower abdomen (mainly appendectomy, repair of inguinal hernia, or staging laparoscopic lymph node dissection), 15% surgery of the upper abdomen (mainly cholecystectomy or gastric surgery), and 3% surgery of the lower and upper abdomen. The median follow-up time was 44 months. Cumulative incidences of late GI toxicity were similar

Fig. 1. Computed tomography scan of a typical patient showing the delineated outer contour of the anorectal wall and the automatically constructed inner wall contour in (A) sagittal view and in transverse view at (B) midrectal and (C) midanal canal level (white lines in A).

that moment on, except for patients with biochemical recurrences or patients with distant metastases only. The clinical variables recorded for every patient were hormonal therapy, age, diabetes mellitus, history of abdominal surgery, smoking, use of acetylsalicylic acids, pretreatment GI symptoms, and maximum acute GI toxicity. The RTOG scoring system for acute GI toxicity was used to score pretreatment GI symptoms (Grade ⬍2 vs. ⱖ2) and acute GI toxicity (Grade ⱕ1, 2 vs. 3). The dose–volume variables used for the analysis of late GI toxicity were the following: relative wall volumes receiving 5–70 Gy or more (V5–V70), maximum dose (Dmax), mean dose (Dmean), wall volume, and volume (including filling). All these variables were derived for the anorectum, the rectum, and the anal canal.

Table 1. Patients’ characteristics All patients (n ⫽ 656) Mean age (years) (range) Randomization arm 1 (68 Gy) Randomization arm 2 (78 Gy) Hospital A B C D Hormonal therapy Treatment group I II III IV History of abdominal surgery Pretreatment gastrointestinal symptoms Acute gastrointestinal toxicity Grade 2 Grade 3

68.7 (49–84) 326 (50%) 330 (50%) 399 (61%) 169 (26%) 69 (11%) 19 (3%) 141 (22%) 107 (16%) 130 (20%) 314 (48%) 105 (16%) 181 (23%) 13 (2%) 285 (43%) 34 (5%)

1154

I. J. Radiation Oncology

● Biology ● Physics

Volume 64, Number 4, 2006

Table 2. Cumulative incidences at 4 years and number of events for all late GI toxicity end points. The p value (log–rank) gives the difference between both treatment arms for each end point Cumulative incidence at 4 years % (standard error)

Number of events

End points

68 Gy n ⫽ 326

78 Gy n ⫽ 330

Total n ⫽ 656

p

68 Gy n ⫽ 326

78 Gy n ⫽ 330

Total n ⫽ 656

RTOG/EORTC GI ⱖ2 RTOG/EORTC GI ⱖ3 Rectal bleeding Incontinence pads Stool frequency Use of steroids Pain/cramps/tenesmus

22.8 (2.7) 3.3 (1.0) 3.6 (1.3) 6.5 (1.6) 6.6 (1.7) 4.7 (1.4) 8.6 (1.8)

28.4 (2.9) 5.1 (1.5) 9.1 (1.9) 12.2 (2.2) 9.2 (1.7) 7.3 (1.8) 10.4 (1.9)

25.7 (2.0) 4.2 (1.0) 6.4 (1.2) 9.4 (1.3) 7.9 (1.2) 6.0 (1.1) 9.5 (1.3)

0.2 0.4 0.02 0.03 0.1 0.4 0.4

65 (20%) 9 (3%) 9 (3%) 18 (6%) 17 (5%) 12 (4%) 24 (7%)

84 (26%) 13 (4%) 22 (7%) 34 (10%) 29 (9%) 17 (5%) 31 (9%)

149 (22%) 22 (3%) 31 (5%) 52 (8%) 46 (7%) 29 (4%) 55 (8%)

Abbreviations: GI ⫽ gastrointestinal; RTOG ⫽ Radiation Therapy Oncology Group; EORTC ⫽ European Organization for Research and Treatment of Cancer.

compared with the previous analysis, where the median follow-up time was 31 months. Table 2 shows for both treatment arms the cumulative incidences of late GI toxicity at 4 years, as well as the crude incidences for all end points. The incidences were always higher in the high-dose treatment arm, but the differences did not reach the significance level of 0.01. The largest differences between both arms were seen for rectal bleeding (p ⫽ 0.02) and incontinence pads (p ⫽ 0.03) (Fig. 2, left column). The cumulative incidence of rectal bleeding at 4 years was more than twice as high in the high-dose treatment arm (9.1%) compared with the low-dose arm (3.6%). For incontinence pads also, there was a large difference between both arms at 4 years (12.2% vs. 6.5%). The cumulative incidence of RTOG/ EORTC Grade ⱖ2 kept rising beyond 3 years of follow-up (Fig. 2A), whereas the cumulative incidence for rectal bleeding seemed to stabilize after 3 to 4 years (Fig. 2B). Scoring for 1 of the 5 indicators led by definition to RTOG/EORTC Grade ⱖ2. The most frequently scored indicators, incontinence pads and pain/cramps/tenesmus, were each scored by approximately one-third of the 149 patients with a Grade ⱖ2 (Table 2, last column). Rectal bleeding and use of steroids were less frequent, and each one was observed in about one-fifth of the patients with Grade ⱖ2. Figure 3 shows the number of indicators per patient that were responsible for scoring RTOG/EORTC GI Grade ⱖ2. Approximately one-third of the patients had more than one complaint, whereas the majority scored for only one indicator. Multivariate analysis with clinical variables A multivariate Cox PHR was performed to test the association of late GI toxicity with the clinical variables and the treatment arm. From all the tested variables, a history of abdominal surgery, the presence of pretreatment GI symptoms, and a higher acute GI toxicity were associated with a significantly higher incidence of toxicity in multivariate analysis (p ⱕ 0.01) (Table 3). Abdominal surgery was most strongly associated with RTOG/EORTC Grade ⱖ2, rectal bleeding, and incontinence pads. Pretreatment GI symptoms

showed a strong correlation with Grade ⱖ2, Grade ⱖ3, and pain/cramps/tenesmus. The hazard ratios for this variable were rather high, but the confidence intervals were also large, because only 13 patients had Grade ⱖ2 pretreatment GI symptoms. Acute GI toxicity was associated with Grade ⱖ2, stool frequency, and pain/cramps/tenesmus. The other investigated clinical variables (including hormonal therapy) were not related to late GI toxicity. The toxicity rate in hospital B, where a simultaneous integrated boost technique was used for some patients, also did not significantly differ from that in the other hospitals. The presence of more than one of the risk factors (Table 3) considerably increased the risk of toxicity. The hazard ratio for RTOG/EORTC Grade ⱖ2, for instance, was 1.8 for patients with a history of abdominal surgery and 1.6 for patients who had acute GI toxicity Grade ⱖ2. This means that for patients with a history of surgery and with acute GI toxicity, the hazard ratio for Grade ⱖ2 rose to 2.9 (i.e., 1.8 multiplied by 1.6) compared with patients without a history of surgery and without acute GI symptoms. Dose–volume characteristics The dose–volume characteristics for the anorectal, rectal, and anal wall are shown in Fig. 4A. There was only a small difference in the relative DVHs of anorectal and rectal wall. The DVH of the anal wall had a larger spread (i.e., larger standard deviation) compared with the DVHs of rectal and anorectal wall (Fig. 4A). The average Dmean of the anorectal wall (46 ⫾ 6 Gy) was nearly equal to that of the rectal wall (48 ⫾ 8 Gy), whereas the average Dmean for the anal wall was much lower (36 ⫾ 13 Gy) The average Dmax of the anorectal wall was identical to that of the rectal wall (73 ⫾ 5 Gy). The average Dmax of the anal wall was somewhat lower (68 ⫾ 10 Gy). The average anorectal, rectal, and anal wall volumes were 34 cm3, 28 cm3, and 6 cm3, respectively. The average anorectal, rectal, and anal volumes (including filling) were 85 cm3, 74 cm3, and 10 cm3. As expected, treatment arm and anorectal Dmax were strongly correlated (correlation coefficient 0.8). The average

Localized volume effects after RT for prostate cancer

● S. T. H. PEETERS et al.

Fig. 2. In the left column, Kaplan–Meier curves for both randomization arms are shown for the end points (A) Radiation Therapy Oncology Group/European Organization for the Research and Treatment of Cancer (RTOG/EORTC) gastrointestinal Grade ⱖ2, (B) rectal bleeding, and (C) incontinence pads. In the right column, the presence of dose–volume effects for the same end points is illustrated for one dosimetric variable divided into quartiles. For each end point, we chose one dosimetric variable that was strongly associated with this end point.

1155

1156

I. J. Radiation Oncology

● Biology ● Physics

Volume 64, Number 4, 2006

Fig. 3. The number and percentages of patients with Radiation Therapy Oncology Group/European Organization for the Research and Treatment of Cancer gastrointestinal Grade ⱖ2 are shown as a function of the number of indicators responsible for scoring this Grade ⱖ2 toxicity.

anorectal DVHs for both treatment arms are depicted in Fig. 4B. The major difference was seen in the high-dose region from 65 to 78 Gy, whereas below 20 Gy no difference in irradiated volume was seen. Between 20 and 65 Gy, the percentage of irradiated anorectal wall volume was slightly higher for patients in the high-dose treatment arm. No significant differences in wall volume, volume including filling, and organ length were seen between both arms. Multivariate analysis including dosimetric parameters Cox PHR analyses were performed with dosimetric variables derived from relative DVHs of anorectal, rectal, and anal wall to study the relationship of the seven late GI toxicity end points with these variables. Results for three of these end points are shown in Fig. 5. First, all models were adjusted for hospital (solid symbols). After that, the clinical variables that were significantly associated with that particular end point (see Table 3) were also included in the models, together with hospital (open symbols). The end point RTOG/EORTC Grade ⱖ2 was significantly associated with the anal parameters in the low- and intermediate-dose regions, as well as with anal Dmean (Fig. 5A). Adding the variables abdominal surgery and pretreatment GI symptoms increased the level of significance. Acute GI toxicity, on the other hand, was not significant

Fig. 4. Average cumulative dose–volume histograms of (A) anorectal, rectal, and anal wall; and of (B) anorectal wall for both treatment arms. Error bars indicate 1 standard deviation.

anymore in the presence of DVH parameters and was therefore not included in these models. To illustrate the impact of this volume effect, we plotted Kaplan–Meier curves of patients divided into four groups, the quartiles of anal Dmean (Fig. 2A, right column). Anal Dmean was chosen, because it was one of the most significant parameters. V5, V10, V30, and V35 were as, or even slightly more, significant, but Dmean is a more robust parameter, because it less dependent on setup variability and internal organ motion. The 4-year cumulative incidence of Grade ⱖ2 almost doubled (16% vs.

Table 3. Results of the multivariate Cox proportional hazards regression analysis showing the hazard ratios (with 95% confidence intervals) of the clinical variables that were significantly (p ⱕ 0.01) associated with one or more end points

RTOG/EORTC Grade ⱖ2 RTOG/EORTC Grade ⱖ3 Rectal bleeding Incontinence pads Stool frequency Use of steroids Pain/cramps/tenesmus Abbreviations as in Table 2.

Abdominal surgery

Pretreatment gastrointestinal symptoms

Acute gastrointestinal toxicity

1.8 (1.3–2.5) – 2.7 (1.3–5.5) 2.2 (1.2–3.9) – – –

4.1 (2.0–8.4) 7.0 (1.6–30.8) – – – – 6.4 (2.5–16.3)

1.6 (1.2–2.0) – – – 2.9 (1.8–4.5) – 1.9 (1.2–2.9)

Localized volume effects after RT for prostate cancer

Fig. 5. The p values of the dosimetric variables (obtained from Cox proportional hazards regression analysis) are plotted, showing the association between three late gastrointestinal toxicity end points and dosimetric variables derived from the anorectal or anal wall. (A) Radiation Therapy Oncology Group/European Organization for the Research and Treatment of Cancer (RTOG/EORTC) Grade ⱖ2 and (C) anal incontinence requiring pads were associated with anal wall parameters, whereas (B) rectal bleeding was related to anorectal wall parameters. Solid symbols ⫽ multivariate analysis adjusted for hospital; open symbols ⫽ multivariate analysis adjusted for hospital and significant clinical variables (mentioned between parentheses).

31%) for an increase of anal Dmean from 19 Gy to 52 Gy. For the anorectal and rectal wall parameters, no significant associations were found with RTOG/EORTC Grade ⱖ2. For rectal bleeding needing treatment, anorectal V55, V60, and V65 were significant (p ⱕ 0.01) (Fig. 5B). The most significant parameter was V65 (p ⫽ 0.004; hazard ratio for each increase of V65 with 1%: 1.052). Most rectal parameters were slightly less significant compared with the corresponding anorectal variables (results not shown). The inclusion of abdominal surgery in the models with the anorectal DVH parameters slightly decreased the level of significance (Fig. 5B, open symbols). The presence of a dose–volume effect for rectal bleeding is illustrated in Fig. 2B (right column). The cumulative incidences are plotted for the quartiles of anorectal V65. An increase of V65 with 24% (from 19% to 43%) resulted in a large increase of the 4-year incidence from 1% to 9%. There was hardly any difference between the curves of quartiles 3 and 4, indicating that keeping V65 beneath approximately 30% considerably reduced the risk of rectal bleeding.

● S. T. H. PEETERS et al.

1157

Regarding the end point incontinence for blood/mucus/ stools requiring pads, all anal dosimetric variables were statistically significant at a level of 0.01, except for Dmax (Fig. 5C). The most significant parameters were anal Dmean (p ⫽ 0.002; hazard ratio for each increase of Dmean by 1 Gy: 1.039) and anal V65 (p ⫽ 0.0004; hazard ratio for each increase of V65 with 1%: 1.033). None of the anorectal and rectal parameters were statistically significant, although there was a trend toward significance (p ⱕ 0.05) for rectal V70 and anorectal V65, V70, and Dmax. The addition of the variable abdominal surgery only slightly altered the significance level of the anal parameters (Fig. 5C, open symbols). The presence of a volume effect is shown for anal Dmean, the most significant parameter besides anal V65 (Fig. 2C, right column). An increase of anal Dmean by 33 Gy corresponded to an increased incidence of incontinence of 12% (5% vs. 17%) at 4 years. Concerning the end point stool frequency (ⱖ6 times a day), only anorectal V40 (p ⫽ 0.007; hazard ratio: 1.031) and anorectal Dmean (p ⫽ 0.01; hazard ratio: 1.062) reached the level of significance of 0.01. There was a trend toward significance (p ⱕ 0.05) for anorectal V25–V35 and V45– V70, rectal V25–V70, and Dmean. After inclusion of acute GI toxicity in the models with anorectal DVH parameters, none of the DVH parameters was significant anymore, whereas acute GI toxicity itself remained statistically significant. For proctitis requiring the use of steroids, a weak association was found with the anal V30 –V65 and anal Dmean (p values between 0.02 and 0.05), with the strongest association with anal V65 (p ⫽ 0.02; hazard ratio: 1.029) and anal Dmean (p ⫽ 0.03; hazard ratio: 1.033). Adding the variable abdominal surgery slightly increased the level of significance. The end points RTOG/EORTC Grade ⱖ3 and pain/ cramps/tenesmus were not found to be associated with any of the DVH parameters. Finally, none of the parameters wall volume, total volume including filling, or organ length was found to be significant in the performed analyses. DISCUSSION In this study, we found several dose–volume effects regarding various late GI complications in prostate cancer patients treated with radiotherapy. Besides the often-described volume effects for late rectal bleeding (Fig. 5B) and RTOG/EORTC GI Grade ⱖ2 (Fig. 5A) (2–11, 13–19), we also found correlations between dosimetric parameters and incontinence pads (Fig. 5C), stool frequency, and use of steroids. Rectal bleeding and stool frequency were related to anorectal wall dose distributions, whereas for RTOG/EORTC Grade ⱖ2, incontinence pads and use of steroids associations were found with the anal wall dose distributions. RTOG/EORTC GI Grade ⱖ2 Late RTOG/EORTC GI Grade ⱖ2 was essentially associated with anal dosimetric parameters, especially with anal wall volumes exposed to low and intermediate doses, and

1158

I. J. Radiation Oncology

● Biology ● Physics

Volume 64, Number 4, 2006

with the mean dose to the anal wall (Fig. 5A). Koper et al. (10) found also a significant correlation between late GI toxicity Grade ⱖ1 and the anal dosimetric variables (anal volume receiving ⱖ59.4 Gy) in multivariate analysis. Comparison with our study is difficult, because the authors investigated DVH parameters of only the high-dose regions of rectum and anus (with filling). We did not find a convincing volume effect for the DVH parameters derived from the anorectal and rectal wall, in contrast to several other studies (7, 8, 13, 16, 19). The reason why we did not find a strong association of anorectal parameters with RTOG/EORTC Grade ⱖ2 may lie in the definition of the end point itself. This type of scoring system includes various symptoms, whereas we have shown that depending on the complication, different parts of the irradiated lower GI tract may play a more important role (Fig. 5). In our opinion, a general toxicity scale such as the RTOG/EORTC should therefore not be used to study dose–volume effects, because this may result in some loss of information. In addition, nearly all patients with a Grade ⱖ2 also scored at least one indicator. This means that with the use of these five indicators, no information is lost compared with an RTOG/EORTC score only, and more details are known. Comparison of the graphs in Fig. 5 reveals that the significance of the anal DVH parameters for the end point RTOG/EORTC Grade ⱖ2 (Fig. 5A) is very likely caused by the volume effect encountered for the end point incontinence pads (Fig. 5C). Of the five indicators investigated, it had the second highest cumulative incidence (Table 2) and may therefore have a higher weight in the overall score. The highest cumulative incidence was seen for pain/cramps/ tenesmus, but in this case no volume effect was seen in any region of the lower GI tract.

thors found a significant association of rectal bleeding with the mean dose (4, 5, 17) and/or the maximum dose (5, 14) to the (ano)rectum. A problem that may hamper comparison of our study with the above-mentioned studies is the definition of the end point rectal bleeding. In most studies, Grade 2 rectal bleeding is defined as intermittent bleeding more than twice a week and Grade 3 as bleeding requiring laser treatment or transfusion (2, 4 – 6, 14, 18). Some authors defined any bleeding as the end point (9, 11). The definition of rectal bleeding we used was more stringent, because it was defined as bleeding requiring a laser treatment or transfusion, which is analogous to the definition used by Boersma et al. (3). Rancati et al. (29) studied the end point bleeding both for moderate/severe and severe bleeding alone (defined as bleeding requiring a single coagulation or transfusion). These two end points had a substantially different volume effect. With the Lyman model, the best fit for the volume parameter n was found with an n value of 0.24 for moderate and severe, whereas for severe bleeding the n value was a factor 4 lower (n ⫽ 0.06). This indicates that the dose– volume dependency of the (ano)rectum might differ, depending on the severity of bleeding. Different definitions of the end point bleeding may therefore lead to different dose– volume effects. This probably explains also the different results found by Heemsbergen et al. (21), because they studied dose–volume effects for “any bleeding.” They found the strongest relationship between any bleeding and the rectum excluding the anal canal, whereas we found a stronger correlation of severe bleeding with the anorectum compared with the rectum. The models with anorectal wall parameters in our study were, however, not statistically significantly better than the models with rectal wall parameters.

Rectal bleeding A frequently studied end point is rectal bleeding (2–5, 9, 14, 16 –18). We found an overall 4-year cumulative incidence of rectal bleeding needing treatment of 6.4%. Bleeding correlated strongly with anorectal wall volumes exposed to high doses, in particular with the percentage of the wall receiving 55 Gy, 60 Gy, and 65 Gy or more (V55, V60, and V65). The associations between bleeding and intermediatedose regions, mean and maximum dose were weaker (Fig. 5B). The hazard ratio for the most significant parameter V65 was 1.052. This means that the risk of rectal bleeding increased by factor of 1.66 (i.e., 1.052 to the power of 10) for an increase of V65 with 10%. The toxicity rate in the studied patient group at 4 years was only 1% when V65 was below 23%, whereas it was around 10% when V65 was 28% or more (Fig. 2B). The association of a higher incidence of rectal bleeding with larger (ano)rectal volumes irradiated to high doses has already been observed in many studies (3– 6, 9, 16 –18). In some of these studies (4, 6, 9, 16, 18), where lower doses were investigated, an association with more intermediate doses around V30 –V50 was found. Furthermore, some au-

Incontinence requiring pads Anal incontinence is another important end point, because it has been shown to impact on the patients’ quality of life (10). We defined this end point as incontinence for stools, mucus, or blood requiring the use of pads more than twice a week. In the present study, the cumulative incidence of incontinence requiring pads was 9.4% at 4 years. We found that incontinence was strongly related to all anal wall dosimetric parameters. For the mean dose to the anal wall, for instance, the hazard ratio was 1.039. This means that for an increase of Dmean with 33 Gy (from 19 Gy to 52 Gy), we would expect an increased incidence of incontinence by factor of 3.5 (i.e., 1.039 to the power 33). This concurs with the data shown in Fig. 2C (right column), where the incidence increased from 5% to 17%. This Fig. 2C shows also that when the mean dose to the anal wall was kept below 46 Gy, the incidence of incontinence requiring pads did not exceed 10%. Because nearly all anal dosimetric parameters were significantly related to incontinence, another anal parameter could have been chosen as well to illustrate the presence of a volume effect. This is due to the very strong correlations

Localized volume effects after RT for prostate cancer

between the anal parameters. We can conclude that, to keep the probability of anal incontinence low, the dose to the anal wall should be minimized whenever possible. We chose to use the parameter Dmean, because this is a rather robust and relatively simple parameter, but other parameters can be used as well. Yeoh et al. (30) have shown that radiotherapy for prostate cancer may cause progressive dysfunction of the anal sphincter and reduction in rectal compliance. They further suggested that these changes might contribute to fecal incontinence. Although fecal incontinence or soiling have a significant impact on quality of life (10, 31), only a few small studies have investigated the relationship between fecal incontinence (or soiling) and DVH parameters of rectum and anus in prostate cancer patients (31–34). Vordermark et al. (31) studied fecal incontinence through the use of a 10-item questionnaire. They found also a trend toward lower continence scores with higher doses or larger volumes irradiated to a certain dose, but only the association of the minimum dose to the anus and severe fecal incontinence was significant. Fokdal et al. (32) found a relationship between fecal incontinence Grade ⱖ1 and the maximum dose to the anal canal, but this study also included bladder cancer patients. In a first study by al-Abany et al. (33), an association was found between fecal leakage (twice a week or more) and the mean dose to the anal canal. A second study by the same group (34) showed the strongest relationship between fecal leakage and the percentage of the anal sphincter volume irradiated to ⱖ45 Gy and ⱖ55 Gy. In our analysis, these correlations were also found, because all anal dosimetric variables were found to be associated with incontinence (Fig. 5). Heemsbergen et al. used a slightly different approach studying dose–area parameters of different regions of the lower GI tract derived from dose maps (21). These dose maps were constructed by virtually unfolding the anorectum and projecting the dose onto a two-dimensional map (35). The authors found that soiling and fecal incontinence were associated with the inferior 40 –50% of a dose map. Because this lower 40 –50% of the dose map is situated in the anal region and the lower part of the rectum, these results are in agreement with the findings of this study. Effect of prescription dose on bleeding and incontinence We already mentioned that patients in the high-dose treatment arm tended to have higher incidences of rectal bleeding (Fig. 2B) and anal incontinence (Fig. 2C). We therefore analyzed also both treatment arms separately for these two end points, to investigate whether the volume effects we found on the whole patient group were present in both arms. This analysis resulted in a loss of power (results not shown). For rectal bleeding, the association of the anorectal DVH parameters was slightly stronger in the low-dose treatment arm compared with the high-dose arm. For incontinence, the association with anal dosimetric parameters was much stronger for patients included in the high-dose treatment arm (results not shown). Hence, the

● S. T. H. PEETERS et al.

1159

predictive value of the dosimetric variables also applies for patients treated to 78 Gy only. And furthermore, comparing the left and right column in Fig. 2 also indicates that the presented dosimetric variables are better predictors for complications compared with the prescribed dose. Stool frequency, use of steroids, and pain/cramps/tenesmus Stool frequency (ⱖ6 times a day) was related to the intermediate-dose region of the anorectum (around V40) and to the mean dose to the anorectum. However, these associations were somewhat weaker compared with the volume effects found for bleeding and incontinence. For the end point pain, cramps, or tenesmus requiring medication, none of the dosimetric variables was found to be significant. The absence of a dose–volume effect for these symptoms may be a real effect, but it is more likely that the end point was inadequate. Different mechanisms may be involved for these symptoms, and analyzing them together may result in loss of information. A similar remark can be made about the end point proctitis requiring the use of steroids, where only a weak association was found with anal DVH parameters. Steroids may have been prescribed for slightly different symptoms caused by different types of damage. Clinical variables predictive for late GI toxicity The importance of a history of abdominal surgery as an independent prognostic factor for developing late GI toxicity has already been stressed (1). We also confirmed that hormonal therapy did not impact on late GI toxicity, which is in accordance with the findings of Vargas et al. (36). It is, however, the first time that we analyzed the influence of acute GI toxicity on late GI toxicity using these patient data. To determine whether acute toxicity was an independent prognostic factor, DVH parameters had to be included in such an analysis. Acute GI toxicity was significantly related to RTOG/ EORTC Grade ⱖ2, pain/cramps/tenesmus, and stool frequency. For stool frequency, acute GI toxicity remained statistically significant after inclusion of significant dosimetric variables. These findings suggest that acute GI damage may contribute to some type of late GI damage, the so-called “consequential late damage” (19). We can, however, not exclude that these patients were predisposed to develop acute as well as late GI complications. Another explanation that cannot be ruled out is the nature of some patients to complain easily, and therefore to report both acute and late complications. Limitations of the study, recommendations, and future work One of the limitations of this study is that the effects of setup variability and internal organ motion during treatment were not taken into account, because the analysis was performed with DVHs obtained from the pretreatment planning CT scans. This may in some cases lead to overestimation or

1160

I. J. Radiation Oncology

● Biology ● Physics

underestimation of the actual exposure (35). This is probably less of a problem for the anal canal, because this structure shows less organ motion compared with the rectum (35). Another observation that can be made is that the delineation of the structures has been done on a CT scan only. CT-derived volumes have been shown to be larger than MR-derived volumes, not only for the prostate, but also for the (ano)rectum (37). The anorectal wall volume and the dose to the wall may therefore have been slightly overestimated in our study. The inner wall of the anorectum, on the other hand, was automatically constructed with the method of Meijer et al. (26). This model has been built on a rectum without the anal canal. For the present study, we assumed that the anal wall behaved in the same way as the rest of the anorectum, which implies that in every cross section, the wall volume is assumed to be the same in the whole anorectum. Recommendations regarding dose constraints can be derived from these results by weighing the possible benefit of a higher dose against the risk of complications that one is willing to accept. Based on the results of this study, an approach would be to use the three following dose–volume constraints during treatment plan optimization: (1) One anorectal DVH parameter from the high-dose region, because exposure to high doses is predictive for bleeding; (2) One anorectal DVH parameter from the intermediate-dose region, because these dose ranges were related to frequency; (3) Mean dose to the anal wall, because this was a strong predictor for incontinence. How to choose the weight of these three possibly conflicting constraints is still a point of discussion. According to the studies of Koper et al. (10) and those of al-Abany et al. (33), patients were most bothered by fecal leakage, but Denham et al. (20) found that both fecal urgency and bleeding had a similar impact on quality of life. Not only should quality of life but also the incidence of the specific complication be taken into account. We restricted this study to the analysis of parameters derived from DVHs. It may be more appropriate to use the equivalent uniform dose and normal tissue complication probability to define constraints, instead of some DVH parameters (38, 39). The equivalent uniform dose condenses the information of the whole DVH into one parameter

Volume 64, Number 4, 2006

taking into account the volumetric dependence of the dose– response relationship, which is represented by the volume parameter n. With a small value of n, the high doses play a more important role (more serially organized organs), whereas when n goes to 1, the mean dose to the organ is more important (parallel organized organs). The equivalent uniform dose does not, however, take into account spatial information. We plan an analysis on the data presented in this paper and fit the parameters of the normal tissue complication probability model for several end points. From the results we found with DVH parameters (Fig. 5), we expect that for the end point incontinence, for instance, the value of n for the anal wall dose distribution will be close to 1, whereas for bleeding, the n value for the anorectal wall is expected to be small. CONCLUSIONS Various localized dose–volume effects were found for different late GI complications. Rectal bleeding was more frequent when larger fractions of the anorectal wall were irradiated to high doses and, to a lesser extent, to intermediate doses. The strongest predictor of bleeding was the percentage of the anorectal wall irradiated to 65 Gy or more (V65). The association of stool frequency with anorectal parameters was a bit weaker compared with bleeding and was most apparent in intermediate-dose regions. Incontinence showed a strong association with anal wall dosimetric parameters. This means that to evaluate the risk of late GI toxicity, not only high doses, and to a lesser extent intermediate doses, to the anorectal wall volume should be taken into account, but also the dose to the anal wall should be included in the cost function during treatment plan optimization. In the analysis on stool frequency, acute GI symptoms remained significant in the presence of dose–volume parameters, suggesting “consequential late damage” may contribute to the occurrence of a higher stool frequency during follow-up. Finally, in our opinion, a general toxicity scale such as the RTOG/EORTC should not be used to investigate dose– volume effects, because this may result in loss of information.

REFERENCES 1. Peeters STH, Heemsbergen WD, van Putten WL, et al. Acute and late complications after radiotherapy for prostate cancer: Results of a multicenter randomized trial comparing 68 Gy to 78 Gy. Int J Radiat Oncol Biol Phys 2005;61:1019 –1034. 2. Akimoto T, Muramatsu H, Takahashi M, et al. Rectal bleeding after hypofractionated radiotherapy for prostate cancer: Correlation between clinical and dosimetric parameters and the incidence of grade 2 or worse rectal bleeding. Int J Radiat Oncol Biol Phys 2004;60:1033–1039. 3. Boersma LJ, van den Brink BM, Bruce AM, et al. Estimation of the incidence of late bladder and rectum complications after high-dose (70 –78 Gy) conformal radiotherapy for prostate cancer, using dose-volume histograms. Int J Radiat Oncol Biol Phys 1998;41:83–92.

4. Cozzarini C, Fiorino C, Ceresoli GL, et al. Significant correlation between rectal DVH and late bleeding in patients treated after radical prostatectomy with conformal or conventional radiotherapy (66.6 –70.2 Gy). Int J Radiat Oncol Biol Phys 2003;55:688 – 694. 5. Fiorino C, Cozzarini C, Vavassori V, et al. Relationships between DVHs and late rectal bleeding after radiotherapy for prostate cancer: Analysis of a large group of patients pooled from three institutions. Radiother Oncol 2002;64:1–12. 6. Fiorino C, Sanguineti G, Cozzarini C, et al. Rectal dosevolume constraints in high-dose radiotherapy of localized prostate cancer. Int J Radiat Oncol Biol Phys 2003;57:953– 962. 7. Greco C, Mazzetta C, Cattani F, et al. Finding dose-volume

Localized volume effects after RT for prostate cancer

8. 9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20. 21.

22.

23.

constraints to reduce late rectal toxicity following 3D-conformal radiotherapy (3D-CRT) of prostate cancer. Radiother Oncol 2003;69:215–222. Huang EH, Pollack A, Levy L, et al. Late rectal toxicity: Dose-volume effects of conformal radiotherapy for prostate cancer. Int J Radiat Oncol Biol Phys 2002;54:1314 –1321. Koper PC, Heemsbergen WD, Hoogeman MS, et al. Impact of volume and location of irradiated rectum wall on rectal blood loss after radiotherapy of prostate cancer. Int J Radiat Oncol Biol Phys 2004;58:1072–1082. Koper PC, Jansen P, van Putten W, et al. Gastro-intestinal and genito-urinary morbidity after 3D conformal radiotherapy of prostate cancer: Observations of a randomized trial. Radiother Oncol 2004;73:1–9. Kupelian PA, Reddy CA, Carlson TP, et al. Dose/volume relationship of late rectal bleeding after external beam radiotherapy for localized prostate cancer: Absolute or relative rectal volume? Cancer J 2002;8:62– 66. Michalski JM, Purdy JA, Winter K, et al. Preliminary report of toxicity following 3D radiation therapy for prostate cancer on 3DOG/RTOG 9406. Int J Radiat Oncol Biol Phys 2000;46: 391– 402. Pollack A, Zagars GK, Starkschall G, et al. Prostate cancer radiation dose response: Results of the M. D. Anderson phase III randomized trial. Int J Radiat Oncol Biol Phys 2002;53: 1097–1105. Skwarchuk MW, Jackson A, Zelefsky MJ, et al. Late rectal toxicity after conformal radiotherapy of prostate cancer (I): Multivariate analysis and dose-response. Int J Radiat Oncol Biol Phys 2000;47:103–113. Storey MR, Pollack A, Zagars G, et al. Complications from radiotherapy dose escalation in prostate cancer: Preliminary results of a randomized trial. Int J Radiat Oncol Biol Phys 2000;48:635– 642. Wachter S, Gerstner N, Goldner G, et al. Rectal sequelae after conformal radiotherapy of prostate cancer: Dose-volume histograms as predictive factors. Radiother Oncol 2001;59:65–70. Zapatero A, Garcia-Vicente F, Modolell I, et al. Impact of mean rectal dose on late rectal bleeding after conformal radiotherapy for prostate cancer: Dose-volume effect. Int J Radiat Oncol Biol Phys 2004;59:1343–1351. Jackson A, Skwarchuk MW, Zelefsky MJ, et al. Late rectal bleeding after conformal radiotherapy of prostate cancer. II. Volume effects and dose-volume histograms. Int J Radiat Oncol Biol Phys 2001;49:685– 698. Vargas C, Martinez A, Kestin LL, et al. Dose-volume analysis of predictors for chronic rectal toxicity after treatment of prostate cancer with adaptive image-guided radiotherapy. Int J Radiat Oncol Biol Phys 2005;62:1297–1308. Denham JW, O’Brien PC, Dunstan RH, et al. Is there more than one late radiation proctitis syndrome? Radiother Oncol 1999;51:43–53. Heemsbergen WD, Hoogeman MS, Hart GA, et al. Gastrointestinal toxicity and its relation to dose distributions in the anorectal region of prostate cancer patients treated with radiotherapy. Int J Radiat Oncol Biol Phys 2005;61:1011–1018. Peeters STH, Hoogeman MS, Heemsbergen WD, et al. Volume and hormonal effects for acute side effects of rectum and bladder during conformal radiotherapy for prostate cancer. Int J Radiat Oncol Biol Phys 2005;63:1142–1152. Partin AW, Yoo J, Carter HB, et al. The use of prostate

24.

25. 26.

27. 28. 29.

30. 31.

32. 33. 34.

35. 36.

37.

38. 39.

● S. T. H. PEETERS et al.

1161

specific antigen, clinical stage and Gleason score to predict pathological stage in men with localized prostate cancer. J Urol 1993;150:110 –114. Bos LJ, Damen EM, de Boer RW, et al. Reduction of rectal dose by integration of the boost in the large-field treatment plan for prostate irradiation. Int J Radiat Oncol Biol Phys 2002;52:254 –265. Rasch C, Remeijer P, Koper PC, et al. Comparison of prostate cancer treatment in two institutions: A quality control study. Int J Radiat Oncol Biol Phys 1999;45:1055–1062. Meijer GJ, van den Brink M, Hoogeman MS, et al. Dose-wall histograms and normalized dose-surface histograms for the rectum: A new method to analyze the dose distribution over the rectum in conformal radiotherapy. Int J Radiat Oncol Biol Phys 1999;45:1073–1080. Vordermark D, Sailer M, Flentje M, et al. Curative-intent radiation therapy in anal carcinoma: Quality of life and sphincter function. Radiother Oncol 1999;52:239 –243. Agarwal KK, Agarwal AK, Singh PJ, et al. Comparative study of anorectal region in mammals. J Anat Soc India 2002;51: 220 –224. Rancati T, Fiorino C, Gagliardi G, et al. Fitting late rectal bleeding data using different NTCP models: Results from an Italian multi-centric study (AIROPROS0101). Radiother Oncol 2004;73:21–32. Yeoh EE, Holloway RH, Fraser RJ, et al. Anorectal dysfunction increases with time following radiation therapy for carcinoma of the prostate. Am J Gastroenterol 2004;99:361–369. Vordermark D, Schwab M, Ness-Dourdoumas R, et al. Association of anorectal dose-volume histograms and impaired fecal continence after 3D conformal radiotherapy for carcinoma of the prostate. Radiother Oncol 2003;69:209 –214. Fokdal L, Honore H, Hoyer M, et al. Dose-volume histograms associated to long-term colorectal functions in patients receiving pelvic radiotherapy. Radiother Oncol 2005;74:203–210. al-Abany M, Helgason AR, Cronqvist AK, et al. Dose to the anal sphincter region and risk of fecal leakage. Acta Oncol 2004;43:117–118. al-Abany M, Helgason AR, Cronqvist AK, et al. Toward a definition of a threshold for harmless doses to the anal-sphincter region and the rectum. Int J Radiat Oncol Biol Phys 2005;61:1035–1044. Hoogeman MS, van Herk M, de Bois J, et al. Quantification of local rectal wall displacements by virtual rectum unfolding. Radiother Oncol 2004;70:21–30. Vargas C, Di Yan FACR, Kestin LL, et al. Phase II dose escalation study of image-guided adaptive radiotherapy for prostate cancer: Use of dose-volume constraints to achieve rectal isotoxicity. Int J Radiat Oncol Biol Phys 2005;63:141– 149. Steenbakkers RJHM, Deurloo KEI, Nowak PJCM, et al. Reduction of dose delivered to the rectum and bulb of the penis using MRI delineation for radiotherapy of the prostate. Int J Radiat Oncol Biol Phys 2003;57:1269 –1279. Mohan R, Mageras GS, Baldwin B, et al. Clinically relevant optimization of 3-D conformal treatments. Med Phys 1992; 19:933–944. Tucker SL, Dong L, Cheung R, et al. Comparison of rectal dose–wall histogram versus dose–volume histogram for modeling the incidence of late rectal bleeding after radiotherapy. Int J Radiat Oncol Biol Phys 2004;60:1589 –1601.