MP12-15 ADDITION OF PREOPERATIVE PROSTATE MRI VARIABLES IMPROVES PREDICTION OF EARLY SOCIAL CONTINENCE AFTER RADICAL PROSTATECTOMY

Abstract

INTRODUCTION AND OBJECTIVE: Limited tools are available to help predict continence outcomes for patients undergoing radical prostatectomy (RP). We sought to determine whether inclusion of preoperative prostate MRI (mpMRI) improves the prediction of early social continence following RP over clinical factors alone. METHODS: The Michigan Urological Surgery Improvement Collaborative (MUSIC) is a consortium of 45 diverse urology practices that maintains a prospective registry of men with prostate cancer. Using mpMRI data from 3 MUSIC practices linked to MUSIC registry data, we developed logistic regression models to predict patient-reported social continence at 3 and 6 months after RP, defined as ≤1 pad/day. We compared models using clinical predictors (e.g., baseline EPIC-26 urinary incontinence domain score, age, clinical T-stage, log PSA, biopsy Gleason grade group), MRI predictors (e.g., membranous urethral length, PI-RADS score, lesion location, number of lesions, presence of EPE, SVI, or abnormal lymph nodes), and both. We assessed model discrimination using 5-fold cross-validated area-under-the-curve (AUC). RESULTS: We identified 638 and 613 patients who had undergone MRI and had continence assessed at 3 and 6 months, respectively. Of these, 437 (68%) were continent at 3 months and 507 (83%) at 6 months. Clinical predictors outperformed MRI predictors at 3 and 6 months. Combining clinical predictors with MRI predictors resulted in improved discrimination at 3 and 6 months (Figure 1). The most important variables in predicting 6 month continence (based on standardized coefficients) were Gleason grade group 5 and baseline urinary domain score. CONCLUSIONS: The addition of objective mpMRI variables to established clinical predictors improves prediction of social continence at 3 and 6 months following RP.Source of Funding: Blue Cross Blue Shield of Michigan