Naturally-occurring canine invasive urothelial carcinoma harbors luminal and basal transcriptional subtypes found in human muscle invasive bladder cancer.

Abstract

There is growing evidence that molecular subtypes (e.g. luminal and basal subtypes) affect the prognosis and treatment response in patients with muscle invasive urinary bladder cancer (invasive urothelial carcinoma, iUC). Modeling these subtypes in pre-clinical animal studies is essential, but it is challenging to produce these subtypes, along with other critical host and tumor features, in experimentally-induced animal models. This study was conducted to determine if luminal and basal molecular subtypes are present in naturally-occurring canine iUC, a cancer that mimics the human condition in other key aspects. RNA sequencing was performed on 29 canine treatment naive iUC tissue samples and on four normal canine bladder mucosal samples. Data were aligned to CanFam 3.1, and differentially expressed genes were identified. Unsupervised hierarchical clustering of these genes revealed two distinct groups (n = 13, n = 16). When genes that distinguish basal and luminal subtypes in human cancer (n = 2015) were used to probe genes differentially expressed between normal canine bladder and iUC, 829 enriched signature genes were identified. Unsupervised hierarchical clustering of these genes revealed two distinct groups comprised of 18 luminal subtype tumors and 11 basal subtype tumors. The enriched genes included MMP9, SERPINE2, CAV1, KRT14, and RASA3 in basal tumors, and PPARG, LY6E, CTSE, CDK3, and TBX2 in luminal tumors. In supervised clustering, additional genes of importance in human iUC were identified in canine iUC associated with claudin-low and infiltrated tumors. A smaller panel of genes (n = 60) was identified that distinguished canine luminal and basal iUC with overall 93.1% accuracy. Immune signature patterns similar to those in human iUC were also identified with the greatest enrichment of immune genes being in the basal subtype tumors. These findings provide additional compelling evidence that naturally-occurring canine iUC is a highly relevant and much needed model of human iUC for translational research.