INTRODUCTION: Endometriosis is a chronic benign gynecological condition with substantial lifestyle impairment that lacks a noninvasive diagnostic option. Symptoms consist of chronic pelvic pain, dysmenorrhea, abnormal uterine bleeding, and infertility, all of which are debilitating and impact the daily life of patients. There is an average delay in diagnosis of 6.7–11 years from the onset of symptoms, emphasizing the importance of developing non-invasive, culturally appropriate, and painless diagnostic methods. Current diagnostic tools for endometriosis include pelvic exams, transvaginal ultrasounds, and magnetic resonance imaging. However, the only way to confirm a diagnosis of endometriosis is laparoscopic surgery. This diagnosis process often spans years and carries substantial costs for both patients and healthcare providers, resulting in numerous barriers to care. Anxiety about procedures, procedure affordability, limited availability of specialized care, and cultural norms conflicting with invasive techniques are all barriers to care that contribute to this delay. OBJECTIVE: The purpose of this study was to identify a metabolic signature from urine samples to identify patients with endometriosis, thereby advancing non-invasive diagnostic techniques. METHODS: The study utilized urine samples from a cohort of 78 women undergoing laparoscopic surgical procedures for benign gynecological conditions, wherein a definitive diagnosis was confirmed by visual assessment and histopathological confirmation of biopsy samples. The patients were classified into endometriosis-positive (n=40) and endometriosis-negative (control) groups (n=38). A global untargeted metabolomic approach at Metabolon, Inc., using ultra-performance liquid chromatography-tandem mass spectrometry, was utilized to quantify metabolites present in samples (n=1527), which were then evaluated as candidate biomarkers for endometriosis. RESULTS: Urine metabolomics identified significantly altered metabolites (n=22) and metabolic pathways that differ between patients with and without endometriosis (n=38), providing further insights into the pathophysiology of endometriosis. Specifically, multiple significant metabolites involved in the androgen and estrogen metabolism pathway (epiandrosterone glucuronide, androsterone glucuronide, androsterone sulfate, etc.) were identified, connecting the hormone-dependent nature of endometriosis to patients’ metabolic signatures. Significantly altered metabolites were identified, belonging mostly to amino acid, lipid, and xenobiotic super pathways. The highest enrichment ratios (ER) belonged to androgen and estrogen metabolism (ER = 6.576), caffeine metabolism (ER = 6.069), and inositol metabolism (ER = 5.267). Age and BMI were identified as significant demographic metrics within the cohort. A multivariable model (n=8) was developed with a sensitivity of 85.0%, specificity of 81.5%, predictive accuracy of 83.3%, kappa value of 0.666, and Youden index of 0.665. An area under the receiver operating characteristic (AUROC) value of 0.913, indicated an excellent potential of multivariable model for diagnosis of endometriosis. CONCLUSIONS: A six-feature model with the inclusion of age and BMI shows promise compared to current diagnostic techniques in terms of sensitivity and specificity. Furthermore, urine diagnostic methods are non-invasive, can be self-collected, culturally appropriate, and minimally impact clinical workflow. Overall, it was found that investigation of urine metabolites as biomarkers for noninvasive diagnostic methods is promising to improve on current approaches. Advancements in endometriosis diagnosis techniques will enable faster diagnosis and treatment, therefore reducing patient suffering, optimizing clinicians’ time, and expanding access to care.
Opferman et al. (Fri,) studied this question.