Concomitant Seminal Vesicle Invasion in pT4a Urothelial Carcinoma of the Bladder with Contiguous Prostatic Infiltration is an Adverse Prognosticator for Cancer-Specific Survival after Radical Cystectomy

PURPOSE: To evaluate the prognostic value of concomitant seminal vesicle invasion (cSVI) in patients with urothelial carcinoma of the bladder (UCB) and contiguous prostatic stromal infiltration in a large cystectomy series.

METHODS: A total of 385 patients with UCB and contiguous prostatic infiltration comprised our study. Patients were divided in two groups according to cSVI. Median follow-up was 36 months (interquartile range 11-74); the primary end point was cancer-specific mortality. The prognostic impact of cSVI was evaluated using multivariable Cox regression analysis. The predictive accuracy was assessed by a receiver operating characteristic analysis.

RESULTS: A total of 229 patients (59.5 %) without cSVI comprised group A, and 156 patients (40.5 %) with cSVI comprised group B. Positive lymph nodes (63 vs. 44 %, p < 0.001) and positive surgical margins (34 % vs. 14 %, p < 0.001) were more common in patients with cSVI. The 5- and 10-year cancer-specific survival rates were 41 % and 32 % (group A) and 21 and 17 % (group B) (p < 0.001). In multivariable analysis, pathological nodal stage (hazard ratio [HR] 2.19, p < 0.001), soft tissue surgical margin (HR 1.57, p = 0.010), clinical tumor stage (HR 1.46, p = 0.010), adjuvant chemotherapy (HR 0.40, p < 0.001), and cSVI (HR 1.69, p < 0.001) independently impacted cancer-specific mortality. The c-indices of the multivariable models with and without inclusion of cSVI were 0.658 (95 % confidence interval 0.60-0.71) and 0.635 (95 % confidence interval 0.58-0.69), respectively, resulting in a predictive accuracy gain of 2.3 % (p = 0.002).

CONCLUSIONS: In patients with UCB and prostatic stromal invasion, cSVI adversely affected cancer-specific survival compared to patients without cSVI. The inclusion of cSVI significantly improved the predictive accuracy of our multivariable model regarding survival.