Rheumatoid arthritis is a highly heterogeneous autoimmune disease characterized by unpredictable disease flares and significant differences in therapeutic response to available treatments. One possible reason for poor efficacy is that it cannot be treated accurately due to no optimal stratification for RA patients.
This study aims to construct an RA classification model by m6A characters and further predict response to medication.
Twenty m6A regulators were used to construct a random forest diagnosis model, and RNA-seq analysis was employed for external validation. The RNA modification patterns mediated by 20 m6A regulators were systematically evaluated in 1191 RA samples and explored different molecular clusters associated with other immune microenvironment characteristics and biological pathways. Then, we established an m6A score model to quantify the m6A modification patterns. The model was applied to patients at baseline to test the association between m6Ascore and infliximab responsiveness.
The m6A diagnosis model showed good discriminatory ability in distinguishing RA. Patients with RA were classified into three clusters with distinct molecular and cellular signatures. Cluster A displayed strongly activated inflammatory cells and pathways. Specific innate lymphocytes occupied cluster B. Cluster C was mainly enriched in prominent adaptive lymphocytes and NK-mediated cytotoxicity signatures with the highest m6A score. Patients with a low m6Ascore exhibited significantly infliximab therapeutic benefits compared with those with a high m6Ascore (p< 0.05).
Our study is the first to provide a comprehensive analysis of m6A modifications in RA, which provides an innovative patient stratification framework and potentially enables improved therapeutic decisions.