Integrated proteomics–driven discovery of synovial fluid biomarkers reflecting structural damage and pain in knee osteoarthritis

Osteoarthritis (OA) is a heterogeneous whole-joint disease, and the frequent mismatch between radiographic damage and pain, together with the lack of biomarkers, has hindered disease-modifying drug development. We aimed to identify synovial fluid (SF) biomarkers associated with both structural severity and symptomatic burden. SF from 43 participants in the OsteoArth cohort was profiled by mass spectrometry, quantifying 412 peptides. Differentially expressed peptides were identified across Kellgren–Lawrence (KL) grades and WOMAC pain subgroups, then screened by 100-bootstrap univariable analyses and refined using elastic net regression. Selected features were evaluated by multivariate logistic regression and further assessed in independent SF and serum proteomics datasets, with additional measurement by ELISA in a larger SF subset. Of 95 differentially expressed peptides, 13 were associated with KL grade and 24 with WOMAC pain. Lumican (LUM) and complement factor I (CFI) were identified across multiple analytical steps and were associated with KL grade and, to a lesser extent, WOMAC pain. A model including LUM and CFI showed comparable or higher performance than CRTAC1 and MMP-3, particularly for low KL vs KL0 (AUC 0.84) and across all grades (KL1–4 vs KL0; AUC 0.76). ELISA analyses supported associations with radiographic severity and pain. LUM and CFI are candidate SF biomarkers associated with structural OA severity, with additional but limited associations with symptom-related outcomes. Further validation in larger and diverse cohorts is required to determine their clinical applicability.