LAUSR

Multivariable prediction models are superior to conventional decision-making based solely on prostate-specific antigen (PSA) testing or digital rectal examination (DRE) in predicting the outcome of prostate biopsies [1]. Therefore, several prostate cancer risk calculators (RCs) have been developed with the aim of minimizing the number of unnecessary biopsies and reducing overdetection and overtreatment of insignificant prostate cancer. External validations have confirmed the utility of several RCs, and thus their use in clinical practice is increasingly recommended [2]. However, the performance of current RCs is still suboptimal, as evidenced by significant variation in RC performance in different patient cohorts [3]. Strategies to improve current RCs include recalibration of existing RCs to adjust for local cohort characteristics [4]. Using data from contemporary clinical cohorts to create novel RCs could be another option for building more accurate up-to-date decision aids. Including biomarker data or results from multiparametric magnetic resonance imaging (mpMRI) are further ways to potentially improve RC performance [5]. In this issue of European Urology, Radtke and colleagues [6] evaluate whether RCs using a combination of clinical parameters and mpMRI data (ie, Prostate Imaging-Data and Reporting System [PI-RADS] v.1.0 score) improves the prediction of significant prostate cancer compared to PIRADS score alone or RCs based only on clinical parameters. They used prospectively collected data for their patient series of 1015 men (660 biopsy-naı̈ve and 355 prebiopsied men) who underwent mpMRI before combined fusion targeted biopsy and transperineal systematic saturation biopsies to develop a risk model (RM) that included clinical parameters and PI-RADS scores. Clinical parameters evaluated for model inclusion were those already used for the RCs