Does hsCRP provide insights into an inflammatory phenotype of knee osteoarthritis?

Abstract

Objective: Posthoc analysis of a canakinumab trial showed a risk reduction of joint replacement in participants with cardiac disease and elevated hsCRP (≥2mg/dL). We determined if hsCRP could serve as a marker to identify an inflammatory OA phenotype characterized by intra-articular synovitis. Method: We used data from the NIH-funded MOST Study, where participants had baseline knee MRIs and hsCRP assays. Inflammation was scored using the Whole Organ MRI Score (WORMS), with presence defined as Hoffa's synovitis or effusion-synovitis score ≥2 in ≥1 of 3 locations. We examined the relationship between hsCRP and inflammation presence using logistic regression, and the average inflammation score with linear regression. Analyses were repeated with hsCRP categorized as ≥2mg/dL vs. <2mg/dL. Cubic spline regression and ROC curve were completed. Analyses were adjusted for age, sex, and body mass index. Results: Out of 792 participants (mean age 62, 52% female), mean hsCRP was 2.98 mg/dL, with 41% having hsCRP ≥2mg/dL. Inflammation was present in 28% of knees. No association was found between hsCRP levels and inflammation presence (odds ratio 0.99, 95% confidence interval 0.96–1.01) or average score, including with dichotomized hsCRP. Cubic spline regression did not reveal non-linear associations. The ROC curve suggested poor predictive ability of hsCRP to identify knees with inflammation on MRI. Conclusion: Despite interest in identifying an inflammatory phenotype in OA, hsCRP levels do not reliably identify knees with inflammation. Further investigation is needed to determine if hsCRP can predict knee response to therapies targeting inflammation not directly related to synovitis.