Abstract

Rheumatoid arthritis (RA) is an autoimmune rheumatic disease, which do not respond well to current treatment partially. Therefore, further in-depth elucidation of the molecular mechanism and pathogenesis of RA is urgently needed for the diagnosis, personalized therapy and drug development. Herein, we collected 111 RA samples from Gene Expression Omnibus (GEO) database, and conducted differentially expressed genes and GESA analysis. Abnormal activation and imbalance of immune cells in RA were observed. WGCNA was utilized to explore the gene modules and CD8+ T cell-related genes (CRGs) were chosen for KEGG and GO analysis. Besides, to explore biomarkers of RA in depth, machine learning algorithms and bioinformatics analysis were used, and we identified GDF15, IGLC1, and IGHM as diagnostic markers of RA, which was confirmed by clinical samples. Next, ssGSEA algorithms were adopted to investigate the differences in immune infiltration of 23 immune cell subsets between RA and healthy control group. Finally, optimal classification analysis based on consensus clustering combined with ssGSEA algorithms were conducted. GDF15 was revealed that to be positively correlated with mast cells and type 2 T helper cells, but negatively correlated with most other immune cells. On the other hand, IGHM and IGLC1 were negatively correlated with CD56dim natural killer cells, while positively associated with other immune cells. Finally, RA samples in subtype A exhibited a higher immune infiltration status. This study could provide guidance for individualized treatment of RA patients and provide new targets for drug design.