The study, published in Frontiers in Animal Science, was led by Andrew Knight of Murdoch University, together with Alexander Bauer and Jenny L. Mace. The team conducted a cross-sectional online survey of companion animal carers, open from May to December 2020, which gathered information on the health of one resident cat per household that had lived there for at least a year. From a larger pool of 4,057 dog and cat carers, 1,397 cat carers met the inclusion criteria, and 1,305 cats were ultimately analysed across the two diet groups of interest: 1,178 cats, or 90.3 percent, fed conventional meat-based diets, and 127 cats, or 9.7 percent, fed vegan diets for at least one year. Carers reported the assessments of their primary veterinarians concerning the cats’ health over the previous twelve months.

The researchers focused on six body weight-related disorders: overweight, underweight, osteoarthritis or arthritis, intervertebral disc disease, other mobility disorders, and other musculoskeletal problems. According to the carer-reported veterinary assessments, only 3.1 percent of vegan-fed cats were classified as overweight, compared with 4.1 percent of meat-fed cats. In unadjusted statistical analyses, cats fed vegan diets had odds of being overweight that were 24 percent lower than their meat-fed counterparts. When the researchers adjusted for ten potential confounding variables, including age, sex, neuter status, activity level and a range of feeding practices, the difference widened to 34 percent lower odds of overweight in the vegan group.

Here is the crucial caveat: neither figure reached statistical significance. The confidence intervals were wide, spanning from a 73 percent reduction to a 116 percent increase in the unadjusted overweight analysis, reflecting the relatively small number of observed cases. Only four vegan-fed cats were classified as overweight at all. The authors interpreted the raw 24 percent effect as a “tendency” and the adjusted 34 percent effect as a “strong tendency,” using heuristic categories they established in earlier work on general health outcomes from the same survey. Following the American Statistical Association’s guidance on p-values, they emphasised effect strength over binary significance thresholds, but they are explicit that the findings should be treated as hypothesis-generating rather than confirmatory.

The pattern extended to underweight as well. Cats fed vegan diets showed 66 percent lower odds of being underweight in the unadjusted models, a difference the authors labelled a strong tendency, though again it was not statistically significant. More striking still were the secondary outcomes: not a single vegan-fed cat in the sample was reported to suffer from osteoarthritis or arthritis, intervertebral disc disease, other mobility disorders, or other musculoskeletal problems. Among the meat-fed group, 15 cats had arthritis impairing mobility, one had intervertebral disc disease, eight had other mobility problems and 14 had other musculoskeletal problems. Because zero cases occurred in the vegan group, the researchers could not run regression models for these outcomes at all, and the low prevalence likely reflects the small size of the vegan subgroup rather than any proven protective effect.

The statistical machinery behind the study was deliberately cautious. The analysis proceeded in three steps: a descriptive stage, an exploratory stage estimating the diet effect with ten separate one-variable regression models, and a final fully adjusted model restricted to outcomes with at least 50 cases across both diet groups, a criterion met only by overweight. That final model produced an area under the curve of 0.78 on a held-out test set, indicating acceptable fit. Two sensitivity analyses bolstered the findings: propensity score matching on age, which matters because vegan-fed cats averaged 6.2 years old versus 8.2 years for meat-fed cats, yielded broadly similar reductions of 21 percent in overweight and 50 percent in underweight odds. And when 64 raw-meat-fed cats were folded into the meat-based comparison group, the results barely moved, with the adjusted overweight effect shifting by just one percentage point.

Why might plant-based diets be associated with leaner cats? The authors cannot answer definitively, since they collected no compositional data, but they outline plausible mechanisms. Nutritional guidelines for weight management in cats favour lower calories, lower fat, higher fibre and higher L-carnitine content. Plant-derived ingredients often contain more fibre and potentially fewer calories or less fat than animal sources. More intriguingly, conventional cat foods are not required to guarantee L-carnitine, a vitamin-like amino acid derivative critical for energy metabolism and the retention of lean body tissue, because manufacturers assume meat supplies adequate quantities. Some nutritionally sound vegan cat foods, by contrast, are explicitly supplemented with L-carnitine, potentially offering cats greater protection against weight gain. There is also a paradox worth noting: vegan-fed cats in this sample were actually more often fed dry food and fed ad libitum, both factors previously associated with increased overweight risk, yet they still showed lower odds of excess weight.

The study also surfaced a striking discrepancy in how weight was judged. Veterinarians reportedly considered 94 percent of meat-fed cats and 96 percent of vegan-fed cats to be at ideal weight, while carers’ own opinions put those figures at 70 percent and 76 percent respectively. Usually the reverse is true, with owners less likely than vets to recognise overweight. The authors suggest several explanations: carer concern may capture perceived adiposity rather than clinically communicated problems; body condition scoring is under-recorded in practice, with a UK study of 129,076 cat visits finding weight recorded in only 10 percent of visits; and surveys may attract more health-conscious carers. The very low 4 percent veterinary-reported overweight prevalence in this sample contrasts sharply with estimates of 30 to 60 percent from veterinary-assessed studies, which may mean the study underestimated overweight overall, a limitation the authors acknowledge candidly.

That said, the limitations extend further. All outcomes relied on carer reports, introducing potential recall and reporting bias, and carer-reported assessments are not equivalent to standardised veterinary body condition scoring. Social desirability bias among vegan-feeding carers, possible confounding by unmeasured health-protective behaviours, and survivorship bias, in which cats harmed by vegan diets are switched back to meat and vanish from the vegan group, cannot be excluded. The analysis also excluded endocrine, urinary, cardiac and dermatological comorbidities of obesity. The fully adjusted model, with only four overweight events in the vegan group, fell short of the traditional ten-events-per-parameter rule of thumb for logistic regression.

What the study does deliver is a consistent direction of effect across every model the authors ran, a robustness to alternative analytical choices, and a clear roadmap for future work. Larger samples, prospective designs, direct extraction of veterinary-recorded body weights and standardised body condition scores, and measurement of actual diet composition and energy intake would all help resolve whether the leaner-cats pattern is real. For now, the message is nuanced: among these 1,305 cats, vegan-fed animals fared no worse on body weight and related disorders than meat-fed ones, and descriptively better, but nobody should yet conclude that vegan diets slim down cats. This is exploratory science, published transparently with its uncertainty on display, precisely so that the next generation of studies can test the hypothesis properly.

Subject of Research: The association between vegan versus meat-based diets and body weight-related disorders in domestic cats

Article Title: Impacts of a vegan diet on body weight and related health disorders in 1,305 cats: an exploratory cross-sectional survey

Article References: Knight, A., Bauer, A., & Mace, J. L. (2026). Impacts of a vegan diet on body weight and related health disorders in 1,305 cats: an exploratory cross-sectional survey. Frontiers in Animal Science, 7, Article 1939965. https://doi.org/10.3389/fanim.2026.1939965

Image Credits: AI Generated

DOI: 10.3389/fanim.2026.1939965

Keywords: cats, vegan cat food, feline obesity, overweight, body condition score, pet nutrition, veterinary epidemiology, cross-sectional survey, logistic regression, obligate carnivores, L-carnitine, pet health

The technology at the heart of the study is the OncoK9 test, a multi-cancer early detection assay developed by PetDx and clinically available for dogs since 2021. The test exploits the biology of circulating tumor DNA: as cancer cells grow and die, they shed fragments of DNA into the bloodstream. These cell-free DNA fragments carry somatic genomic alterations, acquired mutations and chromosomal changes that are hallmarks of malignancy. Using next-generation sequencing on blood drawn from a jugular or peripheral vein, the assay reads these fragments, aligns the sequences against the canine reference genome CanFam3.1, and applies custom bioinformatics pipelines to distinguish cancer-associated signals from background genomic noise. A result is returned to the veterinarian within seven to fourteen days, typically as a binary call: Cancer Signal Detected or Cancer Signal Not Detected.

Enrollment began in December 2021 under two institutional animal care protocols spanning 24 clinical sites across the United States and Canada, a mix of 21 general practices, two academic institutions and one specialty center. In total, 725 presumed cancer-free dogs joined the study, most of them classified as being at increased risk for cancer based on age, breed or body weight. Dogs aged seven years or older were eligible regardless of breed, as were younger dogs belonging to breeds with elevated lifetime cancer risk, and the cohort ranged in age from 1.75 to 20.2 years with a median of 8.4 years. Golden Retrievers dominated the breed list with 68 enrolled dogs, followed by Labrador Retrievers, German Shepherd Dogs, Boxers and Rottweilers. At each routine follow-up visit, conducted annually or every six months depending on the protocol, dogs received a complete physical examination and a blood draw of 14 to 17 milliliters collected into specialized tubes that keep cell-free DNA stable at ambient temperature for a week, requiring no refrigeration or immediate processing.

The first interim analysis focused on the 419 dogs that had been enrolled for at least a year with two or more screening visits, or that had received a cancer diagnosis while on study, with a mean observation period of 422 days. Within this group, 51 dogs were newly diagnosed with cancer, 37 with definitive diagnoses confirmed by pathologist-reviewed histopathology or cytology and 14 with presumptive diagnoses. The most common anatomic locations of disease were the liver, skin, bone, heart, spleen, lung and lymph nodes, a distribution that underscores why genomic screening matters: many of these cancers arise deep within the body where physical examination alone cannot detect them until they are advanced.

The performance metrics of the liquid biopsy test paint a nuanced picture. Of the 51 cancer-diagnosed dogs, 29 received a Cancer Signal Detected result, yielding an observed sensitivity, or detection rate, of 56.9 percent. Specificity was strikingly high at 98.9 percent, meaning only four dogs out of 366 that remained cancer-free received a false positive result, a false positive rate of just 1.1 percent. The positive predictive value reached 87.9 percent, indicating that nearly nine out of ten positive screening results were followed by an actual cancer diagnosis, and the negative predictive value was 94.3 percent. Notably, these figures closely match the test’s performance in its earlier clinical validation study and in a retrospective review of more than 1,500 real-world clinical cases, a consistency the authors attribute to the use of independent training and testing sets during validation, a design strategy intended to guard against statistical overfitting and inflated performance estimates.

Perhaps the most consequential finding concerns how cancer was actually detected. Under conventional veterinary practice, only 12 percent of the cancers in this cohort, six of 51 cases, were caught preclinically through wellness examinations or routine care alone, a figure consistent with a prior retrospective audit of more than 350 dogs in which just 4 percent of cancers were detected at routine visits. When the NGS-based test was added, workups prompted solely by a detected cancer signal led to the preclinical diagnosis of 22 additional cases, lifting the preclinical detection rate to 55 percent, a 4.6-fold increase. Overall, the test accounted for 26 of the 51 cancer diagnoses, effectively doubling the total number of cancers found in the population during the study period.

The clinical pathway following a positive result proved efficient and accessible. Among the 29 true-positive dogs, 86 percent had their first detected signal at a regularly scheduled visit rather than an unscheduled sick visit, and once a positive result triggered confirmatory evaluation, cancer was typically diagnosed within weeks, with a median time to diagnostic resolution of 23 days. More than half of these dogs reached diagnostic resolution within one month, and 92 percent within six months. Critically, in 27 of the 29 true-positive cases, cancer was localized using the diagnostics recommended in the study’s confirmatory evaluation, comprehensive physical exam, routine laboratory tests, three-view thoracic radiographs and abdominal ultrasound, all procedures commonly available in general practice, with only two dogs requiring advanced modalities such as echocardiography or magnetic resonance imaging. Among the 20 definitively diagnosed true-positive dogs, 14 had localized disease confined to the organ of origin, suggesting the test can identify cancer early enough to offer a genuine therapeutic window.

The study also documented rare false positives and a meaningful fraction of false negatives. The four false-positive dogs all underwent substantial confirmatory workups, including laboratory testing, thoracic and abdominal imaging, and in some cases echocardiography, fine needle aspiration or computed tomography, and none was diagnosed with cancer during continued follow-up. The authors caution, however, that some apparent false positives could represent ultra-early detection: human studies have retrospectively identified cancer-derived DNA alterations in plasma three to four years before clinical diagnosis, and the earlier CANDiD validation study reported detecting tumor-associated alterations five months before a hemangiosarcoma diagnosis in one dog. Longer follow-up, they note, will be needed to adjudicate such cases. Among the 22 false-negative dogs, more than a third carried small skin tumors under two centimeters, masses that the test was never intended to replace physical examination or fine needle aspiration in evaluating, and detection rates are known to vary with tumor size, extent of disease and anatomic location, with aggressive cancers like lymphoma and hemangiosarcoma more readily detectable than anatomically isolated tumors of the bladder, prostate or central nervous system.

Owner and veterinarian acceptance of the screening program was overwhelmingly positive. Of 343 owners who responded to satisfaction surveys, average agreement scores with statements about the value of annual or semi-annual screening hovered around 9.3 to 9.6 on a 10-point scale, and 297 of 343 respondents said they were likely or very likely to pay for recommended screening. Veterinarians rated the clarity of the test reports and their usefulness in clinical management at 9.5 or higher on average, though the survey response rate among clinicians was small, with only 10 of 44 questionnaires returned. Safety data were equally reassuring: of 725 enrolled dogs, just 35 adverse events were reported in connection with blood draws, nearly all minor hematomas or bruising at the venipuncture site, and no adverse events occurred during diagnostic workups prompted by positive results.

The study is not without limitations, and the authors are candid about them. The interim analysis covers a relatively short observation window, and key questions remain unresolved, including whether the optimal screening interval is six or twelve months, whether earlier detection ultimately improves stage-specific survival, and how to account for owners who self-select into longer screening intervals. Because dogs lack a standardized national cancer registry analogous to the SEER program in human oncology, precise canine cancer incidence remains uncertain, making the prospectively documented 12 percent period prevalence in this predominantly high-risk cohort a valuable new data point. The study was funded by PetDx, the test’s developer, and PetDx employees participated in the research. Still, the investigators frame CLASSiC as a milestone: the first evidence that incorporating genomic liquid biopsy into regularly scheduled wellness visits can substantially increase preclinical cancer detection in dogs, achieved with a simple blood collection that any clinical practice can perform, and a foundation for the evidence-based screening guidelines that veterinary medicine has long lacked.

Subject of Research: Prospective cancer screening in dogs using next-generation sequencing-based liquid biopsy

Article Title: The Cancer Lifetime Assessment Screening Study in Canines (CLASSiC): interim results of the first prospective cancer screening study in dogs using next-generation sequencing-based liquid biopsy

Article References: Flory, A., Gray, S., McLennan, L. M., Rafalko, J. M., Marshall, M. A., Wotrang, K., Kroll, M., Flesner, B. K., O’Kell, A. L., Cohen, T. A., Ruiz-Perez, C. A., Sandford, E., Clavere-Graciette, A., Phelps-Dunn, A., Motalli-Pepio, R., McCleary-Wheeler, A. L., Nakashe, P., Cristobal, M. A., Anderson, P., … Grosu, D. S. (2026). The Cancer Lifetime Assessment Screening Study in Canines (CLASSiC): interim results of the first prospective cancer screening study in dogs using next-generation sequencing-based liquid biopsy. Veterinary Oncology, 3(1), Article 12. https://doi.org/10.1186/s44356-026-00062-7

Image Credits: AI Generated

DOI: 10.1186/s44356-026-00062-7

Keywords: canine cancer, liquid biopsy, next-generation sequencing, cancer screening, multi-cancer early detection, cell-free DNA, veterinary oncology, OncoK9, preclinical detection, CLASSiC study, prospective study, pet health