Circulating Microrna Profiling for Prediction of Oncological Outcomes in Prostate Cancer Patients Following Radical Prostatectomy

Background Although radical prostatectomy is associated with good long-term oncological outcomes, approximately 30% of patients present biochemical recurrence, whereupon salvage treatments are required. Identification of novel molecular biomarkers to predict cancer behavior is clinically important. Here, we developed a novel microRNA (miRNA)-based prognostic model for patients who underwent radical prostatectomy. Methods We retrospectively investigated the clinical records of 295 patients who underwent radical prostatectomy between 2009 and 2017. We randomly assigned these cases into training or validation sets. The prognostic model was constructed using Fisher linear discriminant analysis in the training set, and we evaluated its performance in the validation set. Results Overall, 72 patients had biochemical recurrence. A prediction model was constructed using a combination of three miRNAs (miR-3147, miR-4513, and miR-4728-5p) and two pathological factors (pathological T stage and Gleason score). In the validation set, the predictive performance of the model was confirmed to be accurate (area under the receiver operating characteristic curve: 0.80; sensitivity: 0.78; specificity: 0.76). Additionally, Kaplan-Meier analysis revealed that the patients with a low prediction index had significantly longer recurrence-free survival than those with a high index (p < 0.001). Conclusions Circulating miRNA profiles can provide information to predict recurrence after prostatectomy. Our model may be helpful for physicians to decide follow-up strategies for patients.