Does imaging-based clinical staging improve the predictive performance of the CAPRA score?

Abstract

Biochemical recurrence (BCR) after radical prostatectomy (RP) remains a major clinical challenge, necessitating accurate preoperative risk stratification. We evaluated whether replacing digital rectal examination (DRE)-based staging with imaging-based staging improves the predictive performance of the CAPRA model. A total of 507 treatment-naïve prostate cancer patients who underwent RP between April 2015 and April 2024 were retrospectively analyzed. All patients had preoperative multiparametric MRI (mpMRI). Imaging was evaluated according to PI-RADS v2.1 by an experienced radiologist. Patients were stratified using both conventional CAPRA and imaging-based CAPRA (ibCAPRA) scores. Imaging-based staging identified a higher proportion of locally advanced disease, with T3 tumors detected in 23.9% of patients. BCR occurred in 159 patients (31.4%). In separate preoperative and postoperative multivariate models, the ibCAPRA score, RP ISUP grade group ≥ 3, and pathological T stage (pT3) were independent predictors of BCR. Median follow-up was 55 months. ROC curve analysis showed comparable discrimination between ibCAPRA (AUC 0.781) and CAPRA (AUC 0.776), with no significant difference between the two models. Kaplan–Meier analysis demonstrated significantly worse oncological outcomes in higher risk groups. Although mpMRI substantially altered clinical stage allocation and was independently associated with BCR, replacing DRE-based staging with imaging-based staging did not significantly improve the predictive performance of the CAPRA model.