Rectal Wall vs. Whole Rectum Dose: Which Volume Better Predicts Gastrointestinal Toxicity from Prostate External Beam Radiotherapy?

Abstract

Radiation-induced gastrointestinal (GI) toxicity is an important complication of prostate cancer (PCa) radiotherapy (RT) that radiation oncologists seek to minimize by respecting dose constraints for the rectum. It remains unclear whether dose to the rectal wall or whole rectum better predicts GI toxicity. This study used prospectively collected toxicity data from a registered phase III randomized trial to address this question. Dose-volume histogram (DVH) data was extracted from patients enrolled in PCS V, a multi-institutional phase III trial that compared conventionally fractionated (76 Gy in 38 fractions) to hypofractionated RT (68 Gy in 25 fractions) in PCa patients. Acute and late GI toxicity was assessed at 0-6 months and 6-24 months, respectively, using the Common Terminology Criteria for Adverse Events (CTCAE v.4). Rectal wall and whole volumes were collected prospectively. Dosimetric values evaluated were the whole rectum and rectal wall V50, V60, V60, V65, V70, V75, Dmax, and Dmean. Spearman’s rho (ρ) was used to correlate each DVH point of both rectum contours to acute and late GI toxicity. A Fisher’s z-transformation was then performed to test the significance of the difference between the rectal wall and whole rectum ρ’s for individual DVH points. Wilcoxon signed rank test was used to determine whether paired differences of the GI toxicity grade associated with rectal wall vs. whole rectum were significant. In total, DVH data was obtained from 155 patients, with 83 patients having been treated conventionally and 72 with hypofractionation. Spearman’s ρ correlation test revealed a strong association between the whole rectum DVH points and acute GI toxicity, namely for V60 (p=0.04), V65 (p=0.01), V70 (p=0.01) and Dmax (p=0.04). When the same analysis was conducted for late toxicity, only the V50 and Dmean points yielded a marginally significant correlation (p=0.049 and 0.042, respectively). These findings were not observed in the hypofractionated arm. There was no significant association between the rectal wall’s DVH points with either acute nor late GI toxicity in both arms. While the Fisher’s z-transformation analysis did not reveal any significant differences between the correlations of individual DVH points obtained from the rectal wall and whole rectum with acute or late GI toxicity, the distribution of Spearman’s ρ was significantly higher for the whole rectum, compared to its wall for acute toxicity (p=0.047) in the conventional arm. Our data suggests that the whole rectum DVH profile better correlates with acute GI toxicity than that of the rectal wall. Furthermore, the whole rectum’s DVH distribution rather than individual DVH points correlated more reliably with acute GI toxicity. Set-up protocols aiming to minimize rectal wall inter- and intra-fractional movement to improve the accuracy of rectal wall dosimetric values to predict toxicity are thus worthy of future study.