458
Background: We evaluated the clinical and prognostic impact of bladder cancer histologic variants (BCHV) using a large population-based cancer database. Methods: Using the Surveillance, Epidemiology, and End Results database (SEER), we identified bladder cancer patients from 2001-2012, and categorized them according to histological differentiation. 5 year disease-specific survival (DSS) was calculated using the Kaplan-Meier method. Cox proportional hazards regression model was used to predict association with disease-specific mortality (DSM). In addition, we fitted multivariate logistic regression models to predict the impact of histological variants on muscle-invasive status (MI), nodal involvement (NI), and metastatic disease (MD). Results: The cohort included 175,544 urothelial (96.3%) and 6,714 non-urothelial (3.7%) cancers. The latter were divided into: 2,382 squamous cell carcinoma, 1,648 adenocarcinoma, 888 small cell, 912 sarcomatoid, 292 signet-ring cell, 314 neuroendocrine and 278 micropapillary bladder tumors. Urothelial cancers overall had the best 5-year DSS. Of the non-urothelial variants, micropapillary and squamous had the best and worst DSS respectively (p < 0.001). On multivariable analysis predicting DSM, micropapillary and squamous variants had the best and worst prognosis respectively (HR 0.79, p = 0.102 and HR 2.63, p < 0.001), compared to urothelial tumors. On multivariable analysis predicting MI, NI, and MD: squamous (OR 22.76, p < 0.001), micropapillary (OR 3.17, p < 0.001) and adenocarcinoma (OR 4.14, p < 0.001), had higher likelihood respectively, compared to urothelial tumors. Conclusions: Despite accounting for a minority of bladder cancers, BCHV are associated with worst outcomes. It is essential to recognize the potential implications of these variants when deciding treatment. Additional studies are warranted to better characterize the clinical impact of these variants.