Comprehensive transcriptomic analysis and machine learning reveal unique gene expression profiles in patients with immune-mediated necrotizing myopathy

Abstract

Background: Immune-mediated necrotizing myopathy (IMNM) is an autoimmune myopathy characterized by severe proximal weakness and muscle fiber necrosis, yet its pathogenesis remains unclear. So far, there are few bioinformatics studies on underlying pathogenic genes and infiltrating immune cell profiles of IMNM. Therefore, we aimed to characterize differentially expressed genes (DEGs) and infiltrating cells in IMNM muscle biopsy specimens, which may be useful for elucidating the pathogenesis of IMNM.

Methods: Three datasets (GSE39454, GSE48280 and GSE128470) of gene expression profiling related to IMNM were obtained from the Gene Expression Omnibus database. Data were normalized, and DEG analysis was performed using the limma package. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses of DEGs were performed using clusterProfiler. The CIBERSORT algorithm was performed to identify infiltrating cells. Machine learning algorithm and gene set enrichment analysis (GSEA) were performed to find distinctive gene signatures and the underlying signaling pathways of IMNM.

Results: DEG analysis identified upregulated and downregulated in IMNM muscle compared to the gene expression levels of other groups. GO and KEGG analysis showed that the pathogenesis of IMNM was notable for the under-representation of pathways that were important in dermatomyositis and inclusion body myositis. Three immune cells (M2 macrophages, resting dendritic cells and resting natural killer cells) with differential infiltration and five key genes (NDUFAF7, POLR2J, CD99, ARF5 and SKAP2) in patients with IMNM were identified through the CIBERSORT and machine learning algorithm. The GSEA results revealed that the key genes were remarkably enriched in diverse immunological and muscle metabolism-related pathways.

Conclusions: We comprehensively explored immunological landscape of IMNM, which is indicative for the research of IMNM pathogenesis.

Keywords: differentially expressed gene; immune characteristics; immune-mediated necrotizing myopathy; machine learning.