Identification of Key Biomarkers for Systemic Lupus Erythematosus Progression and Therapy Response: A Bulk RNA-Sequencing-Based Bioinformatics Study.

BACKGROUND: Systemic lupus erythematosus (SLE) is a chronic autoimmune disease that affects multiple organs and has a higher prevalence in women of reproductive age. The increasing global incidence of this disease poses a significant public health challenge. Current diagnostic methods are delayed because of nonspecific clinical manifestations and the limited specificity of available tests. Furthermore, conventional treatments frequently exhibit considerable side effects and fail to achieve complete disease control in many patients. Therefore, identifying novel diagnostic biomarkers and treatment targets is imperative. This study integrated transcriptomics, protein interaction analysis, and immunological infiltration assays to identify and confirm key biomarkers in SLE. METHODS: We obtained data from the Gene Expression Omnibus database. Weighted gene co-expression network analysis (WGCNA) was used to identify gene modules, and differentially expressed genes (DEGs) were identified using limma software. Functional enrichment analyses were performed using gene ontology and the Kyoto encyclopedia of genes and genomes. Protein-protein interaction (PPI) networks were generated using STRING and illustrated using Cytoscape. Machine learning algorithms, including the least absolute shrinkage and selection operator, random forest, and artificial neural networks (ANN), were employed to identify key diagnostic genes. A diagnostic model was constructed and verified, potential therapeutic compounds were predicted using the Drug Signature Database, and molecular docking was performed for drug-target interaction analysis. RESULTS: The WGCNA revealed a key gene module (red module) comprising SLE-relevant genes. We identified 238 overlapping DEGs enriched in several immune-related biological processes and signaling pathways. Fourteen common hub genes were identified from the PPI network. Four genes, SLC4A1, GATA1, DMTN, and SNCA, emerged as potential diagnostic biomarkers based on machine learning analysis. These genes were significantly correlated with immune cell infiltration patterns in SLE. A diagnostic ANN model that included these genes exhibited high predictive accuracy. A nomogram was constructed and validated. Drug prediction analysis identified N-acetyl-L-cysteine as a viable treatment option, demonstrating a stable binding affinity in molecular docking simulations. CONCLUSIONS: This study demonstrates that SLC4A1, GATA1, DMTN, and SNCA are potential biomarkers for diagnosing SLE and monitoring therapeutic efficacy. The integrated diagnostic approach exhibited robust predictive power, and N-acetyl-L-cysteine was identified as a viable therapeutic agent. These findings provide valuable insights for enhancing the early diagnosis and development of targeted therapies for SLE.