Serum metabolic fingerprinting for diagnosis and therapeutic applications of ovarian endometriosis

Ovarian endometriosis (OvE) is a gynecological disorder with endometrial tissue in the ovaries, for which effective non-invasive diagnosis and curative treatments are currently lacking. Serum samples were collected from both discovery and validation cohorts to examine the metabolomic signatures. Fifty-six differential metabolites between patients with OvE and healthy controls were identified using untargeted metabolomic profiling. Weighted gene co-expression network analysis was further conducted to validate the differential metabolites. Subsequently, twenty-one metabolites were selected for further validation using targeted metabolomic profiling. Five machine learning algorithms confirmed the efficacy and stability of these metabolites for diagnosing OvE. Least absolute shrinkage and selection operator -logit regression identified six serum metabolites and two clinicopathological features with high diagnostic accuracy. Three differential metabolites were found to exhibit therapeutic potential for OvE in an in vivo study. Diagnostic, predictive, and therapeutic potential of serum metabolomes for OvE are provided in this study.