Abstract 462: Results from multiple datasets including the TCGA reveal limited clinical significance of molecular subtypes in bladder cancer

Abstract

Abstract INTRODUCTION AND OBJECTIVE: Molecular subtypes have been suggested to predict outcome in patients with muscle invasive bladder cancer (MIBC). The major subtypes reported have been “luminal” and “basal”; however, there is little consensus regarding subtype-signatures and the clinical parameters used to evaluate outcome. The objective of this study was to examine using multiple datasets, if molecular subtypes are distinct entities within MIBC and whether they predict clinical outcome. METHODS: We analyzed transcriptome data from all 402 MIBC samples in The Cancer Genome Atlas (TCGA) and 151 high-grade MIBC samples from three datasets in Oncomine™. Transcript levels were also measured in 52 bladder tumor specimens with clinical follow-up (cohort-1). Samples were subtype-scored using gene panels (GP-11, GP-30, BASQ) consisting only of those genes common among published studies. Correlation of subtypes to recurrence/progression-free survival (R/PFS), metastasis, cancer-specific survival (CSS) and overall-survival (OS) was examined by logistic regression, Cox proportional Hazards model and Kaplan-Meier analyses. RESULTS: Based on GP-11 subtype-scoring, >75% of the MIBC TCGA-dataset samples were mixed, i.e. neither pure-luminal nor pure-basal. When tumors were categorized as pure-luminal, pure-basal, or mixed, the subtypes could not predict OS or R/PFS. Consistent with published studies, when tumors were categorized as luminal or basal the subtypes predicted OS (P=0.051); sensitivity: 53.9%; specificity: 62%. Subtypes also correlated with tumor-grade (P=0.0005); most low-grade MIBC cases (16/21) in the TCGA-dataset were luminal. However, MIBC is rarely low-grade and subtypes could not predict OS (P=0.131) when only high-grade cases were included. Subtypes were not significant prognosticators in multivariate analyses. GP-30 and BASQ panels validated these results. In the Oncomine-dataset and cohort-1, subtypes could not predict metastasis, CSS, or OS. CONCULUSIONS: Our study of multiple datasets reveals that molecular subtypes may reflect tumor-heterogeneity but are likely not distinct entities within MIBC. Furthermore, molecular subtypes have limited prognostic capability for MIBC patients. Although there is a need to individualize patient care, further examination into the molecular subtypes of MIBC is needed before their incorporation into clinical practice. Citation Format: Sarrah S. Lahorewala, Daley S. Morera, Jiaojiao Wang, Vinata B. Lokeshwar. Results from multiple datasets including the TCGA reveal limited clinical significance of molecular subtypes in bladder cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 462.