Transcriptomic Profiling of Canine Mammary Tumours Reveals Significant Heterogeneity Between and Within Histological Classes

This study presents a comprehensive transcriptomic analysis of 128 canine mammary tumours (CMTs), aiming to characterize their molecular landscape. Differential gene expression analysis (DGE), gene set enrichment analysis (GSEA) and clustering based on the human PAM50 gene panel were applied to explore molecular differences between the different histological categories. The analysis revealed transcriptomic differences between benign and malignant tumours as well as between tumours with high and low mitotic count. Malignant tumours were significantly enriched for gene sets associated with cell cycle regulation, proliferation, inflammatory response, signalling pathways and metabolic processes. Of the malignant tumours, purely epithelial tumours were enriched in gene sets related to proliferation, signalling and metabolic processes when compared to mixed tumours harbouring myoepithelial or mesenchymal components. To assess the relevance of human classification systems, CMTs were clustered on the PAM50 genes. The analysis revealed two clusters resembling human basal-like and luminal A subtypes, while the remaining tumours did not display expression patterns similar to human breast cancer subtypes. This suggests that there are limitations to applying the human molecular classification systems to canine tumours without adaptation. Additionally, CMTs displayed intragroup heterogeneity, where not all tumours within a histopathological category clustered together, indicating that the molecular aspect of CMTs is not necessarily reflected in the pathology. Together, this study highlights the need to develop canine-specific molecular markers based on gene expression, which could enable diagnosis prior to surgical removal of tumours, and ultimately improve treatment strategies and outcomes for dogs with mammary tumours.