High molecular heterogeneity in synchronous canine mammary tumors detected by transcriptomic analysis

Mammary tumors are among the most prevalent and clinically significant neoplastic diseases in dogs, yet their molecular heterogeneity is poorly understood. This study aimed to characterize gene expression in synchronous tumors from the same dog and assess whether these tumors share molecular profiles. A transcriptomic study was conducted using RNA sequencing data from 179 canine mammary tumors to define transcriptomic groups through unsupervised consensus clustering and modules from gene co-expression network analysis. A subset of 45 dogs with synchronous tumors ( n = 90) was then used to evaluate intra-individual transcriptomic concordance by comparing cluster assignment and module expression between tumor pairs, classified as malignant-malignant, benign-benign, and malignant-benign. Additionally, interclass correlation coefficients were applied to identify highly correlated genes within each category. The analysis identified five transcriptomic clusters based on overall gene-expression patterns. GSEA revealed that all five clusters potentially resemble molecular subtypes described in human breast cancer, including luminal-, basal-, and claudin low-like. We found that synchronous tumors were heterogeneous with respect to both cluster assignments and gene-expression patterns, indicating low intra-individual similarity. This heterogeneity was further supported by low interclass correlations across most genes in synchronous tumors, independent of tumor malignancy. Our findings demonstrate substantial molecular heterogeneity of canine mammary tumors, both between and within individuals. The results highlight the value of expression-based approaches for characterizing biologically relevant processes in mammary tumors.