An analysis of 14 molecular markers for monitoring osteoarthritis. Relationship of the markers to clinical end-points

Objective: To investigate whether any of 14 serum and urine molecular marke rs (MMs) used to monitor osteoarthritis (OA) would be associated with parti cular clinical end-points. Design: Thirty-nine OA patients were bled and urine collected at five time points: at baseline visit and at visits 1, 3, 6 and 12 months later. Twelve clinical measurements were made and the concentrations of each of 14 MMs w ere determined. Principal component analysis, stepwise linear regression wi th backward elimination, and logistic regression were used to determine the correlations between MMs and clinical measures. Results: Principal component analysis was used to reduce the 12 clinical me asurements into three independent clinical clusters: baseline clinical asse ssments, changes in clinical assessments and signal joint measurements. The 14 MMs were similarly reduced to five NIM clusters. Each of the three clin ical clusters was correlated with a single but different MM cluster Baselin e clinical assessments were correlated with bone markers typified by hydrox ylysyl pyridinoline (HP) crosslinks, swelling of the signal joint was corre lated with inflammation markers, especially CRP, and the change in clinical assessments over the 1 year evaluation was correlated with TGF beta1. Ther e was no correlation between any of the skeletal markers and the clinical m easures, a situation which draws attention to the need for a direct assessm ent of cartilage damage in OA to validate the use of cartilage markers. Conclusions: This study demonstrates statistical methodology for analysis o f clinical trials using multiple MMs and clinical end-points. The patient n umbers are sufficient to test hypotheses of relationships of single MMs suc h as CRP, TGF beta1 and HP to clinical measures, but larger clinical trials are needed to validate hypotheses. (C) 2001 OsteoArthritis Research Societ y International.