Abstract
BACKGROUND: Neutrophil extracellular traps (NETs) derived from neutrophils are implicated in the pathogenesis of rheumatoid arthritis (RA) pathogenicity, though the underlying mechanisms remain unclear. METHODS: Data were obtained from Gene Expression Omnibus (GEO) database. First, Gene Set Variation Analysis (GSVA) was used to calculate NET scores, and ConsensusClusterPlus was employed to classify RA samples. Subsequently, weighted gene co-expression network analysis (WGCNA) was used to construct co-expression networks. Lasso regression and support vector machine recursive feature elimination (SVM-RFE) were then used to cross-screen biomarkers for RA, with predictive performance evaluated via the timeROC package. Immune infiltration in RA samples was assessed using ssGSEA and MCPcounter methods. Additionally, qRT-PCR was conducted to validate the expression of key genes. Finally, potential therapeutic drugs were predicted through Enrichr using the DSigDB database, and candidate compounds were preprocessed with PyMOL and ChemBioOffice before molecular docking with AutoDockTools. RESULTS: RA patients had significantly higher NET scores than controls, and the samples were divided into C1 and C2. WGCNA combined with differential analysis identified eight key genes, and five biomarkers were screened by two machine learning algorithms, namely, ANGPTL1, CASP8, FNIP2, MEOX2, and ZNF780B. Both the training set and validation set showed an AUC > 0.7. Immunological analysis revealed an association with neutrophil infiltration, while drug prediction and molecular docking revealed that N-Acetyl-L-cysteine and Eckol exhibited favorable binding activity. CONCLUSION: This study provided novel insights into RA progression based on NETs, offering potential signature genes for the prognostic prediction of RA.