Inverse relationship between biochemical outcome and acute toxicity after image-guided radiotherapy for prostate cancer.

PURPOSE: Prostate cancer patients exhibit variability in normal tissue reactions and biochemical failure. With the use of image-guided radiotherapy (IGRT), there is a greater likelihood that the differences in normal tissue and tumor response are due to biological rather than physical factors. We tested the hypothesis that prospectively scored acute toxicity is associated with biochemical failure-free rate (BFFR) in prostate cancer patients treated with IGRT. METHODS AND MATERIALS: We retrospectively analyzed BFFR in 362 patients with localized prostate cancer treated with IGRT. We compared BFFR with prospectively collected Radiation Therapy Oncology Group (RTOG) maximum acute gastrointestinal (GI) and genitourinary (GU) toxicity scores. Median follow-up for all patients was 58.3 months after total radiotherapy doses of 75.6-79.8 Gy. RESULTS: Patients reporting RTOG acute GU or GI toxicity scores of ≥ 2 were considered "sensitive" (n = 141, 39%) and patients reporting scores <2 were considered "nonsensitive" (n = 221, 61%). When calculating biochemical failure (BF) using the American Society for Therapeutic Radiology and Oncology definition at 5 years, 76% (CI 70-82%) of the "nonsensitive" patients were failure free, compared with only 53% (CI 43-62%) of the "sensitive" patients (log-rank test, p < 0.0001). This difference was also observed using the Phoenix definition; "nonsensitive" 5-year BFFR was 81% (CI 74-86%) vs. "sensitive" BFFR was 68% (CI 58-76%; log-rank test p = 0.0012). The difference in BF between cohorts remained significant when controlled for radiation dose (75.6 vs. 79.8 Gy), prognostic stratification (T category, prostate-specific antigen, and Gleason score), and prostate volume. CONCLUSIONS: This study unexpectedly shows that prostate cancer patients who develop ≥ Grade 2 RTOG acute toxicity during radiotherapy are less likely to remain BFF at 5 years. These results deserve further study and, if validated in other large IGRT cohorts, additional models would be required to study interaction between normal tissue and tumor biology in prostate cancer patients.