Plasma MicroRNA Profiles in Patients with Early Rheumatoid Arthritis Responding to Adalimumab plus Methotrexate vs Methotrexate Alone: A Placebo-controlled Clinical Trial

Objective. The aim was to identify plasma (i.e., cell-free) microRNA (miRNA) predicting antitumor necrosis and/or methotrexate (MTX) treatment response in patients enrolled in an investigator-initiated, prospective, double-blinded, placebo-controlled trial (The OPERA study, NCT00660647). Methods. We included 180 disease-modifying antirheumatic drug-naive patients with early rheumatoid arthritis (RA) randomized to adalimumab (ADA; n = 89) or placebo (n = 91) in combination with MTX. Plasma samples before and 3 months after treatment initiation were analyzed for 91 specific miRNA by quantitative reverse transcriptase-polymerase chain reaction on microfluidic dynamic arrays. A linear mixed-effects model was used to test for associations between pretreatment miRNA and changes in miRNA expression and American College of Rheumatology/European League Against Rheumatism (ACR/EULAR) Boolean (28 joints) remission at 3 and 12 months, applying false discovery rate correction for multiple testing. Using leave-one-out cross validation, we built predictive multivariate miRNA models and estimated classification performances using receiver-operating characteristics (ROC) curves. Results. In the ADA group, a higher pretreatment level of miR-27a-3p was significantly associated with remission at 12 months. The level decreased in remitting patients between pretreatment and 3 months, and increased in nonremitting patients. No associations were found in the placebo group receiving only MTX. Two multivariate miRNA models were able to predict response to ADA treatment after 3 and 12 months, with 63% and 82% area under the ROC curves, respectively. Conclusion. We identified miR-27a-3p as a potential predictive biomarker of ACR/EULAR remission in patients with early RA treated with ADA in combination with MTX. We conclude that pretreatment plasma-miRNA profiles may be of predictive value, but the results need confirmation in independent cohorts.