Abstract
Rheumatoid arthritis (RA) causes serious disability and productivity loss, and there is an urgent need for appropriate biomarkers for diagnosis, treatment assessment, and prognosis evaluation. To identify serum markers of RA, we performed mass spectrometry (MS)-based proteomics, and we obtained 24 important markers in normal and RA patient samples using a random forest machine learning model and 11 protein-protein interaction (PPI) network topological analysis methods. Markers were reanalyzed using additional proteomics datasets, immune infiltration status, tissue specificity, subcellular localization, correlation analysis with disease activity-based diagnostic indications, and diagnostic receiver-operating characteristic analysis. We discovered that ORM1 in serum is significantly differentially expressed in normal and RA patient samples, which is positively correlated with disease activity, and is closely related to CD56dim natural killer cell, effector memory CD8+T cell, and natural killer cell in the pathological mechanism, which can be better utilized for future research on RA. This study supplies a comprehensive strategy for discovering potential serum biomarkers of RA and provides a different perspective for comprehending the pathological mechanism of RA, identifying potential therapeutic targets, and disease management.