Identifying functional subtypes and common mechanisms of rheumatoid arthritis and systemic lupus erythematosus

Abstract

Background Although there has been much research on Rheumatoid arthritis (RA) and systemic lupus erythematosus (SLE), few studies focus on the classification and commonality of the two diseases. Therefore, we intends to conduct a joint subtype analysis of RA and SLE based on molecular and immune characteristics, and to investigate the similarity between RA and SLE and heterogeneity between subtypes of the two diseases. Results We analyzed the proportion differences of immune cell subpopulations and identified cell type specific expressed genes based on single-cell RNA sequencing data. Using these genes of abnormal proportion cells as as marker genes for classification, 232 RA patients and 160 SLE patients with bulk RNA sequencing data were grouped into two subtypes by a consensus clustering algorithm. The marker genes showed different expression patterns in two subtypes, and each of the subtype contained both RA and SLE patients. Then we estimated the composition of the immune microenvironment through ssGSEA and Cibersort algorithms, and analyzed the clinical characteristics of RA patients, which verified the heterogeneity between different subtypes. Next, we identified three co-expression modules highly correlated with subtypes by using WGCNA algorithm, and screened for differentially expressed genes (DEGs) between the two disease subtypes. To furter explore the biological mechanisms of different subtypes, functional enrichment analysis of modular genes and DEGs was performed. SubtypeⅠis associated with abnormal activation of phagocytic cells caused by bacterial infection, while subtypeⅡis associated to abnormal activation of lymphatic cells caused by viral infection. To verify accuracy of subtype classification and to test whether the marker genes can be used as subtype signatures and potential drug targets, we constructed LASSO and random forest models, got 10 marker genes between two subtypes, realized the precise subtype classification with average precision rate reaching 90%. Conclusion Two common subtypes of RA and SLE were identified with significant heterogeneity between subtypes, which may provide new insights for the precise diagnosis and treatment of RA and SLE.