Evaluation of cytokine profiles in rheumatoid arthritis patients with clinically active disease and normal inflammatory indices.

Abstract

To assess the potential utility of a cytokine measurement in rheumatoid arthritis (RA) patients with active joint disease but normal acute phase reactants (APR). RA patients in a longitudinal observational registry with available cytokine array data were included. Patients were categorized based on agreement/disagreement of physical examination and APR measurements: concordant high (CH) [high tender and/or swollen joint counts (TJC + SJC > 3) and APR (ESR ≥ 28mm/h + CRP ≥ 1.5mg/L)]; concordant low (CL) [TJC + SJC ≤ 3 and normal APR]. Discordant (D) [TJC + SJC > 3 and normal APR] patients were stratified into low, medium, and high-disease activity (DL, DM, DH). Weighted-average and log-transformed cytokine scores were calculated based on results of a cytokine array. Chi-square tests compared categorical variables by concordance status; t tests, Wilcoxon rank-sum tests, ANOVA models, and ordinary least squares (OLS) regressions were used to compare continuous measures. RA patients (n = 1467) were predominantly male (91%). Compared to CH patients (n = 174), D (n = 434) were younger, less frequently seropositive, with lower TJC, SJC, and DAS28-3v scores (p < 0.001). Cytokine scores for DL, DM, and DH groups were lower than CH patients (p < 0.001) and did not differ between DL, DM, and DH subgroups and were similar to CL (n = 356) patients. In multivariable analyses including CH and D patients, log-cytokine score was associated with higher DAS28-3v scores (p = 0.029). In multivariable analyses including CL patients, concordance status (p = 0.011) and ACPA (p = 0.013) were predictors of higher log cytokine score. In this study, cytokine scores did not identify active joint disease in RA patients with normal APR.