Large Scale Metabolic Profiling identifies Novel Steroids linked to Rheumatoid Arthritis

Recent metabolomics studies of Rheumatoid Arthritis (RA) reported few metabolites that were associated with the disease, either due to small cohort sizes or limited coverage of metabolic pathways. Our objective is to identify metabolites associated with RA and its cofounders using a new untargeted metabolomics platform. Moreover, to investigate the pathomechanism of RA by identifying correlations between RA-associated metabolites. 132 RA patients and 104 controls were analyzed for 927 metabolites. Metabolites were tested for association with RA using linear regression. OPLS-DA was used to discriminate RA patients from controls. Gaussian Graphical Models (GGMs) were used to identify correlated metabolites. 32 metabolites are identified as significantly (Bonferroni) associated with RA, including the previously reported metabolites as DHEAS, cortisol and androstenedione and extending that to a larger set of metabolites in the steroid pathway. RA classification using metabolic profiles shows a sensitivity of 91% and specificity of 88%. Steroid levels show variation among the RA patients according to the corticosteroid treatment; lowest in those taking the treatment at the time of the study, higher in those who never took the treatment, and highest in those who took it in the past. Finally, the GGM reflects metabolite relations from the steroidogenesis pathway.