Abstract
Background: Endometriosis is a serious gynecological disorder characterized by debilitating pain, infertility and the establishment of innervated endometriosis lesions outside the uterus. Early detection and accurate diagnosis are pivotal in endometriosis. The work screened autophagy-related genes (ATGs) as potential biomarkers to reveal new molecular subgroups for the early diagnosis of endometriosis. Materials and Methods: The gene lists of ATGs from five databases were integrated. Then, weighted gene co-expression network analysis (WGCNA) was used to map the genes to the gene profile of endometriosis samples in GSE51981 to obtain functional modules. GO and KEGG analyses were performed on the ATGs from the key modules. Differentially expressed ATGs were identified by the limma R package and further validated in the external datasets of GSE7305 and GSE135485. The DESeq2 R package was utilized to establish multifactorial network. Subsequently, one-way analysis of variance (ANOVA) was performed to identify new molecular subgroups. Real-time quantitative polymerase chain reaction (RT-qPCR) and Western blotting were used to confirm the differential expression of hub ATGs, and the receiver operating characteristic (ROC) curve analysis and Spearman correlation analysis were applied to assess the diagnostic value of hub ATGs in 40 clinical samples and human primary endometrial stromal cells (ESCs). Results: We screened 4 key modules and 12 hub ATGs and found the key genes to be strongly correlated with endometriosis. The pathways of ATGs were mainly enriched in autophagy, apoptosis, ubiquitin-protein ligase binding, and MAPK signaling pathway. The expression levels of EZH2 (Enhancer of Zeste homolog 2) and RND3 (also known as RhoE) had statistically significant changes with higher values in the endometriosis group compared with the controls, both in the tissue samples and primary ESCs. Besides, they also showed higher specificity and sensitivity by the receiver operating characteristic analysis and Spearman correlation analysis for the diagnosis of endometriosis. The TF-mRNA-miRNA-lncRNA multifactorial network was successfully constructed. Four new molecular subgroups were identified, and we preliminarily showed the ability of IQCG to independently differentiate subgroups. Conclusion: EZH2 and RND3 could be candidate biomarkers for endometriosis, which would contribute to the early diagnosis and intervention in endometriosis.