Lectin Microarray-based Glycomics and Machine Learning Identify Shared Osteoarthritis Biomarkers in Humans, Dogs, and Horses

Post-traumatic osteoarthritis is a common sequela of joint injury in humans and companion animals, yet glycosylation changes associated with osteoarthritis remain poorly characterized. Lectin microarray analysis enables high-throughput profiling of glycan features in complex biofluids such as synovial fluid. Here, we used lectin microarrays to characterize synovial fluid glycosylation patterns in healthy and osteoarthritis-affected joints from horses, dogs, and humans. Comparative lectin-binding profiles within and across species revealed both conserved and species-specific glycomic signatures associated with osteoarthritis. Supervised machine learning distinguished osteoarthritis from control joints within a pooled multi-species framework and identified the lectins and their glycan epitopes contributing most strongly to classification, though cross-species generalization was limited and depended on the composition of each species' control cohort. Elevated high-mannose emerged as the shared glycomic signature across species. Together, these findings support the feasibility of synovial fluid glycomics, integrated with machine learning, for identifying conserved osteoarthritis-associated glycan signatures; and inform the design of glycan-based diagnostic and translational strategies in both veterinary and human medicine.