Normal tissue complication probabilities correlated with late effects in the rectum after prostate conformal radiotherapy

Purpose: Radiation therapy of deep-sited tumours will always result in normal tissue doses to some extent. The aim of this study was to calculate different risk estimates of late effects in the rectum for a group of cancer prostate patients treated with conformal radiation therapy (CRT) and correlate these estimates with the occurrences of late effects. Since the rectum is a hollow organ, several ways of generating dose-volume distributions over the organ are possible, and we wanted to investigate two of them. Methods and Materials: A mathematical model, known as the Lyman-Kutcher model, conventionally used to estimate normal tissue complication probabilities (NTCP) associated with radiation therapy, was applied to a material of 52 cancer prostate patients. The patients were treated with a four field box technique, with the rectum as organ at risk. Dose-volume histograms (DVH) were generated for the whole rectum (including the cavity) and of the rectum wall. One to two years after the treatment, the patients completed a questionnaire concerning bowel (rectum) related morbidity quantifying the extent of late effects. Results: A correlation analysis using Spearman's rank correlation coefficient, for NTCP values calculated from the DVHs and the patients' scores, gave correlation coefficients which were not statistically significant at the p < 0.01 level. The correlation coefficients based on histograms of the whole rectum were larger than those derived from histograms of the rectum wall. Also, simpler descriptive measures as D-max, of the whole rectum, correlated better to observed late toxicity than D-max derived from histograms of the rectum wall. Correlation coefficients from "high-dose" measures were larger than those calculated from the NTCP values. Accordingly, as the volume parameter of the Lyman-Kutcher model was reduced, raising the impact of small high-dose volumes on the NTCP values, the correlation between observed effects and NTCP values became significant at p < 0.01 level. Conclusions: 1) High-dose levels corresponding to small volume fractions of the cumulative dose-volume histograms were best correlated with the occurrences of late effects in the rectum as measured with questionnaires. This is compatible with a more serial organisation of the rectal tissue architecture than previously reported. 2) Reducing the Lyman-Kutcher model's volume parameter, thus allowing small high-dose regions to determine the NTCP, improved the correlation, but not beyond that of high-dose levels corresponding to small volume fractions of the cumulative dose-volume histograms. (C) 1999 Elsevier Science Inc.