Identification of baseline gene expression signatures predicting therapeutic effects of TNF inhibitors in rheumatoid arthritis patients

Abstract

【Objective】We have designed a study addressing concerns raised specifically regarding conditions, clinical bias, number of study subjects and approach of identification of biomarkers, in order to identify predictive biomarkers of TNF inhibitors（TNFi）using patients’ blood expression profiles.

【Patients and Methods】All study subjects（n=219）were bDMARDs treatment naïve and considered for co-administration of methotrexate（MTX）and TNFi. It represented the biggest cohort（n=219）ever studied using genome-wide gene expression profiles. To obtain a robust gene panel, we have referred to 6 disease activity criteria（DAS28, CDAI, EULAR, ACR, SJC28 and CRP）to determine treatment response. We designated 2 thresholds for each criterion and this generated 12 pairwise comparisons. Baseline clinical information was matched, and genes that were common in at least 4 out of 12 comparisons were extracted.

【Results】Thirty four genes were shortlisted. Ribosome-related genes were highly expressed in RES while type I interferon related-genes were found to be highly expressed in NRES. A combination of predictive biomarkers and baseline clinical information has an accuracy of 64.5-83.6%, significantly better than baseline clinical information alone.

【Conclusion】Accuracy of prediction improves when combining blood-based predictive factors in this study and clinical data. Reproducibility in a prospective trial is warranted, also to demonstrate causal or consequential relationship between functions of these genes and disease activity of RA.