Prostate cancer-specific mortality after radical prostatectomy: value of preoperative MRI

Background: Although magnetic resonance imaging (MRI) is currently indispensable in the preoperative setting of biopsy-proven prostate cancer, the value of preoperative MRI for predicting prostate cancer-specific mortality (PCSM) is not well known. Purpose: To evaluate the value of MRI for predicting PCSM in patients who underwent radical prostatectomy (RP) for localized prostate cancer. Material and Methods: A total of 318 patients underwent MRI followed by RP. MRI was assessed for the presence of clinically significant cancer using a five-point Likert scale, where >= 4 was considered positive. Cox proportional hazards regression analyses was used to determine the relationship of preoperative factors with PCSM. PCSM was calculated using the Kaplan-Meier method and compared between factors using the log-rank test. Results: After a median follow-up of 104 months, 11 (3.5%) patients died of prostate cancer. One hundred and four (32.7%) patients had clinically significant prostate cancer on MRI. Univariate analysis revealed that Gleason grade, greatest percentage of involved core length (GPCL), and clinically significant cancer on MRI were significantly related to PCSM (P=0.001-0.003). Multivariate analysis showed that GPCL (hazard ratio [HR], 1.028; 95% confidence interval [CI], 1.000-1.057; P=0.048) and clinically significant cancer on MRI (HR, 10.903; 95% CI, 1.287-92.374; P=0.028) were independent predictors of PCSM. The 5- and 10-year PCSM rates were 0.6% and 1.3% in patients with GPCL <50% and 5.1% and 8.6% in those with GPCL >= 50% (P=0.012). Patients without clinically significant cancer on MRI showed 5- and 10-year PCSM rates of 0% and 0.5%, respectively, whereas those with clinically significant cancer on MRI showed rates of 8% and 14.2%, respectively (P<0.001). Conclusion: Preoperative MRI and GPCL may be used to predict PCSM after RP.