Development and External Validation of an Updated Nomogram Including Both MRI and Local PSMA PET/CT Information for the Prediction of Side-specific Extraprostatic Extension in Patients Treated with Radical Prostatectomy

Background and objective The predictive value of prostate-specific membrane antigen (PSMA) positron emission tomography/computed tomography (PET/CT) for extraprostatic extension (EPE) in prostate cancer remains unclear. Objective We assessed whether adding PSMA PET/CT to magnetic resonance imaging (MRI) and clinical factors improves side-specific EPE prediction. Methods Patients undergoing PSMA PET/CT and MRI before radical prostatectomy were analysed retrospectively. The endpoint was side-specific EPE (≥pT3a). Predictors were selected by backward elimination, and the final nomogram was fitted using generalised estimating equations to account for clustering of lobes within patients. Discrimination, calibration, and net benefit were assessed, with external validation in an independent single-centre cohort. Results and limitations Development and validation cohorts included 801 and 287 lobes. The final model comprised prostate-specific antigen density (odds ratio [OR] 1.54 [95% confidence interval {CI} 0.94–2.50]; biopsy International Society of Urological Pathologists grade group (2 vs benign/1: OR 2.01 [95% CI 1.27–3.19]; 3 vs benign/1: OR 3.82 [95% CI 2.30–6.35]; 4 vs benign/1: OR 2.50 [95% CI 1.54–4.08]; 5 vs benign/1: OR 4.05 [95% CI 2.18–7.53]), MRI T-stage (organ-confined with a lesion [OC] vs organ-confined with no lesion [NL]: OR 1.31 [95% CI 0.86–1.99]; EPE vs NL: OR 4.11 [95% CI 2.56–6.60]), PSMA T-stage (OC vs NL: OR 1.67 [95% CI 1.07–2.61]; EPE vs NL: OR 2.59 [95% CI 1.49–4.49]), PSMA total (OR 1.72 [95% CI 1.33–2.23]). In development, the PSMA-MRI nomogram showed superior discrimination (area under the receiver operating characteristic curve [AUC] 0.80 vs 0.71, p < 0.001). On validation, the updated model demonstrated discrimination comparable to the reference model (AUC 0.85 vs 0.86, p = 0.9), with minor risk overestimation and similar net clinical benefit. Limitations include a relatively small validation cohort. Conclusions The PSMA-MRI nomogram appeared to outperform the MRI-only model in development, but this was not confirmed in validation, indicating limited, context-dependent incremental value.