Risk factors for seminal vesicle invasion at radical prostatectomy.

PURPOSE: To comprehensively assess predictors of seminal vesicle invasion (SVI) at radical prostatectomy given the low sensitivity of magnetic resonance imaging (MRI). METHODS: Between 2017 and 2025, 4,448 patients underwent radical prostatectomy alone. Multivariable logistic regression was used to identify clinicopathological and MRI-related predictors of pathologic SVI, including age, Black race, PSA, PSA density, clinical T stage, Grade Group, percentage of positive cores, basal tumor involvement on systematic biopsy, the location of PI-RADS 4/5 lesions (apex/midgland/base), and MRI-suspected extracapsular extension and SVI. The same variables were evaluated in patients without MRI-detected SVI (n = 4,358). Model discrimination was assessed using the AUC. RESULTS: Pathologic SVI prevalence was 6.9% (305/4,448). MRI demonstrated 17% sensitivity and 99% specificity. Independent clinicopathological predictors (all p < 0.04) were Black race (OR 1.48), log-PSA (OR 2.30/1-unit), Grade Groups 3–5 (OR 2.56–4.32), percentage of positive cores (OR 1.21/10%), and basal tumor involvement on biopsy (OR 1.92). MRI-related independent predictors (all p < 0.01) were SVI (OR 15.7), extracapsular extension (OR 1.69), and midgland (OR 1.49) and basal (OR 1.92) PI-RADS 4/5 lesions; apex lesions were inversely associated (OR 0.64). These associations remained significant in the MRI-negative subgroup. The model showed good discrimination (AUC 0.80, 95% CI 0.77–0.82), remaining robust in MRI-negative patients (AUC 0.76, 95% CI 0.73–0.79). Limitation included selection bias toward favorable-risk features by excluding patients with hormonal therapy. CONCLUSION: Multifactorial risk assessment, particularly including tumor location, helps mitigate MRI’s limited sensitivity for SVI detection and could inform surgical or radiation planning in patients at risk of SVI.