Background/Purpose: Biomarkers to identify osteoarthritis (OA) patients at risk for disease progression are needed. Recently, we performed a proteomic analysis of knee synovial fluid from normal and OA patients to identify differentially expressed proteins that could represent biomarkers of disease activity. Mass spectrometry assays were developed to identify representative peptides from several of these proteins in both synovial fluid and peripheral blood. We tested a panel of these putative biomarkers in a cohort of OA patients followed prospectively for 30 months to identify those peptides that correlate with radiographic progression.

Methods: Multiplexed high throughput selected reaction monitoring (SRM) assays were developed to measure multiple peptides representative of the following 9 proteins identified in previous discovery experiments: afamin, clusterin, insulin-like growth factor binding protein, acid labile subunit, lumican, pigment epithelium-derived factor, lubricin, hepatocyte growth factor, kallistatin, cartilage oligomeric matrix protein. Plasma samples obtained from baseline visit of 173 subjects in an observational OA progression cohort were trypsin digested and SRM assays were performed. Linear regression was used to determine association between each biomarker level at the baseline and maximal joint space narrowing  from baseline to 30 months. Correlation matrixes and linear regression models were then performed for the biomarkers, adjusting for age and sex. Finally, a biomarker score was developed for 2 peptides to examine the amount of progression explained by a combination of biomarkers.

Results: For this progression cohort, average age was 61 and average joint space narrowing over 30 months was 0.68 mm. A good correlation between multiple peptides for each individual protein was observed, indicating our assays were correctly measuring their target proteins. From our panel of putative biomarkers, peptides representative of clusterin, lumican and lubricin showed statistically significant associations with joint space narrowing after adjustment for age and sex. Partial R² values from these linear regression models show the clusterin FMETVAEK peptide and lubricin LVEVNPK peptide biomarker explains about 2-3% of the variability of JSN, values similar to the amount explained by age (Table). A biomarker score was developed combining normalized data for both lubricin and clusterin peptides, which increased the model R² to 0.079.

Table. Linear regression models adjusting for age and sex

*Normalized values for Clusterin(F) and PRG4(L) using beta estimates from models 1 and 2

Conclusion: Targeted mass spectrometry assays provide a rapid means to validate novel biomarkers in human samples. Our results suggest that when combined, clusterin and lubricin levels in plasma are as predictive of OA progression as age. Replication of these findings in other prospective OA cohorts is planned.