RESUMO
PURPOSE: Few studies have investigated predictive risk factors of biochemical recurrence (BCR) after radical prostatectomy (RP) in other than Caucasian and Asian populations. We aimed to identify pre- and post-operative predictors of BCR after RP in an Afro-Caribbean population in Guadeloupe (French West Indies). PATIENTS AND METHODS: The study included 964 patients who underwent RP for clinically localized prostate cancer between April 1, 2000 and December 31, 2010 in the University Hospital of Guadeloupe. The hazard ratio (HR) and corresponding 95% confidence interval (CI) for single variable associations with BCR were calculated using the Cox proportional hazards regression. Multiple variable analyses for association with BCR were performed, including all variables that reached statistical significance (P value<0.05) in univariate analysis. A backward selection model was then applied with a P value ≥0.1 for retention in the final model. Sensitivity analysis was performed and restricted to patients with known values for all variables (complete case analysis). RESULTS: With a median follow-up of 4.8 years, the BCR rate was 26.7%. In multivariable analysis, predictors of BCR before surgery were diabetes mellitus type 2 (DT2) (HR: 1.37, 95% CI: 1.02-1.85; P=0.038), pre-operative PSA>7.5ng/ml (1.49, 1.15-1.92; P=0.002), clinical stage T2 (1.55, 1.21-1.98; P=0.0006), Gleason score>7 or 4+3 (2.12, 1.54-2.91; P<0.0001), and percentage of length of biopsy positive scores (1.66, 1.24-2.20; P=0.0006). Predictors of BCR after surgery were DT2 (HR: 1.37, 95% CI: 1.01-1.85; P=0.045), pre-operative PSA>7.5ng/ml (1.37, 1.06-1.79; P=0.018), pathological Gleason score>7 or 4+3 (2.36, 1.74-3.19; P<0.0001), pathological stage pT3b (1.68, 1.15-2.45; P=0.007), positive surgical margins (1.72, 1.32-2.45; P=0.0001), and perioperative blood loss>2000ml (3.74, 1.37-10.2; P=0.01). The results were virtually the same by sensitivity analysis (complete cases), except for DT2, which was associated with BCR with borderline statistical significance in the pre-operative model and not retained in the post-operative model. CONCLUSIONS: Afro-Caribbean populations in French West Indies share the same major clinical and pathological risk factors of BCR after RP identified in other ethnic groups. Perioperative blood loss appears to be an additional and independent predictive factor of BCR. LEVEL OF PROOF: 4.