Researchers unveil 1 million-record dog and cat tumor database

Researchers at the University of Liverpool and the University of Las Palmas de Gran Canaria have announced what they describe as the world’s largest open-source database of canine and feline tumors, containing more than 1 million records. According to the University of Liverpool, the resource is intended to help researchers worldwide study cancer risk factors in dogs and cats, especially patterns that have been difficult to detect when diagnostic data are scattered across separate laboratories and registries. (liverpool.ac.uk)

The announcement builds on an earlier collaboration through SAVSNET, the Small Animal Veterinary Surveillance Network. In 2021, the same research group published a pathology-based registry in Scientific Data built from UK electronic pathology records, using text-mining methods to normalize 109,895 tumors from 93,941 pathology reports gathered from three diagnostic laboratories between April 2018 and June 2019. At the time, the authors described that dataset as the largest and most comprehensive national-level animal pathology tumor registry they were aware of. (pubmed.ncbi.nlm.nih.gov)

That history matters because veterinary oncology has been working around a data problem for years. A 2025 narrative review in Research in Veterinary Science said cancer surveillance in pets remains fragmented after decades of study, with inconsistent diagnostic criteria, limited standardization, and a lack of centralized databases all reducing comparability. The review pointed to newer efforts such as the Global Initiative for Veterinary Cancer Surveillance and Veterinary Cancer Guidelines and Protocols as signs the field is moving toward shared standards, but it also argued that broader data sharing is still needed. A separate species-aware comparative oncology review made a similar point from the translational side, arguing that companion animals are increasingly useful for biomarker discovery and drug screening, but only if cross-species data standards, trial design, and veterinary-centered dosing frameworks improve alongside the science. (sciencedirect.com, mdpi.com)

Liverpool’s announcement suggests this new database is a major scale-up of that earlier model. The university said the resource contains more than 1 million canine and feline tumor records and was assembled to create a “meaningful, research-ready database” from large volumes of diagnostic information. In the university’s description, the payoff is practical: researchers should be better able to study uncommon cancers, examine breed-linked patterns, and generate evidence that was previously hidden in disconnected datasets. (liverpool.ac.uk)

Independent background literature supports why that could be consequential. The earlier SAVSNET registry showed how much signal can be extracted from routine pathology data alone, including common tumor distributions in dogs and cats and broad breed representation across the UK. Other registry efforts in Europe, including Swiss, Portuguese, Danish, and Italian datasets, have helped fill regional gaps, but they remain much smaller than a million-record resource and often reflect specific institutions or countries rather than a broader shared platform. Hospital-based epidemiology also shows the value and the limits of local datasets: a 20-year California study of 150,063 dogs and cats seen at a tertiary hospital identified 26,883 cancer cases and found that older age was the strongest predictor across major tumor groups, while some risks varied by sex and neuter status in more complex, cancer-specific ways. The authors explicitly called for a California-wide companion-animal cancer registry to provide a more complete picture than any single institution can offer. (nature.com, sciencedirect.com)

Recent feline genomics work also helps explain why bigger, better-linked animal cancer datasets matter beyond counting cases. In February, an international team published what it described as the first large-scale genetic map of feline cancer in Science, sequencing 493 tumor-normal pairs across 13 tumor types from cats in five countries and assessing feline versions of roughly 1,000 human cancer genes. The study found recurrent mutations in genes including TP53, FBXW7, CTNNB1, PTEN, and TRAF3, with TP53 mutated in about one-third of tumors overall. In mammary carcinoma, one of the most common and aggressive feline cancers, FBXW7 mutations appeared in more than half of tumors and PIK3CA alterations were also common, reinforcing parallels with human breast cancer biology. (science.org, news.cornell.edu)

That comparative angle is not just theoretical. The feline study reported that tumors with FBXW7 mutations showed increased sensitivity in laboratory testing to certain chemotherapy drugs, and it highlighted potentially actionable alterations in a subset of tumors. Those findings are preliminary and not the same as clinical trial evidence, but they point toward the kind of precision-oncology questions that become easier to ask when surveillance, pathology, and genomics can be connected across large populations. More traditional pathology literature points in the same direction clinically: a recent long-term institutional cohort of male dogs and cats with mammary tumors found these tumors were predominantly malignant in both species, with feline cases skewing more often toward intermediate- and high-grade disease, underscoring the aggressive biology of feline mammary cancer. (science.org, onlinelibrary.wiley.com)

I didn’t find substantial outside expert commentary reacting directly to this specific database announcement, but the direction of travel is clear in the literature. Recent reviews and editorials in veterinary oncology have emphasized that better data infrastructure is now central to comparative oncology, AI-enabled diagnostics, and evidence-based cancer care. At the same time, newer translational studies are showing what that infrastructure could eventually support in practice. In canine diffuse large B-cell lymphoma, for example, researchers analyzing blood samples from dogs enrolled in a prior clinical trial found that immune-gene signatures in circulating blood cells were associated with better or worse treatment response, suggesting a future role for noninvasive blood-based monitoring in personalized cancer care. That makes this database notable not just as a repository, but as a possible backbone for future epidemiology and translational research if access, data quality, and coding standards are robust. This is an inference based on the field’s stated needs and the scale described in the announcement. (sciencedirect.com, newswise.com)

Why it matters: For practicing veterinarians and oncology teams, better registry data can eventually translate into more credible breed- and age-specific risk discussions, stronger benchmarking for common and rare tumors, and improved referral and screening conversations with pet parents. It may also help the profession move closer to the kind of population-level cancer intelligence that human oncology has relied on for decades. Just as importantly, because dogs and cats share household environments with people, stronger cancer surveillance in companion animals may sharpen One Health research into environmental and genetic risk factors. And as feline genomics and canine liquid-biopsy-style studies mature, the value of a large, standardized tumor database may be in helping connect population surveillance with mutation-level and treatment-response data rather than treating those as separate tracks. (sciencedirect.com, science.org, newswise.com)

What to watch: The key next steps are formal publication, clear documentation of how the records were harmonized, and practical terms for outside access. If those pieces follow, the database could become a widely used reference point for veterinary cancer epidemiology; if not, it may remain more promising as a headline than as a working research tool. A second question is whether the platform evolves to accommodate the kinds of linked data now emerging in comparative oncology, including genomics, pathology grading, and longitudinal treatment-response markers. (liverpool.ac.uk, mdpi.com)