Diagnostic Performance of a Salivary Extracellular-Vesicle microRNA Panel for Endometriosis: A Prospective Case-Control Development Study Without Diagnostic Imaging

Abstract

Endometriosis is a chronic inflammatory disease for which diagnosis is often delayed and, in many patients, still relies on invasive confirmation. Consequently, there is an urgent need for non-invasive biomarkers to support triage and clinical assessment. This developmental-phase prospective diagnostic case-control study included 25 women with surgically documented endometriosis and 25 gynecologic control participants. Control status was verified by prior negative laparoscopy where available, or by structured clinical assessment in women without chronic pelvic pain, moderate-to-severe dysmenorrhea, clinical suspicion of deep disease, or elevated CA-125. Unstimulated whole saliva was collected under standardized conditions. Salivary extracellular vesicles (EVs) were isolated using size-exclusion chromatography followed by ultrafiltration, and were characterized via protein quantification, CD63/CD81 ELISAs, hemoglobin absorbance screening, and particle assessment where available. Candidate miRNAs were measured by RT-qPCR, normalized by global-mean ΔCt, and log2-transformed. A LASSO logistic regression model was developed from normalized miRNA expression and refitted as a parsimonious multivariable model. Model performance was summarized using ROC analysis, calibration assessment, decision-curve analysis, and planned bootstrap internal validation. Endometriosis cases exhibited higher salivary EV levels of miR-451a, miR-21-5p, and miR-125b-5p, and lower levels of let-7a-5p, let-7d-5p, and miR-200c-3p compared to controls, while miR-16-5p and miR-191-5p remained relatively stable. The final three-miRNA model retained standardized miR-451a, miR-21-5p, and miR-200c-3p. Apparent discrimination was high, with an AUC of 0.902 (95% CI: 0.814–0.991). At the Youden-index threshold of 0.502, apparent sensitivity was 88% and specificity was 96% in this balanced case-control cohort. Internal validation performed using the observed individual-level clinical dataset yielded a bootstrap optimism-corrected AUC of 0.847 (95% CI: 0.739-0.938), a calibration slope of 0.78 (95% CI: 0.42-1.16), and an optimism-corrected Brier score of 0.142 (95% CI: 0.102-0.188). The salivary EV miRNA panel demonstrates a promising developmental diagnostic signal for endometriosis, alongside biologically coherent associations with pain. However, the balanced case-control design, the use of clinically verified rather than uniformly surgically negative controls, the small sample size, and the risk of overfitting indicate that this panel is not yet ready for stand-alone clinical use. External validation in larger, multicenter, and clinically representative cohorts is required prior to routine diagnostic application.