A five-microRNA model (pCaP) for predicting prostate cancer aggressiveness using cell-free urine.

Improved biomarkers for prostate cancer (PC) risk stratification are urgently needed. Here, we aimed to develop a novel multimarker model for prediction of biochemical recurrence (BCR) after curatively intended radical prostatectomy (RP), based on minimally invasive sampling of blood and urine. We initially measured the levels of 45 selected miRNAs by RT-qPCR in exosome enriched cell-free urine samples collected prior to RP from 215 PC patients (Cohort 1, training). We trained a novel logistic regression model (pCaP), comprising five urine miRNAs (miR-151a-5p, miR-204-5p, miR-222-3p, miR-23b-3p and miR-331-3p) and serum prostate-specific antigen (PSA), which significantly predicted time to BCR in Cohort 1 (univariate Cox regression analysis: HR = 3.12, p < 0.001). Next, using the same exact numeric cutoff for dichotomization as trained in Cohort 1, we tested and successfully validated the prognostic potential of pCaP in two additional cohorts, including 199 (Cohort 2, HR = 2.24, p = 0.002) and 205 (Cohort 3, HR = 2.15, p = 0.004) RP patients, respectively. pCaP remained a significant predictor of BCR, also after adjustment for pathological T-stage, surgical margin status and Gleason grade group (p < 0.05 in multivariate Cox regression analysis: HR = 2.72, 1.94 and 1.83 for Cohorts 1, 2 and 3, respectively). Additionally, pCaP scores correlated positively with the established clinical risk stratification nomogram CAPRA in all three PC cohorts (Pearson's rho: 0.45, 0.39 and 0.44). Together, our results suggest that the minimally invasive pCaP model could potentially be used in the future to improve PC risk stratification and to guide more personalized treatment decisions. Further clinical validation studies are warranted.