Development and External Validation of Nomograms To Predict Adverse Pathological Characteristics After Robotic Prostatectomy: Results of a Prospective, Multi-institutional, Nationwide series

The possibility of predicting pathologic features before surgery can support clinicians in selecting the best treatment strategy for their patients. We sought to develop and externally validate pretreatment nomograms for the prediction of pathological features from a prospective multicentre series of robotic-assisted laparoscopic prostatectomy (RALP) procedures.Between 2009 and 2016, data from 6823 patients undergoing RALP in 25 academic and community hospitals were prospectively collected by the Belgian Cancer Registry. Logistic regression models were applied to predict extraprostatic extension (EPE; pT3a,b-4), seminal vesicle invasion (SVI; pT3b), and high-grade locally advanced disease (HGLA; pT3b-4 and Gleason score [GS] 8-10) using the following preoperative covariates: prostate-specific antigen, clinical T stage, biopsy GS, and percentage of positive biopsy cores. Internal and external validation was performed.The stability of the model was assessed via tenfold cross-validation using 80% of the cohort. The nomograms were independently externally validated using the test cohort. The discriminative accuracy of the nomograms was quantified as the area under the receiver operating characteristic curve and graphically represented using calibration plots.The nomograms predicting EPE, SVI, HGLA showed discriminative accuracy of 77%, 82%, and 88%, respectively. Following external validation, the accuracy remained stable. The prediction models showed excellent calibration properties.We developed and externally validated multi-institutional nomograms to predict pathologic features after RALP. These nomograms can be implemented in the clinical setting or patient selection in clinical trials.We developed novel nomograms using nationwide data to predict postoperative pathologic features and lethal prostate cancer.