The choice of dogs as a comparative model is not arbitrary. Unlike laboratory mice, whose tumors are typically chemically induced and often fail to capture the complexity of human disease, dogs develop cancers spontaneously, with intact immune systems and naturally evolving tumor microenvironments. Over the past decade, accumulating evidence has shown that canine cancers closely resemble human cancers in prevalence, growth patterns, histology and genetic foundations. Yet hepatocellular carcinoma in dogs differs from the human disease in one crucial respect: it is not typically associated with viral hepatitis, cirrhosis or alcohol consumption, the dominant risk factors in people. That distinction offers scientists a rare opportunity to study the intrinsic genetic drivers of liver cancer, stripped of the confounding influences of chronic inflammation and viral infection that dominate the human disease.

To build this cross-species picture, the researchers combined two powerful technologies. First, they performed whole exome sequencing on liver tumors and matched blood samples from ten dogs diagnosed with hepatocellular carcinoma, all confirmed by board-certified veterinary pathologists. The sequencing achieved an average depth of one hundred times coverage, with more than ninety-three percent of reads successfully mapped to the canine reference genome. Second, they applied single nucleus RNA sequencing to flash-frozen tumor biopsies from two of those dogs, profiling nearly twenty-four thousand individual nuclei to map the cellular composition of the canine tumor microenvironment. The exome data revealed a total of 1,895 non-synonymous single nucleotide variants across the cohort, averaging 189 mutations per sample, from which 645 mutations with high, moderate or low predicted impact remained after filtering.

The mutational landscape that emerged contained a familiar villain. The most recurrent missense mutation in the dog tumors was in the CTNNB1 gene, which was altered in five of the ten samples, or fifty percent of the cohort. CTNNB1 encodes beta-catenin, the central component of the Wnt/beta-catenin signaling pathway, which plays a key role in liver cancer development, tumor progression and immune evasion. The gene is also one of the most frequently mutated drivers in human hepatocellular carcinoma, present in roughly twenty-two percent of cases. More remarkably, when the researchers mapped the positions of the CTNNB1 mutations in both species, they found identical amino acid substitutions at the same protein locations: the D32Y, G34V and S37F changes appeared in five, one and one dog respectively, and in forty, fifty-four and thirty-nine human samples in the COSMIC database. A search of the Drug Gene Interaction Database identified sixty-six interactions for CTNNB1, including sixteen involving FDA-approved therapies, hinting at possible treatment avenues pending appropriate validation.

Beyond CTNNB1, the analysis identified ten putative driver genes mutated in at least twenty percent of the dog tumors, including frameshift variants in ARRDC3, found in three dogs but notably absent from human hepatocellular carcinoma data. The team also calculated tumor mutation burden, a metric that in humans helps predict response to immunotherapy. The median burden in the dog cohort was 4.01 mutations per megabase, with a range of 2.31 to 5.96, which sits comfortably within the estimated human hepatocellular carcinoma median range of 2.56 to 5 mutations per megabase. This similarity suggests that the biological determinants of immune visibility may be comparable across species, a finding with direct implications for designing canine immunotherapy trials that could, in turn, inform human studies.

The single nucleus sequencing side of the study painted an equally detailed portrait of the tumor’s cellular ecosystem. After quality filtering and integration, the researchers identified seven major cell types in the dog tumors: fibroblasts, hepatocytes, macrophages, malignant cells, proliferating cells of myeloid origin, endothelial cells and T cells. Malignant cells were distinguished by the overexpression of genes such as CP, APOB, DST and PTGR1, which had previously been reported in malignant cells from human single-cell studies of the disease. To bolster the identification, the team scored each cell type against a published cancer stemness gene signature and found that malignant cells carried significantly higher stemness scores than normal hepatocytes, with a p-value below 2.2 times ten to the minus sixteenth. Both malignant cells and proliferating myeloid cells also showed elevated S and G2M cell cycle phase scores, confirming their proliferative character.

The most consequential comparison came when the researchers integrated the canine data with single nucleus RNA sequencing data from two human patients with non-viral hepatocellular carcinoma. Using the scVI computational framework to align the two species’ transcriptomes, they found that non-epithelial cells integrated far more effectively than epithelial ones. T cells, macrophages, fibroblasts and endothelial cells from dogs and humans formed shared clusters corresponding to their cell types, while hepatocytes remained largely species-specific, and dog malignant cells refused to merge with human epithelial cells at all. A Jaccard similarity analysis confirmed the pattern: human macrophages were most similar to dog macrophages, and conserved gene features included well-established macrophage markers such as MARCO and CD163, along with T cell markers including CD96, SKAP1, ITK and FYN. The authors interpret this conservation as reflecting shared evolutionary pressures on immune surveillance and stromal maintenance, functions that are critical across mammalian species.

Cell-to-cell communication analysis added another layer of insight. Using CellPhoneDB to infer ligand-receptor interactions, the team identified 222 significant pairs in the dog tumors and 227 in the human samples, spanning dozens of signaling pathways. In both species, fibroblasts, endothelial cells and malignant cells emerged as the strongest communicators. Crucially, several key interactions were conserved across species, including IGF1-IGF1R, ANGPT1-TEK, JAG1-NOTCH2 and THBS1-CD36 between malignant cells and fibroblasts, all linked to cancer stemness, angiogenesis and immune evasion, as well as PTPRC-MRC1 and APP-CD74 between macrophages and malignant cells. But the analysis also exposed species-specific crosstalk: dog tumors displayed unique interactions such as IL34-CSF1R and LGALS9-P4HB, both associated with macrophage regulation and tumor-driven immunosuppression, while human tumors showed Chemerin-CMKLR1 and GAS6 interactions with the TAM family receptors Tyro3, Axl and MerTK, known promoters of tumor progression.

The study is not without limitations, and the authors are candid about them. The canine cohort is small, and the single nucleus approach carries technical constraints: copy number variation inference, a common tool for distinguishing malignant from non-malignant cells in human single-cell studies, was limited in the dog data by lower resolution, high dropout rates and the absence of a high-quality normal liver reference genome. These constraints likely explain the clustering proximity between malignant and normal hepatocyte populations in the canine data and caution against over-interpreting some aspects of malignant cell biology. Nevertheless, the work represents the first integrated analysis of potential genetic drivers in canine hepatocellular carcinoma, and the raw sequencing data have been deposited in public repositories under accession numbers PRJNA1237364 and GSE292303, making the resource freely available to the research community.

The broader significance of the study lies in what it sets up. If dogs and humans share the same driver mutations, comparable tumor mutation burdens and conserved immune and stromal communication networks in liver cancer, then canine patients with hepatocellular carcinoma could serve as a genuine translational bridge, testing immunotherapies and targeted agents in a spontaneous disease model that laboratory mice cannot replicate. Conversely, the species-specific differences in malignant cell transcription and signaling interactions serve as a reminder that extrapolation between species demands care. As larger canine cohorts are sequenced and functional validation studies follow, the humble pet dog, sharing our homes and increasingly our cancers, may prove to be one of the most valuable research partners oncology has ever had, accelerating the discovery of therapeutic targets for both veterinary and human liver cancer patients.

Subject of Research: Comparative multi-omics analysis of the tumor microenvironment in canine and human hepatocellular carcinoma

Article Title: Comparative multi-omics evaluation of the tumor microenvironment in dog and human hepatocellular carcinoma

Article References: Nussbaum, Y. I., Bryan, J. N., Li, G., Rodney, A., Culp, W. T. N., Dickinson, P. J., Koutoulas, A., Megquier, K., An, H., Kramer, S. T., Ricemeyer, E. S., Mitchem, J. B., & Warren, W. C. (2025). Comparative multi-omics evaluation of the tumor microenvironment in dog and human hepatocellular carcinoma. Veterinary Oncology, 2(1), Article 25. https://doi.org/10.1186/s44356-025-00037-0

Image Credits: AI Generated

DOI: 10.1186/s44356-025-00037-0

Keywords: hepatocellular carcinoma, comparative oncology, canine cancer, tumor microenvironment, CTNNB1, single nucleus RNA sequencing, whole exome sequencing, tumor mutation burden, cell-cell communication, macrophages, immunotherapy, liver cancer