Development of a model to predict prostate cancer at the apex (PCAP model) in patients undergoing robot-assisted radical prostatectomy

Purpose To develop a model based on preoperative variables to predict apical prostate cancer. Methods We performed a retrospective analysis of 459 patients who underwent a robotic assisted radical prostatectomy (RALP) between January 2016 and September 2017. All patients had a preoperative biopsy and mpMRI of the prostate. Significant apical pathology (SAP) was defined as those patients who had a dominant nodule at the apex with a Gleason score > 6 and/or ECE at the apex. Binary logistic regression analyses were adopted to predict SAP. Variables included in the model were PSA, apical lesions prostate imaging reporting and data system (PI-RADS) score and apical biopsy Gleason score. The area under the curve (AUC) of the model was computed. Results A total of 121 (43.2%) patients had SAP. On univariable analysis, all apex-specific variables investigated emerged as predictors of SAP (all p 3 (all p