Predictors of biochemical recurrence in pT3b prostate cancer after radical prostatectomy without adjuvant radiotherapy.

Abstract

Men with pathologic evidence of seminal vesicle invasion (SVI) at radical prostatectomy (RP) have higher rates of biochemical recurrence (BCR) and mortality. Adjuvant radiotherapy (XRT) has been shown to increase freedom from BCR, but its impact on overall survival is controversial and it may represent overtreatment for some. The present study, therefore, sought to identify men with SVI at higher risk for BCR after RP in the absence of adjuvant XRT. We identified 180 patients in our institutional database who underwent RP from 1990 to 2011 who had pT3bN0-1 disease. The Kaplan-Meier method was used to estimate freedom from BCR for the overall cohort and substratified by Gleason score, PSA, surgical margin status, and lymph node positivity. Cox Proportional Hazards models were used to determine demographic and histopathological factors predictive of BCR. Time-dependent ROC curve analysis was conducted to assess the ability of the UCSF-CAPRA score to predict BCR. Median age was 64 years, and 52.8% of patients were preoperative D'Amico high risk. At RP, 41.4% had a positive surgical margin (PSM), and 12.2% had positive lymph nodes (LN). The most common sites of PSM were the peripheral zone (56.8%) and the apex (32.4%). Positive bladder neck margin (HR = 7.01, P = 0.035) and PSA 10-20 versus ≤10 (HR = 1.63, P = 0.047) predicted higher BCR in multivariable analyses. Median follow-up was 26 months, and 2-, 3-, and 5-year BCR-free rates were 56.1%, 49.0%, and 39.5%. Log rank tests showed that freedom from BCR was significantly less for Gleason 9-10, PSA >20, PSM, and N1 patients. The area under curve (AUC) for CAPRA in predicting BCR was 0.713 at 2 years, 0.692 at 3 years, and 0.641 at 5 years. Increasing CAPRA score was associated with an increased risk of BCR (HR = 1.33, P < 0.001). pT3b prostate cancer is a heterogeneous disease commonly associated with several high-risk features. Stratifying men with SVI by prognostic features (i.e., Gleason, PSA, node status, surgical margin status) and using these features to augment the CAPRA score will improve identification of those at higher risk for BCR that should be strongly considered for adjuvant XRT.