Genomic Evaluation of Canine Prostatic Carcinomas as a Model for the Human Disease: or 'UC or not UC - that is the question'.

Spontaneous canine prostate cancer (PC) is widely considered a pertinent clinical model for the human disease. While over 95% of PC in men are adenocarcinomas, arising from prostatic glandular epithelium, it is increasingly recognised that many canine PC are of urothelial origin, arising within the prostatic urethra or ducts, or through invasion from a primary urinary bladder tumour. At diagnosis, canine prostatic tumours are often poorly differentiated and widely disseminated, masking the primary site and limiting the sensitivity of cellular biomarkers. Consequently, published studies of canine PC show varying representation of glandular versus urothelial tumours, yielding conflicting observations regarding their molecular pathogenesis and clinical behaviour. We characterised DNA sequence mutations and copy number aberrations in 31 canine PC, seeking evidence supporting relevance as a disease model. Only three tumours resembled adenocarcinomas. The remainder were either histologically consistent with urothelial carcinoma (n = 15), showed mixed glandular and urothelial morphology (n = 4), or were carcinomas of undetermined cell type (n = 9). BRAF V588E mutation was detected in 87% of tumours, including all three adenocarcinomas. Urinary bladder involvement was evident in 46% of cases, but none of the adenocarcinomas. Genome-wide DNA copy number instability was apparent throughout the cohort, with chromosome 36 gain significantly associated with urothelial tumours. Hallmark alterations of human PC, such as defects within PI3K and androgen receptor signalling pathways, were not detected. Improved molecular subclassification of canine PC is needed to direct selection of relevant cases for modelling the human disease and to ensure appropriate extrapolation between canine and human studies.