Researchers build largest dog and cat tumor database: full analysis

CURRENT FULL VERSION: Researchers at the University of Liverpool and the University of Las Palmas de Gran Canaria have launched what they describe as the world’s largest open-source database of dog and cat tumors, crossing the 1 million record mark and giving veterinary cancer researchers a much larger shared dataset to study risk patterns in companion animals. The project, built through Liverpool’s Small Animal Veterinary Surveillance Network, or SAVSNET, is positioned as a new infrastructure play for veterinary oncology rather than a single discovery study. (liverpool.ac.uk)

The announcement builds on earlier SAVSNET tumor-registry work. In 2021, the group published a text-mined pathology dataset covering more than 100,000 canine and feline tumors from UK diagnostic laboratories, while project materials later described plans to expand the registry several-fold using additional pathology data and improved natural-language processing. That expansion now appears to have moved from concept to large-scale implementation. (nature.com)

The first major analysis highlighted alongside the announcement focuses on dogs. In a February 2026 paper in Veterinary and Comparative Oncology, the team used 130,998 histologically confirmed tumors from a pathology registry containing more than 1.1 million canine tumor records from 1.02 million dogs in the UK between 2010 and 2023. The paper examined mast cell tumors, melanoma, osteosarcoma, and hemangiosarcoma, and reported strong breed associations, including elevated mast cell tumor odds in Bulldog-related breeds and Retrievers, high melanoma odds in breeds including Rottweilers and Shar Pei, and increased hemangiosarcoma odds in German Shepherd Dogs, Mastiffs, and Bullmastiffs. The authors also reported that neutered dogs generally had higher tumor odds than entire dogs, with the difference often more pronounced in females. (livrepository.liverpool.ac.uk)

That scale matters because veterinary cancer data has historically been fragmented across private diagnostic labs, referral centers, and local registries. In the Liverpool announcement, professor David Killick said the goal was to bring those records into a “meaningful, research-ready database,” while co-author José Rodríguez Torres said the field has lagged human oncology because animal cancer data has been scattered. Independent literature supports that framing: a recent narrative review on cancer registration in dogs and cats argues that better standardization and data sharing are still needed across veterinary oncology. (phys.org)

There’s also a broader industry and research context here. Comparative oncology has been moving toward larger, linked datasets that connect pathology, genomics, and outcomes. That is especially relevant on the feline side, where a February 2026 Science study created the first large-scale genetic map of feline cancer by sequencing 493 tumor-normal tissue pairs across 13 tumor types from cats in five countries and assessing 978 feline orthologs of roughly 1,039 human cancer genes. The study found that the feline oncogenome shares multiple recurrent drivers with human cancers, including TP53, FBXW7, CTNNB1, PTEN, and TRAF3, with TP53 the most frequently mutated gene overall, present in about one-third of tumors. The most represented tumor types included cutaneous squamous cell carcinoma, lung adenocarcinoma, lymphoma, and mammary carcinoma, and the authors reported that 14% of feline tumors carried an oncogenic or likely actionable alteration. (cancerletter.com)

Some of the feline findings are clinically interesting in their own right. Mammary carcinoma, a common and aggressive feline cancer, showed strong parallels to human breast cancer: the study identified seven driver genes in feline mammary tumors, with FBXW7 altered in more than half of cases and PIK3CA mutations also common, including the hotspot p.H1047R. Researchers also reported that mammary carcinoma had the highest mean genome fraction altered among the tumor types studied, while preliminary lab testing suggested FBXW7-mutant mammary tumors could be more sensitive to certain chemotherapy drugs. In T-cell lymphoma, MYC copy-number gain appeared in 57% of cases, far above the all-tumor average, raising the possibility of more targeted therapeutic work over time. These are not practice-changing treatment data yet, but they do show how large-scale feline genomics could become clinically useful if linked to pathology and outcomes datasets. (vetmeduni.ac.at)

That comparative framing is part of why the Liverpool-Las Palmas registry could matter beyond descriptive epidemiology. Cats and dogs develop naturally occurring cancers while sharing many household and environmental exposures with people, which makes them useful comparative models in ways laboratory systems often are not. Recent reporting around the feline oncogenome project emphasized exactly that point, arguing that veterinary oncology is moving toward treating specific mutations and pathways, not just species or tumor labels. A large open pathology registry does not deliver that by itself, but it can provide the case volume and phenotypic backbone needed to connect genomic findings to real-world populations. (vetmeduni.ac.at)

Why it matters: For practicing veterinarians and specialists, a registry like this could eventually improve how risk is discussed with pet parents, when clinicians choose to biopsy or monitor certain lesions more aggressively, and how breed- and sex-associated cancer patterns inform preventive counseling. It may also help researchers design better epidemiology studies and identify cohorts for future translational work. The parallel feline genomics work adds another layer: it suggests that at least some companion-animal tumors may be stratified by actionable molecular features, with potential relevance to both veterinary precision medicine and human oncology. But there’s an important caveat: pathology-based registries are not the same as population-based cancer registries. The earlier SAVSNET data descriptor explicitly warned that pathology submissions should not be used on their own to estimate incidence in the full UK dog and cat population, so clinicians should view these findings as strong surveillance signals and hypothesis-generating evidence, not the final word on absolute risk. (nature.com)

The comparative angle is especially notable given parallel advances in feline cancer genomics. Separate 2026 reporting around a 493-sample feline oncogenome project has underscored how companion-animal cancer datasets are becoming more useful not just for veterinary care, but for cross-species cancer biology and drug development. In that sense, the Liverpool-Las Palmas registry could become more valuable over time if tumor records are increasingly linked with genomic, treatment, and outcome data. That’s an inference based on the direction of the field, but it fits the way comparative oncology is evolving. (vetmeduni.ac.at)

What to watch: Expect follow-on papers from this registry, likely by tumor type and species, and watch for whether the group expands access, publishes more methodological detail, or links the database to genomic and longitudinal clinical data that could make it more useful at the point of care. Feline follow-up work would be especially worth watching in mammary carcinoma, lymphoma, lung adenocarcinoma, and cutaneous squamous cell carcinoma, where the new genomic map already suggests shared drivers and possible therapeutic leads. (liverpool.ac.uk)