Subtyping muscle-invasive bladder cancer to assess clinical response to trimodality therapy.

Abstract

287 Background: Trimodality therapy with TURBT followed by chemoradiation is an acceptable alternative to cystectomy for muscle invasive bladder cancer (MIBC). Recently, genomic profiling has demonstrated MIBC can be divided into three or more subtypes with differing responses to chemotherapy, suggesting genomic subtype may impact therapeutic response. Here, we explore the utility of genomic information to better select patients for bladder-sparing trimodality therapy. Methods: Transcriptome-wide gene expression profiles were generated for 189 MIBC TURBT samples from patients undergoing trimodality therapy at the Massachusetts General Hospital. Of these, 108 passed microarray QC and 100 had complete clinical information. The patient tumors were classified as basal, basal claudin-low, infilrated luminal or luminal subtype. The subtype and the expression of a number of bladder cancer genes were assessed for their association with need for salvage cystectomy and for overall survival. Finally, transcriptome-wide differential expression analysis was used to explore gene set enrichment in trimodality therapy response groups. Results: Our chemoradiation cohort (n = 108) was classified into the four subtypes: basal (n = 45), basal claudin low (n = 13), infiltrated luminal (n = 17) and luminal tumors (n = 33). Survival analysis (n = 100) showed that patients of the luminal subtype trended to better overall survival, but did not reach significance (HR = 0.63, p = 0.1). Fewer patients with infiltrated luminal tumors (12%) required a salvage cystectomy compared to all other subtypes (34+/-1.5%, p = 0.08). We found high expression of BLACAT1 and NORAD (a lncRNA with a role in genome stability) correlated with worse (p = 0.01) and better prognosis (p = 0.008), respectively. Likewise, patients with high levels of the luminal-associated PPARG showed a significant increase in overall survival (p = 0.0002). Gene set enrichment revealed differential regulation of immune pathways in the trimodality therapy responders relative to the non-responders (p < 0.05). Conclusions: Preliminary data exploring MIBC subtyping suggests the possibility of using genomics to predict response to trimodality bladder-sparing therapy.