An international team led by Bram Van Mol of Ghent University and KU Leuven, working with colleagues in Belgium and France, set out to perform joint-specific genome-wide association studies, known as GWAS, and to estimate SNP-based heritabilities for equine osteochondrosis and fetlock osteochondral fragments. Rather than lumping all lesions together, the researchers defined seven base phenotypes with two to four levels of specificity using a novel semi-quantitative scoring system, and paired these carefully stratified definitions with high-density genotypes. The work addressed two long-standing obstacles in the field: the difficulty of assembling a study population large enough to detect genetic signals, and the inconsistent phenotype definitions that have made results across studies hard to compare.

After rigorous quality control, the final study population comprised 716 horses, of which 442 were male and 274 female. Warmblood horses accounted for 99.9 percent of the cohort, and 84 percent were registered with one of three Belgian warmblood studbooks. The mean age at radiographic screening was 36 months, an age chosen to balance the detection of persistent lesions against the age-dependent dynamics that cause some osteochondrosis lesions to resolve on their own as young horses mature. The researchers also had to contend with a subtle source of misclassification: horses whose osteochondral fragments had been surgically removed before screening may show little or no radiographic evidence of the original lesion, which can falsely place affected animals in the control group.

The scale of the phenotyping effort became clear when cases and controls were tallied at the least stringent level of specificity. For total osteochondrosis, 307 cases were compared against 325 controls. When the analysis drilled down to individual joints, the numbers shifted: 187 cases and 478 controls for hock osteochondrosis, 165 against 504 for lesions of the distal intermediate ridge of the tibia, the site classically known as DIRT, 110 against 588 for stifle osteochondrosis, and 51 against 536 for fetlock osteochondrosis. Two additional phenotypes captured fetlock osteochondral fragments at distinct locations within the proximal phalanx: 76 cases of palmaro- or plantaroproximal fragments, abbreviated POF, and 100 cases of dorsoproximal fragments, abbreviated DOF, each compared with more than 580 controls.

Across 23 separate GWAS runs, the team identified 214 suggestive single nucleotide polymorphisms, of which 50 reached the threshold of genome-wide significance. Applying predefined criteria to cluster these signals, the researchers pinpointed eight genomic regions of interest harboring eleven novel candidate genes. For total osteochondrosis, the strongest new candidate was LDB2. Hock osteochondrosis pointed to SCUBE3, MAPK13, and MAPK14. The palmaro- and plantaroproximal fragments of the proximal phalanx yielded the richest harvest, with FGF6, FGF23, IL17A, IL17F, TRAM2, BMP5, and PPP2R5C emerging as candidates. Notably, no candidate genes were detected for DIRT lesions, and for stifle osteochondrosis, fetlock osteochondrosis, and the dorsoproximal fragments, suggestive SNPs failed to define any genomic regions of interest at all. Many of the implicated genes participate in angiogenesis, the formation of new blood vessels, and in bone development, two biological processes long suspected to underlie the failure of endochondral ossification that defines osteochondrosis.

The heritability estimates, which quantify the proportion of variation in a trait attributable to genetic differences, proved even more striking than the gene discoveries. SNP-based heritability ranged from 26.2 to 27.1 percent for total osteochondrosis, but the joint-specific estimates told a far more nuanced story. Hock osteochondrosis showed heritabilities of 53.9 to 61.0 percent, and DIRT lesions reached 49.4 to 61.2 percent, both remarkably high values indicating that these lesions are strongly under genetic control. By contrast, stifle osteochondrosis ranged from just 1.5 to 14.2 percent, fetlock osteochondrosis from 0.0 to 13.7 percent, POF from 10.0 to 54.8 percent, and DOF from 5.7 to 39.6 percent. The interpretation, the authors emphasize, is that even minor changes in phenotype definition can substantially alter both heritability estimates and the detection of genomic associations.

These findings carry immediate practical consequences for horse breeding. The high heritability estimates for hock and DIRT lesions indicate considerable potential for genetic improvement through selection: if more than half of the variation in these conditions is genetic, breeding values estimated from genomic data could meaningfully reduce their prevalence over time. Conversely, the low heritabilities observed for total osteochondrosis, fetlock osteochondrosis, and stifle lesions suggest either that the genotyped SNPs incompletely capture the underlying genetic variation, leading to an underestimation of the true genetic contribution, or that non-genetic factors such as nutrition, exercise, and growth rate genuinely play a larger role in those joints. Distinguishing between these possibilities will require larger, better-characterized datasets.

The study also delivers the first DOF-specific GWAS in horses, establishing a methodological template for dissecting disorders that were previously treated as single diagnostic categories. The semi-quantitative scoring system developed by the team allowed lesions at different anatomical locations within the same joint to be separated, exposing genetic signals that a coarser classification would have diluted or masked. This level of granularity matters because the candidate genes identified for one lesion type were largely absent from others, reinforcing the conclusion that osteochondrosis is not one disease but a family of related developmental disorders with distinct, and partially independent, genetic underpinnings.

Looking forward, the authors argue that future research should prioritize standardized joint-specific phenotyping, the assembly of large multi-population datasets, and validation of the identified loci, candidate genes, and heritability estimates. Only with such rigor, they contend, can genomic information be translated into breeding strategies that improve osteoarticular health in sport horses while preserving genetic diversity within breeds. For an industry in which joint disease remains a leading cause of early retirement and veterinary expense, the message of this study is both cautionary and encouraging: precision in how we define a disease determines whether genetics can help us conquer it, and for at least some forms of equine osteochondrosis, the genetic tools for meaningful improvement are already within reach.

Subject of Research: Genome-wide association and heritability analysis of joint-specific equine osteochondrosis and fetlock osteochondral fragments

Article Title: Leveraging joint-specific phenotypes for genome-wide association studies and SNP heritability estimation of equine osteochondrosis and fetlock osteochondral fragments

Article References: Van Mol, B., Janssens, S., Meyermans, R., Chapard, L., Oosterlinck, M., Pille, F., & Buys, N. (2026). Leveraging joint-specific phenotypes for genome-wide association studies and SNP heritability estimation of equine osteochondrosis and fetlock osteochondral fragments. BMC Genomics. https://doi.org/10.1186/s12864-026-13364-9

Image Credits: AI Generated

DOI: 10.1186/s12864-026-13364-9

Keywords: equine osteochondrosis, fetlock osteochondral fragments, GWAS, SNP-based heritability, joint-specific phenotypes, horse genomics, warmblood, candidate genes, bone development, angiogenesis, animal breeding, BMC Genomics