Abstract
BACKGROUND: Dermatomyositis (DM) is an uncommon autoimmune disease that presents challenges due to the lack of reliable biomarkers in clinical practice. Growing evidence suggests that N6-methyladenosine (m6A) is closely associated with the pathogenesis of autoimmune diseases. METHODS: Microarray gene expression matrix for GSE46239, GSE128314, and GSE142807 were downloaded from the GEO database. Random forest (RF), support vector machine (SVM), and nomogram models were developed, with their performance subsequently compared. The identification of m6A subtypes, based on differentially expressed m6A regulatory genes, was followed by the classification of gene subtypes according to the differently expressed genes between the m6A subtypes. Both classification systems were subjected to m6A scoring analysis and visualized via a Sankey diagram. RESULTS: We retrieved 99 dermatomyositis samples and 14 healthy samples. Using an RF model, we identified five core genes-IGFBP3, ZCCHC4, HNRNPC, WTAP, and RBM15-and constructed a predictive nomogram model. Two m6A clusters were developed. Cluster A exhibited a significant increase of CD56-bright natural killer cells, immature B cells, plasmacytoid dendritic cells, regulatory T cells, and type 1 T helper cells distinct from cluster B (p < 0.05). Based on 32 significantly distinctly expressed genes between m6A subtypes (p < 0.05), we further reproduced two m6A gene subtypes. The Sankey diagram showed significant concordance among m6A scores, m6A subtypes, and m6A gene subtypes. CONCLUSION: m6A regulatory genes significantly influence the pathogenesis of dermatomyositis. In this work, we built a predictive nomogram model, comprehensively evaluated two classification methods, and provided new insights for patient classification.