Identification of diagnostic markers related to inflammatory response and cellular senescence in endometriosis using machine learning and in vitro experiment.

Abstract

To understand the association between chronic inflammation, cellular senescence, and immunological infiltration in endometriosis. Datasets from GEO comprising 108 endometriosis and 97 healthy human samples and the human endometrial stromal cell. Differentially expressed genes were identified using Limma and WGCNA. Inflammatory response-related subtypes were constructed using consensus clustering analysis. The CIBERSORT algorithm and correlation analyses assessed immune cell infiltration. LASSO, SVM-RFE, and RF identified diagnostic genes. Functional enrichment analysis and multifactor regulatory networks established functional effects. Nomograms, internal and external validations, and in vitro experiments validated the diagnostic genes. Inflammatory response subtypes were highly correlated with the immune activities of B and NK cells. Sixteen genes were associated with inflammatory response and cellular senescence and six diagnostic genes (NLK, RAD51, TIMELESS, TBX3, MET, and BTG3) were identified. The six diagnostic gene models had an area under the curve of 0.828 and their expression was significantly downregulated in endometriosis samples. Low expression of NLK and BTG3 promoted the proliferation, migration, and invasion of endometriotic cells. Inflammatory response subtypes were successfully constructed for endometriosis. Six diagnostic genes related to inflammatory response and cellular senescence were identified and validated. Our study provides novel insights for inflammatory response in endometriosis and markers for endometriosis diagnosis and treatment.