Saliva-based microRNA diagnostic signature for the superficial peritoneal endometriosis phenotype

Abstract

Objective: Patients with superficial peritoneal endometriosis (SPE) present with symptoms suggestive of endometriosis but clinical and imaging exams are inconclusive. Consequently, laparoscopy is usually necessary to confirm diagnosis. The present study aimed to evaluate the accuracy of microRNAs (miRNAs) to diagnose patients with SPE from the ENDOmiARN cohort STUDY DESIGN: This prospective study (NCT04728152) included 200 saliva samples obtained between January and June 2021 from women with pelvic pain suggestive of endometriosis. All patients underwent either laparoscopy and/or MRI to confirm the presence of endometriosis. Among the patients with endometriosis, two groups were defined: an SPE phenotype group of patients with peritoneal lesions only, and a non-SPE control group of patients with other endometriosis phenotypes (endometrioma and/or deep endometriosis). Data analysis consisted of two parts: (i) identification of a set of miRNA biomarkers using next-generation sequencing (NGS), and (ii) development of a saliva-based miRNA signature for the SPE phenotype in patients with endometriosis based on a Random Forest (RF) model.

Results: Among the 153 patients with confirmed endometriosis, 10.5 % (n = 16) had an SPE phenotype. Of the 2633 known miRNAs, the feature selection method generated a signature of 89 miRNAs of the SPE phenotype. After validation, the best model, representing the most accurate signature had a 100 % sensitivity, specificity, and AUC.

Conclusion: This signature could constitute a new diagnostic strategy to detect the SPE phenotype based on a simple biological test and render diagnostic laparoscopy obsolete. PRéCIS: We generated a saliva-based signature to identify patients with superficial peritoneal endometriosis which is the most challenging form of endometriosis to diagnose and which is often either misdiagnosed or requires invasive laparoscopy.

Keywords: Diagnostic laparoscopy; Endometriosis phenotype; Machine Learning; Next-generation Sequencing; Superficial peritoneal endometriosis; microRNA.