Subclassification, survival prediction and drug target analyses of chemotherapy-naïve muscle-invasive bladder cancer with a molecular screening

Background: Transcriptome expression studies identified distinct muscle invasive bladder cancer (MIBC) subtypes closely related with breast cancer subclasses. Here we developed a sensitive quantification method for MIBC subclassification (luminal, basal, p53-like). In addition, the subtype specific expression of drug targets has been investigated. Methods: Absolute quantification (nCounter) of a 64-gene panel was performed on MIBC patients (n=47) treated exclusively with radical cystectomy (RC). In conjunction of 170 MIBCs from 3 independent cohorts, a minimal set of consensus genes has been established. Survival of the consensus subtypes has been assessed by multivariate analysis. Relevant drug targets were tested for their subtype specificity in a clustering independent assessment. Results: A reduced 36-gene panel stably clustered into 3 subtypes throughout the cohorts (luminal, basal, infiltrated). Patients treated by RC only, showed worst 8-year disease specific survival (DSS) for the luminal subtype in contrast to the infiltrated subtype (17% vs. 73%, p=0.011). In multivariate analyses, the risk stratification based on luminal versus not-luminal MIBC proved to be an independent predictor for DSS superior to the TNM system in patients with RC. Drug targets (e.g. ERBB2, FGFR, AR, PDGFRB) showed a distinct subtype attribution. The subtypes based on this nCounter screening could further be validated by the TCGA cohort. Conclusion: This MIBC subtype screening predicted survival and allowed an analysis of subtype specific drug targets, thus being a powerful tool for the translation of personalized MIBC treatment concepts.