Postprostatectomy Urinary Continence Prediction: External Validation of a Model Incorporating MRI Derived Membranous Urethral Length.

Abstract

PURPOSE: MRI membranous urethral length is associated with postprostatectomy continence. Nevertheless, it is not routinely used for patient counseling because predictive models have not been externally validated. The objective of this study was to assess the performance of membranous urethral length continence model on an external cohort. METHODS: A continence prediction model was applied to a prospective cohort of patients with a preprostatectomy MRI and validated assessment of postoperative continence between 2015 and 2023. Patients from multiple surgeons underwent open or robotic prostatectomy. Membranous urethral length was retrospectively measured by 2 reviewers. The model included age, nerve spare plan, prostate volume, surgical approach, and membranous urethral length. Continence definitions were 0 pads/24 hours and 0 to 1 pad/24 hours. Model performance was assessed using c-statistics, calibration curves, and decision curve analyses. RESULTS: Among 500 patients, the median membranous urethral length was 11 mm (IQR 8.1-14). One-year postprostatectomy continence (defined as 0 pads/24 hours) occurred in 312 (62%) patients. The c-statistic was 0.64 (95% CI 0.59-0.69). The calibration was poor. The decision curve analysis showed no net-benefit of using the model. Continence defined as 0 to 1 pad/24 hours was achieved in 450 (90%) patients. The c-statistic was 0.69 (95% CI 0.61-0.76), and the calibration was moderate. Decision curves showed a net-benefit of using the model over a range of clinically relevant continence risk thresholds. CONCLUSIONS: This recalibrated prediction model incorporating MRI-derived membranous urethral length provides net-benefit for predicting continence defined as 0 to 1 pad/24 hours. However, the model was not useful for predicting 0 pad use.