Abstract / Summary
Rheumatoid arthritis (RA) is a multifaceted autoimmune disease with numerous pathogenic genes, and their causal roles have yet to be completely revealed. This study aimed to identify and validate signature genes that may be involved in RA progression. We combined transcriptomic data from the Gene Expression Omnibus database (GSE55235 (training set); GSE12021, GSE55457, and GSE77298 (validation sets). Key genes were identified by combining weighted gene co-expression network analysis (WGCNA) with differential expression analysis, followed by screening using protein-protein interaction (PPI) network analysis and a robust machine learning method. Moreover, immune cell infiltration was evaluated using CIBERSORT. The potential causal relationship between the expression of signature genes and the risk of RA was further investigated using two-sample Mendelian randomization (MR) analysis with single-cell eQTL data from the OneK1K project and genome-wide association study summary statistics from the FinnGen project. WGCNA revealed that the MEblue module was significantly associated with RA. Overlap with differentially expressed genes identified 902 overlapping genes enriched in immune and osteoclast-related pathways. Only eight signature genes were identified using machine learning and PPI analysis, including CDC20, MELK, PBK, TPX2, ASPM, CXCL9, CXCL11, and CXCR3. The diagnostic model based on these eight genes demonstrated excellent performance. MR results were significant. Genetic analyses revealed that elevated TPX2 expression in natural killer (NK) cells could be a genetically supported risk factor for RA (odds ratio = 1.302, 95% confidence interval: 1.130–1.501, p = 2.71 × 10 –4 ). Immune infiltration analysis revealed a significant difference between RA and control groups. In this multi-omics study, we established a strong eight-gene signature for RA diagnosis, provided novel statistical genetic evidence supporting a potential causal relationship between NK cell TPX2 expression and RA pathogenesis, and offered new insights into the molecular biology of RA. However, these candidate genes require further functional verification before being translated into clinical diagnosis and therapy targets.