Integrative analysis of hub genes for recurrent pregnancy loss with antiphospholipid syndrome: integrated bioinformatics analysis, machine learning and experimental validation

BackgroundRecurrent Pregnancy Loss (RPL) is significantly associated with Antiphospholipid Syndrome (APS), yet the shared pathogenic mechanism remains unclear. This study identifies potential hub genes for APS-related RPL, providing insights into its pathogenesis and potential diagnostic strategies.MethodsWe retrieved APS and RPL datasets from the GEO database and performed differential expression analysis to identify differentially expressed genes.

Thereafter, Gene Ontology (GO) analysis, pathway enrichment analysis, weighted gene co-expression network analysis (WGCNA), and protein-protein interaction (PPI) analysis were carried out. For hub gene screening, Least Absolute Shrinkage and Selection Operator (LASSO) regression, Gaussian mixture model (GMM), and RandomForest (RF) algorithms were applied to analyze the gene expression profiles of RPL.

The CIBERSORT tool was utilized to assess the immune cell infiltration levels in samples from RPL and APS patients, while the “clusterProfiler” R package was employed to perform single-gene Gene Set Enrichment Analysis (GSEA) for each hub gene in both cohorts. Three hub genes were validated for diagnostic performance via receiver operating characteristic (ROC) curves, with a nomogram model subsequently constructed and its efficacy assessed using ROC and calibration curves.