Profiling N6-methyladenosine (m6A) methylation-related genes in endometriosis towards a diagnostic model

Abstract

Abstract Endometriosis is an inflammatory disease with non-specific symptoms, including chronic pelvic pain and infertility, which affects thousands of women of reproductive age. Early diagnosis of endometriosis remains challenging. We aimed to build a diagnostic model based on m6A methylation-related genes to provide a new perspective on the clinical diagnosis of endometriosis. Two datasets from previous endometriosis studies were selected. GSE51981 was for training and GSE7305 was for validation. The expression of m6A methylation-related genes between proliferative eutopic endometrium from women with and without endometriosis was compared. Most m6A methylation-related genes were down-regulated in eutopic endometrium from women with endometriosis than those without it. The random forest classifier identified 5 significant differentially expressed genes (YTHDF2, NKAP, FTO, ZCCHC4 and HNRNPC) that might be involved in the development of endometriosis by affecting miRNA maturation or immune cell infiltration. These genes were included in a logistic regression to construct a new diagnostic model for endometriosis with an area under the ROC curve of 0.852. The model was tested on another independent dataset（AUC 0.750）and not only diagnosed endometriosis well but also showed how severe it was. We also found that YTHDF2 was very good at diagnosing endometriosis on its own and was correlated with macrophage and neutrophil infiltration that may be important for endometriosis development. In conclusion, this novel diagnostic model using m6A methylation-related genes may be a new method for early non-invasive diagnosis of endometriosis.