Available online 19 January 2022, 101741

Highly efficacious drugs are widely available for treating rheumatoid arthritis (RA). However, accurately selecting a likely effective drug for individual RA patients has been challenging. Biomarkers are required since clinical phenotypes are not reliable to guide the choice of drugs.

Previously identified genetic variants for predicting treatment response have failed in replication in independent cohorts of RA patients. Recent studies aimed at the discovery of biomarkers to predict treatment response have focused on integrative omics analysis, expanded to the microbiome, and further finer definition of synovial pathotypes. Treatment responders and non-responders of RA patients can be distinguished by distinct signatures at baseline in their gut microbiota compositions, peripheral blood transcriptome profiling or histomorphological and molecular pathotypes of synovitis. These distinct biological signatures are promising for developing clinically applicable tools for decision in the selection of drugs for RA, albeit further validations in independent cohorts are required.

Rheumatoid arthritis

Personalized medicine

Prediction of treatment response

Methotrexate

TNF inhibitors

Microbiome

Machine learning

Synovial pathotype

Published by Elsevier Ltd.