Objective: Endometriosis is a chronic inflammatory estrogen-dependent disease with the growth of endometrial tissues outside the uterine cavity. Nevertheless, the etiology of endometriosis is still unclear. Integrated bioinformatics analysis was implemented to reveal the molecular mechanisms underlying this disease. Methods: A total of four gene expression datasets (GSE7305, GSE11691, GSE23339, and GSE25628) were retrieved from the GEO, which were merged into a meta-dataset, followed by the removal of batch effects via the sva package. Weighted gene co-expression network analysis (WGCNA) was implemented, and endometriosis-related genes were screened under normal and endometriosis conditions. Thereafter, characteristic genes were determined via Lasso analysis. The diagnostic performance was estimated via receiver operating characteristic curves, and epigenetic and post-transcriptional modifications were analyzed. Small molecular compounds were predicted. Unsupervised clustering analysis was conducted via non-negative matrix factorization algorithm. The enriched pathways were analyzed via gene set enrichment analysis or GSVA. Immune features were evaluated according to immune-checkpoints, HLA, receptors, chemokines, and immune cells. Results: In total, four characteristic genes (BGN, AQP1, ELMO1, and DDR2) were determined for endometriosis, all of which exhibited the favorable efficacy in diagnosing endometriosis. Their aberrant levels were modulated by epigenetic and post-transcriptional modifications. In total, 51 potential drugs were predicted against endometriosis. The characteristic genes exhibited remarkable associations with immunological function. Three subtypes were classified across endometriosis, with different mechanisms and immune features. Conclusion: Our study reveals the characteristic genes and novel molecular subtyping of endometriosis, contributing to the early diagnosis and intervention in endometriosis.
Keywords: characteristic genes; diagnosis; endometriosis; immune characteristics; molecular subtypes.
Copyright © 2022 Wang, Liu, Li, Lian, Cui, Ng and Zhu.