Accurate Diagnosis of Endometriosis Using Serum MicroRNAs

Endometriosis, a chronic inflammatory disorder causing pain and infertility, affects nearly 10%of reproductive age women. Despite its high prevalence, the average time from onset of symptoms to diagnosis is 5 to 10 years. The criterion standard of diagnosis, laparoscopic surgery, incurs costs, patient risk, and costs and often delays diagnosis resulting in further disease progression and prolonged suffering. There is need for development of an accurate noninvasive diagnostic test to detect the disease at an early stage. Serum microRNAs (miRNAs), which are short, noncoding RNAs, have emerged as potential molecular indicators to noninvasively identify endometriosis. A previous retrospective study reported that in a panel of circulating miRNAs there was at least a 10-fold increase or decrease in expression in serum of women with endometriosis. The aim of this case-control study was to test the feasibility of using a combination of serum miRNAs to diagnose symptomatic endometriosis with accuracy comparable to laparoscopy. It was hypothesized that preoperative evaluation of these miRNA biomarkers could distinguish endometriosis from other benign gynecological conditions, in a diverse patient population. Subjects aged 18 to 49 years undergoing surgery for suspected benign indications were seen at an academic medical center between September 2016 and October 2017. Prior to surgery, serum samples were collected from 100 subjects. Laparoscopy was performed in women selected based on their symptoms to determine the presence or absence of endometriosis. Staging was done according to the revised American Society of Reproductive Medicine classification. Absence of any visual disease at the time of surgery led to categorization of the control group. Quantitative real-time polymerase chain reaction was used tomeasure 6 circulatingmiRNAs (miR-125b-5p, miR-150-5p,miR-342-3p, miR-451a,miR-3613-5p, and let-7b) in a blinded fashion, without knowledge of disease status. Receiver operating characteristic (ROC) analysis was performed on individual miRNAs and combinations of miRNAs to evaluate the diagnostic utility of each miRNA biomarker. To predict the presence or absence of endometriosis on operative findings, an algorithm combining the expression values of these miRNAs was built and tested, using machine learning with a randomforest classifier. The diagnostic validity of the algorithm was then tested in an independent data set obtained from 48 previously identified subjects not included in the training set (24 endometriosis and 24 controls). The mean age of women in the endometriosis group was 34.1 years, and it was 36.9 years in the control group; P > 0.05. Of 100 patients, 41 were categorized as endometriosis and 59 as controls. Subjects in the control group displayed varying pathologies; leiomyoma occurred most often (n = 39). Measured expression levels of the 6 miRNAs among women with endometriosis were significantly higher with 4 serum miRNAs (miR-125b-5p, miR-150-5p, miR-342-3p, and miR-451a) and were significantly decreased for 2 (miR-3613-5p and let-7b). Areas under the ROC curve (AUCs) for individual miRNAs ranged from 0.68 to 0.92. Random forest applied to 6 of the miRNAs yielded an optimal classifier with an AUC of 0.939, when validated in the independent set of subjects not included in the training set; this indicated high accuracy. Based on revised American Society of Reproductive Medicine staging, analysis of the expression levels of each miRNA showed that all miRNAs could distinguish stage I/II from control, and stage III/IV from control, but the difference between these stageswas not significant. In subgroup analysis, neither phase of menstrual cycle or use of hormonal medication significantly impacted expression levels in the miRNAs. This is the first study performed within a diverse population showing the ability of serum miRNA biomarkers to reliably differentiate endometriosis from other gynecologic pathologies, with an AUC greater than 0.9 across 2 independent studies. The validated performance of an algorithm based on previously identified miRNA biomarkers demonstrates their potential to detect endometriosis in a real-world clinical setting, allowing earlier identification and treatment. Larger prospective studies are needed to evaluate the relationships between miRNAs levels in serum and endometriosis and their impact on the accuracy of diagnosis, management, and outcome of the different stages of endometriosis.