Extremely Dose Escalated Radiation Therapy Improves Cancer-Specific Survival Compared With Radical Prostatectomy or Conventionally Dose-Escalated Radiation Therapy in Gleason Score 9-10 Prostate Adenocarcinoma: A Multi-institutional Analysis of 1403 Patie

Purpose/Objective(s): To compare the outcomes of a modern cohort of patients with Gleason Score (GS) 9-10 prostate adenocarcinoma (PCa) following treatment with external beam radiotherapy (EBRT), extremely dose-escalated radiotherapy (exemplified by EBRT with a brachytherapy boost [EBRT+BT]), and radical prostatectomy (RP). Materials/Methods: One thousand four hundred three patients with biopsy GS 9-10 PCa who received definitive treatment between 2000 and 2013 were included (506 treated with EBRT, 392 with EBRT+BT, and 505 with RP). Distant metastasis-free survival (DMFS) and prostate cancer-specific mortality (PCSM) were compared across the three cohorts with four cause-specific Cox regression models: (a) unadjusted, (b) covariate-adjusted for age, ln(iPSA), cT stage, and GS, (c) propensity score adjusted for the same covariates, and (d) in a doubly robust analysis (i.e., propensity score and covariate adjusted). Four analogous Fine and Gray competing risk models were developed to compare cumulative PCSM incidence. Results: The median follow-up period was 5.1 years (5.1, 6.4, and 4.3 years in the EBRT, EBRT + BT, and RP cohorts, respectively). The median doses among EBRT and EBRT+BT patients were isoeffective to 81 Gy and 96 Gy in 1.8 Gy fractions, respectively. Over 90% of patients treated with EBRT or EBRT+BT received ADT (median durations of 23 months and 12 months, respectively). Nearly 40% of RP patients received postoperative RT; 78% of postoperative RT was delivered in the salvage setting. PCSM was significantly lower among EBRT+BT patients than among either EBRT or RP patients in all cause-specific hazard models and Fine and Gray competing risk models, including unadjusted models (P<0.05 in unadjusted models and P<0.01 in all other models). PCSM was not significantly different between EBRT and RP patients in any model. DMFS was also significantly higher among patients treated with EBRT+BT in all cause-specific hazard models. In all adjusted models, DMFS was significantly superior among EBRT patients than RP patients (P<0.05). Five- and 10-year cumulative incidences of PCSM were 3% and 12% with EBRT+BT, 10% and 20% with EBRT, and 9% and 19% with RP, respectively (P