Relationships between DVHs and late rectal bleeding after radiotherapy for prostate cancer: analysis of a large group of patients pooled from three institutions

Background: Accurate modeling of late rectal reactions needs the collection of individual 3D dose-volume data (i.e. DVH) as well as clinical information of large cohorts of patients. The possibility of collecting a large number of patients with many different dose-volume combinations is very suitable for this purpose. Purpose: The purpose of the study is to search for significant correlation between dose-volume histograms/dose statistics of the rectum and late rectum bleeding. Materials and methods: Data from three institutions for 402 patients previously treated for prostate cancer with three to four field techniques, were retrospectively pooled and were collected with a number of clinical and physical parameters, including DVHs of the rectum (including filling). Patients with large air/fecal content in the rectum during planning computerized tomography (CT) scan were excluded from the analysis (n=74). Out of 328 patients, 229 patients received an ICRU dose between 70 and 76 Gy and the current analysis is referred to this subgroup of patients (median follow up: 30 months, range: 12-85 months). Out of these 229 patients, 189 patients were treated with conformal techniques. Rectum was contoured from the anal verge up to the sigmoid flessure by one observer for each institution. Dosimetric and contouring consistencies between the three institutions were previously investigated and the impact on DVHs was found to be quite modest for the purposes of the study. Median/quartile values of all parameters were considered as cut-off values for statistical analysis. We considered as bleeders those patients who experienced grades II-III late bleeding (modified RTOG scoring scale). Results: Twenty two of 229 patients experienced greater than or equal to grade II late bleeding (30 months actuarial incidence: 10.7%). Significant correlation between a number of parameters and late bleeding was found (log-rank test). With regard to DVH, all median and third quartile values for V50-V70 were found to be significantly associated with an increased risk of rectal bleeding, if excepting the median value of V70. Based on the results of univariate analysis, the patients were divided into two groups: 'high risk', with at least one value above quartiles in the range V50-V60 (V50: 70%, V55: 64%, V60: 55%); 'low risk', the remaining patients. The 30 months actuarial rates of bleeding were 19.2 and 5.9% for the 'high' and the 'low' risk group, respectively (P=0.0003 log-rank test). A multivariate analysis (Cox regression model) including 'DVH grouping' and the main remaining variables (age, previous prostatectomy, diabetes, hypertension, adjuvant hormonal therapy, rectum volume and ICRU dose) showed that 'DVH grouping' is the most predictive parameter (P=0.005) together with adjuvant hormonal therapy (P=0.025) and ICRU dose (P=0.06). Conclusions: Our data confirm the role of the rectal DVH in separating groups of patients having prostate radiotherapy in low and high risk of developing late bleeding. Based on these results, V50 below 60-65% and V60 below 50-55% seem to be the robust cut-off values to keep the risk of developing late rectal bleeding reasonably low. However, due to the 'heterogeneity' of the considered population, the results found should be applied with caution in 'more homogeneous' groups of patients. The association of adjuvant hormone deprivation seems to be associated with an increased risk of rectal toxicity; the mechanism for this effect should be a focus of further research. (C) 2002 Elsevier Science Ireland Ltd. All rights reserved.