Reveal the diagnostic value of a neutrophil inflammation- and cell death-associated gene signature in rheumatoid arthritis.

This study aimed to construct an artificial neural network (ANN) diagnostic model for rheumatoid arthritis (RA) based on a neutrophil inflammation- and cell death-associated gene signature derived from a previously reported neutrophil extracellular traps (NETs)-related gene set, and to explore its association with immune infiltration and inflammatory pathways. The GEO dataset GSE110169 was used as the training dataset to identify differentially expressed genes between RA patients and healthy controls. Genes derived from a previously published NETs-related gene set were used as the initial candidate genes. Candidate genes were further screened using a random forest algorithm, and an ANN diagnostic model was constructed using the selected feature genes. Model performance was assessed by 10-fold cross-validation and externally validated in GSE93272. Cell Type Identification By Estimating Relative Samples Of RNA Transcripts was used to estimate immune infiltration, and weighted gene co-expression network analysis, gene ontology, and Kyoto encyclopedia of genes and genomes analyses were performed to explore related biological functions. Eleven selected feature genes were retained for model construction: Wiskott-Aldrich syndrome protein-like actin nucleation promoting factor, MFN1, enolase-1, optic atrophy 1, CLEC7A, interleukin-8, S100A8, ACTN4, RIPK1, CASP1, and integrin-linked kinase. The ANN model achieved an area under the curve of 0.923 in the training dataset and 0.722 in the external validation dataset. Immune infiltration analysis suggested that these genes were associated with gamma delta T cells, macrophages M0, memory B cells, resting dendritic cells, and activated dendritic cells. Functional enrichment analysis indicated involvement in lysosome, osteoclast differentiation, hematopoietic cell lineage, chemokine signaling pathway, and Fc gamma R-mediated phagocytosis. This study developed an ANN diagnostic model based on an neutrophil inflammation- and cell death-associated gene signature with potential diagnostic value for RA. However, because direct NET markers were not experimentally evaluated, the selected genes should not be interpreted as NETs-specific biomarkers. Further experimental validation is required.